Biomarkers for predicting activity of multiple sclerosis disease

Abstract

The present invention relates to biomarkers for predicting multiple sclerosis disease activity. Disclosed herein are methods for analyzing quantitative expression values of biomarkers of a biomarker panel to determine disease activity in a human subject. Also disclosed herein are kits for measuring quantitative expression values of biomarkers and computer system and software embodiments for determining a predictive model of disease activity in a human subject based on the quantitative expression values of the markers.

Description

[0001] This application is a divisional application of the patent application No. 202080076653.1 with a filing date of September 4, 2020. The original application entered the Chinese national phase on April 29, 2022. The application number of the PCT international application is PCT / US2020 / 049375, and the publication data is WO2021 / 046329.

[0002] Cross-reference to related applications

[0003] This application claims the benefit and priority of U.S. Provisional Patent Application No. 62 / 896,430, filed on September 5, 2019, the entire disclosure of which is hereby incorporated by reference in its entirety for all purposes. Summary of the invention

[0004] Disclosed herein are methods for predicting multiple sclerosis disease activity using a multivariate biomarker panel. Also disclosed herein are non-transitory computer-readable media for predicting multiple sclerosis disease activity using a multivariate biomarker panel. Also disclosed herein is a kit containing a set of reagents for determining the expression levels of multivariate biomarkers that provide information for predicting the disease activity of multiple sclerosis. Also disclosed herein is a system for predicting multiple sclerosis disease activity using a multivariate biomarker panel.

[0005] Advantages of using a multivariate biomarker panel to predict disease activity include:

[0006] · Increased sensitivity: Multibiomarker testing improves performance (AUC, accuracy), especially by eliminating false negatives that cannot be detected by a single biomarker.

[0007] · Minor disease activity: Multivariate models can better help distinguish samples through one gadolinium (Gd) lesion than a single biomarker alone.

[0008] · Specificity: Single biomarkers are often differentially expressed in other neurological conditions. Multibiomarker testing will help distinguish specific disease activity / progression in multiple sclerosis.

[0009] · Predictive ability: Multivariate models that combine shifts in biomarker levels can identify increases or decreases in active lesions in patients (with stronger performance compared to a single biomarker alone).

[0010] The present disclosure relates to a method for determining multiple sclerosis activity in a subject, the method comprising: obtaining or having obtained a data set comprising expression levels of a plurality of biomarkers, wherein the plurality of biomarkers comprises each biomarker selected from at least one of group 1, group 2, and group 3, wherein group 1 comprises biomarker 1, biomarker 2, biomarker 3, biomarker 4, biomarker 5, biomarker 6, biomarker 7, and biomarker 8, wherein biomarker 1 is NEFL, MOG, CADM3, or GFAP, wherein biomarker 2 is MOG, CADM3, KLK6, BCAN, OMG, or GFAP, wherein biomarker 3 is CD6, CD5, CRTAM, CD244, or TNFRSF9, wherein biomarker 4 is CXCL9, CXCL10, IL-12B, CXCL11, or GFAP, wherein biomarker 5 is OPG, TFF3, or ENPP2, wherein biomarker 6 is OPN, OMD, MEPE, or GFAP, wherein biomarker 7 is CXCL13, NOS3, or MMP-2, and wherein biomarker 8 is GFAP, NEFL, OPN, CXCL9, MOG, or CHI3L1, and wherein group 2 comprises biomarker 9, biomarker 10, biomarker 11, biomarker 12, biomarker 13, biomarker 14, biomarker 15, biomarker 16, and biomarker 17, wherein biomarker 9 is CDCP1, IL-18BP, IL-18, GFAP, or MSR1, wherein biomarker 10 is CCL20, CCL3, or TWEAK, wherein biomarker 11 is IL-12B, IL12A, or CXCL9, wherein biomarker 12 is APLP1, SEZ6L, BCAN, DPP6, NCAN, or KLK6, wherein biomarker 13 is TNFRSF10A, TNFRSF11A, SPON2, CHI3L1, or IFI30, wherein biomarker 14 is COL4A1, IL6, NOTCH3, or PCDH17, wherein biomarker 15 is SERPINA9, TNFRSF9, or CNTN4, wherein biomarker 16 is FLRT2, DDR1, NTRK2, CDH6, MMP-2, and wherein biomarker 17 is TNFSF13B, CXCL16, ALCAM, or IL-18, wherein group 3 comprises biomarker 18, biomarker 19, biomarker 20, and biomarker 21, wherein biomarker 18 is GH, GH2, or IGFBP-1, wherein biomarker 19 is VCAN, TINAGL1, CANT1, NECTIN2, MMP-9, or NPDC1, wherein biomarker 20 is PRTG, NTRK2, NTRK3, or CNTN4,and wherein biomarker 21 is CNTN2, DPP6, GDNFR-α-3, or SCARF2; and generating a prediction of multiple sclerosis disease activity by applying a prediction model to the expression levels of the plurality of biomarkers.,

[0011] In various embodiments, the plurality of biomarkers includes each biomarker in Group 1, wherein biomarker 1 is NEFL, wherein biomarker 2 is MOG, wherein biomarker 3 is CD6, wherein biomarker 4 is CXCL9, wherein biomarker 5 is OPG, wherein biomarker 6 is OPN, wherein biomarker 7 is CXCL13, and wherein biomarker 8 is GFAP. In various embodiments, the performance of the prediction model is characterized by an area under the curve (AUC) ranging from 0.667 to 0.869. In various embodiments, the performance of the prediction model is characterized by a positive predictive value ranging from 0.617 to 0.861. In various embodiments, the plurality of biomarkers includes each biomarker in Group 2, wherein biomarker 9 is CDCP1, wherein biomarker 10 is CCL20, wherein biomarker 11 is IL-12B, wherein biomarker 12 is APLP1, wherein biomarker 13 is TNFRSF10A, wherein biomarker 14 is COL4A1, wherein biomarker 15 is SERPINA9, wherein biomarker 16 is FLRT2, and wherein biomarker 17 is TNFSF13B. In various embodiments, the performance of the prediction model is characterized by an area under the curve (AUC) ranging from 0.595 to 0.761. In various embodiments, the performance of the prediction model is characterized by a positive predictive value ranging from 0.523 to 0.769. In various embodiments, the plurality of biomarkers further includes each biomarker in Group 2, wherein biomarker 9 is CDCP1, wherein biomarker 10 is CCL20, wherein biomarker 11 is IL-12B, wherein biomarker 12 is APLP1, wherein biomarker 13 is TNFRSF10A, wherein biomarker 14 is COL4A1, wherein biomarker 15 is SERPINA9, wherein biomarker 16 is FLRT2, and wherein biomarker 17 is TNFSF13B. In various embodiments, the performance of the prediction model is characterized by an area under the curve (AUC) ranging from 0.693 to 0.892. In various embodiments, the performance of the prediction model is characterized by a positive predictive value ranging from 0.613 to 0.843.

[0012] In various embodiments, the plurality of biomarkers includes each biomarker in Group 3, wherein biomarker 18 is GH, wherein biomarker 19 is VCAN, wherein biomarker 20 is PRTG, and wherein biomarker 21 is CNTN2. In various embodiments, the performance of the prediction model is characterized by an area under the curve (AUC) ranging from 0.566 to 0.644. In various embodiments, the performance of the prediction model is characterized by a positive predictive value ranging from 0.370 to 0.742. In various embodiments, the plurality of biomarkers further includes each biomarker in Group 3, wherein biomarker 18 is GH, wherein biomarker 19 is VCAN, wherein biomarker 20 is PRTG, and wherein biomarker 21 is CNTN2. In various embodiments, the performance of the prediction model is characterized by an area under the curve (AUC) ranging from 0.686 to 0.889. In various embodiments, the performance of the prediction model is characterized by a positive predictive value ranging from 0.648 to 0.835.

[0013] Also disclosed herein is a method for determining multiple sclerosis activity in a subject, the method comprising: obtaining or having obtained a data set comprising expression levels of a plurality of biomarkers, the plurality of biomarkers including NEFL, MOG, CD6, CXCL9, OPG, OPN, CXCL13, and GFAP; and generating a prediction of multiple sclerosis disease activity by applying a prediction model to the expression levels of the plurality of biomarkers. In various embodiments, the plurality of biomarkers further includes CDCP1, CCL20, IL-12B, APLP1, TNFRSF10A, COL4A1, SERPINA9, FLRT2, and TNFSF13B. In various embodiments, the plurality of biomarkers further includes GH, VCAN, PRTG, and CNTN2.

[0014] The present disclosure further provides a method for generating quantitative data for a subject, comprising: performing at least one immunoassay on a sample obtained from the subject to generate a data set comprising the quantitative data, wherein the quantitative data represents at least eight protein biomarkers, including: NEFL, MOG, CD6, CXCL9, OPG, OPN, CXCL13, and GFAP, wherein the subject has or is suspected of having multiple sclerosis. In various embodiments, the quantitative data further represents at least nine additional protein biomarkers, including: CDCP1, CCL20, IL-12B, APLP1, TNFRSF10A, COL4A1, SERPINA9, FLRT2, and TNFSF13B. In various embodiments, the quantitative data further represents at least four additional protein biomarkers, including: GH, VCAN, PRTG, and CNTN2. In various embodiments, the method further comprises: determining multiple sclerosis activity in the subject, wherein the determining comprises applying a prediction model to the quantitative data. In various embodiments, applying the determination further comprises comparing a score output by the prediction model with a reference score. In various embodiments, the reference score corresponds to any one of the following: A) a patient at a baseline time point, at which the patient does not exhibit disease activity, B) a patient clinically diagnosed as having no disease activity, or C) a healthy patient.

[0015] The present disclosure further provides a method for determining multiple sclerosis activity in a subject, the method comprising: obtaining or having obtained a data set comprising expression levels of a plurality of biomarkers, the plurality of biomarkers including: one or more neurodegeneration biomarkers selected from the group consisting of NEFL, APLP1, OPG, SERPINA9, PRTG, GFAP, CNTN2, and FLRT2; one or more inflammation biomarkers selected from the group consisting of CCL20, GH, CXCL13, IL-12B, VCAN, TNFRSF10A, TNFSF13B, CD6, and CXCL9; one or more immunomodulation biomarkers selected from the group consisting of CDCP1 and OPN; one or more myelin integrity biomarkers selected from the group consisting of COL4A1 and MOG; and generating a prediction of multiple sclerosis disease activity by applying a prediction model to the expression levels of the plurality of biomarkers.

[0016] In various embodiments, the one or more neurodegeneration biomarkers include NEFL, OPG, and GFAP, wherein the one or more inflammation biomarkers include CXCL13, CD6, and CXCL9, wherein the one or more immunomodulation biomarkers include OPN, and wherein the one or more myelin integrity biomarkers include MOG. In various embodiments, the one or more neurodegeneration biomarkers further include APLP1, SERPINA9, and FLRT2, wherein the one or more inflammation biomarkers further include CCL20, IL-12B, TNFRSF10A, and TNFSF13B, wherein the one or more immunomodulation biomarkers further include CDCP1, and wherein the one or more myelin integrity biomarkers further include COL4A1. In various embodiments, the one or more neurodegeneration biomarkers further include PRTG and CNTN2, and wherein the one or more inflammation biomarkers include GH and VCAN. In various embodiments, the plurality of biomarkers are protein biomarkers.

[0017] In various embodiments, the multiple sclerosis disease activity is any one of the following: the presence of general disease activity, the presence of minor disease activity, a shift (increase or decrease) in disease activity, the severity of MS, a relapse or episode associated with MS, the relapse rate, the MS status, the response to MS therapy, the degree of MS disability, or the risk of developing MS. In various embodiments, the general disease activity is the presence or absence of one or more gadolinium-enhanced MRI lesions, and wherein the minor disease activity is the presence of one gadolinium-enhanced MRI lesion. In various embodiments, the severity of MS corresponds to the number of gadolinium-enhanced MRI lesions. In various embodiments, the MS status is a worsening or quiescent state of multiple sclerosis.

[0018] Also disclosed herein is a method for determining multiple sclerosis activity in a subject, the method comprising: obtaining or having obtained a data set comprising the expression levels of a plurality of biomarkers, wherein the plurality of biomarkers include two or more of the following: NEFL, MOG, CD6, CXCL9, OPG, OPN, CXCL13, GFAP, CDCP1, CCL20, IL-12B, APLP1, TNFRSF10A, COL4A1, SERPINA9, FLRT2, TNFSF13B, GH, VCAN, PRTG, and CNTN2; and generating a prediction of multiple sclerosis disease activity by applying a prediction model to the expression levels of the plurality of biomarkers.

[0019] The present disclosure further provides a method for preparing biomarker fractions from a sample obtained from a subject, the method comprising: (a) extracting biomarkers from the sample; (b) generating fractions of the extracted biomarkers after (a), wherein the fractions of the extracted biomarkers after (b) comprise a plurality of biomarkers, and wherein the plurality of biomarkers includes two or more of the following: NEFL, MOG, CD6, CXCL9, OPG, OPN, CXCL13, GFAP, CDCP1, CCL20, IL-12B, APLP1, TNFRSF10A, COL4A1, SERPINA9, FLRT2, TNFSF13B, GH, VCAN, PRTG, and CNTN2; (c) analyzing the plurality of biomarkers in the fractions of the extracted biomarkers generated in (b). In various embodiments, the two or more biomarkers include NEFL and at least one other biomarker. In various embodiments, the two or more biomarkers are biomarker pairs selected from any one of the following: A) NEFL and CD6, B) NEFL and MOG, C) NEFL and CXCL9, and D) NEFL and TNFRSF10A. In various embodiments, the two or more biomarkers are biomarker triples selected from any one of the following: A) NEFL, CD6, and CXCL9, and B) NEFL, TNFRSF10A, and COL4A1. In various embodiments, the two or more biomarkers are biomarker quadruples selected from any one of the following: A) NEFL, MOG, CD6, and CXCL9, B) NEFL, CXCL9, TNFRSF10A, and COL4A1, C) NEFL, CD6, CXCL9, and CXCL13, and D) NEFL, TNFRSF10A, COL4A1, and CCL20. In various embodiments, the two or more biomarkers do not include NEFL. In various embodiments, the two or more biomarkers are biomarker pairs selected from any one of the following: A) MOG and IL-12B, B) CXCL9 and CD6, C) MOG and CXCL9, D) MOG and CD6, E) CXCL9 and COL4A1, and F) CD6 and VCAN. In various embodiments, the two or more biomarkers are biomarker triples selected from any one of the following: A) MOG, IL-12B, and APLP1, B) MOG, CD6, and CXCL9, C) CXCL9, COL4A1, and VCAN, D) MOG, IL-12B, and CNTN2, and E) CD6, CCL20, and VCAN.In various embodiments, the two or more biomarkers are a biomarker quadruple selected from any of the following: A) MOG, CXCL9, IL-12B, APLP1, B) CXCL9, COL4A1, OPG, and VCAN, C) CXCL9, OPG, APLP1, and OPN, D) MOG, IL-12B, OPN, and CNTN2, and E) CD6, COL4A1, CCL20, and VCAN. In various embodiments, the multiple sclerosis activity is a deviation of disease activity.

[0020] In various embodiments, the two or more biomarkers are a biomarker pair selected from any of the following: A) NEFL and TNFSF13B, B) NEFL and CNTN2, and C) NEFL and CXCL9. In various embodiments, the two or more biomarkers are a biomarker triple selected from any of the following: A) NEFL, CNTN2, and TNFSF13B, B) NEFL, APLP1, and TNFSF13B, and C) NEFL, TNFRSF10A, and TNFSF13B. In various embodiments, the two or more biomarkers are a biomarker quadruple selected from any of the following: A) NEFL, TNFRSF10A, CNTN2, and TNFSF13B, B) NEFL, COL4A1, CNTN2, and TNFSF13B, and C) NEFL, TNFRSF10A, APLP1, and TNFSF13B. In various embodiments, the two or more biomarkers are a biomarker pair selected from any of the following: A) MOG and CDCP1, B) MOG and TNFSF13B, and C) MOG and CXCL9. In various embodiments, the two or more biomarkers are a biomarker triple, and one biomarker in the biomarker triple is MOG. In various embodiments, the biomarker triple is selected from any of the following: A) MOG, CXCL9, and TNFSF13B, B) MOG, OPG, and TNFSF13B, and C) MOG, CCL20, and TNFSF13B. In various embodiments, the two or more biomarkers are a biomarker quadruple, and one biomarker in the biomarker quadruple is MOG. In various embodiments, the biomarker quadruple is selected from any of the following: A) MOG, CXCL9, APLP1, and TNFSF13B, B) MOG, CXCL9, OPG, and TNFSF13B, and C) MOG, CXCL9, OPG, and CNTN2. In various embodiments, the multiple sclerosis activity is the presence or absence of multiple sclerosis.

[0021] In various embodiments, the two or more biomarkers are biomarker pairs selected from any of the following: A) NEFL and TNFSF13B, B) NEFL and SERPINA9, and C) NEFL and GH. In various embodiments, the two or more biomarkers are biomarker triples selected from any of the following: A) NEFL, SERPINA9, and TNFSF13B, B) NEFL, CNTN2, and TNFSF13B, and C) NEFL, APLP1, and TNFSF13B. In various embodiments, the two or more biomarkers are biomarker quadruples selected from any of the following: A) NEFL, CCL20, SERPINA9, and TNFSF13B, B) NEFL, APLP1, SERPINA9, and TNFSF13B, and C) NEFL, CCL20, APLP1, and TNFSF13B. In various embodiments, the two or more biomarkers are biomarker pairs selected from any of the following: A) MOG and TNFSF13B, B) MOG and CXCL9, and C) MOG and IL-12B. In various embodiments, the two or more biomarkers are biomarker triples selected from any of the following: A) MOG, CXCL9, and TNFSF13B, B) MOG, SERPINA9, and TNFSF13B, and C) MOG, OPG, and TNFSF13B. In various embodiments, the two or more biomarkers are biomarker quadruples selected from any of the following: A) MOG, CXCL9, OPG, and TNFSF13B, B) MOG, CXCL9, OPG, SERPINA9, and TNFSF13B, C) MOG, CXCL9, SERPINA9, and TNFSF13B. In various embodiments, the multiple sclerosis activity is disease severity based on the predicted number of gadolinium-enhanced lesions.

[0022] In various embodiments, the two or more biomarkers are biomarker pairs selected from any one of the following: A) GFAP and MOG, B) GFAP and NEFL, C) APLP1 and GFAP, D) NEFL and MOG, or E) CXCL9 and OPG. In various embodiments, the two or more biomarkers are biomarker triplets selected from any one of the following: A) GFAP, NEFL, and MOG, B) NEFL, MOG, and GH, C) NEFL, MOG, and SERPINA9, D) CXCL9, OPG, and SERPINA9, or E) CXCL9, OPG, and TNFRSF10A. In various embodiments, the two or more biomarkers are biomarker quadruplets selected from any one of the following: A) GFAP, NEFL, MOG, and IL12B, B) GFAP, NEFL, MOG, and PRTG, C) GFAP, NEFL, MOG, and APLP1, D) NEFL, MOG, GH, and SERPINA9, E) NEFL, MOG, GH, and TNFRSF10A, F) MOG, CXCL9, OPG, and SERPINA9, or G) CD6, IL12B, APLP1, and CCL20. In various embodiments, the multiple sclerosis activity is disease progression.

[0023] In various embodiments, the method further includes classifying the subject based on the prediction. In various embodiments, generating the prediction of the multiple sclerosis disease activity includes comparing a score output by the prediction model with a reference score. In various embodiments, the reference score corresponds to any one of the following: A) a patient at a baseline time point at which the patient does not exhibit disease activity, B) a patient clinically diagnosed as having no disease activity, or C) a healthy patient. In various embodiments, the expression levels of the multiple biomarkers are determined from a test sample obtained from the subject. In various embodiments, the test sample is a blood or serum sample. In various embodiments, the subject has multiple sclerosis, is suspected of having multiple sclerosis, or has been previously diagnosed with multiple sclerosis. In various embodiments, obtaining or having obtained the data set includes performing an immunoassay to determine the expression levels of the multiple biomarkers. In various embodiments, the immunoassay is a proximity extension assay (PEA) or a LUMINEX xMAP multiplex assay. In various embodiments, performing the immunoassay includes contacting the test sample with a plurality of reagents comprising antibodies. In various embodiments, the antibodies include one of monoclonal antibodies and polyclonal antibodies. In various embodiments, the antibodies include both monoclonal antibodies and polyclonal antibodies.

[0024] In various embodiments, the method further includes selecting a therapy for administration to the subject based on the prediction of multiple sclerosis disease activity. In various embodiments, the method further includes determining the therapeutic efficacy of a therapy previously administered to the subject based on the prediction of multiple sclerosis disease activity. In various embodiments, determining the therapeutic efficacy of the therapy includes comparing the prediction to a previous prediction determined for the subject at a previous time point. In various embodiments, determining the therapeutic efficacy of the therapy includes determining that the therapy is efficacious in response to a difference between the prediction and the previous prediction. In various embodiments, determining the therapeutic efficacy of the therapy includes determining that the therapy is not efficacious in response to a lack of difference between the prediction and the previous prediction. In various embodiments, the method further includes determining a differential diagnosis of multiple sclerosis based on the prediction of multiple sclerosis disease activity. In various embodiments, the differential diagnosis of multiple sclerosis includes at least one of the following: relapsing-remitting multiple sclerosis (RRMS), secondary progressive multiple sclerosis (SPMS), primary progressive multiple sclerosis (PPMS), progressive relapsing multiple sclerosis (PRMS), and clinically isolated syndrome (CIS).

[0025] The present disclosure further provides a non-transitory computer-readable medium for determining multiple sclerosis activity in a subject, the non-transitory computer-readable medium comprising instructions that, when executed by a processor, cause the processor to perform the following steps: obtaining a data set comprising the expression levels of a plurality of biomarkers, wherein the plurality of biomarkers comprises each biomarker selected from at least one of Group 1, Group 2, and Group 3, wherein Group 1 comprises Biomarker 1, Biomarker 2, Biomarker 3, Biomarker 4, Biomarker 5, Biomarker 6, Biomarker 7, and Biomarker 8, wherein Biomarker 1 is NEFL, MOG, CADM3, or GFAP, wherein Biomarker 2 is MOG, CADM3, KLK6, BCAN, OMG, or GFAP, wherein Biomarker 3 is CD6, CD5, CRTAM, CD244, or TNFRSF9, wherein Biomarker 4 is CXCL9, CXCL10, IL-12B, CXCL11, or GFAP, wherein Biomarker 5 is OPG, TFF3, or ENPP2, wherein Biomarker 6 is OPN, OMD, MEPE, or GFAP, wherein Biomarker 7 is CXCL13, NOS3, or MMP-2, and wherein Biomarker 8 is GFAP, NEFL, OPN, CXCL9, MOG, or CHI3L1, and wherein Group 2 comprises Biomarker 9, Biomarker 10, Biomarker 11, Biomarker 12, Biomarker 13, Biomarker 14, Biomarker 15, Biomarker 16, and Biomarker 17, wherein Biomarker 9 is CDCP1, IL-18BP, IL-18, GFAP, or MSR1, wherein Biomarker 10 is CCL20, CCL3, or TWEAK, wherein Biomarker 11 is IL-12B, IL12A, or CXCL9, wherein Biomarker 12 is APLP1, SEZ6L, BCAN, DPP6, NCAN, or KLK6, wherein Biomarker 13 is TNFRSF10A, TNFRSF11A, SPON2, CHI3L1, or IFI30, wherein Biomarker 14 is COL4A1, IL6, NOTCH3, or PCDH17, wherein Biomarker 15 is SERPINA9, TNFRSF9, or CNTN4, wherein Biomarker 16 is FLRT2, DDR1, NTRK2, CDH6, MMP-2, and wherein Biomarker 17 is TNFSF13B, CXCL16, ALCAM, or IL-18, wherein Group 3 comprises Biomarker 18, Biomarker 19, Biomarker 20, and Biomarker 21, wherein Biomarker 18 is GH, GH2, or IGFBP-1, wherein Biomarker 19 is VCAN, TINAGL1, CANT1, NECTIN2, MMP-9, or NPDC1,wherein biomarker 20 is PRTG, NTRK2, NTRK3, or CNTN4, and wherein biomarker 21 is CNTN2, DPP6, GDNFR-α-3, or SCARF2; and generating a prediction of multiple sclerosis disease activity by applying a prediction model to the expression levels of the plurality of biomarkers.

[0026] In various embodiments, the plurality of biomarkers includes each biomarker in Group 1, wherein biomarker 1 is NEFL, wherein biomarker 2 is MOG, wherein biomarker 3 is CD6, wherein biomarker 4 is CXCL9, wherein biomarker 5 is OPG, wherein biomarker 6 is OPN, wherein biomarker 7 is CXCL13, and wherein biomarker 8 is GFAP. In various embodiments, the performance of the prediction model is characterized by an area under the curve (AUC) ranging from 0.667 to 0.869. In various embodiments, the performance of the prediction model is characterized by a positive predictive value ranging from 0.617 to 0.861.

[0027] In various embodiments, the plurality of biomarkers includes each biomarker in Group 2, wherein biomarker 9 is CDCP1, wherein biomarker 10 is CCL20, wherein biomarker 11 is IL-12B, wherein biomarker 12 is APLP1, wherein biomarker 13 is TNFRSF10A, wherein biomarker 14 is COL4A1, wherein biomarker 15 is SERPINA9, wherein biomarker 16 is FLRT2, and wherein biomarker 17 is TNFSF13B. In various embodiments, the performance of the prediction model is characterized by an area under the curve (AUC) ranging from 0.595 to 0.761. In various embodiments, the performance of the prediction model is characterized by a positive predictive value ranging from 0.523 to 0.769.

[0028] In various embodiments, the plurality of biomarkers further includes each biomarker in Group 2, wherein biomarker 9 is CDCP1, wherein biomarker 10 is CCL20, wherein biomarker 11 is IL-12B, wherein biomarker 12 is APLP1, wherein biomarker 13 is TNFRSF10A, wherein biomarker 14 is COL4A1, wherein biomarker 15 is SERPINA9, wherein biomarker 16 is FLRT2, and wherein biomarker 17 is TNFSF13B. In various embodiments, the performance of the prediction model is characterized by an area under the curve (AUC) ranging from 0.693 to 0.892. In various embodiments, the performance of the prediction model is characterized by a positive predictive value ranging from 0.613 to 0.843.

[0029] In various embodiments, the plurality of biomarkers includes each biomarker in Group 3, wherein biomarker 18 is GH, wherein biomarker 19 is VCAN, wherein biomarker 20 is PRTG, and wherein biomarker 21 is CNTN2. In various embodiments, the performance of the predictive model is characterized by an area under the curve (AUC) ranging from 0.566 to 0.644. In various embodiments, the performance of the predictive model is characterized by a positive predictive value ranging from 0.370 to 0.742. In various embodiments, the plurality of biomarkers further includes each biomarker in Group 3, wherein biomarker 18 is GH, wherein biomarker 19 is VCAN, wherein biomarker 20 is PRTG, and wherein biomarker 21 is CNTN2. In various embodiments, the performance of the predictive model is characterized by an area under the curve (AUC) ranging from 0.686 to 0.889. In various embodiments, the performance of the predictive model is characterized by a positive predictive value ranging from 0.648 to 0.835.

[0030] Also disclosed herein is a non-transitory computer-readable medium for determining multiple sclerosis activity in a subject, the non-transitory computer-readable medium comprising instructions that, when executed by a processor, cause the processor to perform the following steps: obtaining or having obtained a data set comprising expression levels of a plurality of biomarkers, the plurality of biomarkers including NEFL, MOG, CD6, CXCL9, OPG, OPN, CXCL13, and GFAP; and generating a prediction of multiple sclerosis disease activity by applying a predictive model to the expression levels of the plurality of biomarkers. In various embodiments, the plurality of biomarkers further includes CDCP1, CCL20, IL-12B, APLP1, TNFRSF10A, COL4A1, SERPINA9, FLRT2, and TNFSF13B. In various embodiments, the plurality of biomarkers further includes GH, VCAN, PRTG, and CNTN2.

[0031] The present disclosure further provides a non-transitory computer-readable medium for determining multiple sclerosis activity in a subject, the non-transitory computer-readable medium comprising instructions that, when executed by a processor, cause the processor to perform the following steps: obtaining a data set comprising expression levels of a plurality of biomarkers, the plurality of biomarkers comprising: one or more neurodegeneration biomarkers selected from the group consisting of NEFL, APLP1, OPG, SERPINA9, PRTG, GFAP, CNTN2, and FLRT2; one or more inflammation biomarkers selected from the group consisting of CCL20, GH, CXCL13, IL-12B, VCAN, TNFRSF10A, TNFSF13B, CD6, and CXCL9; one or more immunomodulation biomarkers selected from the group consisting of CDCP1 and OPN; one or more myelin integrity biomarkers selected from the group consisting of COL4A1 and MOG; and generating a prediction of multiple sclerosis disease activity by applying a prediction model to the expression levels of the plurality of biomarkers.

[0032] In various embodiments, the one or more neurodegeneration biomarkers include NEFL, OPG, and GFAP, wherein the one or more inflammation biomarkers include CXCL13, CD6, and CXCL9, wherein the one or more immunomodulation biomarkers include OPN, and wherein the one or more myelin integrity biomarkers include MOG. In various embodiments, the one or more neurodegeneration biomarkers further include APLP1, SERPINA9, and FLRT2, wherein the one or more inflammation biomarkers further include CCL20, IL-12B, TNFRSF10A, and TNFSF13B, wherein the one or more immunomodulation biomarkers further include CDCP1, and wherein the one or more myelin integrity biomarkers further include COL4A1. In various embodiments, the one or more neurodegeneration biomarkers further include PRTG and CNTN2, and wherein the one or more inflammation biomarkers include GH and VCAN. In various embodiments, the plurality of biomarkers are protein biomarkers. In various embodiments, the multiple sclerosis disease activity is any one of the following: the presence of general disease activity, the presence of mild disease activity, a shift (increase or decrease) in disease activity, the severity of MS, a relapse or episode associated with MS, the relapse rate, the MS status, the response to MS therapy, the degree of MS disability, or the risk of developing MS. In various embodiments, the general disease activity is the presence or absence of one or more gadolinium-enhanced MRI lesions, and wherein the mild disease activity is the presence of one gadolinium-enhanced MRI lesion. In various embodiments, the severity of the MS corresponds to the number of gadolinium-enhanced MRI lesions. In various embodiments, the MS status is a worsening or quiescent state of multiple sclerosis.

[0033] Also disclosed herein is a non-transitory computer-readable medium for determining multiple sclerosis activity in a subject, the non-transitory computer-readable medium comprising instructions that, when executed by a processor, cause the processor to perform the following steps: obtaining a data set comprising the expression levels of a plurality of biomarkers, wherein the plurality of biomarkers comprises two or more of the following: NEFL, MOG, CD6, CXCL9, OPG, OPN, CXCL13, GFAP, CDCP1, CCL20, IL-12B, APLP1, TNFRSF10A, COL4A1, SERPINA9, FLRT2, TNFSF13B, GH, VCAN, PRTG, and CNTN2; and generating a prediction of multiple sclerosis disease activity by applying a prediction model to the expression levels of the plurality of biomarkers. In various embodiments, the two or more biomarkers include NEFL and at least one other biomarker.

[0034] In various embodiments, the two or more biomarkers are biomarker pairs selected from any one of the following: A) NEFL and CD6, B) NEFL and MOG, C) NEFL and CXCL9, and D) NEFL and TNFRSF10A. In various embodiments, the two or more biomarkers are biomarker triplets selected from any one of the following: A) NEFL, CD6, and CXCL9 and B) NEFL, TNFRSF10A, and COL4A1. In various embodiments, the two or more biomarkers are biomarker quadruplets selected from any one of the following: A) NEFL, MOG, CD6, and CXCL9, B) NEFL, CXCL9, TNFRSF10A, and COL4A1, C) NEFL, CD6, CXCL9, and CXCL13, and D) NEFL, TNFRSF10A, COL4A1, and CCL20. In various embodiments, the two or more biomarkers do not include NEFL. In various embodiments, the two or more biomarkers are biomarker pairs selected from any one of the following: A) MOG and IL-12B, B) CXCL9 and CD6, C) MOG and CXCL9, D) MOG and CD6, E) CXCL9 and COL4A1, and F) CD6 and VCAN. In various embodiments, the two or more biomarkers are biomarker triplets selected from any one of the following: A) MOG, IL-12B, and APLP1, B) MOG, CD6, and CXCL9, C) CXCL9, COL4A1, and VCAN, D) MOG, IL-12B, and CNTN2, and E) CD6, CCL20, and VCAN. In various embodiments, the two or more biomarkers are biomarker quadruplets selected from any one of the following: A) MOG, CXCL9, IL-12B, APLP1, B) CXCL9, COL4A1, OPG, and VCAN, C) CXCL9, OPG, APLP1, and OPN, D) MOG, IL-12B, OPN, and CNTN2, and E) CD6, COL4A1, CCL20, and VCAN. In various embodiments, the multiple sclerosis activity is an offset of the disease activity.

[0035] In various embodiments, the two or more biomarkers are biomarker pairs selected from any one of the following: A) NEFL and TNFSF13B, B) NEFL and CNTN2, and C) NEFL and CXCL9. In various embodiments, the two or more biomarkers are biomarker triples selected from any one of the following: A) NEFL, CNTN2, and TNFSF13B, B) NEFL, APLP1, and TNFSF13B, and C) NEFL, TNFRSF10A, and TNFSF13B. In various embodiments, the two or more biomarkers are biomarker quadruples selected from any one of the following: A) NEFL, TNFRSF10A, CNTN2, and TNFSF13B, B) NEFL, COL4A1, CNTN2, and TNFSF13B, and C) NEFL, TNFRSF10A, APLP1, and TNFSF13B. In various embodiments, the two or more biomarkers are biomarker pairs selected from any one of the following: A) MOG and CDCP1, B) MOG and TNFSF13B, and C) MOG and CXCL9. In various embodiments, the two or more biomarkers are biomarker triples, and one of the biomarkers in the biomarker triple is MOG. In various embodiments, the biomarker triple is selected from any one of the following: A) MOG, CXCL9, and TNFSF13B, B) MOG, OPG, and TNFSF13B, and C) MOG, CCL20, and TNFSF13B. In various embodiments, the two or more biomarkers are biomarker quadruples, and one of the biomarkers in the biomarker quadruple is MOG. In various embodiments, the biomarker quadruple is selected from any one of the following: A) MOG, CXCL9, APLP1, and TNFSF13B, B) MOG, CXCL9, OPG, and TNFSF13B, and C) MOG, CXCL9, OPG, and CNTN2. In various embodiments, the multiple sclerosis activity is the presence or absence of multiple sclerosis.

[0036] In various embodiments, the two or more biomarkers are biomarker pairs selected from any one of the following: A) NEFL and TNFSF13B, B) NEFL and SERPINA9, and C) NEFL and GH. In various embodiments, the two or more biomarkers are biomarker triplets selected from any one of the following: A) NEFL, SERPINA9, and TNFSF13B, B) NEFL, CNTN2, and TNFSF13B, and C) NEFL, APLP1, and TNFSF13B. In various embodiments, the two or more biomarkers are biomarker quadruplets selected from any one of the following: A) NEFL, CCL20, SERPINA9, and TNFSF13B, B) NEFL, APLP1, SERPINA9, and TNFSF13B, and C) NEFL, CCL20, APLP1, and TNFSF13B. In various embodiments, the two or more biomarkers are biomarker pairs selected from any one of the following: A) MOG and TNFSF13B, B) MOG and CXCL9, and C) MOG and IL-12B. In various embodiments, the biomarker triplets are selected from any one of the following: A) MOG, CXCL9, and TNFSF13B, B) MOG, SERPINA9, and TNFSF13B, and C) MOG, OPG, and TNFSF13B. In various embodiments, the biomarker quadruplets are selected from any one of the following: A) MOG, CXCL9, OPG, and TNFSF13B, B) MOG, CXCL9, OPG, SERPINA9, and TNFSF13B, C) MOG, CXCL9, SERPINA9, and TNFSF13B. In various embodiments, the multiple sclerosis activity is disease severity based on the predicted number of gadolinium-enhanced lesions.

[0037] In various embodiments, the two or more biomarkers are biomarker pairs selected from any one of the following: A) GFAP and MOG, B) GFAP and NEFL, C) APLP1 and GFAP, D) NEFL and MOG, or E) CXCL9 and OPG. In various embodiments, the two or more biomarkers are biomarker triplets selected from any one of the following: A) GFAP, NEFL, and MOG, B) NEFL, MOG, and GH, C) NEFL, MOG, and SERPINA9, D) CXCL9, OPG, and SERPINA9, or E) CXCL9, OPG, and TNFRSF10A. In various embodiments, the two or more biomarkers are biomarker triplets selected from any one of the following: A) GFAP, NEFL, MOG, and IL12B, B) GFAP, NEFL, MOG, and PRTG, C) GFAP, NEFL, MOG, and APLP1, D) NEFL, MOG, GH, and SERPINA9, E) NEFL, MOG, GH, and TNFRSF10A, F) MOG, CXCL9, OPG, and SERPINA9, or G) CD6, IL12B, APLP1, and CCL20. In various embodiments, the multiple sclerosis activity is disease progression.

[0038] In various embodiments, the step further includes classifying the subject based on the prediction. In various embodiments, generating the prediction of multiple sclerosis disease activity includes comparing a score output by the prediction model with a reference score. In various embodiments, the reference score corresponds to any one of the following: A) a patient at a baseline time point, at which the patient does not exhibit disease activity, B) a patient clinically diagnosed as having no disease activity, or C) a healthy patient.

[0039] In various embodiments, the expression levels of the multiple biomarkers are determined from a test sample obtained from the subject. In various embodiments, the test sample is a blood or serum sample. In various embodiments, the subject has multiple sclerosis, is suspected of having multiple sclerosis, or has been previously diagnosed with multiple sclerosis. In various embodiments, the data set is obtained from an immunoassay performed. In various embodiments, the immunoassay is a proximity extension assay (PEA) or a LUMINEX xMAP multiplex assay. In various embodiments, performing the immunoassay includes contacting the test sample with a plurality of reagents comprising antibodies. In various embodiments, the antibodies include one of monoclonal antibodies and polyclonal antibodies. In various embodiments, the antibodies include both monoclonal antibodies and polyclonal antibodies.

[0040] In various embodiments, the step further comprises: selecting a therapy for administration to the subject based on the prediction of multiple sclerosis disease activity. In various embodiments, the step further comprises: determining the therapeutic efficacy of a therapy previously administered to the subject based on the prediction of multiple sclerosis disease activity. In various embodiments, determining the therapeutic efficacy of the therapy comprises comparing the prediction to a previous prediction determined for the subject at a previous time point. In various embodiments, determining the therapeutic efficacy of the therapy comprises determining that the therapy exhibits efficacy in response to a difference between the prediction and the previous prediction. In various embodiments, determining the therapeutic efficacy of the therapy comprises determining that the therapy lacks efficacy in response to a lack of difference between the prediction and the previous prediction. In various embodiments, the step further comprises determining a differential diagnosis of multiple sclerosis based on the prediction of multiple sclerosis disease activity. In various embodiments, the differential diagnosis of multiple sclerosis comprises any one of the following: relapsing-remitting multiple sclerosis (RRMS), secondary progressive multiple sclerosis (SPMS), primary progressive multiple sclerosis (PPMS), progressive relapsing multiple sclerosis (PRMS), and clinically isolated syndrome (CIS).

[0041] The present disclosure further provides a kit for assessing multiple sclerosis disease activity in a subject, the kit comprising: a set of reagents for determining the expression levels of a plurality of biomarkers in a test sample from the subject, wherein the plurality of biomarkers includes each biomarker selected from at least one of Group 1, Group 2, and Group 3, wherein Group 1 includes Biomarker 1, Biomarker 2, Biomarker 3, Biomarker 4, Biomarker 5, Biomarker 6, Biomarker 7, and Biomarker 8, wherein Biomarker 1 is NEFL, MOG, CADM3, or GFAP, wherein Biomarker 2 is MOG, CADM3, KLK6, BCAN, OMG, or GFAP, wherein Biomarker 3 is CD6, CD5, CRTAM, CD244, or TNFRSF9, wherein Biomarker 4 is CXCL9, CXCL10, IL-12B, CXCL11, or GFAP, wherein Biomarker 5 is OPG, TFF3, or ENPP2, wherein Biomarker 6 is OPN, OMD, MEPE, or GFAP, wherein Biomarker 7 is CXCL13, NOS3, or MMP-2, and wherein Biomarker 8 is GFAP, NEFL, OPN, CXCL9, MOG, or CHI3L1, and wherein Group 2 includes Biomarker 9, Biomarker 10, Biomarker 11, Biomarker 12, Biomarker 13, Biomarker 14, Biomarker 15, Biomarker 16, and Biomarker 17, wherein Biomarker 9 is CDCP1, IL-18BP, IL-18, GFAP, or MSR1, wherein Biomarker 10 is CCL20, CCL3, or TWEAK, wherein Biomarker 11 is IL-12B, IL12A, or CXCL9, wherein Biomarker 12 is APLP1, SEZ6L, BCAN, DPP6, NCAN, or KLK6, wherein Biomarker 13 is TNFRSF10A, TNFRSF11A, SPON2, CHI3L1, or IFI30, wherein Biomarker 14 is COL4A1, IL6, NOTCH3, or PCDH17, wherein Biomarker 15 is SERPINA9, TNFRSF9, or CNTN4, wherein Biomarker 16 is FLRT2, DDR1, NTRK2, CDH6, MMP-2, and wherein Biomarker 17 is TNFSF13B, CXCL16, ALCAM, or IL-18, wherein Group 3 includes Biomarker 18, Biomarker 19, Biomarker 20, and Biomarker 21, wherein Biomarker 18 is GH, GH2, or IGFBP-1, wherein Biomarker 19 is VCAN, TINAGL1, CANT1, NECTIN2, MMP-9, or NPDC1, wherein Biomarker 20 is PRTG, NTRK2, NTRK3, or CNTN4,and wherein biomarker 21 is CNTN2, DPP6, GDNFR-α-3, or SCARF2; and instructions for using the set of reagents to determine the expression level of the biomarker from the test sample.,

[0042] In various embodiments, the plurality of biomarkers includes each biomarker in Group 1, wherein biomarker 1 is NEFL, wherein biomarker 2 is MOG, wherein biomarker 3 is CD6, wherein biomarker 4 is CXCL9, wherein biomarker 5 is OPG, wherein biomarker 6 is OPN, wherein biomarker 7 is CXCL13, and wherein biomarker 8 is GFAP. In various embodiments, the performance of the prediction model is characterized by an area under the curve (AUC) in the range of 0.667 to 0.869. In various embodiments, the performance of the prediction model is characterized by a positive predictive value in the range of 0.617 to 0.861.

[0043] In various embodiments, the plurality of biomarkers includes each biomarker in Group 2, wherein biomarker 9 is CDCP1, wherein biomarker 10 is CCL20, wherein biomarker 11 is IL-12B, wherein biomarker 12 is APLP1, wherein biomarker 13 is TNFRSF10A, wherein biomarker 14 is COL4A1, wherein biomarker 15 is SERPINA9, wherein biomarker 16 is FLRT2, and wherein biomarker 17 is TNFSF13B. In various embodiments, the prediction model is characterized by an area under the curve (AUC) in the range of 0.595 to 0.761. In various embodiments, the performance of the prediction model is characterized by a positive predictive value in the range of 0.523 to 0.769.

[0044] In various embodiments, the plurality of biomarkers further includes each biomarker in Group 2, wherein biomarker 9 is CDCP1, wherein biomarker 10 is CCL20, wherein biomarker 11 is IL-12B, wherein biomarker 12 is APLP1, wherein biomarker 13 is TNFRSF10A, wherein biomarker 14 is COL4A1, wherein biomarker 15 is SERPINA9, wherein biomarker 16 is FLRT2, and wherein biomarker 17 is TNFSF13B. In various embodiments, the performance of the prediction model is characterized by an area under the curve (AUC) in the range of 0.693 to 0.892. In various embodiments, the performance of the prediction model is characterized by a positive predictive value in the range of 0.613 to 0.843. In various embodiments, the plurality of biomarkers includes each biomarker in Group 3, wherein biomarker 18 is GH, wherein biomarker 19 is VCAN, wherein biomarker 20 is PRTG, and wherein biomarker 21 is CNTN2. In various embodiments, the performance of the prediction model is characterized by an area under the curve (AUC) in the range of 0.566 to 0.644. In various embodiments, the performance of the prediction model is characterized by a positive predictive value in the range of 0.370 to 0.742. In various embodiments, the plurality of biomarkers further includes each biomarker in Group 3, wherein biomarker 18 is GH, wherein biomarker 19 is VCAN, wherein biomarker 20 is PRTG, and wherein biomarker 21 is CNTN2. In various embodiments, the performance of the prediction model is characterized by an area under the curve (AUC) in the range of 0.686 to 0.889. In various embodiments, the performance of the prediction model is characterized by a positive predictive value in the range of 0.648 to 0.835.

[0045] Also disclosed herein is a kit for assessing multiple sclerosis disease activity in a subject, the kit comprising: a set of reagents for determining the expression levels of a plurality of biomarkers in a test sample from the subject, wherein the plurality of biomarkers includes NEFL, MOG, CD6, CXCL9, OPG, OPN, CXCL13, and GFAP; and instructions for using the set of reagents to determine the expression levels of the biomarkers in the test sample. In various embodiments, the plurality of biomarkers further includes CDCP1, CCL20, IL-12B, APLP1, TNFRSF10A, COL4A1, SERPINA9, FLRT2, and TNFSF13B. In various embodiments, the plurality of biomarkers further includes GH, VCAN, PRTG, and CNTN2.

[0046] The present disclosure further provides a kit for assessing multiple sclerosis disease activity in a subject, the kit comprising: a set of reagents for determining the expression levels of a plurality of biomarkers in a test sample from the subject, wherein the plurality of biomarkers includes: one or more neurodegeneration biomarkers selected from the group consisting of NEFL, APLP1, OPG, SERPINA9, PRTG, GFAP, CNTN2, and FLRT2; one or more inflammation biomarkers selected from the group consisting of CCL20, GH, CXCL13, IL-12B, VCAN, TNFRSF10A, TNFSF13B, CD6, and CXCL9; one or more immunomodulation biomarkers selected from the group consisting of CDCP1 and OPN; one or more myelin integrity biomarkers selected from the group consisting of COL4A1 and MOG; and instructions for using the set of reagents to determine the expression levels of the biomarkers in the test sample. In various embodiments, the one or more neurodegeneration biomarkers include NEFL, OPG, and GFAP, wherein the one or more inflammation biomarkers include CXCL13, CD6, and CXCL9, wherein the one or more immunomodulation biomarkers include OPN, and wherein the one or more myelin integrity biomarkers include MOG. In various embodiments, the one or more neurodegeneration biomarkers further include APLP1, SERPINA9, and FLRT2, wherein the one or more inflammation biomarkers further include CCL20, IL-12B, TNFRSF10A, and TNFSF13B, wherein the one or more immunomodulation biomarkers further include CDCP1, and wherein the one or more myelin integrity biomarkers further include COL4A1. In various embodiments, the one or more neurodegeneration biomarkers further include PRTG and CNTN2, and wherein the one or more inflammation biomarkers include GH and VCAN. In various embodiments, the plurality of biomarkers are protein biomarkers. In various embodiments, the multiple sclerosis disease activity is any one of the following: the presence of general disease activity, the presence of minor disease activity, a shift (increase or decrease) in disease activity, the severity of MS, a relapse or episode associated with MS, the relapse rate, the MS status, the response to MS therapy, the degree of MS disability, or the risk of developing MS. In various embodiments, the general disease activity is the presence or absence of one or more gadolinium-enhanced MRI lesions, and wherein the minor disease activity is the presence of one gadolinium-enhanced MRI lesion. In various embodiments, the severity of MS corresponds to the number of gadolinium-enhanced MRI lesions. In various embodiments, the MS status is a worsening or quiescent state of multiple sclerosis.

[0047] The present disclosure further provides a kit for determining multiple sclerosis activity in a subject, the kit comprising: a set of reagents for determining the expression levels of a plurality of biomarkers in a test sample from the subject, the plurality of biomarkers including two or more of the following: NEFL, MOG, CD6, CXCL9, OPG, OPN, CXCL13, GFAP, CDCP1, CCL20, IL-12B, APLP1, TNFRSF10A, COL4A1, SERPINA9, FLRT2, TNFSF13B, GH, VCAN, PRTG, and CNTN2; and instructions for using the set of reagents to determine the expression levels of the biomarkers in the test sample. In various embodiments, the two or more biomarkers include NEFL and at least one other biomarker. In various embodiments, the two or more biomarkers are a biomarker pair selected from any one of the following: A) NEFL and CD6, B) NEFL and MOG, C) NEFL and CXCL9, and D) NEFL and TNFRSF10A. In various embodiments, the two or more biomarkers are a biomarker triplet selected from any one of the following: A) NEFL, CD6, and CXCL9 and B) NEFL, TNFRSF10A, and COL4A1. In various embodiments, the two or more biomarkers are a biomarker quadruplet selected from any one of the following: A) NEFL, MOG, CD6, and CXCL9, B) NEFL, CXCL9, TNFRSF10A, and COL4A1, C) NEFL, CD6, CXCL9, and CXCL13, and D) NEFL, TNFRSF10A, COL4A1, and CCL20. In various embodiments, the two or more biomarkers do not include NEFL. In various embodiments, the two or more biomarkers are a biomarker pair selected from any one of the following: A) MOG and IL-12B, B) CXCL9 and CD6, C) MOG and CXCL9, D) MOG and CD6, E) CXCL9 and COL4A1, and F) CD6 and VCAN. In various embodiments, the two or more biomarkers are a biomarker triplet selected from any one of the following: A) MOG, IL-12B, and APLP1, B) MOG, CD6, and CXCL9, C) CXCL9, COL4A1, and VCAN, D) MOG, IL-12B, and CNTN2, and E) CD6, CCL20, and VCAN.In various embodiments, the two or more biomarkers are a biomarker quadruple selected from any one of the following: A) MOG, CXCL9, IL-12B, APLP1, B) CXCL9, COL4A1, OPG, and VCAN, C) CXCL9, OPG, APLP1, and OPN, D) MOG, IL-12B, OPN, and CNTN2, and E) CD6, COL4A1, CCL20, and VCAN. In various embodiments, the multiple sclerosis activity is a deviation of disease activity.

[0048] In various embodiments, the two or more biomarkers are a biomarker pair selected from any one of the following: A) NEFL and TNFSF13B, B) NEFL and CNTN2, and C) NEFL and CXCL9. In various embodiments, the two or more biomarkers are a biomarker triple selected from any one of the following: A) NEFL, CNTN2, and TNFSF13B, B) NEFL, APLP1, and TNFSF13B, and C) NEFL, TNFRSF10A, and TNFSF13B. In various embodiments, the two or more biomarkers are a biomarker quadruple selected from any one of the following: A) NEFL, TNFRSF10A, CNTN2, and TNFSF13B, B) NEFL, COL4A1, CNTN2, and TNFSF13B, and C) NEFL, TNFRSF10A, APLP1, and TNFSF13B. In various embodiments, the two or more biomarkers are a biomarker pair selected from any one of the following: A) MOG and CDCP1, B) MOG and TNFSF13B, and C) MOG and CXCL9. In various embodiments, the two or more biomarkers are a biomarker triple, and one biomarker in the biomarker triple is MOG. In various embodiments, the biomarker triple is selected from any one of the following: A) MOG, CXCL9, and TNFSF13B, B) MOG, OPG, and TNFSF13B, and C) MOG, CCL20, and TNFSF13B. In various embodiments, the two or more biomarkers are a biomarker quadruple, and one biomarker in the biomarker quadruple is MOG. In various embodiments, the biomarker quadruple is selected from any one of the following: A) MOG, CXCL9, APLP1, and TNFSF13B, B) MOG, CXCL9, OPG, and TNFSF13B, and C) MOG, CXCL9, OPG, and CNTN2. In various embodiments, the multiple sclerosis activity is the presence or absence of multiple sclerosis.

[0049] In various embodiments, the two or more biomarkers are biomarker pairs selected from any one of the following: A) NEFL and TNFSF13B, B) NEFL and SERPINA9, and C) NEFL and GH. In various embodiments, the two or more biomarkers are biomarker triples selected from any one of the following: A) NEFL, SERPINA9, and TNFSF13B, B) NEFL, CNTN2, and TNFSF13B, and C) NEFL, APLP1, and TNFSF13B. In various embodiments, the two or more biomarkers are biomarker quadruples selected from any one of the following: A) NEFL, CCL20, SERPINA9, and TNFSF13B, B) NEFL, APLP1, SERPINA9, and TNFSF13B, and C) NEFL, CCL20, APLP1, and TNFSF13B. In various embodiments, the two or more biomarkers are biomarker pairs selected from any one of the following: A) MOG and TNFSF13B, B) MOG and CXCL9, and C) MOG and IL-12B. In various embodiments, the two or more biomarkers are biomarker triples selected from any one of the following: A) MOG, CXCL9, and TNFSF13B, B) MOG, SERPINA9, and TNFSF13B, and C) MOG, OPG, and TNFSF13B. In various embodiments, the two or more biomarkers are biomarker quadruples selected from any one of the following: A) MOG, CXCL9, OPG, and TNFSF13B, B) MOG, CXCL9, OPG, SERPINA9, and TNFSF13B, C) MOG, CXCL9, SERPINA9, and TNFSF13B. In various embodiments, the multiple sclerosis activity is disease severity based on the predicted number of gadolinium-enhanced lesions.

[0050] In various embodiments, the two or more biomarkers are biomarker pairs selected from any one of the following: A) GFAP and MOG, B) GFAP and NEFL, C) APLP1 and GFAP, D) NEFL and MOG, or E) CXCL9 and OPG. In various embodiments, the two or more biomarkers are biomarker triplets selected from any one of the following: A) GFAP, NEFL, and MOG, B) NEFL, MOG, and GH, C) NEFL, MOG, and SERPINA9, D) CXCL9, OPG, and SERPINA9, or E) CXCL9, OPG, and TNFRSF10A. In various embodiments, the two or more biomarkers are biomarker triplets selected from any one of the following: A) GFAP, NEFL, MOG, and IL12B, B) GFAP, NEFL, MOG, and PRTG, C) GFAP, NEFL, MOG, and APLP1, D) NEFL, MOG, GH, and SERPINA9, E) NEFL, MOG, GH, and TNFRSF10A, F) MOG, CXCL9, OPG, and SERPINA9, or G) CD6, IL12B, APLP1, and CCL20. In various embodiments, the multiple sclerosis activity is disease progression.

[0051] In various embodiments, the specification further includes instructions for generating the prediction of the multiple sclerosis disease activity by applying a prediction model to the expression levels of the multiple biomarkers. In various embodiments, generating the prediction of the multiple sclerosis disease activity includes comparing a score output by the prediction model with a reference score. In various embodiments, the reference score corresponds to any one of the following: A) a patient at a baseline time point, at which the patient does not exhibit disease activity, B) a patient clinically diagnosed as having no disease activity, or C) a healthy patient. In various embodiments, the test sample is a blood or serum sample. In various embodiments, the subject has multiple sclerosis, is suspected of having multiple sclerosis, or has been previously diagnosed with multiple sclerosis. In various embodiments, the instructions for using the set of reagents include instructions for performing an immunoassay to determine the expression levels of the multiple biomarkers. In various embodiments, the immunoassay is a proximity extension assay (PEA) or a LUMINEX xMAP multiplex assay. In various embodiments, performing the immunoassay includes contacting a test sample with a plurality of reagents comprising antibodies. In various embodiments, the antibodies include one of monoclonal antibodies and polyclonal antibodies. In various embodiments, the antibodies include both monoclonal antibodies and polyclonal antibodies.

[0052] In various embodiments, the kit further comprises instructions for performing: selecting a therapy for administration to the subject based on the prediction of multiple sclerosis disease activity. In various embodiments, the kit further comprises instructions for performing: determining the therapeutic efficacy of a therapy previously administered to the subject based on the prediction of multiple sclerosis disease activity. In various embodiments, determining the therapeutic efficacy of the therapy comprises comparing the prediction to a previous prediction determined for the subject at a previous time point. In various embodiments, determining the therapeutic efficacy of the therapy comprises determining that the therapy is efficacious in response to a difference between the prediction and the previous prediction. In various embodiments, determining the therapeutic efficacy of the therapy comprises determining that the therapy is not efficacious in response to a lack of difference between the prediction and the previous prediction. In various embodiments, the kit further comprises instructions for determining a differential diagnosis of multiple sclerosis based on the prediction of multiple sclerosis disease activity. In various embodiments, the differential diagnosis of multiple sclerosis comprises any of the following: relapsing-remitting multiple sclerosis (RRMS), secondary progressive multiple sclerosis (SPMS), primary progressive multiple sclerosis (PPMS), progressive relapsing multiple sclerosis (PRMS), and clinically isolated syndrome (CIS).

[0053] The present disclosure further provides a system for assessing multiple sclerosis disease activity in a subject, the system comprising: a set of reagents for determining the expression levels of a plurality of biomarkers in a test sample from the subject, wherein the plurality of biomarkers includes each biomarker selected from at least one of Group 1, Group 2, and Group 3, wherein Group 1 includes Biomarker 1, Biomarker 2, Biomarker 3, Biomarker 4, Biomarker 5, Biomarker 6, Biomarker 7, and Biomarker 8, wherein Biomarker 1 is NEFL, MOG, CADM3, or GFAP, wherein Biomarker 2 is MOG, CADM3, KLK6, BCAN, OMG, or GFAP, wherein Biomarker 3 is CD6, CD5, CRTAM, CD244, or TNFRSF9, wherein Biomarker 4 is CXCL9, CXCL10, IL-12B, CXCL11, or GFAP, wherein Biomarker 5 is OPG, TFF3, or ENPP2, wherein Biomarker 6 is OPN, OMD, MEPE, or GFAP, wherein Biomarker 7 is CXCL13, NOS3, or MMP-2, and wherein Biomarker 8 is GFAP, NEFL, OPN, CXCL9, MOG, or CHI3L1, and wherein Group 2 includes Biomarker 9, Biomarker 10, Biomarker 11, Biomarker 12, Biomarker 13, Biomarker 14, Biomarker 15, Biomarker 16, and Biomarker 17, wherein Biomarker 9 is CDCP1, IL-18BP, IL-18, GFAP, or MSR1, wherein Biomarker 10 is CCL20, CCL3, or TWEAK, wherein Biomarker 11 is IL-12B, IL12A, or CXCL9, wherein Biomarker 12 is APLP1, SEZ6L, BCAN, DPP6, NCAN, or KLK6, wherein Biomarker 13 is TNFRSF10A, TNFRSF11A, SPON2, CHI3L1, or IFI30, wherein Biomarker 14 is COL4A1, IL6, NOTCH3, or PCDH17, wherein Biomarker 15 is SERPINA9, TNFRSF9, or CNTN4, wherein Biomarker 16 is FLRT2, DDR1, NTRK2, CDH6, MMP-2, and wherein Biomarker 17 is TNFSF13B, CXCL16, ALCAM, or IL-18, wherein Group 3 includes Biomarker 18, Biomarker 19, Biomarker 20, and Biomarker 21, wherein Biomarker 18 is GH, GH2, or IGFBP-1, wherein Biomarker 19 is VCAN, TINAGL1, CANT1, NECTIN2, MMP-9, or NPDC1, wherein Biomarker 20 is PRTG, NTRK2, NTRK3, or CNTN4,and wherein biomarker 21 is CNTN2, DPP6, GDNFR-α-3, or SCARF2, and a device configured to receive a mixture of one or more reagents from the group and the test sample and measure an expression level of the biomarker from the test sample; and a computer system communicatively coupled to the device to obtain a data set comprising expression levels of a plurality of biomarkers from the test sample and generate a prediction of multiple sclerosis disease activity by applying a prediction model to the expression levels of the plurality of biomarkers.,

[0054] In various embodiments, the plurality of biomarkers includes each biomarker in Group 1, wherein biomarker 1 is NEFL, wherein biomarker 2 is MOG, wherein biomarker 3 is CD6, wherein biomarker 4 is CXCL9, wherein biomarker 5 is OPG, wherein biomarker 6 is OPN, wherein biomarker 7 is CXCL13, and wherein biomarker 8 is GFAP. In various embodiments, the performance of the prediction model is characterized by an area under the curve (AUC) in the range of 0.667 to 0.869. In various embodiments, the performance of the prediction model is characterized by a positive predictive value in the range of 0.617 to 0.861.

[0055] In various embodiments, the plurality of biomarkers includes each biomarker in Group 2, wherein biomarker 9 is CDCP1, wherein biomarker 10 is CCL20, wherein biomarker 11 is IL-12B, wherein biomarker 12 is APLP1, wherein biomarker 13 is TNFRSF10A, wherein biomarker 14 is COL4A1, wherein biomarker 15 is SERPINA9, wherein biomarker 16 is FLRT2, and wherein biomarker 17 is TNFSF13B. In various embodiments, the performance of the prediction model is characterized by an area under the curve (AUC) in the range of 0.595 to 0.761. In various embodiments, the performance of the prediction model is characterized by a positive predictive value in the range of 0.523 to 0.769.

[0056] In various embodiments, the plurality of biomarkers further includes each biomarker in Group 2, wherein biomarker 9 is CDCP1, wherein biomarker 10 is CCL20, wherein biomarker 11 is IL-12B, wherein biomarker 12 is APLP1, wherein biomarker 13 is TNFRSF10A, wherein biomarker 14 is COL4A1, wherein biomarker 15 is SERPINA9, wherein biomarker 16 is FLRT2, and wherein biomarker 17 is TNFSF13B. In various embodiments, the performance of the predictive model is characterized by an area under the curve (AUC) ranging from 0.693 to 0.892. In various embodiments, the performance of the predictive model is characterized by a positive predictive value ranging from 0.613 to 0.843.

[0057] In various embodiments, the plurality of biomarkers includes each biomarker in Group 3, wherein biomarker 18 is GH, wherein biomarker 19 is VCAN, wherein biomarker 20 is PRTG, and wherein biomarker 21 is CNTN2. In various embodiments, the performance of the predictive model is characterized by an area under the curve (AUC) ranging from 0.566 to 0.644. In various embodiments, the performance of the predictive model is characterized by a positive predictive value ranging from 0.370 to 0.742.

[0058] In various embodiments, the plurality of biomarkers further includes each biomarker in Group 3, wherein biomarker 18 is GH, wherein biomarker 19 is VCAN, wherein biomarker 20 is PRTG, and wherein biomarker 21 is CNTN2. In various embodiments, the performance of the predictive model is characterized by an area under the curve (AUC) ranging from 0.686 to 0.889. In various embodiments, the performance of the predictive model is characterized by a positive predictive value ranging from 0.648 to 0.835.

[0059] The present disclosure further provides a system for assessing multiple sclerosis disease activity in a subject, the system comprising: a set of reagents for determining the expression levels of a plurality of biomarkers in a test sample from the subject, wherein the plurality of biomarkers include NEFL, MOG, CD6, CXCL9, OPG, OPN, CXCL13, and GFAP; and a device configured to receive a mixture of one or more of the reagents in the set and the test sample and measure the expression levels of the biomarkers from the test sample; and a computer system communicatively coupled to the device to obtain a data set comprising the expression levels of the plurality of biomarkers from the test sample and generate a prediction of multiple sclerosis disease activity by applying a prediction model to the expression levels of the plurality of biomarkers. In various embodiments, the plurality of biomarkers further include CDCP1, CCL20, IL-12B, APLP1, TNFRSF10A, COL4A1, SERPINA9, FLRT2, and TNFSF13B. In various embodiments, the plurality of biomarkers further include GH, VCAN, PRTG, and CNTN2.

[0060] The present disclosure further provides a system for determining multiple sclerosis activity in a subject, the system comprising: a set of reagents for determining the expression levels of a plurality of biomarkers in a test sample from the subject, wherein the plurality of biomarkers include: obtaining or having obtained a data set comprising the expression levels of a plurality of biomarkers, wherein the plurality of biomarkers include: one or more neurodegeneration biomarkers selected from the group consisting of NEFL, APLP1, OPG, SERPINA9, PRTG, GFAP, CNTN2, and FLRT2; one or more inflammation biomarkers selected from the group consisting of CCL20, GH, CXCL13, IL-12B, VCAN, TNFRSF10A, TNFSF13B, CD6, and CXCL9; one or more immunomodulation biomarkers selected from the group consisting of CDCP1 and OPN; one or more myelin integrity biomarkers selected from the group consisting of COL4A1 and MOG; and a device configured to receive a mixture of one or more of the reagents in the set and the test sample and measure the expression levels of the biomarkers from the test sample; and a computer system communicatively coupled to the device to obtain a data set comprising the expression levels of the plurality of biomarkers from the test sample and generate a prediction of multiple sclerosis disease activity by applying a prediction model to the expression levels of the plurality of biomarkers.

[0061] In various embodiments, the one or more neurodegeneration biomarkers include NEFL, OPG, and GFAP, wherein the one or more inflammation biomarkers include CXCL13, CD6, and CXCL9, wherein the one or more immunomodulation biomarkers include OPN, and wherein the one or more myelin integrity biomarkers include MOG. In various embodiments, the one or more neurodegeneration biomarkers further include APLP1, SERPINA9, and FLRT2, wherein the one or more inflammation biomarkers further include CCL20, IL-12B, TNFRSF10A, and TNFSF13B, wherein the one or more immunomodulation biomarkers further include CDCP1, and wherein the one or more myelin integrity biomarkers further include COL4A1. In various embodiments, the one or more neurodegeneration biomarkers further include PRTG and CNTN2, and wherein the one or more inflammation biomarkers include GH and VCAN. In various embodiments, the plurality of biomarkers are protein biomarkers. In various embodiments, the multiple sclerosis disease activity is any of the following: presence of general disease activity, presence of minor disease activity, shift (increase or decrease) in disease activity, severity of MS, relapse or episode associated with MS, relapse rate, MS status, response to MS therapy, degree of MS disability, or risk of developing MS. In various embodiments, the general disease activity is the presence or absence of one or more gadolinium-enhanced MRI lesions, and wherein the minor disease activity is the presence of one gadolinium-enhanced MRI lesion. In various embodiments, the severity of the MS corresponds to the number of gadolinium-enhanced MRI lesions. In various embodiments, the MS status is a worsening or quiescent state of multiple sclerosis.

[0062] The present disclosure further provides a system for determining multiple sclerosis activity in a subject, the system comprising: a set of reagents for determining the expression levels of a plurality of biomarkers in a test sample from the subject, wherein the plurality of biomarkers comprises two or more of the following: NEFL, MOG, CD6, CXCL9, OPG, OPN, CXCL13, GFAP, CDCP1, CCL20, IL-12B, APLP1, TNFRSF10A, COL4A1, SERPINA9, FLRT2, TNFSF13B, GH, VCAN, PRTG, and CNTN2; and a device configured to receive a mixture of one or more reagents from the set and the test sample and measure the expression levels of the biomarkers from the test sample; and a computer system communicatively coupled to the device to obtain a data set comprising the expression levels of the plurality of biomarkers from the test sample and generate a prediction of multiple sclerosis disease activity by applying a prediction model to the expression levels of the plurality of biomarkers. In various embodiments, the two or more biomarkers comprise NEFL and at least one other biomarker. In various embodiments, the two or more biomarkers are a biomarker pair selected from any one of the following: A) NEFL and CD6, B) NEFL and MOG, C) NEFL and CXCL9, and D) NEFL and TNFRSF10A. In various embodiments, the two or more biomarkers are a biomarker triplet selected from any one of the following: A) NEFL, CD6, and CXCL9 and B) NEFL, TNFRSF10A, and COL4A1. In various embodiments, the two or more biomarkers are a biomarker quadruplet selected from any one of the following: A) NEFL, MOG, CD6, and CXCL9, B) NEFL, CXCL9, TNFRSF10A, and COL4A1, C) NEFL, CD6, CXCL9, and CXCL13, and D) NEFL, TNFRSF10A, COL4A1, and CCL20. In various embodiments, the two or more biomarkers do not include NEFL. In various embodiments, the two or more biomarkers are a biomarker pair selected from any one of the following: A) MOG and IL-12B, B) CXCL9 and CD6, C) MOG and CXCL9, D) MOG and CD6, E) CXCL9 and COL4A1, and F) CD6 and VCAN.In various embodiments, the two or more biomarkers are a biomarker triplet selected from any of the following: A) MOG, IL-12B, and APLP1, B) MOG, CD6, and CXCL9, C) CXCL9, COL4A1, and VCAN, D) MOG, IL-12B, and CNTN2, and E) CD6, CCL20, and VCAN. In various embodiments, the two or more biomarkers are a biomarker quadruplet selected from any of the following: A) MOG, CXCL9, IL-12B, APLP1, B) CXCL9, COL4A1, OPG, and VCAN, C) CXCL9, OPG, APLP1, and OPN, D) MOG, IL-12B, OPN, and CNTN2, and E) CD6, COL4A1, CCL20, and VCAN. In various embodiments, the multiple sclerosis activity is a deviation of disease activity.

[0063] In various embodiments, the two or more biomarkers are biomarker pairs selected from any of the following: A) NEFL and TNFSF13B, B) NEFL and CNTN2, and C) NEFL and CXCL9. In various embodiments, the two or more biomarkers are biomarker triples selected from any of the following: A) NEFL, CNTN2, and TNFSF13B, B) NEFL, APLP1, and TNFSF13B, and C) NEFL, TNFRSF10A, and TNFSF13B. In various embodiments, the two or more biomarkers are biomarker quadruples selected from any of the following: A) NEFL, TNFRSF10A, CNTN2, and TNFSF13B, B) NEFL, COL4A1, CNTN2, and TNFSF13B, and C) NEFL, TNFRSF10A, APLP1, and TNFSF13B. In various embodiments, the two or more biomarkers are biomarker pairs selected from any of the following: A) MOG and CDCP1, B) MOG and TNFSF13B, and C) MOG and CXCL9. In various embodiments, the two or more biomarkers are biomarker triples, and one of the biomarkers in the biomarker triple is MOG. In various embodiments, the biomarker triple is selected from any of the following: A) MOG, CXCL9, and TNFSF13B, B) MOG, OPG, and TNFSF13B, and C) MOG, CCL20, and TNFSF13B. In various embodiments, the two or more biomarkers are biomarker quadruples, and one of the biomarkers in the biomarker quadruple is MOG. In various embodiments, the biomarker quadruple is selected from any of the following: A) MOG, CXCL9, APLP1, and TNFSF13B, B) MOG, CXCL9, OPG, and TNFSF13B, and C) MOG, CXCL9, OPG, and CNTN2. In various embodiments, the multiple sclerosis activity is the presence or absence of multiple sclerosis.

[0064] In various embodiments, the two or more biomarkers are biomarker pairs selected from any one of the following: A) NEFL and TNFSF13B, B) NEFL and SERPINA9, and C) NEFL and GH. In various embodiments, the two or more biomarkers are biomarker triples selected from any one of the following: A) NEFL, SERPINA9, and TNFSF13B, B) NEFL, CNTN2, and TNFSF13B, and C) NEFL, APLP1, and TNFSF13B. In various embodiments, the two or more biomarkers are biomarker quadruples selected from any one of the following: A) NEFL, CCL20, SERPINA9, and TNFSF13B, B) NEFL, APLP1, SERPINA9, and TNFSF13B, and C) NEFL, CCL20, APLP1, and TNFSF13B. In various embodiments, the two or more biomarkers are biomarker pairs selected from any one of the following: A) MOG and TNFSF13B, B) MOG and CXCL9, and C) MOG and IL-12B. In various embodiments, the two or more biomarkers are biomarker triples selected from any one of the following: A) MOG, CXCL9, and TNFSF13B, B) MOG, SERPINA9, and TNFSF13B, and C) MOG, OPG, and TNFSF13B. In various embodiments, the two or more biomarkers are biomarker quadruples selected from any one of the following: A) MOG, CXCL9, OPG, and TNFSF13B, B) MOG, CXCL9, OPG, SERPINA9, and TNFSF13B, C) MOG, CXCL9, SERPINA9, and TNFSF13B. In various embodiments, the multiple sclerosis activity is disease severity based on the predicted number of gadolinium-enhanced lesions.

[0065] In various embodiments, the two or more biomarkers are biomarker pairs selected from any one of the following: A) GFAP and MOG, B) GFAP and NEFL, C) APLP1 and GFAP, D) NEFL and MOG, or E) CXCL9 and OPG. In various embodiments, the two or more biomarkers are biomarker triplets selected from any one of the following: A) GFAP, NEFL, and MOG, B) NEFL, MOG, and GH, C) NEFL, MOG, and SERPINA9, D) CXCL9, OPG, and SERPINA9, or E) CXCL9, OPG, and TNFRSF10A. In various embodiments, the two or more biomarkers are biomarker triplets selected from any one of the following: A) GFAP, NEFL, MOG, and IL12B, B) GFAP, NEFL, MOG, and PRTG, C) GFAP, NEFL, MOG, and APLP1, D) NEFL, MOG, GH, and SERPINA9, E) NEFL, MOG, GH, and TNFRSF10A, F) MOG, CXCL9, OPG, and SERPINA9, or G) CD6, IL12B, APLP1, and CCL20. In various embodiments, the multiple sclerosis activity is disease progression.

[0066] In various embodiments, generating the prediction of multiple sclerosis disease activity includes comparing the score output by the prediction model with a reference score. In various embodiments, the reference score corresponds to any one of the following: A) a patient at a baseline time point at which the patient does not exhibit disease activity, B) a patient clinically diagnosed as having no disease activity, or C) a healthy patient.

[0067] In various embodiments, the test sample is a blood or serum sample. In various embodiments, the subject has multiple sclerosis, is suspected of having multiple sclerosis, or has been previously diagnosed with multiple sclerosis. In various embodiments, the device is configured to perform an immunoassay to determine the expression levels of the multiple biomarkers. In various embodiments, the immunoassay is a proximity extension assay (PEA) or a LUMINEX xMAP multiplex assay. In various embodiments, performing the immunoassay includes contacting the test sample with a plurality of reagents comprising antibodies. In various embodiments, the antibodies include one of monoclonal antibodies and polyclonal antibodies. In various embodiments, the antibodies include both monoclonal antibodies and polyclonal antibodies.

[0068] In various embodiments, the computer system is further configured to select a therapy for administration to the subject based on the prediction of multiple sclerosis disease activity. In various embodiments, the computer system is further configured to determine the therapeutic efficacy of a therapy previously administered to the subject based on the prediction of multiple sclerosis disease activity. In various embodiments, determining the therapeutic efficacy of the therapy includes comparing the prediction to a previous prediction determined for the subject at a previous time point. In various embodiments, determining the therapeutic efficacy of the therapy includes determining that the therapy is efficacious in response to a difference between the prediction and the previous prediction. In various embodiments, determining the therapeutic efficacy of the therapy includes determining that the therapy is not efficacious in response to a lack of difference between the prediction and the previous prediction. In various embodiments, the computer system is further configured to determine a differential diagnosis of multiple sclerosis based on the prediction of multiple sclerosis disease activity. In various embodiments, the differential diagnosis of multiple sclerosis includes any of the following: relapsing-remitting multiple sclerosis (RRMS), secondary progressive multiple sclerosis (SPMS), primary progressive multiple sclerosis (PPMS), progressive relapsing multiple sclerosis (PRMS), and clinically isolated syndrome (CIS). BRIEF DESCRIPTION OF THE DRAWINGS

[0069] These and other features, aspects, and advantages of the present invention will be better understood with reference to the following description and drawings.

[0070] Figure 1A An overview of an environment for evaluating disease activity in a subject by an activity prediction system according to an embodiment is shown.

[0071] Figure 1B is an exemplary block diagram of an activity prediction system according to an embodiment.

[0072] Figure 1C An exemplary set of training data according to an embodiment is shown.

[0073] Figure 2A Feature sequence forward selection using samples from the F6 study is shown.

[0074] Figure 2B Feature sequence forward selection using samples from the F4 study is shown.

[0075] Figure 3A The ROC curve of a multivariate model (training and testing) compared to a univariate model using neurofilament light chain as a single feature is shown.

[0076] Figure 3B The confusion matrix of the multivariate model is shown.

[0077] Figure 4A Shows the sequential forward selection of biomarkers for cross-sectional classification of the presence / absence of radiologically defined disease activity.

[0078] Figure 4B Shows the ROC curves of the training models for predicting disease activity (mild, moderate, and extreme disease activity).

[0079] Figure 4C Shows the confusion matrices for each of the mild disease model, moderate disease model, and extreme disease model.

[0080] Figure 4D Shows the sequential forward selection of biomarkers for predicting disease severity based on the predicted number of lesions.

[0081] Figure 5A Shows the sequential forward selection of features for constructing a model for predicting the annualized relapse rate (ARR).

[0082] Figure 5B Shows the ROC curves of the training models for predicting HIGH (>=0.8) or LOW (<0.3) annualized relapse rate.

[0083] Figure 6A Shows the sequential forward selection of features for constructing a model for classifying the deteriorating and quiescent disease states of two independent patient cohorts.

[0084] Figure 6B Shows the ROC curves of the training models for predicting clinical-defined disease as deteriorating or quiescent.

[0085] Figure 7 Shows the sequential forward selection of features based on the absolute quantitative data of the Expanded Disability Status Scale (EDSS).

[0086] Figure 8 Shows for implementing Figure 1A 、 1B and the exemplary computer of the entity shown in 1C. Detailed Description

[0087] I. Definition

[0088] Unless otherwise specified, the terms used in the claims and the specification are defined as set forth below.

[0089] The term "subject" encompasses cells, tissues, or organisms, human or non-human, whether in vivo, ex vivo, or in vitro, male or female.

[0090] The term "mammal" encompasses both human and non-human mammals, and includes, but is not limited to, humans, non-human primates, canines, felines, murine, bovines, equines, and porcines.

[0091] The term "sample" can include a single cell or multiple cells, or cell debris, or an aliquot of a body fluid, such as a blood sample, obtained from a subject by means including venipuncture, excretion, ejaculation, massage, biopsy, needle aspiration, lavage sample, scraping, surgical incision or intervention, or other means known in the art. Examples of body fluid aliquots include amniotic fluid, aqueous humor, bile, lymphatic fluid, breast milk, interstitial fluid, blood, plasma, cerumen (ear wax), Cowper's fluid (pre-ejaculatory fluid), chyle, chyme, female ejaculate, menstrual blood, mucus, saliva, urine, vomit, tears, vaginal lubrication fluid, sweat, serum, semen, sebum, pus, pleural fluid, cerebrospinal fluid, synovial fluid, intracellular fluid, and vitreous humor.

[0092] The term "disease activity" encompasses the disease activity of any neurodegenerative disease, including multiple sclerosis, Parkinson's disease, Lewy body disease, Alzheimer's disease, amyotrophic lateral sclerosis (ALS), motor neuron disease, Huntington's disease, spinal muscular atrophy, Friedreich's ataxia, and Batten disease.

[0093] The term "multiple sclerosis" or "MS" encompasses all forms of multiple sclerosis, including relapsing-remitting multiple sclerosis (RRMS), secondary progressive multiple sclerosis (SPMS), primary progressive multiple sclerosis (PPMS), progressive relapsing multiple sclerosis (PRMS), and clinically isolated syndrome (CIS).

[0094] As used herein, the term "multiple sclerosis disease activity" or "disease activity of multiple sclerosis" refers to any diagnosis of multiple sclerosis (MS): the presence or absence of MS (e.g., general disease, mild disease), a shift in disease activity (e.g., increase or decrease), disease progression, the severity of MS, a relapse or episode associated with MS, a future or impending relapse or episode, relapse rate (e.g., annualized relapse rate), MS status (e.g., worsening or quiescent), confirmation of the absence of evidence of disease status, the response of a subject diagnosed with multiple sclerosis to a therapy, the degree of multiple sclerosis disability, the risk (e.g., likelihood) that a subject will develop multiple sclerosis at a subsequent time, changes in multiple sclerosis disease compared to a previous measurement (e.g., longitudinal changes in a patient relative to a baseline measurement), a measurement that provides disease activity information, or a differential diagnosis of the type of multiple sclerosis, the types including relapsing-remitting multiple sclerosis (RRMS), secondary progressive multiple sclerosis (SPMS), primary progressive multiple sclerosis (PPMS), progressive relapsing multiple sclerosis (PRMS), and clinically isolated syndrome (CIS).

[0095] In one embodiment, the term "multiple sclerosis disease activity" or "disease activity of multiple sclerosis" refers to a diagnosis of multiple sclerosis (MS). In one embodiment, the term "multiple sclerosis disease activity" or "disease activity of multiple sclerosis" refers to the presence or absence of MS (e.g., general disease, mild disease). In one embodiment, the term "multiple sclerosis disease activity" or "disease activity of multiple sclerosis" refers to a deviation in disease activity (e.g., increase or decrease). In one embodiment, the term "multiple sclerosis disease activity" or "disease activity of multiple sclerosis" refers to the severity of MS. In one embodiment, the term "multiple sclerosis disease activity" or "disease activity of multiple sclerosis" refers to a relapse or episode associated with MS. In one embodiment, the term "multiple sclerosis disease activity" or "disease activity of multiple sclerosis" refers to a future or impending relapse or episode. In one embodiment, the term "multiple sclerosis disease activity" or "disease activity of multiple sclerosis" refers to the relapse rate (e.g., annualized relapse rate). In one embodiment, the term "multiple sclerosis disease activity" or "disease activity of multiple sclerosis" refers to the MS status (e.g., worsening or quiescent). In one embodiment, the term "multiple sclerosis disease activity" or "disease activity of multiple sclerosis" refers to confirmation of the absence of evidence of disease status. In one embodiment, the term "multiple sclerosis disease activity" or "disease activity of multiple sclerosis" refers to the response of a subject diagnosed with multiple sclerosis to a therapy. In one embodiment, the term "multiple sclerosis disease activity" or "disease activity of multiple sclerosis" refers to the degree of multiple sclerosis disability. In one embodiment, the term "multiple sclerosis disease activity" or "disease activity of multiple sclerosis" refers to the risk (e.g., likelihood) that a subject will develop multiple sclerosis at a subsequent time. In one embodiment, the term "multiple sclerosis disease activity" or "disease activity of multiple sclerosis" refers to a change in multiple sclerosis disease compared to a previous measurement (e.g., longitudinal change of a patient relative to a baseline measurement). In one embodiment, the term "multiple sclerosis disease activity" or "disease activity of multiple sclerosis" refers to a measurement that provides information on disease activity.

[0096] In various embodiments, measurements that provide information on MS disease activity include measurements of: minor disease activity (e.g., presence or absence of a specific number of gadolinium-enhancing lesions, e.g., exactly one lesion), general disease activity (e.g., presence or absence of 1 or more gadolinium-enhancing lesions), deviation of disease activity (e.g., appearance or disappearance of active gadolinium-enhancing lesions), and severity of disease activity (e.g., number of gadolinium-enhancing lesions, where more gadolinium-enhancing lesions indicate an increase in disease severity). In one embodiment, measurements that provide information on MS disease activity include measurements of minor disease activity (e.g., presence or absence of a specific number of gadolinium-enhancing lesions, e.g., exactly one lesion). In one embodiment, measurements that provide information on MS disease activity include measurements of general disease activity (e.g., presence or absence of 1 or more gadolinium-enhancing lesions). In one embodiment, measurements that provide information on MS disease activity include measurements of deviation of disease activity (e.g., appearance or disappearance of active gadolinium-enhancing lesions). In one embodiment, measurements that provide information on MS disease activity include measurements of severity of disease activity (e.g., number of gadolinium-enhancing lesions, where more gadolinium-enhancing lesions indicate an increase in disease severity).

[0097] In one embodiment, the term "multiple sclerosis disease activity" or "disease activity of multiple sclerosis" refers to the progression of MS (e.g., MS disease progression). In one embodiment, measurements that provide information on MS disease activity include measurements of disease progression (e.g., disease progression as represented by the Expanded Disability Status Scale (EDSS)). In some embodiments, the term "multiple sclerosis disease activity" or "disease activity of multiple sclerosis" does not include the progression of MS (e.g., MS disease progression). Specifically, in such embodiments as disclosed herein, the biomarker panel for predicting "multiple sclerosis disease activity" is different from the biomarker panel for predicting "multiple sclerosis disease progression".

[0098] The terms "marker", "markers", "biomarker", and "biomarkers" encompass but are not limited to lipids, lipoproteins, proteins, cytokines, chemokines, growth factors, peptides, nucleic acids, genes, and oligonucleotides, and their associated complexes, metabolites, mutations, variants, polymorphisms, modifications, fragments, subunits, degradation products, elements, and other analytes or sample-derived measurements. Markers may also include such mutations, copy number variations, and / or transcriptional variants in cases where mutant proteins, mutant nucleic acids, copy number changes, and / or transcriptional variants can be used to generate a predictive model or can be used in a predictive model developed using related markers (e.g., non-mutated versions of proteins or nucleic acids, alternative transcripts, etc.).

[0099] The term "antibody" is used in the broadest sense and specifically encompasses monoclonal antibodies (including full-length monoclonal antibodies), polyclonal antibodies, multispecific antibodies (e.g., bispecific antibodies), and antigen-binding antibody fragments, provided they exhibit the desired biological activity, e.g., an antibody or its antigen-binding fragment.

[0100] As used herein, "antibody fragment" and all its grammatical variants are defined as a portion of a full antibody that contains the antigen-binding site or variable region of the full antibody, wherein the portion does not contain the constant heavy chain domains of the full antibody Fc region (i.e., CH2, CH3, and CH4, depending on the antibody isotype). Examples of antibody fragments include Fab, Fab', Fab'-SH, F(ab') 2 and Fv fragments; diabodies; any antibody fragment that is a polypeptide having a primary structure consisting of an uninterrupted sequence of contiguous amino acid residues (referred to herein as a "single-chain antibody fragment" or "single-chain polypeptide").

[0101] The term "biomarker panel" refers to a set of biomarkers that provides information for predicting disease activity in multiple sclerosis. For example, the expression levels of a set of biomarkers in a biomarker panel can provide information for predicting disease activity in multiple sclerosis (e.g., predicting MS relapse). In various embodiments, a biomarker panel can include two, three, four, five, six, seven, eight, nine, ten, eleven, twelve, thirteen, fourteen, fifteen, sixteen, seventeen, eighteen, nineteen, twenty, twenty-one, twenty-two, twenty-three, twenty-four, or twenty-five biomarkers.

[0102] The term "obtaining a data set associated with a sample" encompasses obtaining a set of data determined from at least one sample. Obtaining a data set encompasses obtaining a sample and processing the sample to experimentally determine the data. The phrase also encompasses receiving, e.g., from a third party, a set of data that the third party has processed to experimentally determine the data. Additionally, the phrase encompasses mining data from at least one database or at least one publication or a combination of a database and a publication. A person of ordinary skill in the art can obtain a data set by a variety of known means including those stored on a memory.

[0103] It must be noted that, unless the context clearly dictates otherwise, the singular forms "a / an" and "the" as used in this specification include plural referents.

[0104] II. Overview of System Environment

[0105] Figure 1AAn overview of a system environment 100 for assessing disease activity in a subject according to an embodiment is shown. The system environment 100 provides context for introducing a biomarker quantification assay 120 and an activity prediction system 130.

[0106] In various embodiments, a test sample is obtained from a subject 110. The sample can be obtained by an individual or a third party such as a healthcare professional. Examples of healthcare professionals include physicians, emergency medical technicians, nurses, first responders, psychologists, phlebotomists, medical physicists, nurse practitioners, surgeons, dentists, and any other apparent healthcare professionals known to those skilled in the art.

[0107] By performing a biomarker quantification assay 120, the test sample is tested to determine the value of one or more biomarkers. The biomarker quantification assay 120 determines the quantitative expression values of one or more biomarkers from the test sample. The biomarker quantification assay 120 can be an immunoassay, and more specifically a multiplex immunoassay, examples of which are described in more detail below. The expression levels of individual biomarkers can be obtained in a single run using a single test sample obtained from the subject 110. The quantitative expression values of the biomarkers are provided to the activity prediction system 130.

[0108] Generally, the activity prediction system 130 includes one or more computers, embodied as a computer system 700 as discussed below with respect to Figure 8 Thus, in various embodiments, the steps described with reference to the activity prediction system 130 are performed by computer simulation. The activity prediction system 130 analyzes the biomarker expression values received from the biomarker quantification assay 120 to generate an assessment of the disease activity 140 in the subject 110.

[0109] In various embodiments, the biomarker quantification assay 120 and the activity prediction system 130 can be used by different parties. For example, a first party performs the biomarker quantification assay 120 and then provides the results to a second party that implements the activity prediction system 130. For example, the first party can be a clinical laboratory that obtains a test sample from the subject 110 and performs the assay 120 on the test sample. The second party receives the expression values of the biomarkers obtained by performing the assay 120 and analyzes the expression values using the activity prediction system 130.

[0110] Now refer to Figure 1B , which shows a block diagram of the computer logic components of an activity prediction system 130 according to an embodiment. Specifically, the activity prediction system 130 can include a model training module 150, a model deployment module 160, and a training data memory 170.

[0111] The following describes each component of the activity prediction system 130 in two phases: 1) a training phase and 2) a deployment phase. More specifically, the training phase refers to constructing and training one or more prediction models based on training data, which includes quantitative expression values obtained from individuals known to be healthy, at rest, in remission, or in a state of remission, or individuals known to have disease activity, in a deteriorating state, or in a relapsing state. Thus, the prediction models are trained to predict disease activity in a subject based on the quantitative biomarker expression values. During the deployment phase, the prediction models are applied to the quantitative biomarker expression values of test samples obtained from the subject of interest to generate a prediction of the disease activity of the subject of interest.

[0112] In some embodiments, the components of the activity prediction system 130 are applied during one of the training phase and the deployment phase. For example, the model training module 150 and the training data memory 170 (indicated by the dashed lines in Figure 1B are applied during the training phase, while the model deployment module 160 is applied during the deployment phase. In various embodiments, the training phase and the deployment phase can be carried out to implement a continuously trained model. For example, the model training module 150 can train a model that the model deployment module 160 can then deploy. The same model can be subjected to additional training by the model training module 150 (e.g., continuously trained using, for example, newly obtained training data). Thus, as the model is continuously trained, it can exhibit improved predictive ability when analyzing samples during deployment.

[0113] In various embodiments, the components of the activity prediction system 130 can be carried out by different parties, depending on whether the component is applied during the training phase or the deployment phase. In this case, the training and deployment of the prediction model are carried out by different parties. For example, the model training module 150 and the training data memory 170 applied during the training phase can be used by a first party (e.g., to train the prediction model), and the model deployment module 160 applied during the deployment phase can be carried out by a second party (e.g., to deploy the prediction model).

[0114] III. Prediction Model

[0115] III.A. Training the Prediction Model

[0116] During the training phase, the model training module 150 trains one or more prediction models using training data that includes biomarker expression values. Refer to Figure 1B, the training data can be stored in the training data memory 170. In various embodiments, the activity prediction system 130 generates training data including biomarker expression values by analyzing the biomarker expression values in test samples. In various embodiments, the activity prediction system 130 obtains training data including biomarker expression values from a third party. The third party may have analyzed the test samples to determine the biomarker expression values.

[0117] In various embodiments, the training data including biomarker expression values is sourced from clinical subjects. For example, the training data can be biomarker expression values measured from test samples obtained from clinical subjects. Examples of biomarker expression values sourced from clinical subjects include biomarker expression values obtained through clinical studies such as the CLIMB study in multiple sclerosis (e.g., the Comprehensive Longitudinal Investigation of Multiple Sclerosis at Brigham and Women's Hospital), the Accelerated Cure Project (ACP) in multiple sclerosis, and the Expression, Proteomics, Imaging, Clinical (EPIC) study at UCSF and the University Hospital Basel cohort (UHBC).

[0118] In various embodiments, the training data further includes reference ground truth values indicating disease activity such as multiple sclerosis disease activity. As an example, the training data includes reference ground truth values that identify the presence or absence of multiple sclerosis (MS), relapses or episodes associated with MS, relapse rate (e.g., annualized relapse rate), MS status (e.g., worsening or quiescent), response of a subject diagnosed with multiple sclerosis to a therapy, degree of multiple sclerosis disability, risk (e.g., likelihood) that a subject will develop multiple sclerosis at a subsequent time, or measurement of minor disease activity (e.g., presence or absence of a specific number of gadolinium-enhanced lesions, e.g., one, two, three, or four lesions), or measurement of general disease activity (e.g., presence or absence of 1 or more gadolinium-enhanced lesions).

[0119] Reference Figure 1C , which shows an exemplary set of training data 190 according to an embodiment. As Figure 1C shown, the training data 190 includes data corresponding to multiple individuals (e.g., the first column describes individuals 1, 2, 3, 4, …). For each individual, the training data 190 includes quantitative expression values of different biomarkers obtained from the corresponding individual (e.g., A1, B1, A2, B2, etc.). In some embodiments, the quantitative expression values are determined by the biomarker quantification assay 120 shown in FIG. 1. Although Figure 1CFour individuals and two different markers (Marker A and Marker B) are shown, but the training data 190 can include dozens, hundreds, or thousands of individuals and dozens, hundreds, or thousands of markers.

[0120] As Figure 1C shown, the first training instance (e.g., the first row) of the training data refers to the quantitative expression values of individual 1 and the corresponding Marker A (e.g., A1) and the quantitative expression value of Marker B (e.g., B1). Similarly, the second training instance (e.g., the second row) of the training data refers to the quantitative expression values of individual 2 and the corresponding Marker A (e.g., A2) and the quantitative expression value of Marker B (e.g., B2). Individuals 3 and 4 have corresponding marker values, as Figure 1C shown.

[0121] As Figure 1C shown, the training data 190 also includes reference true values ("Indication" column), which identify whether the corresponding individual has a positive or negative indication regarding disease activity. As an example, each indication can be an indication of disease activity in a patient. For example, referring to the first training instance (e.g., the first row), a "positive" indication can reflect the presence of disease activity in individual 1. For example, an MRI scan of individual 1 may show the presence of one or more gadolinium-enhanced lesions. Similarly, an indication of a negative result (e.g., individual 3 or individual 4) reflects a negative indication of disease activity in the corresponding individual. For example, MRI scans of individuals 3 and 4 may show the absence of gadolinium-enhanced lesions.

[0122] In some embodiments, the model training module 150 retrieves the training data from the training data memory 170 and randomly partitions the training data into a training set and a test set. As an example, 80% of the training data can be partitioned into the training set, while the other 20% can be partitioned into the test set. Other ratios of the training set and the test set can be implemented. Thus, the training set is used to train the prediction model, while the test set is used to validate the prediction model.

[0123] In various embodiments, the prediction model is any of the following: a regression model (e.g., linear regression, logistic regression, or polynomial regression), a decision tree, a random forest, a support vector machine, a naive Bayes model, k-means clustering, or a neural network (e.g., a feedforward network, a convolutional neural network (CNN), a deep neural network (DNN), an autoencoder neural network, a generative adversarial network, or a recurrent network (e.g., a long short-term memory network (LSTM), a bidirectional recurrent network, a deep bidirectional recurrent network)) or any combination thereof. For example, the prediction model can be a stacked classifier that includes both linear regression and a decision tree.

[0124] Methods implementable using machine learning can be used to train a prediction model, such as any of the following: linear regression algorithm, logistic regression algorithm, decision tree algorithm, support vector machine classification, naive Bayes classification, K-nearest neighbor classification, random forest algorithm, deep learning algorithm, gradient boosting algorithm, and dimensionality reduction techniques, such as manifold learning, principal component analysis, factor analysis, autoencoder regularization, and independent component analysis, or combinations thereof. In various embodiments, a supervised learning algorithm, an unsupervised learning algorithm, a semi-supervised learning algorithm (e.g., partially supervised), weak supervision, transfer, multi-task learning, or any combination thereof is used to train the cell disease model.

[0125] In various embodiments, the prediction model has one or more parameters, such as hyperparameters or model parameters. Hyperparameters are typically established before training. Examples of hyperparameters include learning rate, depth or leaves of a decision tree, number of hidden layers in a deep neural network, number of clusters in k-means clustering, penalty in a regression model, and regularization parameters related to a cost function. Model parameters are generally adjusted during training. Examples of model parameters include weights associated with nodes in a neural network layer, support vectors in a support vector machine, and coefficients in a regression model. The model parameters of the cell disease model are trained (e.g., adjusted) using training data to improve the prediction ability of the cell disease model.

[0126] The model training module 150 trains one or more prediction models, and each prediction model receives one or more biomarkers as input. In various embodiments, the model training module 150 constructs a prediction model that receives the expression values of two biomarkers as input. In various embodiments, the model training module 150 constructs a prediction model that receives the expression values of three biomarkers as input. In various embodiments, the model training module 150 constructs a prediction model that receives the expression values of four biomarkers as input. In some embodiments, the model training module 150 constructs a prediction model for more than four biomarkers. For example, the prediction model receives the expression values of 8 biomarkers (e.g., 8 biomarkers classified as layer 1 in Table 2 or any corresponding alternative biomarkers in Table 10) as input. As another example, the prediction model receives the expression values of 17 biomarkers (e.g., 17 biomarkers classified as layer 1 or layer 2 in Table 2 or any corresponding alternative biomarkers in Table 10) as input. As another example, the prediction model receives the expression values of 21 biomarkers (e.g., 21 biomarkers classified as layer 1, layer 2, or layer 3 in Table 2 or any corresponding alternative biomarkers in Table 10) as input.

[0127] In various embodiments, the model training module 150 identifies a set of biomarkers to be used in training a prediction model. The model training module 150 may start with a list of candidate biomarkers that are promising for predicting disease activity. In one embodiment, the candidate biomarkers may be biomarkers identified through a literature management program. In some embodiments, the candidate biomarkers may be biomarkers whose expression values in test samples obtained from individuals positive for disease activity (e.g., having MS, being in a worsening state, etc.) are statistically significant compared to the biomarker expression values in test samples obtained from individuals negative for disease activity.

[0128] In one embodiment, the model training module 150 performs a feature selection process to identify a set of biomarkers to be included in the biomarker panel. For example, the model training module 150 performs sequential forward feature selection based on the expression values of the biomarkers and their importance in predicting a specific endpoint (e.g., disease activity). For example, a candidate biomarker determined to be highly correlated with a specific disease activity endpoint will be considered very important and thus likely to be included in the biomarker panel compared to other biomarkers that are not highly correlated with the disease activity endpoint.

[0129] In some embodiments, the importance of each biomarker for the disease activity endpoint is determined by using a method including one of random forest (RF), gradient boosting (GBM), extreme gradient boosting (XGB), or LASSO algorithms. For example, if the random forest algorithm is used, the model training module 150 may generate a variable importance plot showing the importance of each candidate biomarker. Specifically, the random forest algorithm may provide for each candidate biomarker 1) the average decrease in model accuracy and 2) the average decrease in the Gini coefficient, which is a measure of the degree of contribution of each candidate biomarker to the homogeneity of nodes and leaves in the random forest. In one case, the importance of each candidate biomarker depends on one or both of the average decrease in model accuracy and the average decrease in the Gini coefficient. Each of GBM, XGB, and LASSO may also be used to rank the importance of each candidate biomarker based on the influence value. Thus, the model training module 150 may use one of the methods including RF, GBM, XGB, or LASSO to generate a ranking for each candidate biomarker.

[0130] Each prediction model is iteratively trained using the quantitative expression values of the markers for each individual as input. For example, referring again to Figure 1C, one iteration involves providing a training instance (e.g., a row of training data) that includes quantitative expression values of biomarkers (e.g., "A1" and "B1") for a specific individual (e.g., individual 1). Each predictive model is trained on reference ground truth data that includes an indication (e.g., a positive or negative outcome). In various embodiments, during the training iteration, each predictive model is trained (e.g., its parameters are adjusted) to minimize the prediction error between the prediction of MS activity output by the predictive model and the ground truth data. In various embodiments, the prediction error is calculated based on a loss function, examples of which include an L1 regularization (Lasso regression) loss function, an L2 regularization (Ridge regression) loss function, or a combination of L1 and L2 regularization (ElasticNet).

[0131] III.B. Deploying the Prediction Model

[0132] During the deployment phase, the model deployment module 160 (as Figure 1B shown) analyzes the quantitative biomarker expression values from a test sample obtained from a subject of interest by applying the trained predictive model. In some embodiments, the subject has not previously been diagnosed with a disease, and thus, the deployment of the predictive model enables the diagnosis of the disease by computer simulation based on the quantitative biomarker expression values derived from the subject. In some embodiments, the subject has previously been diagnosed with a disease. Here, the deployment of the predictive model enables the prediction of disease activity by computer simulation based on the quantitative biomarker expression values derived from the subject.

[0133] In various embodiments, quantitative biomarker expression values are provided as input to the predictive model. The predictive model analyzes the quantitative biomarker expression values and outputs an assessment of disease activity.

[0134] In various embodiments, the assessment of disease activity is a predictive score that provides information about disease activity in a subject. In various embodiments, the predictive score output by the predictive model is compared to one or more reference scores to determine a measure of disease activity. A reference score refers to a previously determined score, further described below as a "healthy score" or "diseased score", corresponding to a diseased or non-diseased patient. For example, the one or more scores can be a "healthy score" that corresponds to a healthy patient, the patient's own baseline at a previous time point when the patient did not exhibit disease activity (e.g., longitudinal analysis), a patient clinically diagnosed with the disease but not exhibiting disease activity, or a threshold score (e.g., cut-off value). As another example, the one or more scores can be a "diseased score" that corresponds to a diseased patient, the patient's own score indicating disease activity at a previous time point, or a threshold score (e.g., cut-off value). As an example, a threshold score can correspond to a healthy patient and can be generated by training the predictive model using biomarker expression values from healthy patients. As another example, a threshold score can correspond to a diseased patient and can be generated by training the predictive model using biomarker expression values from diseased patients.

[0135] In various embodiments, the assessment of disease activity corresponds to the presence or absence of disease. In one embodiment, the predictive score output by the predictive model can be compared to a healthy score. If the predictive score of the subject is significantly different from the healthy score (e.g., p-value < 0.05), the subject can be classified as having the disease. In one embodiment, the predictive score output by the predictive model can be compared to a diseased score. If the predictive score of the subject is significantly different from the diseased score (e.g., p-value < 0.05), the subject can be classified as not having the disease. In some embodiments, the predictive score output by the predictive model is compared to both a healthy score and a diseased score. For example, if the predictive score of the subject is significantly different from the healthy score (e.g., p-value < 0.05) and not significantly different from the diseased score of a patient diagnosed with the disease (e.g., p-value > 0.05), the subject can be classified as having the disease. In various embodiments, depending on the classification of the subject, the subject can receive treatment. In other words, the assessment can guide the treatment of the subject. For example, if the subject is classified as having the disease, a treatment intervention can be administered to the subject to treat the disease.

[0136] In various embodiments, the assessment of disease activity corresponds to the presence or absence of mild disease. In one embodiment, the predicted score output by the prediction model can be compared to the score corresponding to an individual previously determined to have mild disease (e.g., a specific number of gadolinium-enhanced lesions on an MRI scan, e.g., exactly one lesion). If the predicted score of the subject does not differ significantly (e.g., p-value > 0.05) compared to the score corresponding to an individual previously determined to have mild disease, then the subject can be classified as having mild disease. If the predicted score of the subject differs significantly (e.g., p-value < 0.05) compared to the score corresponding to an individual previously determined to not have mild disease, then the subject can be classified as having mild disease. In one embodiment, the predicted score output by the prediction model is compared to the score corresponding to an individual without mild disease (e.g., zero gadolinium-enhanced lesions on an MRI scan). If the predicted score of the subject does not differ significantly (e.g., p-value > 0.05) compared to the score corresponding to an individual without mild disease (e.g., zero gadolinium-enhanced lesions on an MRI scan), then the subject can be classified as not having mild disease. Alternatively, if the predicted score of the subject differs significantly (e.g., p-value < 0.05) compared to the score corresponding to an individual without mild disease (e.g., zero gadolinium-enhanced lesions on an MRI scan), then the subject can be classified as having mild disease.

[0137] In some embodiments, the predicted score output by the prediction model is compared to both the score corresponding to an individual previously determined to have mild disease (e.g., a specific number of gadolinium-enhanced lesions on an MRI scan) and the score corresponding to an individual without mild disease (e.g., zero gadolinium-enhanced lesions on an MRI scan). For example, if the predicted score of the subject differs significantly (e.g., p-value < 0.05) compared to the score corresponding to an individual without mild disease (e.g., zero gadolinium-enhanced lesions on an MRI scan) and does not differ significantly (e.g., p-value > 0.05) compared to the score corresponding to an individual previously determined to have mild disease (e.g., a specific number of gadolinium-enhanced lesions on an MRI scan, e.g., exactly one gadolinium-enhanced lesion), then the subject can be classified as having mild disease.

[0138] In various embodiments, the assessment of disease activity corresponds to the presence or absence of a general disease. In one embodiment, the predicted score output by the prediction model can be compared to the scores of individuals corresponding to a previously determined presence of a general disease (e.g., one or more gadolinium-enhanced lesions on an MRI scan). If the predicted score of the subject is significantly different (e.g., p-value < 0.05) compared to the scores of individuals corresponding to a previously determined absence of a general disease, the subject can be classified as having the general disease. If the predicted score of the subject is not significantly different (e.g., p-value > 0.05) compared to the scores of individuals corresponding to a previously determined absence of a general disease, the subject can be classified as not having the general disease. In one embodiment, the predicted score output by the prediction model is compared to the scores of individuals corresponding to an absence of a general disease (e.g., zero gadolinium-enhanced lesions on an MRI scan). If the predicted score of the subject is not significantly different (e.g., p-value > 0.05) compared to the scores of individuals corresponding to an absence of a general disease (e.g., zero gadolinium-enhanced lesions on an MRI scan), the subject can be classified as not having the general disease. If the predicted score of the subject is significantly different (e.g., p-value < 0.05) compared to the scores of individuals corresponding to an absence of a general disease (e.g., zero gadolinium-enhanced lesions on an MRI scan), the subject can be classified as having the general disease. In some embodiments, the predicted score output by the prediction model is compared to both the scores of individuals corresponding to a previously determined presence of a general disease (e.g., one or more gadolinium-enhanced lesions on an MRI scan) and the scores of individuals corresponding to an absence of a general disease (e.g., zero gadolinium-enhanced lesions on an MRI scan). For example, if the predicted score of the subject is significantly different (e.g., p-value < 0.05) compared to the scores of individuals corresponding to an absence of a general disease (e.g., zero gadolinium-enhanced lesions on an MRI scan) and not significantly different (e.g., p-value > 0.05) compared to the scores of individuals corresponding to a previously determined presence of a general disease (e.g., one or more gadolinium-enhanced lesions on an MRI scan), the subject can be classified as having the general disease.

[0139] In various embodiments, the assessment of disease activity corresponds to a shift in the direction of predicted increased or decreased disease activity based on the number of gadolinium-enhanced lesions. In one embodiment, the predicted score output by the prediction model can be compared to the score corresponding to an individual who has previously been determined to have experienced an increase in disease activity (e.g., an increase in the number of gadolinium-enhanced lesions on an MRI scan). If the predicted score of the subject is significantly different (e.g., p-value < 0.05) compared to the score corresponding to an individual who has not previously experienced an increase in disease activity, the subject can be classified as likely to experience an increase in disease activity. If the predicted score of the subject is not significantly different (e.g., p-value > 0.05) compared to the score corresponding to an individual who has not previously experienced an increase in disease activity, the subject can be classified as unlikely to experience an increase in disease activity. In one embodiment, the predicted score output by the prediction model can be compared to the score corresponding to an individual who has previously been determined to have experienced a decrease in disease activity (e.g., a decrease in the number of gadolinium-enhanced lesions on an MRI scan). If the predicted score of the subject is not significantly different (e.g., p-value > 0.05) compared to the score corresponding to an individual who has experienced a decrease in disease activity, the subject can be classified as likely to experience a decrease in disease activity. If the predicted score of the subject is significantly different (e.g., p-value < 0.05) compared to the score corresponding to an individual who has not experienced a decrease in disease activity, the subject can be classified as likely to experience a decrease in disease activity. In some embodiments, the predicted score output by the prediction model can be compared to both the score corresponding to an individual who has previously been determined to have experienced an increase in disease activity (e.g., an increase in the number of gadolinium-enhanced lesions on an MRI scan) and the score corresponding to an individual who has experienced a decrease in disease activity (e.g., a decrease in the number of gadolinium-enhanced lesions on an MRI scan). For example, if the predicted score of the subject is significantly different (e.g., p-value < 0.05) compared to the score corresponding to an individual who has experienced a decrease in disease activity (e.g., a decrease in the number of gadolinium-enhanced lesions on an MRI scan) and not significantly different (e.g., p-value > 0.05) compared to the score corresponding to an individual who has experienced an increase in disease activity (e.g., an increase in the number of gadolinium-enhanced lesions on an MRI scan), the subject can be classified as likely to experience an increase in disease activity. In various embodiments, if the predicted score of the subject is not significantly different (e.g., p-value > 0.05) compared to both the score corresponding to an individual who has experienced an increase in disease activity and the score corresponding to an individual who has experienced a decrease in disease activity, the subject can be classified as unlikely to experience an increase or decrease in disease activity (e.g., the subject's disease activity is stable).

[0140] In various embodiments, the assessment of disease activity corresponds to the disease state of a subject. For example, if the disease is MS, the disease state of the subject is either quiescent or worsening. In one embodiment, the predicted score output by the prediction model can be compared to the score corresponding to an individual previously determined to be in a quiescent state (e.g., clinically determined to be in a quiescent state). If the predicted score of the subject is significantly different (e.g., p-value < 0.05) compared to the score corresponding to an individual previously determined not to be in a quiescent state, the subject can be classified as being in a quiescent state. If the predicted score of the subject is not significantly different (e.g., p-value > 0.05) compared to the score corresponding to an individual previously determined to be in a quiescent state, the subject can be classified as not being in a quiescent state. In one embodiment, the predicted score output by the prediction model is compared to the score corresponding to an individual previously determined to be in a worsening state. If the predicted score of the subject is not significantly different (e.g., p-value > 0.05) compared to the score corresponding to an individual previously determined to be in a worsening state, the subject can be classified as being in a worsening state. If the predicted score of the subject is significantly different (e.g., p-value < 0.05) compared to the score corresponding to an individual previously determined to be in a worsening state, the subject can be classified as not being in a worsening state. In some embodiments, the predicted score output by the prediction model is compared to both the score corresponding to an individual previously determined to be in a quiescent state and the score corresponding to an individual previously determined to be in a worsening state. For example, if the predicted score of the subject is significantly different (e.g., p-value < 0.05) compared to the score corresponding to an individual in a quiescent state and not significantly different (e.g., p-value > 0.05) compared to the score corresponding to an individual previously determined to be in a worsening state, the subject can be classified as being in a worsening state.

[0141] In various embodiments, the assessment of disease activity corresponds to the likely response to a therapy provided to a subject. In one embodiment, the predicted score output by a prediction model can be compared to scores corresponding to individuals previously determined to be responsive to the therapy (e.g., clinically determined to be responsive to the therapy). If the predicted score of the subject is significantly different (e.g., p-value < 0.05) compared to scores corresponding to individuals previously determined to be non-responsive to the therapy, the subject can be classified as a responder. If the predicted score of the subject is not significantly different (e.g., p-value > 0.05) compared to scores corresponding to individuals previously determined to be responsive to the therapy, the subject can be classified as a responder. In one embodiment, the predicted score output by a prediction model can be compared to scores corresponding to individuals previously determined to be non-responders. If the predicted score of the subject is not significantly different (e.g., p-value > 0.05) compared to scores corresponding to individuals previously determined to be non-responders, the subject can be classified as a non-responder. If the predicted score of the subject is significantly different (e.g., p-value < 0.05) compared to scores corresponding to individuals previously determined to be responders, the subject can be classified as a non-responder. In some embodiments, the predicted score output by a prediction model is compared to both scores corresponding to individuals previously determined to be responders and scores corresponding to individuals previously determined to be non-responders. For example, if the predicted score of the subject is significantly different (e.g., p-value < 0.05) compared to scores corresponding to individuals previously determined to be non-responders and not significantly different (e.g., p-value > 0.05) compared to scores corresponding to individuals previously determined to be responders, the subject can be classified as a responder.

[0142] In one embodiment, the assessment of disease activity is an assessment of disease progression and can correspond to the MS disability level of a subject diagnosed with multiple sclerosis. In one embodiment, the MS disability level corresponds to the EDSS. In various embodiments, the assessment corresponding to the subject (e.g., predicted score) is compared to a plurality of scores. Each score can correspond to a group of individuals clinically classified by disability level. For example, a first score can correspond to individuals clinically classified according to an EDSS score of 1. Additional scores can correspond to groups of individuals clinically classified according to scores of 1.5, 2.0, 2.5, 3.0, 3.5, 4.0, 4.5, 5.0, 5.5, 6.0, 6.5, 7.0, 7.5, 8.0, 8.5, 9.0, 9.5, and 10.0. In one case, if the predicted score of the subject output by the prediction model is not significantly different (e.g., p-value > 0.05) compared to one group and is significantly different (e.g., p-value < 0.05) compared to all other groups, the subject can be classified as one of the EDSS scores. The subject can be treated according to a clinical protocol based on the classification.

[0143] In one embodiment, the assessment of disease activity corresponds to the risk (e.g., likelihood) that a subject will develop the disease at a subsequent time. In various embodiments, an assessment corresponding to the subject (e.g., a prediction score) is compared to a plurality of scores. Each score may correspond to a group of individuals in a risk group that is clinically classified as having a specific risk of developing MS. As an example, the risk groups may be divided into a high-risk group, a medium-risk group, and a low-risk group. In one case, if the prediction score of a subject does not differ significantly (e.g., p-value > 0.05) compared to one group and differs significantly (e.g., p-value < 0.05) compared to other groups, the subject may be classified into a risk group. Thus, a subject may make lifestyle and / or treatment changes based on the prediction of the risk / likelihood of developing MS.

[0144] In various embodiments, the measurement of disease activity predicted by a prediction model provides additional utility for managing a patient's disease activity. As an example, the measurement of disease activity predicted by a prediction model can be used to select a candidate therapeutic agent or to determine the effectiveness of a previously administered therapeutic agent.

[0145] In various embodiments, the measurement of disease activity predicted by a prediction model for a patient can be compared to a previous measurement of disease activity to determine whether a therapeutic agent administered to the patient is showing efficacy. As an example, the previous measurement of disease activity can be a prediction determined for the same patient (e.g., a baseline measurement of disease activity). Thus, in this example, the comparison of the measurement of disease activity to the previous measurement of disease activity is a longitudinal analysis of a patient being treated with a therapeutic agent. Thus, the difference or lack of difference between the measurement of disease activity and the previous measurement of disease activity can indicate that the therapeutic agent has an effect or lacks an effect. As another example, the previous measurement of disease activity can be a measurement determined for a group of patients (e.g., a reference group of patients). In this example, the comparison of the measurement of disease activity to the previous measurement of disease activity can reveal whether a patient is experiencing an effect due to the therapeutic agent as demonstrated by the measurement of disease activity compared to the previous measurement of disease activity of the group of patients.

[0146] In various embodiments, if a comparison of a measurement of disease activity to a previous measurement of disease activity indicates that the currently administered therapeutic agent is not showing an effect, or is not showing an effect to the desired degree, then the treatment of the patient can be changed. In one embodiment, the therapeutic dose of the currently administered therapeutic agent can be changed to affect the patient's response. For example, the dose of the currently administered therapeutic agent can be increased. In one embodiment, a candidate therapeutic agent can be selected for administration to the patient. In various embodiments, the candidate therapeutic agent can be administered to the patient in place of the currently administered therapeutic agent, or in addition to the currently administered therapeutic agent, the candidate therapeutic agent can also be administered to the patient.

[0147] As another example, measurements of disease activity can be used to support symptom and medication tracking, care interventions, laboratory monitoring, and scheduled longitudinal MRI reporting. In such cases, measurements of disease activity can reduce unplanned healthcare utilization (e.g., unplanned physician office visits), thereby increasing patient and physician satisfaction.

[0148] IV. Biomarker Panel

[0149] In various embodiments, the assessment of disease activity involves implementing a univariate biomarker panel. Thus, a univariate biomarker panel contains one biomarker. In other embodiments, the assessment of disease activity involves implementing a multivariate biomarker panel. In such embodiments, a multivariate biomarker panel contains more than one biomarker. In various embodiments, a multivariate biomarker panel contains two biomarkers. In various embodiments, a multivariate biomarker panel contains 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, or 50 biomarkers. In certain embodiments, a multivariate biomarker panel contains 2 biomarkers. In certain embodiments, a multivariate biomarker panel contains 3 biomarkers. In certain embodiments, a multivariate biomarker panel contains 4 biomarkers. In certain embodiments, a multivariate biomarker panel contains 7 biomarkers. In certain embodiments, a multivariate biomarker panel contains 8 biomarkers. In certain embodiments, a multivariate biomarker panel contains 17 biomarkers. In certain embodiments, a multivariate biomarker panel contains 21 biomarkers.

[0150] In various embodiments described herein, a biomarker panel is implemented to assess or predict disease progression, such as MS disease progression. In various embodiments, the assessment of disease progression involves implementing a univariate biomarker panel. Thus, a univariate biomarker panel contains one biomarker. In other embodiments, the assessment of disease progression involves implementing a multivariate biomarker panel. In such embodiments, the multivariate biomarker panel for assessing disease progression contains more than one biomarker. In various embodiments, the multivariate biomarker panel for assessing disease progression contains two biomarkers. In various embodiments, the multivariate biomarker panel for assessing disease progression contains 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, or 50 biomarkers. In certain embodiments, the multivariate biomarker panel contains 2 biomarkers. In certain embodiments, the multivariate biomarker panel contains 3 biomarkers. In certain embodiments, the multivariate biomarker panel contains 4 biomarkers. In certain embodiments, the multivariate biomarker panel contains 7 biomarkers. In certain embodiments, the multivariate biomarker panel contains 8 biomarkers. In certain embodiments, the multivariate biomarker panel contains 17 biomarkers. In certain embodiments, the multivariate biomarker panel contains 21 biomarkers.

[0151] In one embodiment, the biomarkers in the biomarker panel may include one or more of the following: 6Ckine, adiponectin, adrenomedullin (ADM), alpha-1 antitrypsin (AAT), alpha-1-microglobulin (A1Micro), alpha-2-macroglobulin (A2Macro), alpha-fetoprotein (AFP), amphiregulin (AR), angiopoietin, angiopoietin 1 (ANG-1), angiopoietin 2 (ANG-2), angiotensin converting enzyme (ACE), antileukoprotease (ALP), antithrombin III (ATIII), apolipoprotein A (Apo-A), apolipoprotein D (Apo-D), apolipoprotein E (Apo-E), AXL receptor tyrosine kinase (AXL), B cell activating factor (BAFF), B lymphocyte chemoattractant (BLC), beta-amyloid (1-40) (AB-40), beta-amyloid (1-42) (AB-42), beta-2 microglobulin (B2M), betacellulin (BTC), brain-derived neurotrophic factor (BDNF), C-reactive protein (CRP), cadherin 1 (E-Cad), calbindin, cancer antigen 125 (CA-125), cancer antigen 15-3 (CA 15-3), cancer antigen 19-9 (CA 19-9), carbonic anhydrase 9 (CA-9), carcinoembryonic antigen (CEA), carcinoembryonic antigen-related cell adhesion molecule 1 (CEACAM1), cathepsin D, CD40 ligand (CD40-L), CD163, ceruloplasmin, chemokine CC-4 (HCC-4), chromogranin A (CgA), ciliary neurotrophic factor (CNTF), clusterin (CLU), complement C3 (C3), complement factor H (CFH), complement factor H-related protein 1 (CFHR1), cystatin B, cystatin C, decorin, Dickkopf-related protein 1 (DKK-1), dopamine beta-hydroxylase (DBH), E-selectin, advanced glycation end product (EN-RAGE), eosinophil chemotactic factor-1 (Eotaxin-1), eosinophil chemotactic factor-2, eosinophil chemotactic factor-3, epidermal growth factor (EGF), epidermal growth factor receptor (EGFR), epithelial regulatory protein (EPR), epithelial-derived neutrophil activating protein 78 (ENA-78), erythropoietin (EPO), factor VII, Fas ligand (FasL), FASLG receptor (FAS), ferritin (FRTN), fibrinogen, fibrin 1C (Fib1C), ficolin 3, follicle-stimulating hormone (FSH), gastric inhibitory polypeptide (GIP), gelsolin, glucagon-like peptide-1 (GLP-1), glycogen phosphorylase isoenzyme BB (GPBB), granulocyte colony-stimulating factor (GCSF),Granulocyte macrophage colony-stimulating factor (GM-CSF), growth differentiation factor 15 (GDF-15), growth hormone (GH), growth-regulated alpha protein (GROα), haptoglobin, heat shock protein 70 (HSP-70), heparin-binding EGF-like growth factor (HB-EGF), hepatocyte growth factor (HGF), human chorionic gonadotropin beta (hCG), immunoglobulin A (IgA), immunoglobulin E (IgE), immunoglobulin M (IgM), insulin, insulin-like growth factor-binding protein 2 (IGFBP2), intercellular adhesion molecule 1 (ICAM-1), interferon alpha (IFN-α), interferon gamma (IFN-γ), interferon gamma-induced protein 10 (IP-10), interferon-inducible T cell alpha chemoattractant (ITAC), interleukin 1 alpha (IL-1α), interleukin 1 beta (IL-1β), interleukin 1 receptor antagonist (IL1ra), interleukin 2 (IL-2), interleukin 2 receptor alpha (IL2 receptor α), interleukin 3 (IL-3), interleukin 4 (IL-4), interleukin 5 (IL-5), interleukin 6 (IL-6), interleukin 6 receptor (IL6r), interleukin 6 receptor subunit beta (IL6Rβ), interleukin 7 (IL-7), interleukin 8 (IL-8), interleukin 10 (IL-10), interleukin 12 subunit p40 (IL12p40), interleukin 12 subunit p70 (IL12p70), interleukin 13 (IL13), interleukin 15 (IL15), interleukin 16 (IL16), interleukin 17 (IL17), interleukin 18 (IL18), interleukin 18 binding protein (IL18bp), interleukin 22 (IL22), interleukin 23 (IL23), interleukin 31 (IL31), kidney injury molecule 1 (KIM-1), lactoferrin (LTF), latency-associated peptide of transforming growth factor beta1 (LAP TGF b1), leptin, leptin receptor (leptin R), leucine-rich alpha-2 glycoprotein (LRG1), luteinizing hormone (LH), macrophage colony-stimulating factor 1 (M-CSF), macrophage-derived chemokine (MDC), macrophage inflammatory protein 1 alpha (MIP1-α), macrophage inflammatory protein 1 beta (MIP1-β), macrophage inflammatory protein 3 alpha (MIP3-α), macrophage inflammatory protein 3 beta (MIP3-β), macrophage migration inhibitory factor (MIF), macrophage stimulating protein (MSP), mast / stem cell growth factor receptor (SCFR), matrix metalloproteinase 1 (MMP-1), matrix metalloproteinase 2 (MMP-2), matrix metalloproteinase 3 (MMP-3), matrix metalloproteinase 7 (MMP-7), matrix metalloproteinase 9 (MMP-9), total matrix metalloproteinase 9 (total MMP-9), matrix metalloproteinase 10 (MMP-10), microalbumin, monocyte chemoattractant protein 1 (MCP-1)Monocyte chemoattractant protein 2 (MCP-2), Monocyte chemoattractant protein 3 (MCP-3), Monocyte chemoattractant protein 4 (MCP-4), Monokine induced by gamma interferon (MIG), Myeloid progenitor inhibitory factor 1 (MPIF-1), Myeloperoxidase (MPO), Myoglobin, Nerve growth factor beta (NGF-β), Neurofilament heavy polypeptide (NF-H), Neuron-specific enolase (NSE), Neuron cell adhesion molecule (NrCAM), Neuropilin-1, Neutrophil-activating peptide 2 (NAP-2), Omentin, Osteocalcin, Osteopontin, Osteoprotegerin (OPG), P-selectin, Pancreatic polypeptide (PPP), Trypsin inhibitor, pancreatic secretory (TATI), Paraoxonase-1 (PON1), Pepsinogen-I (PGI), Periostin, Pigment epithelium-derived factor (PEDF), Placenta growth factor (PLGF), Plasminogen activator inhibitor 1 (PAI-1), Platelet endothelial cell adhesion molecule (PECAM-1), Platelet-derived growth factor BB (PDGF-BB), Prolactin (PRL), Prostate-specific antigen-free (PSA-f), DJ-1 protein (DJ-1), Pulmonary and activation-regulated chemokine (PARC), Pulmonary surfactant-associated protein D (SP-D), Receptor for advanced glycation end products (RAGE), Resistin, S100 calcium-binding protein B (S100B), Serum amyloid A (SAA), Serum amyloid P component (SAP), Sex hormone-binding globulin (SHBG), Sortilin, ST2, Stem cell factor (SCF), Stromal cell-derived factor 1 (SDF-1), Soluble superoxide dismutase 1 (SOD-1), T cell-specific protein RANTES (RANTES), T lymphocyte-secreted protein I 309 (I309), Tamm Horsfall glycoprotein (THP), Tenascin C (TN-C), Tetranectin, Thrombin-activatable fibrinolysis inhibitor (TAFI), Thrombospondin-1, Thymus and activation-regulated chemokine (TARC), Thyroid-stimulating hormone (TSH), Thyroxine-binding globulin (TBG), Tissue inhibitor of metalloproteinase 1 (TIMP-1), Tissue inhibitor of metalloproteinase 2 (TIMP-2), TNF-related apoptosis-inducing ligand receptor 3 (TRAIL-R3), Transferrin receptor protein 1 (TFR1), Transforming growth factor beta3 (TGF-beta3), Tumor necrosis factor alpha (TNF-α), Tumor necrosis factor beta (TNF-β), Tumor necrosis factor ligand superfamily member 12 (Tweak), Tumor necrosis factor ligand superfamily member 13 (APRIL), Tumor necrosis factor receptor I (TNF-RI), Tumor necrosis factor receptor 2 (TNFR2), Vascular cell adhesion molecule 1 (VCAM-1), Vascular endothelial growth factor (VEGF)Visceral adipose tissue-derived serine protease inhibitor (serpin) A12 (Vaspin), visfatin, vitamin D binding protein (VDBP), vitronectin, von Willebrand factor (vWF), or YKL-40.

[0152] In some embodiments, the biomarkers in the biomarker panel include the biomarkers shown in Tables 8 to 10. In some embodiments, the biomarker may include one or more of the following: neurofilament light polypeptide chain (NEFL), myelin oligodendrocyte glycoprotein (MOG), cluster of differentiation 6 (CD6), chemokine (C-X-C motif) ligand 9 (CXCL9), osteoprotegerin (OPG), osteopontin (OPN), matrix metallopeptidase 9 (MMP-9), glial fibrillary acidic protein (GFAP), CUB domain-containing protein 1 (CDCP1), C-C motif chemokine ligand 20 (CCL20 / MIP 3-α), interleukin-12 subunit beta (IL-12B), amyloid beta precursor-like protein 1 (APLP1), tumor necrosis factor receptor superfamily member 10A (TNFRSF10A), type IV collagen alpha-1 (COL4A1), Serpin family A member 9 (SERPINA9), leucine-rich fibronectin transmembrane protein 2 (FLRT2), chemokine (C-X-C motif) ligand 13 (CXCL13), growth hormone (GH), versican core protein (VCAN), protogenin (PRTG), contactin-2 (CNTN2). In some embodiments, the biomarker further includes growth hormone (GH2), interleukin-18 (IL18), matrix metalloproteinase-2 (MMP-2), gamma-interferon-inducible lysosomal thiol reductase (IFI30), and chitinase-3-like protein 1 (CHI3L1 / YkL40).

[0153] In some embodiments, the biomarker may include one or more of the following: cell adhesion molecule 3 (CADM3), kallikrein-related peptidase 6 (KLK6), brevican (BCAN), oligodendrocyte myelin glycoprotein (OMG), CD5 molecule (CD5), cytotoxic and regulatory T cell molecule (CRTAM), CD244 molecule (CD244), tumor necrosis factor receptor superfamily member 9 (TNFRSF9), proteinase 3 (PRTN3), follistatin-like protein 3 (FSTL3), C-X-C motif chemokine ligand 10 (CXCL10), C-X-C motif chemokine ligand 11 (CXCL11), interleukin-18 binding protein (IL-18BP), macrophage scavenger receptor 1 (MSR1), C-C motif chemokine ligand 3 (CCL3), tumor necrosis factor ligand superfamily member 12 (TWEAK), trefoil factor 3 (TFF3), ectonucleotide pyrophosphatase / phosphodiesterase 2 (ENPP2), insulin-like growth factor binding protein 1 (IGFBP-1), interleukin 12A (IL12A), seizure-related 6 homolog-like (SEZ6L), dipeptidyl peptidase-like 6 (DPP6), neurocan (NCAN), tubulointerstitial nephritis antigen-like protein 1 (TINAGL1), calcium-activated nuclease 1 (CANT1), nectin cell adhesion molecule 2 (NECTIN2), neural cell proliferation and differentiation regulatory protein 1 (NPDC1), tumor necrosis factor receptor superfamily member 11A (TNFRSF11A), contactin 4 (CNTN4), neurotrophic receptor tyrosine kinase 2 (NTRK2), neurotrophic receptor tyrosine kinase 3 (NTRK3), cadherin 6 (CDH6), carcinoembryonic antigen-related cell adhesion molecule 8 (CEACAM8), mitotic arrest deficient 1-like protein 1 (MAD1L1), IgA Fc fragment receptor (FCAR), myeloperoxidase (MPO), osteomodulin (OMD), matrix extracellular phosphoglycoprotein (MEPE), GDNF family receptor alpha 3 (GDNFR-α-3), scavenger receptor class F member 2 (SCARF2), CD40 ligand (IgM), tumor necrosis factor receptor superfamily member 1B (TNF-R2), programmed cell death 1 ligand (PD-L1), Notch 3 (NOTCH3), contactin 1 (CNTN1), oncostatin M (OSM), transforming growth factor alpha (TGF-α), peptidoglycan recognition protein 1 (PGLYRP1), nitric oxide synthase 3 (NOS3).

[0154] In certain embodiments, a biomarker panel useful for generating predictions (e.g., predictions of MS disease activity or predictions of MS disease progression) comprises biomarkers identified as Layer A in Table 1, Layer 1 in Table 2, or any corresponding alternative biomarkers thereof in Table 10. For example, the biomarker panel comprises NEFL, MOG, CD6, CXCL9, OPG, OPN, MMP-9, and GFAP (Layer A in Table 1). As another example, the biomarker panel comprises NEFL, MOG, CD6, CXCL9, OPG, OPN, CXCL13, and GFAP (Layer 1 in Table 2).

[0155] In certain embodiments, a biomarker panel useful for generating predictions (e.g., predictions of MS disease activity or predictions of MS disease progression) comprises biomarkers identified as Layer B in Table 1, Layer 2 in Table 2, or any corresponding alternative biomarkers thereof in Table 10. For example, the biomarker panel comprises CDCP1, CCL20 / MIP 3-α, IL-12B, APLP1, TNFRSF10A, COL4A1, SERPINA9, FLRT2, and CXCL13 (Layer B in Table 1). As another example, the biomarker panel comprises CDCP1, CCL20 / MIP 3-α, IL-12B, APLP1, TNFRSF10A, COL4A1, SERPINA9, FLRT2, and TNFSF13B (Layer 2 in Table 2).

[0156] In certain embodiments, a biomarker panel useful for generating predictions (e.g., predictions of MS disease activity or predictions of MS disease progression) comprises biomarkers identified as Layer C in Table 1, Layer 3 in Table 2, or any corresponding alternative biomarkers thereof in Table 10. For example, the biomarker panel comprises GH, VCAN, PRTG, and CNTN2 (Layer C in Table 1 and Layer 3 in Table 2).

[0157] In certain embodiments, a biomarker panel useful for generating predictions (e.g., predictions of MS disease activity or predictions of MS disease progression) comprises biomarkers identified as Tier A and Tier B in Table 1, Tier 1 and Tier 2 in Table 2, or any corresponding alternative biomarkers thereof in Table 10. For example, the biomarker panel comprises NEFL, MOG, CD6, CXCL9, OPG, OPN, MMP-9, GFAP, CDCP1, CCL20 / MIP 3-α, IL-12B, APLP1, TNFRSF10A, COL4A1, SERPINA9, FLRT2, and CXCL13 (Tier A and Tier B in Table 1). As another example, the biomarker panel comprises NEFL, MOG, CD6, CXCL9, OPG, OPN, CXCL13, GFAP, CDCP1, CCL20 / MIP 3-α, IL-12B, APLP1, TNFRSF10A, COL4A1, SERPINA9, FLRT2, and TNFSF13B (Tier 1 and Tier 2 in Table 2).

[0158] In certain embodiments, a biomarker panel useful for generating predictions (e.g., predictions of MS disease activity or predictions of MS disease progression) comprises biomarkers identified as Tier A, Tier B, or Tier C in Table 1, Tier 1, Tier 2, or Tier 3 in Table 2, or any corresponding alternative biomarkers thereof in Table 10. For example, the biomarker panel comprises NEFL, MOG, CD6, CXCL9, OPG, OPN, MMP-9, GFAP, CDCP1, CCL20 / MIP 3-α, IL-12B, APLP1, TNFRSF10A, COL4A1, SERPINA9, FLRT2, CXCL13, GH, VCAN, PRTG, and CNTN2 (Tier A, Tier B, and Tier C in Table 1). As another example, the biomarker panel comprises NEFL, MOG, CD6, CXCL9, OPG, OPN, CXCL13, GFAP, CDCP1, CCL20 / MIP 3-α, IL-12B, APLP1, TNFRSF10A, COL4A1, SERPINA9, FLRT2, TNFSF13B, GH, VCAN, PRTG, CNTN2, GH2, IL18, MMP-2, IFI30, and CHI3L1 / YkL40. (Tier 1, Tier 2, and Tier 3 in Table 2).

[0159] In various embodiments, biomarker panels for generating predictions (e.g., predictions of disease activity or predictions of disease progression) comprise minimal predictive biomarker sets, such as biomarker pairs, biomarker triples, or biomarker quadruples. In various embodiments, at least one biomarker in a biomarker pair, biomarker triple, or biomarker quadruple is NEFL. In various embodiments, at least one biomarker in a biomarker pair, biomarker triple, or biomarker quadruple is MOG. In various embodiments, a biomarker pair, biomarker triple, or biomarker quadruple does not comprise NEFL. In such embodiments, a biomarker pair, biomarker triple, or biomarker quadruple that does not comprise NEFL does comprise MOG.

[0160] Examples of biomarker pairs that can be used to generate predictions (e.g., predictions of MS disease activity or predictions of MS disease progression) include: 1) NEFL and MOG, 2) NEFL and CD6, 3) NEFL and CXCL9, 4) NEFL and TNFRSF10A, 5) MOG and IL-12B, 6) CXCL9 and CD6, 7) MOG and CXCL9, 8) MOG and CD6, 9) CXCL9 and COL4A1, and 10) CD6 and VCAN. Additional examples of biomarker pairs that can predict multiple sclerosis disease activity include: 1) NEFL and TNFSF13B, 2) NEFL and CNTN2, 3) NEFL and CXCL9, 4) MOG and CDCP1, 5) MOG and TNFSF13B, and 6) MOG and CXCL9. Additional examples of biomarker pairs that can predict multiple sclerosis disease activity include: 1) NEFL and TNFSF13B, 2) NEFL and SERPINA9, 3) NEFL and GH, 4) MOG and TNFSF13B, 5) MOG and CXCL9, and 6) MOG and IL-12B.

[0161] Examples of biomarker triplets that can be used to generate predictions (e.g., predictions of MS disease activity or predictions of MS disease progression) include: 1) MOG, IL-12B, and APLP1, 2) MOG, CD6, and CXCL9, 3) CXCL9, COL4A1, and VCAN, 4) NEFL, CD6, and CXCL9, 5) NEFL, TNFRSF10A, and COL4A1, 6) MOG, IL-12B, and CNTN2, and 7) CD6, CCL20, and VCAN. Additional examples of biomarker triplets that can predict multiple sclerosis disease activity include: 1) NEFL, CNTN2, and TNFSF13B, 2) NEFL, APLP1, and TNFSF13B, 3) NEFL, TNFRSF10A, and TNFSF13B, 4) MOG, CXCL9, and TNFSF13B, 5) MOG, OPG, and TNFSF13B, and 6) MOG, CCL20, and TNFSF13B. Additional examples of biomarker triplets that can predict multiple sclerosis disease activity include: 1) NEFL, SERPINA9, and TNFSF13B, 2) NEFL, CNTN2, and TNFSF13B, 3) NEFL, APLP1, and TNFSF13B, 4) MOB, CXCL9, and TNFSF13B, 5) MOG, SERPINA9, and TNFSF13B, and 6) MOG, OPG, and TNFSF13B.

[0162] Examples of biomarker quartets that can be used to generate predictions (e.g., predictions of MS disease activity or predictions of MS disease progression) include: 1) NEFL, MOG, CD6, and CXCL9, 2) NEFL, CXCL9, TNFRSF10A, and COL4A1, 3) MOG, CXCL9, IL-12B, and APLP1, 4) CXCL9, COL4A1, OPG, and VCAN, 5) CXCL9, OPG, APLP1, and OPN, 6) NEFL, CD6, CXCL9, and CXCL13, 7) NEFL, MOG, CD6, and CXCL9, 8) NEFL, TNFRSF10A, COL4A1, and CCL20, 9) MOG, IL-12B, OPN, and CNTN2, and 10) CD6, COL4A1, CCL20, and VCA. Additional examples of biomarker quartets that can predict multiple sclerosis disease activity include: 1) NEFL, TNFRSF10A, CNTN2, and TNFSF13B, 2) NEFL, COL4A1, CNTN2, and TNFSF13B, 3) NEFL, TNFRSF10A, APLP1, and TNFSF13B, 4) MOG, CXCL9, APLP1, and TNFSF13B, 5) MOG, CXCL9, OPG, and TNFSF13B, and 6) MOG, CXCL9, OPG, and CNTN2. Additional examples of biomarker quartets that can predict multiple sclerosis disease activity include: 1) NEFL, CCL20, SERPINA9, and TNFSF13B, 2) NEFL, APLP1, SERPINA9, and TNFSF13B, 3) NEFL, CCL20, APLP1, and TNFSF13B, 4) MOG, CXCL9, OPG, and TNFSF13B, 5) MOG, OPG, SERPINA9, and TNFSF13B, and 6) MOG, CXCL9, SERPINA9, and TNFSF13B.

[0163] V. Biomarker

[0164] Dysregulation of the biomarkers disclosed herein may contribute to the development and / or progression of disease activity such as that of neurodegenerative diseases, which include multiple sclerosis, Parkinson's disease, Lewy body disease, Alzheimer's disease, amyotrophic lateral sclerosis (ALS), motor neuron disease, Huntington's disease, spinal muscular atrophy, Friedreich's ataxia, Batten disease, and the like. The biomarkers and their corresponding classifications are shown in Table 11 below. Exemplary categories include: neurodegeneration, myelin integrity, neurite growth and neurogenesis, inflammation, immune regulation, cell regulation, cell adhesion, gut-brain axis, metabolism, and neuroregulation categories. Additionally, the biomarkers and their involvement in specific locations (e.g., brain, blood-brain barrier, or blood) and cell types are shown in Tables 12, 13A, and 13B.

[0165] NEFL is a 68 kDa biomarker that can reflect axonal damage in the microenvironment. In other words, NEFL often serves as a proxy for axonal degeneration. Additionally, NEFL interacts with other biomarkers such as MAP2, protein kinase N1, and tuberous sclerosis complex 1 (TSC1).

[0166] COL4A1 is a 26 kDa biomarker that is involved in cell proliferation, migration, extracellular matrix formation, and inhibits endothelial cell proliferation, migration, and lumen formation. COL4A1 is involved in the growth of hippocampal embryonic neurons and further participates in myelin integrity. Type IV collagen is a major structural component of the glomerular basement membrane (GBM), which forms a chicken-wire sieve together with laminin, proteoglycan, and nidogen / entactin. It contains a C-terminal NC1 domain that inhibits angiogenesis and tumor formation. The C-terminal half is found to have anti-angiogenic activity. Type IV collagen also inhibits endothelial cell proliferation, migration, and lumen formation, as well as inhibits the expression of hypoxia-inducible factor 1α and the activation of ERK1 / 2 and p38 MAPK. COL4A1 mutations are associated with a wide range of phenotypes, including ischemic and hemorrhagic stroke, migraine, leukomalacia, nephropathy, hematuria, chronic muscle spasm, and anterior segment diseases, which include congenital cataract, glaucoma, and Axenfeld-Rieger anomaly. Case Rep Neurol. May - Aug 2015;7(2):142 - 147. Epub Jun 2, 2015. doi:10.1159 / 000431309.

[0167] APLP1 is a 72 kDa biomarker that is involved in the regulation of synapse maturation and neurite growth during cortical development. APLP1 is one of two homologs: amyloid precursor-like protein 1 and amyloid precursor-like protein 2, or APLP1 and APLP2. The gene encoding APLP1 is a member of the highly conserved amyloid precursor protein gene family. The encoded protein is a membrane-associated glycoprotein that is cleaved by secretases in a manner similar to the cleavage of amyloid beta A4 precursor protein. This cleavage releases an intracellular cytoplasmic fragment that can act as a transcriptional activator. APLP1 also plays a role in synapse maturation during cortical development. It can regulate neurite growth by binding to extracellular matrix components such as heparin and collagen I. APLP1 is widely expressed in humans. Functions attributed to APLP1 include neurite growth and synaptogenesis, protein transport along axons, cell adhesion, calcium metabolism, neuronal injury, synaptic dysfunction, and signal transduction.

[0168] MMP2 (72 kDa) and MMP9 (78 - 92 kDa) are gelatinases, proteolytic enzymes involved in the breakdown of the extracellular matrix. MMP2 and MMP9 play roles in physiological processes such as embryonic development, reproduction, and tissue remodeling. Serum MMP-2 and MMP-9 are elevated in different multiple sclerosis subtypes. Avolio, C., et al. Serum MMP-2 and MMP-9 are elevated in different multiple sclerosis subtypes, J. Neuroimmunol. Mar;136(1 - 2):46 - 53. The integrity of the blood - brain barrier, which is the major structural interface between the periphery and the brain, seems to play an important role in MS. Reducing the secretion of proteolytic matrix metalloproteinases (MMPs), such as MMP2 and / or MMP9, which are disruptors of blood - brain barrier integrity, may have profound effects on MS. Proschinger et al. “Influence of combined functional resistance and endurance exercise over 12 weeks on matrix metalloproteinase - 2 serum concentration in persons with relapsing - remitting multiple sclerosis – a community - based randomized controlled trial.” BMC Neurol 19, 314 (2019).

[0169] FLRT2 is a 74 kDa biomarker and is a member of the leucine-rich fibronectin transmembrane protein family, which plays a role in cell adhesion and / or receptor signaling. FLRT2 is expressed in the brain, as well as in the heart and several other organs, and is involved in fibroblast growth factor-mediated signaling cascades. In the heart, it is essential for the normal organization of the cardiac basement membrane during embryogenesis and for normal embryonic epicardium and heart morphogenesis. Neurologically, FLRT2 plays a role in cell-cell adhesion, cell migration, and axon guidance. It may play a role in the migration of cortical neurons during brain development through its interaction with UNC5D. FLRT2 is also involved in glutamate excitotoxicity, neuronal cell death, and synapse formation and plasticity.

[0170] VCAN (a biomarker >200 kDa) is involved in cell motility, cell growth and differentiation, cell adhesion, cell proliferation, cell migration, and angiogenesis. VCAN is further involved in myelin protection, astrocyte excitotoxicity, and is a pro-inflammatory mediator secretion. VCAN becomes a key factor in inflammation by interacting with adhesion molecules on the surface of inflammatory leukocytes and with chemokines involved in recruiting inflammatory cells. In the adult central nervous system, versican is present in the perineuronal net, where it can stabilize synaptic connections. Versican can also inhibit nervous system regeneration and axon growth after central nervous system injury.

[0171] TNFSF13B, also known in this text as B cell-activating factor (BAFF), is a biomarker involved in T cell-independent B cell activation and the formation of ectopic lymphoid follicles.

[0172] CHI3L1 is a 40 kDa biomarker that plays a role in inflammation, the innate immune system, tissue remodeling, and the cell's ability to respond and adapt to changes in its environment. CHIL3L1 further plays a role in type 2 T helper cell (Th2) inflammatory responses and IL-13-induced inflammation, regulation of allergen sensitization, inflammatory cell apoptosis, dendritic cell accumulation, and M2 macrophage differentiation. CHI3L1 promotes the invasion of pathogenic intestinal bacteria into the colonic mucosa and lymphoid organs, activates the AKT1 signaling pathway and subsequent production of IL8 in colonic epithelial cells, and promotes the antibacterial response in the lung by helping to kill macrophage bacteria, control bacterial spread, and enhance host tolerance. CHI3L1 also regulates hyperoxia-induced lung injury, inflammation, and epithelial cell apoptosis.

[0173] IL-12B is a 40 kDa biomarker that represents one subunit of the IL-12 heterodimer. IL-12A (35 kDa) represents the other subunit of the IL-12 heterodimer. IL-12B is involved in innate and adaptive immunity and the regulation of memory / effector Th1 cells. IL-12B is a growth factor that activates T cells and NK cells. IL-12B binds to IL23A to form interleukin-23, a heterodimeric cytokine that functions in innate and adaptive immunity. Polymorphisms in the genes encoding the interleukin-23 receptor (IL23R) and the IL-12 / 23 (IL12B) p40 subunit are associated with the risk of multiple sclerosis (MS). Huang et al., “Meta-analysis of the IL23R and IL12B polymorphisms in multiple sclerosis.” Int. J. of Neuroscience, 126:3, 205-212 (2016).

[0174] IFI30 is a 30-35 kDa biomarker that plays a role in antigen processing by promoting the complete unfolding of proteins destined for lysosomal degradation. IFI30 promotes the generation of MHC class II-restricted epitopes from disulfide-containing antigens through endocytic reduction of disulfide bonds. IFI30 promotes MHC class I-restricted recognition of exogenous disulfide-containing antigens by CD8+ T cells or cross-presentation. IFI30 is constitutively expressed in antigen-presenting cells and is induced by inflammatory cytokines.

[0175] SERPINA9 is a 42 kDa biomarker that is a member of the serpin family of serine protease inhibitors. SERPINA9 is involved in neuronal injury. The expression of SERPINA9 may be restricted to germinal center B cells and lymphoid malignancies. SERPINA9 may function as an efficient inhibitor of trypsin-like proteases in the germinal center in vivo.

[0176] IL18 is involved in immune responses and inflammatory processes. IL18 is a pro-inflammatory cytokine that is mainly involved in the immune responses of polarized T helper 1 (Th1) cells and natural killer (NK) cells. It acts as an inhibitor of the early Th1 cytokine response. It further plays a role in the Th-1 response through its ability to induce IFN-γ production in T cells and NK cells. The levels of IL-18 in cerebrospinal fluid and serum were significantly higher than those in patients without enhancing lesions. The results suggest that IL-18 is involved in the immunopathogenesis of MS, especially during the active phase of the disease. Losy, J., et al. IL-18 in patients with multiple sclerosis. Acta Neurologica Scandinavica, 104:171-173 (2001). Additionally, higher serum levels of IL-18 and significantly different frequencies of two polymorphisms of IL-18 were found in MS patients. Jahanbani-Ardakani, H., et al., Interleukin 18 polymorphisms and its serum level in patients with multiple sclerosis, Ann Indian Acad Neurol; 22:474-76 (2019).

[0177] CDCP1 is a biomarker of 90-140 kDa that is involved in T cell migration, cell adhesion, and cell-matrix binding. CDCP1 can play a role in the regulation of anchorage and migration or proliferation and differentiation through its phosphorylation. CDCP1 is expressed in cells with phenotypes reminiscent of mesenchymal stem cells and neural stem cells. Additionally, CDCP1 is a ligand of CD6, a receptor molecule expressed on certain T cells that can play a role in their migration and chemotaxis.

[0178] CNTN2 is a 113 kDa biomarker that is involved in cell adhesion, proliferation, migration, neuronal axon guidance, neuronal injury, and axon-dendrite rearrangement. CNTN2 is a member of the contactin protein family and is part of the immunoglobulin superfamily of cell adhesion molecules. CNTN2 is a glycosylphosphatidylinositol (GPI)-anchored neuronal membrane protein and plays a role in the proliferation, migration, and axon guidance of developing cerebellar neurons. Mutations in CNTN2 may be associated with adult myoclonic epilepsy. Binding to another transmembrane protein, CNTNAP2, CNTNAP2 contributes to the organization of the axonal domain at the nodes of Ranvier by maintaining voltage-gated potassium channels in the paranodal region.

[0179] GFAP is a 50 kDa biomarker that is involved in demyelination, degeneration, and axonal injury. Astrocyte activation is associated with the activation of the immune cascade and is thought to play a role in the demyelination and axonal injury observed in MS. Glial fibrillary acidic protein (GFAP) is a major component of glial scar formation. GFAP is used as a marker to distinguish astrocytes from other glial cells during development. Higher serum concentrations of GFAP and NEFL are associated with higher EDSS, older age, longer disease duration, progressive disease course, and MRI pathology. H. et al. Serum glialfibrillary acidic protein correlates with multiple sclerosis disease severity. Multiple Sclerosis Journal, 26(13) 2018.

[0180] MOG is a 28 kDa membrane protein that is expressed on the surface of oligodendrocytes and the outermost surface of myelin sheaths. Due to this localization, it acts as a cell surface receptor or cell adhesion molecule and is a major target antigen involved in immune-mediated demyelination. This protein may be involved in the integrity and maintenance of myelin sheaths as well as cell-to-cell communication. Diseases associated with MOG include narcolepsy and rubella. Its related pathways include the neural stem cell differentiation pathway and lineage-specific markers. The paralog of the MOG gene is BTN1A1.

[0181] CD6 is a 90 - 130 kDa biomarker that is involved in the development of the central nervous system. CD6 is a cell adhesion molecule involved in blood-brain barrier disruption and T cell-mediated acute inflammatory responses. Recent studies have identified CD6 as a risk gene for multiple sclerosis (MS), a disease in which autoreactive T cells are fully involved. DeJager, P.L. et al., Meta-analysis of genome scans and replication identify CD6, IRF8 and TNFRSF1A as new multiple sclerosis susceptibility loci. Nat Genet. 2009;41(7):776–782. CD6 is present on the outer membrane of T lymphocytes and is involved in the migration of leukocytes across the blood-brain barrier.

[0182] CXCL9 is a 12 kDa biomarker that is involved in immune responses and inflammatory processes. CXCL9 is a cytokine that affects the growth, motility, or activation state of cells involved in immune and inflammatory responses. CXCL9 (MIG) is a chemokine that elicits chemotactic activity for T cells upon binding to its receptor CXCR3 and is involved in inflammatory responses. CXCL9 is not constitutively expressed but is induced by IFN-γ. CXCL9 has been described as being involved in a variety of inflammation-related diseases, such as hepatitis C, skin inflammation, rheumatoid arthritis, and pharyngitis. Consistent with this observation is the upregulation of the ELR-CXC chemokines CXCL9, CXCL10, and CXCL11, which are upregulated in the CNS of mice affected by EAE induced by Th1 cell transfer. Lovett-Racke, A. et al. Th1 versus Th17: Are T cell cytokines relevant in multiple sclerosis? Biochimca et Biophysica Acta (BBA) – Molecular Basis of Disease. 1812(2):246-251(2011).

[0183] CXCL13 is a biomarker that is involved in cell growth, cell proliferation, regeneration, and inflammatory responses. CXCL13 belongs to the CXC chemokine family and has selective chemotaxis for B cells. It interacts with the chemokine receptor CXCR5 and regulates the organization of B cells through CXCR5. Serum levels of CXCL13 are associated with multiple sclerosis. Festa, E. et al. Serum levels of CXCL13 are elevated in active multiple sclerosis. Multiple Sclerosis Journal, 15(11):1271-1279(2009).

[0184] CCL20 is an 11 kDa biomarker that is involved in axon guidance and the chemotaxis of dendritic cells. CCL20 is a chemokine that is involved in immune regulation and inflammatory processes (e.g., acute inflammatory responses) and is expressed in the epithelial cells of the human brain choroid plexus. It acts as a cognate ligand for CCR6.

[0185] OPG is a biomarker of 55 - 60 kDa that is involved in inflammation, apoptosis, and T cell activation processes. OPG is a decoy receptor for the cytokine TNFSF11 (RANKL) and potentially for TNFSF10 (TRAIL), and belongs to the TNF receptor superfamily. OPG is upregulated by estrogen and increased calcium concentration and plays a role in transcriptional regulation in inflammation, innate immunity, and cell survival and differentiation; for example, the binding of OPG to TNFSF11 inhibits the differentiation of osteoclast precursors into mature osteoclasts, and OPG has been experimentally used in the treatment of osteoporosis. OPG has been described as being involved in several inflammation - related diseases such as rheumatoid arthritis, inflammatory bowel disease, and periodontitis.

[0186] OPN is a biomarker of 33 - 44 kDa that is involved in inflammation and immune regulation. OPN is a pleiotropic integrin - binding protein that plays a role in cell - mediated immunity, inflammation, tissue repair, and cell survival. OPN also plays a role in biomineralization.

[0187] PRTG is a biomarker of 180 kDa that is involved in neurogenesis, neurotrophin binding, neuron survival, and demyelination. It may play a role in anteroposterior axis elongation. PRTG is a membrane protein and is a member of the immunoglobulin superfamily. It is considered to be mainly a developmental protein that has some association with neuralgia, demyelinating diseases, and dyslexia.

[0188] TNFRSF10A is a biomarker of 50 kDa that is a member of the TNF receptor superfamily. TNFRSF10A is involved in inflammation and neurodegenerative processes. This receptor is activated by tumor necrosis factor - related apoptosis - inducing ligand (TNFSF10 / TRAIL), thereby transducing cell death signals and inducing apoptosis.

[0189] GH, also known as growth hormone or somatotropin, is a neuroendocrine biomarker that stimulates growth, cell proliferation, and regeneration in humans and other animals. It regulates energy balance and metabolism. It is a mitogen that is specific only for certain types of cells. Previous studies have shown that it is decreased in the serum of patients with severe MS. Gironi, M., et al. Growth hormone and Disease Severity in Early Stage of Multiple Sclerosis. Multiple Sclerosis International 2013:(2013).

[0190] GH2, also known as growth hormone 2, placental-specific growth hormone, and growth hormone variant, is a biomarker involved in the control of myoblast growth, differentiation, and proliferation. It regulates energy balance and metabolism. It is produced and secreted by the placenta during pregnancy and is the major form of growth hormone during gestation.

[0191] VCAM-1 is an 80 kDa transmembrane biomarker that is normally expressed in blood vessels mediating cell adhesion to vascular endothelium. VCAM-1 is characterized by its multiple immunoglobulin domains. VCAM-1 is associated with multiple sclerosis. Peterson, J. et al., VCAM-1-Positive Microglia Target Oligodendrocytes at the Border of Multiple Sclerosis Lesions, Journal of Neuropathology & Experimental Neurology, Vol. 61, No. 6, June 2002, pp. 539–546. Matsuda, M. et al. Increased levels of soluble vascular cell adhesion molecule-1 (VCAM-1) in the cerebrospinal fluid and sera of patients with multiple sclerosis and human T lymphotropic virus type-1-associated myelopathy. J. Neuroimmunology, 59(1-2):35-40(1995).

[0192] In various embodiments, the biomarker panel can include more additional biomarkers as described herein. In various embodiments, these more additional biomarkers serve as surrogate biomarkers for the aforementioned biomarkers.In various embodiments, additional biomarkers include: cell adhesion molecule 3 (CADM3), kallikrein-related peptidase 6 (KLK6), brevican (BCAN), oligodendrocyte myelin glycoprotein (OMG), CD5 molecule (CD5), cytotoxic and regulatory T cell molecule (CRTAM), CD244 molecule (CD244), tumor necrosis factor receptor superfamily member 9 (TNFRSF9), proteinase 3 (PRTN3), follistatin-like protein 3 (FSTL3), C-X-C motif chemokine ligand 10 (CXCL10), C-X-C motif chemokine ligand 11 (CXCL11), interleukin-18 binding protein (IL-18BP), macrophage scavenger receptor 1 (MSR1), C-C motif chemokine ligand 3 (CCL3), tumor necrosis factor ligand superfamily member 12 (TWEAK), trefoil factor 3 (TFF3), ectonucleotide pyrophosphatase / phosphodiesterase 2 (ENPP2), insulin-like growth factor binding protein 1 (IGFBP-1), interleukin 12A (IL12A), seizure-related 6 homolog-like (SEZ6L), dipeptidyl peptidase-like 6 (DPP6), neurocan (NCAN), tubulointerstitial nephritis antigen-like protein 1 (TINAGL1), calcium-activated nuclease 1 (CANT1), Nectin cell adhesion molecule 2 (NECTIN2), neural cell proliferation and differentiation regulatory protein 1 (NPDC1), tumor necrosis factor receptor superfamily member 11A (TNFRSF11A), contactin 4 (CNTN4), neurotrophic receptor tyrosine kinase 2 (NTRK2), neurotrophic receptor tyrosine kinase 3 (NTRK3), cadherin 6 (CDH6), carcinoembryonic antigen-related cell adhesion molecule 8 (CEACAM8), mitotic arrest deficient 1-like protein 1 (MAD1L1), IgA Fc fragment receptor (FCAR), myeloperoxidase (MPO), osteomodulin (OMD), matrix extracellular phosphoglycoprotein (MEPE), GDNF family receptor alpha 3 (GDNFR-α-3), scavenger receptor class F member 2 (SCARF2), CD40 ligand (IgM), tumor necrosis factor receptor superfamily member 1B (TNF-R2), programmed cell death 1 ligand (PD-L1), Notch 3 (NOTCH3), contactin 1 (CNTN1), oncostatin M (OSM), transforming growth factor alpha (TGF-α), peptidoglycan recognition protein 1 (PGLYRP1), nitric oxide synthase 3 (NOS3), discoidin domain receptor tyrosine kinase 1 (DDR1), C-X-C motif chemokine ligand 16 (CXCL16), CD166 antigen (ALCAM), spondin-2 (SPON2), and protocadherin-17 (PCDH17).

[0193] CADM3 is involved in cell-cell adhesion and interacts with any one of IGSF4, NECTIN1, NECTIN 3, and EPB41L1. CADM3 is involved in the biological processes of adherens junction organization, heterophilic cell-cell adhesion, homophilic cell adhesion, and protein localization.

[0194] KLK6 is a serine protease that exhibits activity against proteins such as α-synuclein, amyloid precursor protein, myelin basic protein, gelatin, casein, and extracellular matrix proteins (such as fibronectin, laminin, vitronectin, and collagen). KLK6 is involved in the biological processes of amyloid precursor protein metabolism, CNS development, myelination, protein autoprocessing, regulation of cell differentiation and neuronal development, wound response, and tissue regeneration.

[0195] BCAN is a proteoglycan and a member of the lectican protein family. BCAN is involved in the biological processes of cell adhesion, CNS development, chondroitin sulfate biosynthesis and catabolism, extracellular matrix organization, and axon synapse maturation.

[0196] OMG is a cell adhesion molecule involved in central nervous system myelination. OMG is involved in the biological processes of cell adhesion, regulation of axonogenesis, and neuronal projection regeneration.

[0197] CD5 is a signal transduction molecule that can be expressed on the surface of cells such as T lymphocytes. CD5 is involved in the biological processes of apoptotic signaling pathway, cell recognition, T cell proliferation, and T cell co-stimulation.

[0198] CRTAM is a transmembrane protein of the immunoglobulin superfamily and is involved in the biological processes of adaptive immune response, cell recognition, detection of stimuli or cells, and regulation of immune response. CRTAM is further involved in heterophilic cell-cell adhesion that regulates the activation, differentiation, and tissue retention of various T cell subsets.

[0199] CD244 is a member of the signaling lymphocytic activation molecule expressed on natural killer cells. CD244 is involved in the biological processes of immune response (adaptive and innate), leukocyte migration, regulation of cytokine secretion, and signal transduction. CD244 regulates the activation and differentiation of multiple immune cells and thus is involved in the regulation and interplay of innate and adaptive immune responses.

[0200] TNFRSF9 is a member of the tumor necrosis factor receptor family and is involved in the biological processes of TNFR signaling pathway, cell proliferation, and apoptosis. TNFRSF9 is expressed by activated T cells.

[0201] PRTN3 is a serine protease expressed by neutrophils. PRTN3 is involved in biological processes such as antibacterial humoral responses, blood coagulation, neutrophil activity, proteolysis, and cytokine-mediated signal transduction pathways.

[0202] FSTL3 is a secreted glycoprotein of the follistatin module protein family. FSTL3 is involved in biological processes such as activin / fibronectin binding, organ development, osteogenesis, ossification, and regulation of intercellular adhesion.

[0203] Both CXCL10 and CXCL11 are cytokines in the CXC chemokine family. CXCL10 and CXCL11 are involved in biological processes such as immune responses, inflammatory responses, cell signaling, chemotaxis, T cell recruitment, and cell proliferation.

[0204] IL-18BP is a protein that acts as an inhibitor of the pro-inflammatory cytokine IL18. IL-18BP is involved in biological processes such as cytokine stimulation, IL-18-mediated signal transduction pathways, and immune responses.

[0205] MSR1 is a membrane glycoprotein expressed by macrophages. MSR1 is involved in endocytosis and biological processes of cholesterol transport and storage, and may be related to the pathological deposition of cholesterol on the arterial wall during the formation of atherosclerosis.

[0206] CCL3 is a monokine with inflammatory and chemokinetic properties that binds to CCR1, CCR4, and CCR5. CCL3 is involved in biological processes such as cell migration (such as lymphocyte and macrophage chemotaxis), calcium-mediated signal transduction, intercellular signal transduction, cytokine secretion, and inflammatory responses.

[0207] TWEAK is a cytokine in the tumor necrosis factor ligand family. TWEAK is involved in biological processes such as angiogenesis, cell differentiation, immune responses, signal transduction, apoptosis, and TNF-mediated signal transduction pathways. TWEAK further promotes the proliferation and migration of endothelial cells.

[0208] TFF3 is a 6 kDa glycoprotein that is usually produced by goblet cells and is involved in the gastrointestinal tract. TFF3 is involved in biological processes such as the maintenance and healing of the gastrointestinal epithelium and the regulation of glucose metabolism.

[0209] ENPP2 is a phosphodiesterase involved in the generation of the lipid signaling molecule lysophosphatidic acid. ENPP2 hydrolyzes lysophospholipids and is involved in biological processes such as cell movement, chemotaxis, immune responses, and angiogenesis.

[0210] IGFBP-1 is a member of the insulin-like growth factor binding protein family. It binds insulin-like growth factor (IGF) I and II. IGFBP-1 is involved in biological processes such as aging, cellular metabolic processes, cell growth, signal transduction, and tissue regeneration.

[0211] IL-12A is a subunit that forms the IL-12 heterodimer together with other IL-12B subunits. IL-12A is involved in biological processes such as cell migration, cell proliferation, cell adhesion, cell differentiation, and regulation of the activation of immune cells (e.g., T cells, dendritic cells, natural killer cells).

[0212] SEZ6L is a protein mainly located in the endoplasmic reticulum membrane, and it regulates the endoplasmic reticulum function in neurons. SEZ6L is involved in biological processes such as synaptic maturation, adult motor behavior, and regulation of protein kinase C signaling.

[0213] DPP6 is a membrane protein that is a member of the serine protease peptidase S9B family. DPP6 is involved in biological processes such as regulation of potassium ion channels and protein localization to the plasma membrane, and it can affect the susceptibility to amyotrophic lateral sclerosis.

[0214] NCAN is a protein that is a member of the lectican / chondroitin sulfate proteoglycan family. NCAN is involved in biological processes such as cell adhesion, CNS development, organization of the ECM, and synthesis of chondroitin and dermatan sulfate. NCAN is involved in neuronal adhesion and neurite outgrowth during development by binding to neural cell adhesion molecules.

[0215] TINAGL1 is an extracellular matrix protein involved in biological processes such as cell adhesion, proliferation, migration, and differentiation. TINAGL1 further plays a role in endocytosis and endosomal trafficking.

[0216] CANT1 is a UDP-preferring calcium-dependent nucleotidase. CANT1 is involved in biological processes such as regulation of calcium ion binding, neutrophil degranulation, regulation of NF-κB signaling, and proteoglycan biosynthesis. CANT1 regulates metabolic processes such as nucleotide metabolism.

[0217] NECTIN2 is a membrane glycoprotein that serves as a component of adherens junctions. NECTIN2 is involved in biological processes such as intercellular adhesion (through tissue adherens junctions), organization of the tissue cytoskeleton, viral receptor activity, and regulation of NK cell and T cell activities.

[0218] NPDC1 is a protein mainly expressed in the brain. NPDC1 is involved in biological processes such as regulation of immune responses and neural cell development and proliferation. NPDC1 inhibits oncogenic transformation in neural and non-neural cells and downregulates neural cell proliferation.

[0219] TNFRSF11A is a member of the TNF receptor superfamily. TNFRSF11A is involved in biological processes such as intercellular signaling, immune response, monocyte chemotaxis, and TNF-mediated signaling pathways. In addition, TNFSF11A plays a role in osteoclastogenesis.

[0220] CNTN4 is a member of the immunoglobulin contactin family. CNTN4 is involved in biological processes such as axon guidance and development, synaptogenesis, cell surface interactions during nervous system development, brain development, neuronal cell adhesion, neuronal projection, neuronal differentiation, and regulation of synaptic plasticity.

[0221] NTKR2 is a receptor tyrosine kinase that binds to brain-derived neurotrophic factor (part of the neurotrophic tyrosine receptor kinase family). NTKR2 is involved in biological processes such as neuronal survival, proliferation, migration, differentiation, synapse formation, and synaptic plasticity.

[0222] NTKR3 is a receptor tyrosine kinase that binds to neurotrophin-3 (part of the neurotrophic tyrosine receptor kinase family). NTKR3 is involved in biological processes such as regulation of GTPase and MAPK activity, regulation of astrocyte differentiation, nervous system development, and neuronal migration.

[0223] CDH6 is a member of the cadherin superfamily that mediates intercellular adhesion. CDH6 is involved in biological processes such as intercellular adhesion (adherens junction organization), cell morphogenesis, and the Notch signaling pathway. CDH6 mediates heterotypic cell-cell contacts through its interaction with CD6, and also mediates homotypic cell-cell contacts. CDH6 is further involved in axon extension and axon guidance.

[0224] CEACAM8 is a cell surface glycoprotein belonging to the carcinoembryonic antigen (CEA) superfamily. CEACAM8 is involved in biological processes such as regulation of immune response, leukocyte migration, neutrophil degranulation, and cell-cell adhesion.

[0225] MAD1L1 is a mitotic spindle assembly checkpoint protein. MAD1L1 is involved in biological processes such as cell division and the mitotic cell cycle checkpoint.

[0226] FCAR is a transmembrane glycoprotein on the surface of immune cells such as neutrophils, monocytes, and macrophages. FCAR is involved in biological processes such as regulation of immune response, neutrophil activation / degranulation, and response to cytokines (e.g., interferon, interleukin, TNF).

[0227] MPO is a heme protein (enzyme) expressed in neutrophils. MPO is involved in immune responses, neutrophil degranulation, and biological processes related to chromatin / heme / heparin binding.

[0228] OMD is thought to be involved in the process of biomineralization and binding to osteoblasts. OMD participates in biological processes of cell adhesion and regulation of bone mineralization.

[0229] MEPE is a calcium-binding phosphoprotein in the small integrin-binding ligand N-linked glycoprotein (SIBLING) family. MEPE is involved in biological processes of extracellular matrix binding / regulation, biomineral tissue development, skeletal system development, and bone and cartilage mineralization. MEPE is involved in renal phosphate excretion and inhibits intestinal phosphate absorption. MEPE is further involved in the proliferation and differentiation of dental pulp stem cells.

[0230] GDNFR-α-3 is a glial cell line-derived neurotrophic factor and a member of the GDNF receptor family, and binds to artemin (ARTN). GDNFR-α-3 is involved in biological processes of axon guidance, nervous system development, neuronal migration, GDNF receptor activity, and signal transduction receptor activity and binding.

[0231] SCARF2 is a member of the scavenger receptor class F family. SCARF2 is an adhesion protein and is involved in biological processes of scavenger receptor activity and cell-cell adhesion.

[0232] CD40 ligand is mainly expressed on activated T cells and is a member of the TNF molecular superfamily. CD40 ligand is involved in biological processes of B cell differentiation and proliferation, inflammatory responses, leukocyte cell-cell adhesion, platelet activation, T cell co-stimulation, and TNF-mediated signal transduction pathways.

[0233] TNF-R2 is a membrane receptor that binds tumor necrosis factor-α. TNF-R2 is involved in biological processes of TNF-mediated signal transduction pathways, neutrophil degranulation, regulation of neuroinflammatory responses, and cell signaling. TNF-R2 protects neurons from apoptosis by stimulating the antioxidant pathway.

[0234] PD-L1 is a ligand that binds to PD-1 and is an important target in cancer immunotherapy checkpoint inhibitor research. PD-L1 is involved in biological processes of immune responses, interferon regulation, T cell proliferation, T cell co-stimulation, cell migration, and cytokine production.

[0235] NOTCH3 is a protein in the NOTCH receptor family that is involved in the Notch signaling pathway. NOTCH3 participates in the biological processes of gene activation, calcium ion binding, signal transduction receptor activity, neuronal fate commitment, and brain development. NOTCH3 further regulates cell fate determination.

[0236] CNTN1 is a neuronal membrane protein that plays a role in cell adhesion. CNTN1 participates in the biological processes of neuronal projection development, brain development, cell adhesion, and Notch signaling.

[0237] OSM is a cytokine in the interleukin-6 cytokine family. OSM participates in the biological processes of regulating immune responses, cell proliferation / division, inflammatory responses, and cytokine activities (e.g., MAPK and STAT pathways).

[0238] TGFA is a mitogenic polypeptide and is part of the epidermal growth factor family. TGFA participates in the biological processes of growth factor activity, signal transduction pathways (e.g., MAPK, EGF), cell division / proliferation, and signal transduction.

[0239] PGLYRP1 is a peptidoglycan-binding protein and participates in the biological processes of innate immune responses, inflammatory responses, neutrophil degranulation, and peptidoglycan immune receptor activity.

[0240] NOS3 regulates the production of nitric oxide. NOS3 participates in the biological processes of angiogenesis, vascular remodeling, endothelial cell migration, vasodilation, vascular smooth muscle relaxation, and promoting blood clotting by activating platelets.

[0241] DDR1 regulates cell attachment to the extracellular matrix, remodeling of the extracellular matrix, cell migration, differentiation, and survival and cell proliferation. DDR1 further promotes smooth muscle cell migration.

[0242] CXCL16 plays a role in immune regulation and acts as a scavenger receptor on macrophages.

[0243] IL6 is a cytokine that is involved in the differentiation of B cells, lymphocytes, and monocytes.

[0244] ALCAM participates in axon guidance, embryonic and induced pluripotent stem cell differentiation pathways and lineage-specific markers, and L1CAM interactions.

[0245] NTRK2 participates in the development and maturation of the central and peripheral nervous systems by regulating neuronal survival, proliferation, migration, differentiation, and synapse formation and plasticity.

[0246] SPON2 is involved in the growth of hippocampal embryonic neurons.

[0247] NTRK3 is involved in regulating the development of the nervous system and the heart.

[0248] PCDH17 is involved in the establishment and function of specific intercellular connections in the brain.

[0249] VI. Assay

[0250] As Figure 1A shown, the system environment 100 involves implementing a biomarker quantification assay 120 for evaluating the expression levels of one or more biomarkers. Examples of assays for one or more biomarkers (e.g., biomarker quantification assay 120) include DNA assays, microarrays, polymerase chain reaction (PCR), RT-PCR, Southern blotting, Northern blotting, antibody binding assays, enzyme-linked immunosorbent assay (ELISA), flow cytometry, protein assays, Western blotting, turbidimetry, nephelometry, chromatography, mass spectrometry, immunoassays, for example, including but not limited to RIA, immunofluorescence, immunochemiluminescence, immunoelectrochemiluminescence or competitive immunoassays, immunoprecipitation, and the assays described in the Examples section below. Information from the assays can be quantitative and sent to the computer system of the present invention. Such information can also be qualitative, observing patterns or fluorescence, which can be automatically converted to quantitative measurements by the user or by a reader or computer system.

[0251] Various immunoassays (including multiplex assays) designed to quantify biomarkers can be used for screening. Measuring the concentration of a target biomarker in a sample or its fraction can be accomplished by a variety of specific assays. For example, traditional sandwich assays can be used in the form of arrays, ELISA, RIA, etc. Other immunoassays include Ouchterlony plates that provide a simple assay for antibody binding. Additionally, detection systems specific for the biomarker can be used as needed, and labeling methods can be conveniently used for Western blotting on protein gels or protein spots on filters.

[0252] Protein-based assays using antibodies that specifically bind to a polypeptide (e.g., a biomarker) can be used to quantify the level of the biomarker in a test sample obtained from a subject. In various embodiments, the antibody that binds to the biomarker can be a monoclonal antibody. In various embodiments, the antibody that binds to the biomarker can be a polyclonal antibody. For multiplex assays of biomarkers, an array containing one or more biomarker affinity reagents (e.g., antibodies) can be generated. Such an array can be constructed to contain antibodies against the biomarkers. Detection can utilize one or a set of biomarker affinity reagents, such as a set or mixture of affinity reagents specific for one, two, three, four, five, six, seven, eight, nine, ten, eleven, twelve, thirteen, fourteen, fifteen, sixteen, seventeen, eighteen, nineteen, twenty, twenty-one or more biomarkers.

[0253] In various embodiments, multiplex assays involve the use of oligonucleotide-labeled antibody probes that bind to target biomarkers and allow subsequent quantification of the biomarkers. An example of a multiplex assay involving oligonucleotide-labeled antibody probes is the proximity extension assay (PEA) technology (Olink Proteomics). Briefly, a pair of oligonucleotide-labeled antibodies bind to a biomarker, where the two oligonucleotide sequences are complementary to each other. Thus, the oligonucleotide sequences will hybridize only when both antibodies are bound to the target biomarker. Mismatched oligonucleotide sequences (occurring due to non-specific binding of the antibodies or cross-reactivity of the antibodies) will not hybridize and thus will not result in a readout. The hybridized oligonucleotide sequences are subjected to nucleic acid extension and amplification and then quantified using microfluidic qPCR. The level of quantification is related to the quantitative expression value of each biomarker.

[0254] In various embodiments, multiplex assays involve the use of bead-conjugated antibodies (e.g., capture antibodies) that can bind to and detect biomarkers. An example of a multiplex assay involving bead-conjugated antibodies is the technology of Luminex. Here, the bead-conjugated antibody is added to the sample together with a biotinylated detection antibody. Both antibodies are specific for the biomarker of interest and thus form an antibody-antigen sandwich. Streptavidin is further added, which binds to the biotinylated detection antibody and enables detection of the complex. The Luminex 200 TM or analyzer is used to identify and quantify the number of biomarkers in the sample. In various embodiments, the multiplex assay represents an improvement over the technology of Luminex, such as the multi-analyte profiling (MAP) technology of Myriad Rules Based Medicine (RBM).

[0255] In various embodiments, a sample obtained from a subject may be processed prior to performing a biomarker quantification assay 120 (e.g., an immunoassay). In various embodiments, the sample is processed such that performance of the biomarker quantification assay 120 enables more accurate assessment of the expression level of one or more biomarkers in the sample.

[0256] In various embodiments, a sample from a subject may be processed to extract a biomarker from the sample. In one embodiment, the sample may be phase separated to separate the biomarker from other portions of the sample. For example, the sample may be centrifuged (e.g., pelleted or density gradient centrifuged) to separate larger and / or denser entities (e.g., cells and other macromolecules) in the sample from the biomarker. Other examples include filtration (e.g., ultrafiltration) to separate the biomarker from other portions of the sample.

[0257] In various embodiments, a sample from a subject may be processed to produce a subsample containing a biomarker fraction present in the sample. In various embodiments, generating a biomarker fraction may involve performing a protein fractionation procedure. An example of a protein fractionation procedure includes chromatography (e.g., gel filtration, ion exchange, hydrophobic chromatography, or affinity chromatography). In certain embodiments, the protein fractionation procedure involves affinity purification or immunoprecipitation, where the biomarker binds to a specific antibody. Such antibodies may be immobilized on a support, such as magnetic particles or nanoparticles or a plate.

[0258] In various embodiments, a sample from a subject is processed to extract a biomarker from the sample and the sample is further processed to produce a subsample containing the extracted biomarker fraction. Overall, this enables a purified biomarker subsample of the biomarker of particular interest. Thus, performing an assay (e.g., an immunoassay) for assessing the expression level of the biomarker of particular interest can be more accurate and of higher quality. In various embodiments, a particular biomarker may be a biomarker in a biomarker panel, embodiments of which are described herein. As an example, the biomarkers in a biomarker panel may include two or more of the following: NEFL, MOG, CD6, CXCL9, OPG, OPN, CXCL13, GFAP, CDCP1, CCL20, IL-12B, APLP1, TNFRSF10A, COL4A1, SERPINA9, FLRT2, TNFSF13B, GH, VCAN, PRTG, and CNTN2.

[0259] VII. Therapeutic Agents and Combinations of Therapeutic Agents

[0260] In various embodiments, a therapeutic agent is provided to an individual before and / or after obtaining a sample from the individual and determining a quantitative expression value of one or more markers in the obtained sample. As an example, a predictive model that receives the quantitative expression value predicts that the individual will be diagnosed with multiple sclerosis and a therapeutic agent will be provided. In another example, the predictive model predicts that the provided therapeutic agent is showing therapeutic efficacy against multiple sclerosis in previously diagnosed individuals.

[0261] In various embodiments, the therapeutic agent is a biological agent, such as a cytokine, an antibody, a soluble cytokine receptor, an antisense oligonucleotide, siRNA, etc. Such biological agents encompass mutant proteins and derivatives of biological agents, which may include, for example, fusion proteins, polyethylene glycolated derivatives, cholesterol-conjugated derivatives, etc., known in the art. Also included are antagonists of cytokines and cytokine receptors, such as traps and monoclonal antagonists, such as IL-1Ra, IL-1Trap, sIL-4Ra, etc. Also included are biosimilars or bioequivalent drugs of the active agents described herein.

[0262] Therapeutic agents for multiple sclerosis include corticosteroids, plasma exchange, ocrelizumab IFN-β Glatiramer acetate Anti-VLA4 (Tysabri, natalizumab), dimethyl fumarate Teriflunomide Monomethyl fumarate (Bafiertam TM )、Ozanimod Siponimod Fingolimod Anti-CD52 antibody (e.g., alemtuzumab Mitoxantrone Methotrexate, cladribine Simvastatin and cyclophosphamide. In addition to or in place of the therapeutic agent, other treatments for multiple sclerosis include lifestyle changes, such as physical therapy or dietary changes. The method also provides combination therapy and / or additional treatments with one or more therapeutic agents, wherein the combination may provide additional or synergistic benefits.

[0263] A pharmaceutical composition administered to an individual comprises an active agent, such as the aforementioned therapeutic agent. The active ingredient is present in a therapeutically effective amount, i.e., an amount sufficient to treat the disease or medical condition mediated thereby upon administration. The composition may also contain various other agents to enhance delivery and efficacy, for example, to enhance the delivery and stability of the active ingredient. Thus, for example, depending on the desired formulation, the composition may also contain a pharmaceutically acceptable non-toxic carrier or diluent, which are defined as the vehicles commonly used to formulate pharmaceutical compositions for administration to animals or humans. The diluent is selected so as not to affect the biological activity of the combination. Examples of such diluents are distilled water, buffered water, saline, PBS, Ringer's solution, dextrose solution, and Hank's solution. Additionally, the pharmaceutical composition or formulation may contain other carriers, adjuvants, or non-toxic non-therapeutic, non-immunogenic stabilizers, excipients, etc. The composition may also contain additional substances close to physiological conditions, such as pH adjusters and buffers, toxicity adjusters, wetting agents, and detergents. The composition may also contain any of a variety of stabilizers, such as antioxidants.

[0264] The pharmaceutical compositions described herein can be administered in a variety of different ways. Examples include administering a composition containing a pharmaceutically acceptable carrier by oral, intranasal, rectal, topical, intraperitoneal, intravenous, intramuscular, subcutaneous, subdermal, transdermal, intracapsular, or intracranial methods.

[0265] Such pharmaceutical compositions can be administered for prophylactic (e.g., before diagnosing a patient with multiple sclerosis) or therapeutic (e.g., after diagnosing a patient with multiple sclerosis) purposes. As used in the context of the present invention, prevention of a disease or disorder (Preventing / prophylaxis / prevention) refers to administering a composition to prevent the occurrence or onset of multiple sclerosis or some or all of the symptoms of multiple sclerosis, or to reduce the likelihood of the onset of the disease or disorder. Treatment (treating / treatment) or therapy of multiple sclerosis refers to slowing, halting, or reversing the progression of the disease by administering a treatment according to the present invention. In a preferred embodiment, treating multiple sclerosis means reversing the progression of the disease, ideally to the extent of eliminating the disease itself.

[0266] VIII. Disease Activity of the Subject

[0267] The methods described herein focus on assessing the disease activity of a subject by using the quantitative expression level of a biomarker as an input to a prediction model. In various embodiments, a subject is classified into a category based on a predictive assessment of disease activity. To classify a subject, the prediction for the subject can be compared to the outcomes of individuals who have previously been classified into clinically diagnosed categories. For example, an individual can be clinically classified based on one of the following: an MS diagnosis (e.g., the presence of MS), a classification of being in a quiescent or worsening state, a classification of the level of disability according to the Expanded Disability Status Scale (EDSS), an identified clinical response to a therapy, and a clinical identification of the risk of developing MS. Any one of the Multiple Sclerosis Functional Composite (MSFC), Timed 25-Foot Walk (T25Fw), Nine-Hole Peg Test (9HPT), or Patient-Derived Disability Status Scale (PDDS / MSSS) can also be used to determine the clinical category. An individual can be clinically classified based on a measurable value of MS disease activity, such as a specific number of gadolinium-enhanced lesions (e.g., mild disease activity) or the presence of at least one gadolinium-enhanced lesion (e.g., moderate disease activity). Clinical classification can also be based on other radiological measurements, including T2 lesions (new or enlarging), slowly expanding lesions, marginally expanding lesions, brain parenchymal fraction (BPF) and percent change, gray matter fraction, white matter fraction, thalamic volume, cortical gray matter volume, deep gray matter volume, or ancillary features (e.g., Dawson's fingers) noted by a radiologist. A previous individual can be classified according to clinical criteria.

[0268] The clinical diagnosis of MS can be made by a variety of methods. As an example, the clinical diagnosis of MS can be made by identifying lesions or plaques caused by MS through magnetic resonance imaging (MRI) of the brain and spinal cord. The McDonald criteria can be used to make the diagnosis. The clinical diagnosis of MS can also be made by lumbar puncture (spinal tap), which observes an abnormality in the concentration of antibodies in the spinal fluid due to the presence of MS. The clinical diagnosis of MS can also be made by evoked potential testing, in which electrical signals generated by neurons in the nervous system in response to a stimulus are recorded. Impaired transmission indicates the presence of MS.

[0269] The clinical classification of MS patients previously diagnosed as being in a quiescent or worsening state can depend on a variety of factors. That is, a patient can be clinically classified as being in a worsening state after the emergence of a new disease associated with MS (e.g., a comorbidity or symptom such as clinical depression or optic neuritis). As another example, if a patient experiences a significant worsening of symptoms, the patient is clinically classified as being in a worsening state. Examples can include a worsening of balance and / or mobility, vision, eye pain, fatigue, and / or heart-related problems. If a patient does not present a new disease or a change or worsening of symptoms, a patient previously diagnosed with MS can be clinically classified as being in a quiescent state.

[0270] Determining whether a patient previously diagnosed with MS responds to a therapy can depend on a variety of clinical variables. For example, response to a therapy can be determined based on the occurrence or non-occurrence of relapses. If no relapse occurs, the patient can be considered to respond to the therapy. Response to a therapy can also be determined based on the total number of relapses, the time to the first relapse, the patient's EDSS score, the change in the patient's EDSS score (e.g., an increase in score corresponds to a lack of response to the therapy), the change in MRI status (e.g., the appearance of additional lesions or plaques corresponds to a lack of response to the therapy).

[0271] Patients can be clinically classified by disability level, which can serve as a measure of disease progression. For example, the EDSS can be used to determine the severity of a patient's MS. Thus, patients are classified into categories corresponding to EDSS scores at intervals of 0.5 between 1.0 and 10.0. Generally, an EDSS score of 1.0 to 4.5 refers to MS patients who are able to walk without any assistance. An EDSS score of 5.0 to 9.5 refers to MS patients with impaired walking ability, where a higher score indicates a higher degree of impairment.

[0272] IX. Computer-Implemented Manner

[0273] The method of the present invention, including a method for assessing multiple sclerosis activity in an individual, is, in some embodiments, carried out on one or more computers.

[0274] For example, the construction and deployment of predictive models and database memories can be implemented in hardware or software or a combination of both. In one embodiment of the present invention, a machine-readable storage medium is provided, the medium containing data storage material encoded with machine-readable data, which, when used with a machine programmed with instructions regarding the use of the data, can display any data set, execution, and results of the predictive model of the present invention. Such data can be used for a variety of purposes, such as patient monitoring, treatment considerations, etc. The present invention can be implemented in a computer program executable on a programmable computer, the computer including a processor, a data storage system (including volatile and non-volatile memory and / or storage elements), a graphics adapter, a pointing device, a network adapter, at least one input device, and at least one output device. A display is coupled to the graphics adapter. The program code is applied to the input data to perform the above functions and generate output information. The output information is applied to one or more output devices in a known manner. The computer can be, for example, a personal computer, a microcomputer, or a workstation of traditional design.

[0275] Each program can be implemented in a high-level program or an object-oriented programming language to communicate with a computer system. However, if desired, the program can be implemented in assembly language or machine language. In any case, the language can be a compiled language or an interpreted language. Each such computer program is preferably stored on a storage medium or device readable by a general or special purpose programmable computer (e.g., ROM or magnetic floppy disk) for configuring and operating the computer when the computer reads the storage medium or device, thereby executing the program described herein. The system can also be considered to be implemented as a computer-readable storage medium configured with a computer program, where such a configured storage medium causes the computer to operate in a specific and predefined manner to perform the functions described herein.

[0276] Signature patterns and their databases can be provided in a variety of media to facilitate their use. "Media" refers to the manufacturer that contains the signature pattern information of the present invention. The database of the present invention can be recorded on a computer-readable medium, such as any medium directly readable and accessible by a computer. Such media include, but are not limited to: magnetic storage media, such as floppy disks, hard disk storage media, and magnetic tapes; optical storage media, such as CD-ROMs; electrical storage media, such as RAM and ROM; and mixtures of these categories, such as magnetic / optical storage media. Those skilled in the art can readily understand how to use any currently known computer-readable medium to create a manufacture containing a record of the information in this database. "Recording" refers to the process of storing information on a computer-readable medium using any such method known in the art. Depending on the means for accessing the stored information, any convenient data storage structure can be selected. Various data processor programs and formats can be used for storage, such as word processing text files, database formats, etc.

[0277] In some embodiments, the methods of the invention, including methods for assessing multiple sclerosis activity in an individual, are performed on one or more computers in a distributed computing system environment (e.g., a cloud computing environment). In this specification, "cloud computing" is defined as a model for enabling on-demand network access to a shared pool of configurable computing resources. Cloud computing can be used to provide on-demand access to the shared pool of configurable computing resources. The shared pool of configurable computing resources can be rapidly provisioned via virtualization and released with a low management effort or service provider interaction, and then scaled accordingly. The cloud computing model can consist of various characteristics such as on-demand self-service, broad network access, resource pooling, rapid elasticity, measurable service, etc. The cloud computing model can also expose various service models such as software as a service ("SaaS"), platform as a service ("PaaS"), and infrastructure as a service ("IaaS"). The cloud computing model can also be deployed using different deployment models (such as private cloud, community cloud, public cloud, hybrid cloud, etc.). In this specification and the claims, a "cloud computing environment" is an environment that employs cloud computing.

[0278] VIII.A. Exemplary Computer

[0279] Figure 8 An exemplary computer 800 for implementing the entities shown in FIGS. 1 and 3 is shown. Computer 800 includes at least one processor 802 coupled to a chipset 804. Chipset 804 includes a memory controller hub 820 and an input / output (I / O) controller hub 822. Memory 806 and a graphics adapter 812 are coupled to the memory controller hub 820, and a display 818 is coupled to the graphics adapter 812. A storage device 808, an input device 814, and a network adapter 816 are coupled to the I / O controller hub 822. Other embodiments of computer 800 have different architectures.

[0280] The storage device 808 is a non-transitory computer-readable storage medium such as a hard disk drive, a compact disc read-only memory (CD-ROM), a DVD, or a solid-state memory device. Memory 806 holds instructions and data used by the processor 802. The input interface 814 is a touchscreen interface, a mouse, a trackball, or other type of pointing device, a keyboard, or some combination thereof, and is used to input data into computer 800. In some embodiments, computer 800 can be configured to receive input (e.g., commands) from the input interface 814 via a user's gestural actions. The graphics adapter 812 displays images and other information on the display 818. The network adapter 816 couples computer 800 to one or more computer networks.

[0281] The computer 800 is adapted to execute computer program modules to provide the functions described herein. As used herein, the term "module" refers to computer program logic for providing a specified function. Thus, a module can be implemented in hardware, firmware, and / or software. In one embodiment, the program modules are stored on the storage device 808, loaded into the memory 806, and executed by the processor 802.

[0282] The type of computer 800 used by the entities of FIG. 1 can vary according to the embodiment and the processing capabilities required by the entity. For example, the activity prediction system 130 can run on a single computer 800 or on multiple computers 800 that communicate with each other over a network, such as in a server farm. The computer 800 can lack some of the components described above, such as the graphics adapter 812 and the display 818.

[0283] X. Kit-Implemented Manner

[0284] Also disclosed herein are kits for assessing disease activity (e.g., multiple sclerosis activity) in an individual. Such kits can include reagents for detecting the expression level of one or more biomarkers and instructions for assessing disease activity based on the detected expression levels.

[0285] The detection reagents can be provided as part of a kit. Thus, the present invention also provides a kit for detecting the presence of a panel of biomarkers of interest in a biological test sample. The kit can include a set of reagents for generating a data set by performing at least one protein detection assay (e.g., immunoassay) on a test sample from a subject. In various embodiments, the set of reagents is capable of detecting the quantitative expression level of a biomarker from any of Tables 8 to 10. In a particular embodiment, the set of reagents is capable of detecting the quantitative expression level of a biomarker classified as a biomarker in Layer A, Layer B, or Layer C in Table 1. In a particular embodiment, the set of reagents is capable of detecting the quantitative expression level of a biomarker classified as a biomarker in Tier 1, Tier 2, or Tier 3 in Table 2. In some aspects, the reagents include one or more antibodies that bind to one or more markers. The antibodies can be monoclonal antibodies or polyclonal antibodies. In some aspects, the reagents can include reagents for performing an ELISA, including buffers and detection agents.

[0286] The kit may include instructions for use of a set of reagents. For example, the kit may include instructions for performing at least one biomarker detection assay, such as immunoassays, protein-binding assays, antibody-based assays, antigen-binding protein-based assays, protein-based arrays, enzyme-linked immunosorbent assay (ELISA), flow cytometry, protein arrays, blotting, Western blotting, turbidimetry, nephelometry, chromatography, mass spectrometry, enzyme activity, proximity extension assay, and immunoassay methods selected from RIA, immunofluorescence, immunochemiluminescence, immuno-electrochemiluminescence, immunoelectrophoresis, competitive immunoassay, and immunoprecipitation.

[0287] In various embodiments, the kit includes instructions for practicing the methods disclosed herein (e.g., methods for training or deploying a predictive model to predict disease activity assessment). These instructions can exist in a variety of forms in the subject kit, and one or more of them can be present in the kit. One form in which these instructions can exist is as printed information on a suitable medium or substrate (e.g., one or more sheets of paper on which information is printed), in the packaging of the kit, in a package insert, etc. Another means is a computer-readable medium on which information is recorded, such as a floppy disk, CD, hard drive, network data storage, etc. Another means that can exist is a website address, which can be accessed via the Internet to obtain information at the transfer site. Any convenient means can be present in the kit.

[0288] XI. System

[0289] Also disclosed herein is a system for analyzing the quantitative expression level of a biomarker to assess disease activity. In various embodiments, such a system may include a set of reagents for detecting the expression level of a biomarker in a biomarker panel; a device configured to receive a mixture of the set of reagents and a test sample obtained from a subject to measure the expression level of a soluble medium; and a computer system communicatively coupled to the device to obtain the measured expression level and implement the predictive model to assess disease activity.

[0290] The set of reagents is capable of detecting the quantitative expression level of a biomarker in a biomarker panel. In various embodiments, the set of reagents relates to reagents for performing assays such as the assays described above or immunoassays. For example, the reagents include one or more antibodies that bind to one or more biomarkers. The antibodies can be monoclonal antibodies or polyclonal antibodies. As another example, the reagents may include reagents for performing ELISA, including buffers and detection agents.

[0291] The device is configured to detect the expression level of a biomarker in a mixture of a reagent and a test sample. For example, the device can determine the quantitative expression level of the biomarker through immunoassay or nucleic acid detection assays. The mixture of the reagent and the test sample can be provided to the device through various conduits, examples of which include wells of a microplate (e.g., 96-well microplate), vials, tubes, and integrated fluid circuits. Accordingly, the device can have an opening (e.g., slot, chamber, opening, sliding tray) that can receive a container containing the reagent-test sample mixture and perform a reading to generate a quantitative expression value of the biomarker. Examples of the device include plate readers (e.g., luminescence plate reader, absorbance plate reader, fluorescence plate reader), spectrometers, and spectrophotometers.

[0292] The computer system, such as Figure 8 the exemplary computer 800 described in, communicates with the device to receive the quantitative expression value of the biomarker. The computer system implements a prediction model through computer simulation to analyze the quantitative expression value of the biomarker, thereby predicting the assessment of disease activity.

[0293] Examples

[0294] The following are examples for carrying out specific embodiments of the present invention. The examples are provided for illustrative purposes only and are not intended to limit the scope of the present invention in any way. Efforts have been made to ensure the accuracy of the numbers used (e.g., amounts, temperatures, etc.), but some experimental errors and deviations should be allowed.

[0295] Example 1: Human Clinical Study

[0296] The development of this biomarker panel used multiple human clinical studies, including: ACP = Accelerated Cure Project (study code: F2), CLIMB = Comprehensive Longitudinal Investigation of Multiple Sclerosis at Brigham and Women's Hospital (study codes: F3A, F3B, F3C, and F4), EPIC = Expression, Proteomics, Imaging, Clinical studies at UCSF (study code: F5), and UHBC = University Hospital Basel Cohort (study code: F6). ACP (n = 124) focused on clinically defined worsening events, while AIM1 (n = 60) and unpaired F4 (n = 326) focused on the annualized relapse rate (ARR). Unpaired AIM3 (n = 58), unpaired F4 (n = 326), and F5 EPIC (n = 180) focused on the cross-sectional perspective of gadolinium (Gd)-enhanced MRI lesion endpoints, while paired AIM3 (n = 58), paired F4 (n = 196), and F6 (n = 205) were also the same by longitudinal analysis (e.g., considering patients to establish baseline normality). Additionally, in some cases, samples from different studies with the same endpoint were combined for biomarker analysis. For example, study code F4 and study code F5, which had the presence or absence of gadolinium-enhanced lesions as a common endpoint, were combined for biomarker analysis. For each of the three major categories, each sample was given equal weight in 7 study paradigms (i.e., studies with more samples had proportionally greater weight). The study codes and other information used for analysis are recorded in Tables 8A and 8B below. More than 1,300 proteins in more than 1,000 individual samples were screened using 2 immunoassay platforms (Rules Based Medicine (RBM) and Olink biomarker panel). Multiple endpoints were studied, including the presence / absence of gadolinium-enhanced lesions, clinical relapse status, EDSS, annualized relapse rate, and T2 volume.

[0297] Example 2: Univariate Analysis

[0298] Three different statistical measures were calculated for univariate analysis of individual biomarkers.

[0299] 1) Univariate population - p-values from standard statistical tests

[0300] · The p-values were converted back to their t-statistics using the traditional inverse normal function and then the statistics were combined using the Stauffer method according to their respective sample sizes. The final p-value / test statistic reflecting cumulative power was then normalized to the range of [0,1].

[0301] 2) Univariate separation - AUC value of the integral of true positive rate and false positive rate on the ROC curve

[0302] · Calculate the AUC for each individual biomarker on the selected dataset. Then normalize them to the range of [0,1], and then convert the cumulative separation ability into a single value between 0 and 1 using their weighted sum (based on the sample size below).

[0303] 3) Univariate regression - the adjusted r-squared value of OLS between the lesion burden (trimming the higher number of lesions (e.g., 5 lesions) to exclude outliers) and the distribution of the normalized protein expression values.

[0304] · This was done independently for each cohort in addition to the mixed cohort divided by training / testing.

[0305] Example 3: Multivariate Analysis

[0306] The following were done for the multivariate biomarker analysis:

[0307] 1) Multivariate biomarker ranking - Combine the importance of accuracy-weighted aggregation among millions of simulated support vector machines, logistic regression, random forest, linear discriminant analysis, and stochastic gradient descent models with different feature sizes. Hyperparameter tuning (e.g., the choice of regularization or numerical solver techniques) was exhaustively searched in grid search. Construct a multivariate model through thousands of simulated forward selections, and then aggregate the most frequently selected biomarkers on different cross-validation slices of the dataset. Forward selection iteratively combines features according to the optimization metrics (e.g., AUC, F1 in classification, and Adj-R2, RMSE in regression). Due to involving multiple data cohorts, the importance of each study was weighted by the sample size, and Gd was listed as the primary endpoint, and then finally determine the spatially limited 21 plexuses. Next, some features were sequentially deleted, and after deleting the said features, the remaining features still had predictive power for each endpoint / study. Use cross-validation and bootstrapping to reduce overfitting. Use the test holdout set as much as possible.

[0308] 2) Study the binary (2 features) model for orthogonal signal improvement

[0309] 3) Interaction terms (product, ratio, quadratic term)

[0310] To ensure the reproducibility of the results:

[0311] · The model was trained on the CLIMB dataset and tested on the EPIC dataset (before / after batch normalization and demographic adjustment)

[0312] · Paired samples (baseline-normalized signals) between AIM3 and F4

[0313] Example 4: Univariate APLP1 Biomarker Analysis for Predicting Disease Activity in Multiple Sclerosis

[0314] According to the method described in Example 2, univariate analysis of APLP1 was performed in different human clinical studies. The statistical measurements (p-value, area under the curve (AUC), and correlation value (R-squared)) of the univariate analysis are shown below.

[0315]

[0316] Example 5: Univariate CCL20 Biomarker Analysis for Predicting Disease Activity in Multiple Sclerosis

[0317] According to the method described in Example 2, univariate analysis of CCL20 was performed in different human clinical studies. The statistical measurements (p-value, area under the curve (AUC), and correlation value (R-squared)) of the univariate analysis are shown below.

[0318]

[0319] Example 6: Univariate CD6 Biomarker Analysis for Predicting Disease Activity in Multiple Sclerosis

[0320] According to the method described in Example 2, univariate analysis of CD6 was performed in different human clinical studies. The statistical measurements (p-value, area under the curve (AUC), and correlation value (R-squared)) of the univariate analysis are shown below.

[0321]

[0322] Example 7: Univariate CDCP1 Biomarker Analysis for Predicting Disease Activity in Multiple Sclerosis

[0323] According to the method described in Example 2, univariate analysis of CDCP1 was performed in different human clinical studies. The statistical measurements (p-value, area under the curve (AUC), and correlation value (R-squared)) of the univariate analysis are shown below.

[0324]

[0325] Example 8: Univariate CNTN2 Biomarker Analysis for Predicting Disease Activity in Multiple Sclerosis

[0326] According to the method described in Example 2, univariate analysis of CNTN2 was performed in different human clinical studies. The statistical measurements (p-value, area under the curve (AUC), and correlation value (R-squared)) of the univariate analysis are shown below.

[0327]

[0328] Example 9: Univariate COL4A1 Biomarker Analysis for Predicting Disease Activity in Multiple Sclerosis

[0329] According to the method described in Example 2, univariate analysis of COL4A1 was performed in different human clinical studies. The statistical measurements (p-value, area under the curve (AUC), and correlation value (R-squared)) of the univariate analysis are shown below.

[0330]

[0331] Example 10: Univariate CXCL9 Biomarker Analysis for Predicting Disease Activity in Multiple Sclerosis

[0332] According to the method described in Example 2, univariate analysis of CXCL9 was performed in different human clinical studies. The statistical measurements (p-value, area under the curve (AUC), and correlation value (R-squared)) of the univariate analysis are shown below.

[0333]

[0334] Example 11: Univariate FLRT2 Biomarker Analysis for Predicting Disease Activity in Multiple Sclerosis

[0335] According to the method described in Example 2, univariate analysis of CDCP1 was performed in different human clinical studies. The statistical measurements (p-value, area under the curve (AUC), and correlation value (R-squared)) of the univariate analysis are shown below.

[0336]

[0337]

[0338] Example 12: Univariate Growth Hormone Biomarker Analysis for Predicting Disease Activity in Multiple Sclerosis

[0339] According to the method described in Example 2, univariate analysis of growth hormone (GH) was performed in different human clinical studies. The statistical measurements (p-value, area under the curve (AUC), and correlation value (R-squared)) of the univariate analysis are shown below.

[0340]

[0341] Example 13: Univariate IFI30 Biomarker Analysis for Predicting Disease Activity in Multiple Sclerosis

[0342] According to the method described in Example 2, univariate analysis of IFI30 was performed in different human clinical studies. The statistical measurements (p-value, area under the curve (AUC), and correlation value (R-squared)) of the univariate analysis are shown below.

[0343]

[0344] Example 15: Univariate MOG Biomarker Analysis for Predicting Disease Activity in Multiple Sclerosis

[0345] According to the method described in Example 2, univariate analysis of MOG was performed in different human clinical studies. The statistical measurements (p-value, area under the curve (AUC), and correlation value (R-squared)) of the univariate analysis are shown below.

[0346]

[0347] Example 16: Univariate NEFL Biomarker Analysis for Predicting Disease Activity in Multiple Sclerosis

[0348] According to the method described in Example 2, univariate analysis of NEFL was performed in different human clinical studies. The statistical measurements of the univariate analysis (p-value, area under the curve (AUC), and correlation value (R-squared)) are shown below.

[0349]

[0350] Example 17: Univariate OPG Biomarker Analysis for Predicting Disease Activity in Multiple Sclerosis

[0351] According to the method described in Example 2, univariate analysis of OPG was performed in different human clinical studies. The statistical measurements of the univariate analysis (p-value, area under the curve (AUC), and correlation value (R-squared)) are shown below.

[0352]

[0353]

[0354] Example 18: Univariate OPN Biomarker Analysis for Predicting Disease Activity in Multiple Sclerosis

[0355] According to the method described in Example 2, univariate analysis of OPN was performed in different human clinical studies. The statistical measurements of the univariate analysis (p-value, area under the curve (AUC), and correlation value (R-squared)) are shown below.

[0356]

[0357] Example 19: Univariate PRTG Biomarker Analysis for Predicting Disease Activity in Multiple Sclerosis

[0358] According to the method described in Example 2, univariate analysis of PRTG was performed in different human clinical studies. The statistical measurements of the univariate analysis (p-value, area under the curve (AUC), and correlation value (R-squared)) are shown below.

[0359]

[0360] Example 20: Univariate SERPINA9 Biomarker Analysis for Predicting Disease Activity in Multiple Sclerosis

[0361] According to the method described in Example 2, univariate analysis of SERPINA9 was performed in different human clinical studies. The statistical measurements of the univariate analysis (p-value, area under the curve (AUC), and correlation value (R-squared)) are shown below.

[0362]

[0363] Example 21: Univariate TNFRSF10A Biomarker Analysis for Predicting Disease Activity in Multiple Sclerosis

[0364] According to the method described in Example 2, univariate analysis of TNFRSF10A was performed in different human clinical studies. The statistical measurements of the univariate analysis (p-value, area under the curve (AUC), and correlation value (R-squared)) are shown below.

[0365]

[0366] Example 22: Univariate VCAN Biomarker Analysis for Predicting Disease Activity in Multiple Sclerosis

[0367] According to the method described in Example 2, univariate analysis of VCAN was performed in different human clinical studies. The statistical measurements of the univariate analysis (p-value, area under the curve (AUC), and correlation value (R-squared)) are shown below.

[0368]

[0369]

[0370] Example 23: Univariate CHI3L1 Biomarker Analysis for Predicting Disease Activity in Multiple Sclerosis

[0371] According to the method described in Example 2, univariate analysis of CHI3L1 was performed in different human clinical studies. The statistical measurements of the univariate analysis (p-value, area under the curve (AUC), and correlation value (R-squared)) are shown below.

[0372]

[0373] Example 24: Univariate CXCL13 Biomarker Analysis for Predicting Disease Activity in Multiple Sclerosis

[0374] According to the method described in Example 2, univariate analysis of CXCL13 was performed in different human clinical studies. The statistical measurements of the univariate analysis (p-value, area under the curve (AUC), and correlation value (R-squared)) are shown below.

[0375]

[0376] Example 25: Univariate Growth Hormone 2 Biomarker Analysis for Predicting Disease Activity in Multiple Sclerosis

[0377] According to the method described in Example 2, univariate analysis of growth hormone 2 (GH2) was performed in different human clinical studies. The statistical measurements of the univariate analysis (p-value, area under the curve (AUC), and correlation value (R-squared)) are shown below.

[0378]

[0379] Example 26: Univariate IL-12B Biomarker Analysis for Predicting Disease Activity in Multiple Sclerosis

[0380] According to the method described in Example 2, univariate analysis of IL-12B was performed in different human clinical studies. The statistical measurements of the univariate analysis (p-value, area under the curve (AUC), and correlation value (R-squared)) are shown below.

[0381]

[0382] Example 27: Univariate IL18 Biomarker Analysis for Predicting Multiple Sclerosis Disease Activity

[0383] According to the method described in Example 2, univariate analysis of IL18 was performed in different human clinical studies. The statistical measurements of the univariate analysis (p-value, area under the curve (AUC), and correlation value (R-squared)) are shown below.

[0384]

[0385]

[0386] Example 28: Univariate MMP-2 Biomarker Analysis for Predicting Multiple Sclerosis Disease Activity

[0387] According to the method described in Example 2, univariate analysis of MMP-2 was performed in different human clinical studies. The statistical measurements of the univariate analysis (p-value, area under the curve (AUC), and correlation value (R-squared)) are shown below.

[0388]

[0389] Example 29: Univariate MMP-9 Biomarker Analysis for Predicting Multiple Sclerosis Disease Activity

[0390] According to the method described in Example 2, univariate analysis of MMP-9 was performed in different human clinical studies. The statistical measurements of the univariate analysis (p-value, area under the curve (AUC), and correlation value (R-squared)) are shown below.

[0391]

[0392] Example 30: Univariate VCAM-1 Biomarker Analysis for Predicting Multiple Sclerosis Disease Activity

[0393] According to the method described in Example 2, univariate analysis of VCAM-1 was performed in different human clinical studies. The statistical measurements of the univariate analysis (p-value, area under the curve (AUC), and correlation value (R-squared)) are shown below.

[0394]

[0395] Example 31: Univariate TNFSF13B Biomarker Analysis for Predicting Multiple Sclerosis Disease Activity

[0396] According to the method described in Example 2, univariate analyses of TNFSF13B were performed in different human clinical studies. The statistical measurements (p-value, area under the curve (AUC), and correlation value (R-squared)) of the univariate analyses are shown below.

[0397]

[0398] Example 32: Univariate GFAP Biomarker Analysis for Predicting Multiple Sclerosis Disease Activity

[0399] According to the method described in Example 2, univariate analyses of GFAP were performed in different human clinical studies. The statistical measurements (p-value, area under the curve (AUC), and correlation value (R-squared)) of the univariate analyses are shown below.

[0400]

[0401] Example 33: Univariate Biomarker Analysis for Predicting Different Types of Multiple Sclerosis Disease Activity

[0402] According to the method described in Example 2, univariate analyses of various biomarkers were performed in different human clinical studies. The statistical p-values and R 2 .

[0403]

[0404] Example 34: Model Training and Validation - Baseline Normalization Shift

[0405] A linear regression model (L2 ridge regularization) was trained to predict the baseline normalized offset of Gd analysis on two different populations (F4 and F6). Here, the model was trained to classify pairs of samples (both sera were drawn near the MRI) with and without MRI lesions showing Gd enhancement relative to the baseline.

[0406] A logistic regression classification algorithm was applied to the positive and negative offsets of the Gd count. Since the predictors are the offsets of biomarker protein levels (starting from the baseline samples with no Gd activity in the corresponding MRI), there is no need for demographic correction as the within-patient variation has been taken into account. The dataset was split into a training set and a 5-fold cross-validation set, and the parameters of the model were hyper-tuned to optimize the model; all biomarkers from the custom assay panel (tiers 1, 2, and 3 in Table 2 below) were included as features in the model. The area under the receiver operating characteristic curve (AUROC) generated by the 5-fold cross-validation for classifying Gd offsets was 0.958 + / - 0.034.

[0407] Figure 2A and Figure 2BThe sequence forward selection performance curves for the F6 and F4 studies are shown separately.

[0408] Next, independent training / test holdouts of the 4-feature model were trained and tested. Here, a logistic regression (L1 regularization) model was trained on the F6 data and tested on the F4 data. The model feature space was limited to the top 4 features in the analysis. The logistic regression classification algorithm was applied to the positive and negative changes in Gd counts. Since the predictors are offsets in biomarker protein levels (starting from a baseline of no Gd activity), no demographic correction was necessary as within-patient variation was accounted for. Bridge normalization was applied to the dataset using a re-run set of overlapping samples to use the same model coefficients for the training and test sets. The parameters of the model were hyper-tuned to optimize the model; all biomarkers in the custom assay panel were included as features in the model. This analysis was performed to evaluate the multivariate prediction performance (relative to the univariate performance of serum neurofilament light chain (or sNfL)), as Figure 3A shown. The performance exhibited by the model was AUROC: 0.91 (sNfL - 0.88), accuracy: 0.84 (sNfL - 0.77), sensitivity: 0.76 (sNfL - 0.70), specificity: 0.91 (sNfL - 0.83), and Youden statistic = 0.67 (sNfL – 0.53). Figure 3B The corresponding confusion matrix for the multivariate classifier is shown, which established a high true positive rate (TPR) = 0.757 and a high true negative rate (TNR) = 0.905.

[0409] Example 35: Model Training and Validation - Cross-Sectional Classification of Presence / Absence of Disease

[0410] A logistic regression model (L1 regularization) was trained to predict the presence / absence of disease based on Gd counts. Specifically, the model was trained to predict: mild disease (corresponding to 1 lesion present or absent on each MRI), moderate disease (corresponding to lesions present or absent), and extreme disease (corresponding to more than three lesions present or absent). The model was built using 321 samples from F4, 180 samples from F5, and 155 samples from F6. Figure 4A The sequence forward selection performance curve is shown.

[0411] To build the model, bridge normalization and demographic correction were applied to the dataset that combined the three studies. The dataset was split into a training set and a 5-fold cross-validation set, and the parameters of the model were hyper-tuned to optimize the model; all biomarkers from the custom assay panel were included as features in the model. Figure 4BThe ROC curves of three models are shown. Specifically, the AUROC obtained by 5-fold cross-validation is: 0.741 ± 0.017 for the mild model; 0.785 ± 0.004 for the moderate model, and 0.903 ± 0.009 for the severe model.

[0412] In addition, Figure 4C The normalized confusion matrices of each model are shown. The Gd+ classification comparisons are as follows, using 1) a univariate NFL model, 2) a multivariate model including NFL, and 3) a multivariate model not including NFL.

[0413]

[0414] Example 36: Model Training and Validation - Predicting Disease Severity by Predicting Lesion Number

[0415] 321 samples from F4, 180 samples from F5, and 155 samples from F6 were used to train and test a linear regression model (L2 regularization). The ridge regression algorithm was applied to the Gd count. Bridge normalization and demographic correction were applied to the dataset combining three studies. The dataset was divided into a training set and a 5-fold cross-validation set; all biomarkers from the custom assay panel were included as features in the model, and a forward fitting procedure was applied to estimate which feature combinations would result in higher model performance. Figure 4D Feature sequence forward selection is shown. The corresponding best regression plots of the predicted and actual Gd lesion counts (i.e., an indicator of the MS disease activity burden in this analysis) were generated. Based on R 2 The best performance was obtained using 7 features (NEFL, MOG, CDCP1, OPG, APLP1, and COL4A1): 0.219 ± 0.050. Looking at the Spearman rank (Spearman r) scores, the best model was obtained from 4 features (NEFL, CDCP1, APLP1, and TNFSF13B), with a Spearman rank score of 0.496 ± 0.038.

[0416] Example 37: Model Training and Validation - Predicting Annualized Relapse Rate

[0417] A logistic regression model (Elasticnet 0.2) was trained and tested on 282 samples, where 161 LOW samples < 0.3 ARR, and 121 HIGH samples > 0.8 ARR. Figure 5A Feature sequence forward selection is shown. Applying the logistic regression classification algorithm, after forward fitting the features and overshoot parameters, the best model showed an AUROC of 0.672 + / - 0.053, which had the following 5 features: NEFL, MOG, CDCP1, IL-12B, and TNFRSF10A. Figure 5BThe ROC curve for predicting ARR is shown.

[0418] Example 38: Model Training and Validation - Predicting Clinically Defined Relapses

[0419] A logistic regression model was built using the LBFGS solver with L2 regularization (C = 1.0) and balanced class weights. The recurrence status was set by the clinical definition criteria in the ACP study ("deterioration" vs. "quiescence") and by whether the patients in F6 had a recurrence within 90 days after blood sampling. Since there were no bridging samples between ACP and F6, two independent analyses were performed for these two studies. The NPX values of both studies were corrected for demographics, and the studies in F6 that were more than 30 days from MRI were excluded (to maintain consistency with the Gd study).

[0420] Figure 6A Feature-sequence forward selection using the F6 study (n = 155, 136 patients without recurrence, 19 with recurrence within the first 90 days) and the ACP study (n = 124, 64 quiescent MS samples, 60 deteriorating MS samples) is shown. As the figure moves from left to right, new features are cumulatively added to the total. Figure 6B The performance of each model is shown. Specifically, the model trained in the F6 study exhibited an AUROC of 0.915 ± 0.053, while the model trained in the ACP study exhibited an AUROC of 0.845 ± 0.061.

[0421] Example 39: Model Training and Validation - Predicting Disease Progression

[0422] Figure 7 Feature-sequence forward selection of the absolute quantitative data according to the Expanded Disability Status Scale (EDSS) is shown. Here, the biomarkers and their absolute quantification measurements of 163 samples with all data available were considered. The samples were from the F6 (University of Basel) study cohort. For each protein, the serum measurements were directly correlated with the respective endpoints (EDSS, T2-weighted lesion volume). Only the serum drawn within 30 days after MRI was considered for the radiological endpoint (i.e., T2-weighted lesion volume). 5-fold cross-validation was used to train and test the logistic model (L2 regularization) and evaluate the performance. The optimization metric used was R 2 (the square of the Pearson correlation coefficient R), and the normalization coefficient is shown. The values below show the EDSS R of different biomarkers 2 . Notably, the raw serum GFAP was correlated with the EDSS, with an R 2 value of 0.201, and the raw serum OPG was correlated with the EDSS, with an R 2 value of 0.204.

[0423] Biomarker <![CDATA[EDSS R 2 > <![CDATA[T2 weighted volume R 2 > APLP1 0.000227 0.037332 CCL20 0.041031 0.091351 CD6 0.004626 0.07696 CDCP1 0.03931 0.001367 CNTN2 0.001594 0.023916 COL4A1 0.00246 0.011448 CXCL13 0.02188 0.053104 CXCL9 0.019516 0.113253 FLRT2 0.00237 0.042253 GFAP 0.201107 0.176669 GH 0.001453 0.005613 IL12B 0.004874 0.000005 MOG 0.001258 0.016593 NEFL 0.054577 0.077404 OPG 0.204099 0.081336 OPN 0.052497 0.055554 PRTG 0.00736 0.001347 SERPINA9 0.015369 0.000716 TNFRSF10A 0.038961 0.026022 TNFSF13B 0.046195 0.063893 VCAN 0.043875 0.030339

[0424] The values shown below represent the absolute quantification (log-transformed, pg / mL) and relative quantification (normalized protein expression (or NPX)) data for n = 205 samples in the F6 Basel cohort. AvN1 represents the Pearson R 2 value, which corresponds to the correlation between the absolute quantification and the NPX values measured on the discovery panel, to guide initial research and development. N2vN1 represents the same metric between the relative NPX of the absolute quantification (before fitting a standard curve to map the values to concentrations) and the previously measured discovery panel NPX. *There is no corresponding relative quantification measurement for GFAP because the exploratory analysis did not exist at that time. This example is used to show that the quantification information of the markers on the panel, regardless of its source format, can be used to predict MS disease activity.

[0425] Biomarker <![CDATA[AvN1 R 2 > <![CDATA[N2vN1 R 2 > APLP1 0.616 0.618 CCL20 0.950 0.951 CD6 0.792 0.793 CDCP1 0.847 0.525 CNTN2 0.751 0.751 COL4A1 0.537 0.538 CXCL13 0.811 0.811 CXCL9 0.915 0.916 FLRT2 0.349 0.346 GFAP 0.999 N / A GH 0.909 0.915 IL12B 0.786 0.788 MOG 0.802 0.802 NEFL 0.808 0.813 OPG 0.718 0.718 OPN 0.713 0.710 PRTG 0.479 0.478 SERPINA9 0.885 0.892 TNFRSF10A 0.695 0.695 TNFSF13B 0.733 0.732 VCAN 0.398 0.398

[0426] Example 40: Multivariate Biomarker Panels (Biomarkers in Layers A, B, C) for Predicting Multiple Sclerosis Disease Activity Example 41: Multivariate Biomarker Panels for Predicting Multiple Sclerosis Disease Activity

[0427] Biomarkers are selected for the multivariate custom panel based on their correlation with various endpoints (e.g., mild disease (e.g., 0 to 1 Gd-enhancing lesion), general disease (e.g., 0 to at least 1 Gd-enhancing lesion), annualized relapse rate, and worsening vs. quiescence).

[0428] Based on the method described in Example 3, a multivariate analysis of the biomarker panel is performed in different human clinical studies. Specifically, the biomarkers are classified into different strata (e.g., Stratum A, Stratum B, and Stratum C). The biomarker panel is constructed from one or more strata (e.g., a single Stratum A, a single Stratum B, a single Stratum C, Stratum A + Stratum B, or Stratum A + Stratum B + Stratum C). The total of 21 biomarkers evaluated through this multivariate example are:

[0429] Table 1: Panel of 21 biomarkers divided into Stratum A, Stratum B, and Stratum C.

[0430]

[0431] The training and cross-validation of the linear regression model (L1 regularization) were performed independently on each dataset (AIM1-ARR, ACP-E, and Q), except for F4+F5, which was mixed with the bridging samples normalized to the primary endpoint Gd. Where possible, for classification problems, disease activity was classified as mild (0 and 1 Gd lesions), moderate (0 and any number of Gd lesions), and extreme (0 and 3+ lesions). The same model construction strategy was redeployed to progressively larger subsets / layers of markers on the panel to report AUC / PPV. The statistical measures (area under the curve (AUC) and positive predictive value (PPV)) of the multivariate analysis are shown below.

[0432] Generally, biomarker panels using the biomarkers in each of Layer A, Layer B, and Layer C (a total of 21 biomarkers) correspond to prediction models that exhibit better predictive ability at different disease activity endpoints (e.g., mild disease activity, moderate disease activity, extreme disease activity, annualized relapse rate, or disease status). Specifically, the AUC for these different disease activity endpoints ranges from 0.771 to 0.961, while the PPV ranges from 0.687 to 0.895. The AUC values obtained for biomarker panels using Layer A and Layer B biomarkers (a total of 17 biomarkers) at different disease endpoints range from 0.737 to 0.968, while the PPV ranges from 0.620 to 0.896. The AUC values obtained for biomarker panels using only Layer A biomarkers (a total of 8 biomarkers) at different disease endpoints range from 0.763 to 0.880, while the PPV ranges from 0.716 to 0.871. Biomarker panels using only Layer B biomarkers or only Layer C biomarkers are still predictive, but are significantly lower than those using Layer A biomarkers or combinations of Layer A+Layer A or Layer A+Layer B+Layer C. Specifically, the AUC values obtained for biomarker panels using only Layer B biomarkers at different disease endpoints range from 0.562 to 0.841, while the PPV ranges from 0.462 to 0.999. The AUC values obtained for biomarker panels using only Layer C biomarkers at different disease endpoints range from 0.589 to 0.779, while the PPV ranges from 0.410 to 0.781.

[0433]

[0434]

[0435]

[0436]

[0437]

[0438] Example 42: Additional Multivariate (Pairs, Triples, and

[0439] According to the method described in Example 3, a multivariate analysis of the biomarker panel was performed in different human clinical studies. Specifically, the biomarkers were classified into different strata (e.g., Stratum 1, Stratum 2, and Stratum 3). The biomarker panel was constructed from one or more strata (e.g., individual Stratum 1, individual Stratum 2, individual Stratum 3, Stratum 1 + Stratum 2, or Stratum 1 + Stratum 2 + Stratum 3). A total of 21 biomarkers evaluated through this multivariate example are the 21 biomarkers shown in Table 2. Additional alternative biomarkers that can be used in place of any one of these 21 biomarkers were identified as the Stratum 4 biomarkers in Table 2.

[0440] Table 2: Stratified biomarkers for predicting multiple sclerosis disease activity.

[0441]

[0442]

[0443] Training and cross-validation of the linear regression model (L1 regularization) were performed independently on each dataset (AIM1-ARR, ACP-E, and Q) (except for F4+F5, which was mixed by normalizing and bridging samples with the primary endpoint Gd). Whenever possible, for classification problems, disease activity was classified as mild (0 and 1 Gd lesion), moderate (0 and any number of Gd lesions), and extreme (0 and 3+ lesions). The same model construction strategy was redeployed to successively larger subsets / strata of markers on the panel to report AUC / PPV. Statistical measures of the multivariate analysis (area under the curve (AUC) and positive predictive value (PPV)) are shown below.

[0444] Typically, biomarker panels using biomarkers from each of Layer 1, Layer 2, and Layer 3 (a total of 21 biomarkers) correspond to predictive models that exhibit better predictive ability at different disease activity endpoints (e.g., mild disease activity, general disease activity, extreme disease activity, annualized relapse rate, or disease status). Specifically, the AUC ranges for these different disease activity endpoints are from 0.686 to 0.889, while the PPV ranges are from 0.648 to 0.835. The AUC values obtained for biomarker panels using Layer 1 and Layer 2 biomarkers (a total of 17 biomarkers) at different disease endpoints range from 0.693 to 0.892, while the PPV ranges are from 0.613 to 0.843. The AUC values obtained for biomarker panels using only Layer 1 biomarkers (a total of 8 biomarkers) at different disease endpoints range from 0.667 to 0.869, while the PPV ranges are from 0.617 to 0.861. Biomarker panels using only Layer 2 biomarkers or only Layer 3 biomarkers are still predictive, but are significantly lower than those using Layer 1 biomarkers or the Layer 1 + Layer 2 or Layer 1 + Layer 2 + Layer 3 combinations. Specifically, the AUC values obtained for biomarker panels using only Layer 2 biomarkers at different disease endpoints range from 0.595 to 0.761, while the PPV ranges are from 0.462 to 0.769. The AUC values obtained for biomarker panels using only Layer 3 biomarkers at different disease endpoints range from 0.566 to 0.626, while the PPV ranges are from 0.370 to 0.634.

[0445]

[0446]

[0447]

[0448]

[0449]

[0450] Quads) Biomarker Panels for Predicting Directional Shifts in Multiple Sclerosis Disease Activity Example 43: Additional Multivariate (Pairs, Triples, and Quads)

[0451] Multivariate analyses of the minimal biomarker sets (e.g., pairs, triplets, and quadruplets) were performed in different human clinical studies. Specifically, the ability of the minimal biomarker sets to predict shifts in the direction of MS disease activity (e.g., increasing or decreasing disease activity, as measured by the number of gadolinium-enhanced lesions) was analyzed.

[0452] To analyze the minimal predictive biomarker sets in a longitudinal patient sample, combinations of biomarker pairs, triplets, and quadruplets were analyzed in an offset (baseline normalization, paired differences) analysis to predict increases or decreases in Gd-enhancing lesions. The primary procedures for the analysis were as follows:

[0453] 1. Read the biomarker list.

[0454] 2. Read the data for these biomarkers, their accompanying demographics, and the clinical characteristics data for each study from the data lake.

[0455] 3. Exclude all samples for which the associated MRI scans were collected over a threshold time (e.g., over 30 days).

[0456] 4. Calculate the differences in NPX values between all sample pairs with increased or decreased lesion activity.

[0457] a. For the true baseline normalization method, only filter out sample pairs for which one of the samples has 0 Gd lesions in the associated MRI.

[0458] 5. For each study:

[0459] a. Use increase / decrease or no Gad+→Gad+ / Gad+→no Gad+ as the endpoint,

[0460] b. Perform a five-fold cross-validation split,

[0461] c. Select a logistic regression model configuration that performed well in past multivariate analyses,

[0462] d. Create a feature matrix from each combination of two, three, and four proteins (there is no need to use demographics / clinical characteristics as covariates since baseline normalization is incorporated into the process),

[0463] e. Train a copy of the logistic regression model on four out of the five splits and evaluate its performance on the fifth split,

[0464] 6. Average the AUROC for each model across the five splits and three studies. Use the standard deviation of the performance to calculate the uncertainty measure (repeat for PPV).

[0465] 7. Sort the models by the AUROC of the model and create an ROC curve for the model corresponding to the highest-performing feature set for each study.

[0466] We performed the above analysis in two ways:

[0467] The above analysis was performed in two ways:

[0468] A. Include the complete list of biomarkers, and

[0469] B. Subsequent analysis excluding the best-performing biomarker (NEFL).

[0470] Tables 3 and 4 below show that biomarker panels containing two, three, or four biomarkers can predict the deviation in the direction of MS disease activity.

[0471] Specifically, for biomarker pairs, the area under the receiver operating characteristic curve (AUROC) obtained for the panel containing the NEFL and CD6 biomarkers was 0.860 and the average PPV was 0.77. Additionally, the AUROC obtained for the biomarker panel containing the NEFL and CXCL9 biomarkers was 0.85 and the average PPV was 0.74. Other biomarker pairs (excluding NEFL) were also predictive. For example, the AUROC obtained for the biomarker panel containing MOG and CXCL9 was 0.761 and the PPV was 0.686. The AUROC obtained for the biomarker panel containing CD6 and CXCL9 was 0.766 and the PPV was 0.699. The AUROC obtained for the biomarker panel containing MOG and CD6 was 0.745 and the PPV was 0.705.

[0472] Specifically, for biomarker triples, the AUROC obtained for the panel containing NEFL, CXCL9, and CD6 was 0.885 and the average PPV was 0.79. Additionally, the AUROC obtained for the biomarker panel containing MOG, CD6, and CXCL9 was 0.798 and the PPV was 0.71.

[0473] Specifically, for biomarker quadruples, the AUROC obtained for the biomarker panel containing NEFL, MOG, CXCL9, and CD6 was 0.884 and the PPV was 0.763. The AUROC obtained for the biomarker panel containing NEFL, CXCL9, CD6, and CXCL13 was 0.889 and the PPV was 0.764. The AUROC obtained for the biomarker panel containing NEFL, TNFRSF10A, COL4A1, and CCL20 was 0.836 and the PPV was 0.725. Other biomarker quadruples (excluding NEFL) also showed predictability. For example, the combination of MOG, CXCL9, IL-12B, and APLP1 obtained an AUROC of 0.795 and a PPV of 0.67. The combination of CXCL9, OPG, APLP1, and OPN obtained an AUROC of 0.741 and a PPV of 0.65. The combination of MOG, IL-12B, OPN, and CNTN2 obtained an AUROC of 0.765 and a PPV of 0.69.

[0474] Table 3: Baseline-normalized offset prediction: Biomarker panels for predicting an offset (e.g., increase or decrease) in MS disease activity.

[0475]

[0476]

[0477] Table 4: Biomarker panels for predicting a decrease in MS disease activity.

[0478]

[0479]

[0480] Biomarker Panels for Predicting the Presence or Absence of Multiple Sclerosis Example 44: Additional Multivariate (Pairs, Triples, and Quads)

[0481] Multivariate analyses of minimal biomarker sets (e.g., pairs, triples, and quadruples) were performed in different human clinical studies. Specifically, the ability of minimal biomarker sets to predict the presence or absence of general MS disease (e.g., the presence of at least 1 Gd-enhancing lesion indicates presence and 0 Gd-enhancing lesions indicates absence) was analyzed for classification.

[0482] To analyze the minimal predictive biomarker sets, combinations of biomarker pairs, triples, and quadruples were analyzed in a cross-sectional analysis to predict the presence or absence of Gd-enhancing lesions ("general disease activity" (or GDA)). The primary procedures for the analysis were:

[0483] 1. Read the biomarker list.

[0484] 2. Read from the data lake the data on these biomarkers and their accompanying demographics, as well as data on the clinical characteristics of each study.

[0485] 3. Exclude all samples for which the MRI scans associated with them were collected over a threshold time (e.g., over 30 days).

[0486] 4. Perform optimized demographic and clinical adjustments.

[0487] A. Run an OLS regression between the best subset of demographic / clinical characteristics and the NPX levels of each individual biomarker value, using only Gd samples (no disease activity) from a given study, and filtering out outliers.

[0488] a. DiseaseDuration: Years since MS diagnosis.

[0489] b. Age: The age of the patient, in years.

[0490] c.Sample_Age: The storage time of the sample before formal analysis.

[0491] d.Delta_BloodMinusDiagnosis: The number of days between the MRI and the corresponding blood sample (must be between 30).

[0492] B. Extract the residuals from this procedure and use them as input to the model building procedure.

[0493] 5. For each study:

[0494] a. Construct the GDA endpoint,

[0495] b. Perform a five-fold cross validation split,

[0496] c. Select a logistic regression model configuration that has performed well in past multivariate analyses,

[0497] d. Create a signature matrix from each combination of two, three, and four proteins (and it is not necessary to include demographic / clinical characteristics as covariates),

[0498] e. Train a copy of the logistic regression model on four of the five splits and evaluate its performance on the fifth split,

[0499] 6. AUROC for each model was averaged across the five splits and three studies.

[0500] 7. Sort the models by their AUROC and create a ROC curve for each model studied that corresponds to the highest performing feature set.

[0501] The above analysis is performed in two ways:

[0502] A. Include a complete list of biomarkers, and

[0503] B. Subsequent analysis excluding the best performing biomarker (NEFL).

[0504] Table 5 below shows that biomarker panels containing two, three or four biomarkers can predict the presence or absence of MS. Specifically, for biomarker pairs, the panel containing NEFL and TNFSF13B obtained an AUROC of 0.788 and a PPV of 0.708, the panel containing NEFL and CNTN2 obtained an AUROC of 0.777 and a PPV of 0.732, and the panel containing NEFL and CXCL9 obtained an AUROC of 0.777 and a PPV of 0.713. Additional biomarker pairs (excluding NEFL) can also predict the presence or absence of multiple sclerosis. The panel containing MOG and CDCP1 obtained an AUROC of 0.672 and a PPV of 0.597, the panel containing MOG and TNFSF13B obtained an AUROC of 0.672 and a PPV of 0.599, and the panel containing MOG and CXCL9 obtained an AUROC of 0.670 and a PPV of 0.606.

[0505] Specifically, for the biomarker triplet, the panel containing NEFL, CNTN2, and TNFSF13B obtained an AUROC of 0.794 and a PPV of 0.745. Here, replacing CNTN2 in the biomarker triplet with APLP1 or TNFRSF10A, similar AUROC values ​​of 0.794 and 0.792 were obtained, respectively. For the biomarker triplet without NEFL, the panel containing MOG, CXCL9, and TNFSF13B obtained an AUROC of 0.690 and a PPV of 0.623, the panel containing MOG, OPG, and TNFSF13B obtained an AUROC of 0.685 and a PPV of 0.637, and the panel containing MOG, CCL20, and TNFSF13B obtained an AUROC of 0.685 and a PPV of 0.664. APLP1, CCL20, and CNTN2 are the next group of proteins that contribute to the improved biomarker triplet signal. For the biomarker quadruple, the panel including NEFL, TNFRSF10A, CNTN2, and TNFSF13B obtained an AUROC of 0.798 and a PPV of 0.742.

[0506] Table 5: Biomarker panel used to determine the presence or absence of gadolinium-enhancing lesions on relevant MRI based on the protein signature of serum from a single blood draw within 30 days after MRI.

[0507]

[0508]

[0509] Biomarker Panels for Predicting Multiple Sclerosis Severity Example 45: Additional Multivariate (Pairs, Triples, and Quads)

[0510] Multivariate analysis of minimal biomarker panels (e.g., pairs, triplets, and quadruplets) in different human clinical studies was performed according to the methods described in Example 3. Specifically, the ability of minimal biomarker panels to predict the number of gadolinium-enhancing lesions (e.g., a measure of mild MS disease) was analyzed.

[0511] Similar to the cross-sectional and shift analyses, regression analysis was performed as follows:

[0512] 1. Read the biomarker list.

[0513] 2. Read these biomarkers and their accompanying demographic and clinical characteristics data for each study from the data lake.

[0514] 3. Eliminate all samples whose associated MRI scans were collected for more than a threshold time (e.g., more than 30 days).

[0515] 4. Perform optimized demographic and clinical adjustments.

[0516] A. OLS regression was run between the optimal subset of demographic / clinical characteristics and NPX levels for each individual biomarker value, using only the Gd sample (without disease activity) from a given study, filtering out outliers.

[0517] a.DiseaseDuration: Number of years since MS diagnosis.

[0518] b.Age: The patient's age in years.

[0519] c.Sample_Age: The storage time of the sample before formal analysis.

[0520] d.Delta_BloodMinusDiagnosis: The number of days between the MRI and the corresponding blood sample (must be between 30).

[0521] B. Extract the residuals from this procedure and use them as input to the model building procedure.

[0522] 5. For each study:

[0523] a. Using Gd lesion counts (cut at a maximum of 5) as regression endpoint,

[0524] b. Perform a five-fold cross validation split,

[0525] c. Select a ridge regression model configuration that has worked well in past multivariate analyses,

[0526] d. Create a feature matrix from each combination of two, three, and four proteins (and there is no need to use demographic / clinical characteristics as covariates),

[0527] e. Train copies of the ridge regression model on four out of five splits and evaluate its performance on the fifth split,

[0528] 6. For each model in the five splits and three studies, take the average of the adjusted R 2 Take the average.

[0529] 7. Sort the models by the adjusted R 2 and create a correlation plot between the predictions and the endpoints for the model corresponding to the highest-performing feature set for each study.

[0530] The above analysis was performed in two ways:

[0531] A. Include the complete list of biomarkers, and

[0532] B. A subsequent analysis excluding the best-performing biomarker (NEFL).

[0533] Table 6 below shows that biomarker panels containing two, three, or four biomarkers can generally predict the determination of mild MS disease activity.

[0534] Specifically, for biomarker pairs, the Spearman rank coefficient value obtained for the panel containing NEFL and TNFSF13B was 0.524, the value obtained for the panel containing NEFL and SERPINA9 was 0.501, and the value obtained for the panel containing NEFL and GH was 0.505. Additional biomarker pairs (excluding NEFL) can also predict the determination of mild MS disease activity. For example, the Spearman rank coefficient value obtained for the panel containing MOG and TNFSF13B was 0.286, while the value obtained for the panel containing MOG and CXCL9 was 0.290.

[0535] For biomarker triplets, the Spearman rank coefficient value obtained for the panel containing NEFL, SERPINA9, and TNFSF13B was 0.533, the Spearman rank coefficient value obtained for the panel containing NEFL, CNTN2, and TNFSF13B was 0.525, and the Spearman rank coefficient value obtained for the panel containing NEFL, APLP1, and TNFSF13B was 0.537. Additionally, the Spearman rank coefficient value for the AUC obtained for the panel of MOG, CXCL9, and TNFSF13B was 0.306, the Spearman rank coefficient value obtained for the panel of MOG, CCL20, and COL4A1 was 0.210, and the Spearman rank coefficient value obtained for the panel of CXCL13, APLP1, and FLRT2 was 0.143.

[0536] For biomarker quadruplets, the Spearman rank coefficient value obtained for the panel containing NEFL, CXCL13, CCL20, and TNFSF13B was 0.520. Additionally, the Spearman rank coefficient value obtained for the panel containing NEFL, CXCL13, CXCL9, and TNFSF13B was 0.513, and the Spearman rank coefficient value obtained for the panel containing NEFL, CXCL13, SERPINA9, and TNFSF13B was 0.513. The Spearman rank coefficient value obtained for the biomarker quadruplet (without NEFL) MOG, CXCL9, OPG, SERPINA9, and TNFSF13B was 0.302.

[0537] Table 6: Biomarker panels for predicting the number of lesions on the relevant MRI based on protein signatures in a single blood draw within 30 days after MRI.

[0538]

[0539]

[0540] Biomarker Panels for Predicting Multiple Sclerosis Disease Progression ​

[0541] According to the method described in Example 3, multivariate analysis of the minimum biomarker sets (e.g., pairs, triplets, and quadruplets) was performed on n = 205 samples from the University Hospital Basel. Specifically, the ability of the minimum biomarker sets to predict the primary disease endpoint of disease progression (e.g., the association between serum protein measurements and the Expanded Disability Status Scale (EDSS)) was analyzed. All sets used mathematical combinations (such as logistic regression models or decision trees) to combine the individual biomarker levels into a multivariate score.

[0542] The Ridge (L2) regularized linear model was evaluated for all exhaustive combinations of protein subset pairs, triples, and quadruples by 5-fold cross-validation (estimating mean and uncertainty). Then the performance was ranked by Spearman R2 (according to the principle above), and this procedure was repeated:

[0543] 1. Build the model using only the log-transformed protein concentration.

[0544] 2. Build the model by incorporating covariates with the log-transformed protein concentration plus age, sex, and disease duration as eligible features.

[0545] 3. After extracting the residuals from the OLS regression procedure, build the model using the log-transformed protein concentration to predict the concentration of each corresponding biomarker from the demographic information (age, sex, and disease duration) included only in samples without disease activity (i.e., having 0 Gd lesions on their corresponding MRI).

[0546] Table 7 below records the best-performing minimal predictive biomarker groups (e.g., pairs, triples, and quadruples) according to EDSS regression, which predict the number of lesions on the relevant MRI based on protein signatures in serum from a single blood draw within 30 days after MRI. Notably, the biomarker triple GFAP, NEFL, and MOG showed the highest overall adjusted R 2 (showing a measurable improvement compared to the best single feature GFAP in the field). Additionally, the biomarker pairs A) GFAP and MOG, B) GFAP and NEFL, and C) APLP1 and GFAP, and the biomarker quadruple GFAP, NEFL, MOG, and IL-12B / PRTG / APLP1 also showed predictivity.

[0547] The best biomarker groups with covariates (age, sex, disease duration) adjusted Log(pg / mL) include: the biomarker pair NEFL and MOG, which showed the highest overall adjusted R 2 . The biomarker triples NEFL, MOG, and GH, and NEFL, MOG, and SERPINA9 also predicted EDSS in an improved manner. The biomarker quadruples NEFL, MOG, GH, and SERPINA9, and NEFL, MOG, GH, and TNFRSF10A were further improved. Additionally, the biomarker triple CXCL9, OPG, and SERPINA9 could predict disease progression and was further improved when MOG was added to form a biomarker quadruple.

[0548] The optimised biomarker sets adjusted for demographics (age, sex, disease duration) included the biomarker pair CXCL9 and OPG and the biomarker triplet CXCL9, OPG and TNFRSF10A. In any demographically adjusted model, the biomarker quartet CD6, IL-12B, APLP1 and CCL20 formed the highest Pearson R 2 correlation.

[0549] Table 7: Biomarker panels for predicting disease progression (e.g., predicting the number of lesions on a relevant MRI based on protein signatures in serum from a single blood draw within 30 days of an MRI).

[0550]

[0551]

[0552] Additional tables

[0553] Table 8A: Study codes for biomarker analysis

[0554]

[0555]

[0556]

[0557] Table 8B: Additional descriptions of the study (e.g., joint study)

[0558]

[0559] Table 9: Additional biomarkers for predicting multiple sclerosis disease activity.

[0560]

[0561]

[0562]

[0563] Table 10: Alternative biomarkers for predicting multiple sclerosis disease activity.

[0564]

[0565]

[0566] Table 11: Classification of biomarkers

[0567]

[0568]

[0569]

[0570]

[0571] Table 12: Biomarker involvement at specific locations (brain, blood-brain barrier, blood). The numerical scale ranges from 1 - 5, where 1 indicates the lowest amount of the corresponding biomarker found at the location, and 5 indicates the highest amount of the corresponding biomarker found at the location.

[0572]

[0573] Table 13A: Biomarker involvement in specific cell types. The numerical scale ranges from 1 - 4, where 1 indicates the lowest amount of the corresponding biomarker found in the cell type, and 4 indicates the highest amount of the corresponding biomarker found in the cell type.

[0574]

[0575] Table 13B: Biomarker involvement in specific cell types. The numerical scale ranges from 1 - 4, where 1 indicates the lowest amount of the corresponding biomarker found in the cell type, and 4 indicates the highest amount of the corresponding biomarker found in the cell type.

[0576] The disclosure of the present application also includes the following embodiments:

[0577] 1. A method for determining multiple sclerosis activity in a subject, the method comprising:

[0578] Obtaining or having obtained a data set comprising expression levels of a plurality of biomarkers, wherein the plurality of biomarkers comprises each biomarker selected from at least one of Group 1, Group 2, and Group 3,

[0579] wherein Group 1 comprises Biomarker 1, Biomarker 2, Biomarker 3, Biomarker 4, Biomarker 5, Biomarker 6, Biomarker 7, and Biomarker 8,

[0580] wherein Biomarker 1 is NEFL, MOG, CADM3, or GFAP,

[0581] wherein Biomarker 2 is MOG, CADM3, KLK6, BCAN, OMG, or GFAP,

[0582] wherein Biomarker 3 is CD6, CD5, CRTAM, CD244, or TNFRSF9,

[0583] wherein biomarker 4 is CXCL9, CXCL10, IL-12B, CXCL11 or GFAP,

[0584] wherein biomarker 5 is OPG, TFF3 or ENPP2,

[0585] wherein biomarker 6 is OPN, OMD, MEPE or GFAP,

[0586] wherein biomarker 7 is CXCL13, NOS3 or MMP-2, and

[0587] wherein biomarker 8 is GFAP, NEFL, OPN, CXCL9, MOG or CHI3L1, and

[0588] wherein Group 2 includes biomarker 9, biomarker 10, biomarker 11, biomarker 12, biomarker 13, biomarker 14, biomarker 15, biomarker 16 and biomarker 17,

[0589] wherein biomarker 9 is CDCP1, IL-18BP, IL-18, GFAP or MSR1,

[0590] wherein biomarker 10 is CCL20, CCL3 or TWEAK,

[0591] wherein biomarker 11 is IL-12B, IL12A or CXCL9,

[0592] wherein biomarker 12 is APLP1, SEZ6L, BCAN, DPP6, NCAN or KLK6,

[0593] wherein biomarker 13 is TNFRSF10A, TNFRSF11A, SPON2, CHI3L1 or IFI30,

[0594] wherein biomarker 14 is COL4A1, IL6, NOTCH3 or PCDH17,

[0595] wherein biomarker 15 is SERPINA9, TNFRSF9 or CNTN4,

[0596] wherein biomarker 16 is FLRT2, DDR1, NTRK2, CDH6, MMP-2, and

[0597] wherein biomarker 17 is TNFSF13B, CXCL16, ALCAM or IL-18,

[0598] wherein Group 3 includes biomarker 18, biomarker 19, biomarker 20 and biomarker 21,

[0599] where biomarker 18 is GH, GH2, or IGFBP-1,

[0600] where biomarker 19 is VCAN, TINAGL1, CANT1, NECTIN2, MMP-9, or NPDC1,

[0601] where biomarker 20 is PRTG, NTRK2, NTRK3, or CNTN4, and

[0602] where biomarker 21 is CNTN2, DPP6, GDNFR-α-3, or SCARF2; and

[0603] a prediction of multiple sclerosis disease activity is generated by applying a prediction model to the expression levels of the plurality of biomarkers.

[0604] 2. The method according to embodiment 1, wherein the plurality of biomarkers includes each biomarker in group 1, where biomarker 1 is NEFL, where biomarker 2 is MOG, where biomarker 3 is CD6, where biomarker 4 is CXCL9, where biomarker 5 is OPG, where biomarker 6 is OPN, where biomarker 7 is CXCL13, and where biomarker 8 is GFAP.

[0605] 3. The method according to any one of embodiments 1-2, wherein the performance of the prediction model is characterized in that the area under the curve (AUC) ranges from 0.667 to 0.869.

[0606] 4. The method according to any one of embodiments 1-2, wherein the performance of the prediction model is characterized in that the positive predictive value ranges from 0.617 to 0.861.

[0607] 5. The method according to embodiment 1, wherein the plurality of biomarkers includes each biomarker in group 2, where biomarker 9 is CDCP1, where biomarker 10 is CCL20, where biomarker 11 is IL-12B, where biomarker 12 is APLP1, where biomarker 13 is TNFRSF10A, where biomarker 14 is COL4A1, where biomarker 15 is SERPINA9, where biomarker 16 is FLRT2, and where biomarker 17 is TNFSF13B.

[0608] 6. The method according to embodiment 5, wherein the performance of the prediction model is characterized in that the area under the curve (AUC) ranges from 0.595 to 0.761.

[0609] 7. The method according to embodiment 5, wherein the performance of the prediction model is characterized in that the positive predictive value ranges from 0.523 to 0.769.

[0610] 8. The method according to embodiment 2, wherein the plurality of biomarkers further includes each biomarker in group 2, wherein biomarker 9 is CDCP1, wherein biomarker 10 is CCL20, wherein biomarker 11 is IL-12B, wherein biomarker 12 is APLP1, wherein biomarker 13 is TNFRSF10A, wherein biomarker 14 is COL4A1, wherein biomarker 15 is SERPINA9, wherein biomarker 16 is FLRT2, and wherein biomarker 17 is TNFSF13B.

[0611] 9. The method according to embodiment 8, wherein the performance of the prediction model is characterized in that the area under the curve (AUC) ranges from 0.693 to 0.892.

[0612] 10. The method according to embodiment 8, wherein the performance of the prediction model is characterized in that the positive predictive value ranges from 0.613 to 0.843.

[0613] 11. The method according to embodiment 1, wherein the plurality of biomarkers includes each biomarker in group 3, wherein biomarker 18 is GH, wherein biomarker 19 is VCAN, wherein biomarker 20 is PRTG, and wherein biomarker 21 is CNTN2.

[0614] 12. The method according to embodiment 11, wherein the performance of the prediction model is characterized in that the area under the curve (AUC) ranges from 0.566 to 0.644.

[0615] 13. The method according to embodiment 11, wherein the performance of the prediction model is characterized in that the positive predictive value ranges from 0.370 to 0.742.

[0616] 14. The method according to embodiment 8, wherein the plurality of biomarkers further includes each biomarker in group 3, wherein biomarker 18 is GH, wherein biomarker 19 is VCAN, wherein biomarker 20 is PRTG, and wherein biomarker 21 is CNTN2.

[0617] 15. The method according to embodiment 11, wherein the performance of the prediction model is characterized in that the area under the curve (AUC) ranges from 0.686 to 0.889.

[0618] 16. The method according to embodiment 11, wherein the performance of the prediction model is characterized in that the positive predictive value ranges from 0.648 to 0.835.

[0619] 17. A method for determining multiple sclerosis activity in a subject, the method comprising:

[0620] obtaining or having obtained a data set comprising the expression levels of a plurality of biomarkers, the plurality of biomarkers comprising NEFL, MOG, CD6, CXCL9, OPG, OPN, CXCL13 and GFAP; and

[0621] generating a prediction of multiple sclerosis disease activity by applying a prediction model to the expression levels of the plurality of biomarkers.

[0622] 18. The method according to embodiment 17, wherein the plurality of biomarkers further comprises CDCP1, CCL20, IL-12B, APLP1, TNFRSF10A, COL4A1, SERPINA9, FLRT2 and TNFSF13B.

[0623] 19. The method according to embodiment 18, wherein the plurality of biomarkers further comprises GH, VCAN, PRTG and CNTN2.

[0624] 20. A method for generating quantitative data for a subject, comprising:

[0625] performing at least one immunoassay on a sample obtained from the subject to generate a data set comprising the quantitative data, wherein the quantitative data represents at least eight protein biomarkers, comprising: NEFL, MOG, CD6, CXCL9, OPG, OPN, CXCL13 and GFAP,

[0626] wherein the subject has or is suspected of having multiple sclerosis.

[0627] 21. The method according to embodiment 20, wherein the quantitative data further represents at least nine additional protein biomarkers, comprising: CDCP1, CCL20, IL-12B, APLP1, TNFRSF10A, COL4A1, SERPINA9, FLRT2 and TNFSF13B.

[0628] 22. The method according to embodiment 21, wherein the quantitative data further represents at least four additional protein biomarkers, comprising: GH, VCAN, PRTG and CNTN2.

[0629] 23. The method according to any one of embodiments 20-22, further comprising:

[0630] Determine the multiple sclerosis activity in the subject, wherein the determination includes applying a prediction model to the quantitative data.

[0631] 24. The method according to embodiment 23, wherein applying the determination further includes comparing the score output by the prediction model with a reference score.

[0632] 25. The method according to embodiment 24, wherein the reference score corresponds to any one of the following:

[0633] A) A patient at a baseline time point, at which the patient does not exhibit disease activity,

[0634] B) A patient clinically diagnosed as having no disease activity, or

[0635] C) A healthy patient.

[0636] 26. A method for determining multiple sclerosis activity in a subject, the method comprising:

[0637] Obtain or have obtained a data set comprising the expression levels of a plurality of biomarkers, the plurality of biomarkers comprising:

[0638] One or more neurodegeneration biomarkers selected from the group consisting of NEFL, APLP1, OPG, SERPINA9, PRTG, GFAP, CNTN2, and FLRT2;

[0639] One or more inflammation biomarkers selected from the group consisting of CCL20, GH, CXCL13, IL-12B, VCAN, TNFRSF10A, TNFSF13B, CD6, and CXCL9;

[0640] One or more immunomodulation biomarkers selected from the group consisting of CDCP1 and OPN;

[0641] One or more myelin integrity biomarkers selected from the group consisting of COL4A1 and MOG; and

[0642] Generate a prediction of multiple sclerosis disease activity by applying a prediction model to the expression levels of the plurality of biomarkers.

[0643] 27. The method according to embodiment 26, wherein the one or more neurodegeneration biomarkers include NEFL, OPG, and GFAP, wherein the one or more inflammation biomarkers include CXCL13, CD6, and CXCL9, wherein the one or more immunomodulation biomarkers include OPN, and wherein the one or more myelin integrity biomarkers include MOG.

[0644] 28. The method according to embodiment 27, wherein the one or more neurodegeneration biomarkers further include APLP1, SERPINA9, and FLRT2, wherein the one or more inflammation biomarkers further include CCL20, IL-12B, TNFRSF10A, and TNFSF13B, wherein the one or more immunomodulation biomarkers further include CDCP1, and wherein the one or more myelin integrity biomarkers further include COL4A1.

[0645] 29. The method according to embodiment 28, wherein the one or more neurodegeneration biomarkers further include PRTG and CNTN2, and wherein the one or more inflammation biomarkers include GH and VCAN.

[0646] 30. The method according to any one of embodiments 1-29, wherein the plurality of biomarkers are protein biomarkers.

[0647] 31. The method according to any one of embodiments 1-29, wherein the multiple sclerosis disease activity is any one of the following:

[0648] The presence of general disease activity, the presence of mild disease activity, a shift (increase or decrease) in disease activity, the severity of MS, a relapse or episode associated with MS, the relapse rate, the MS status, the response to MS therapy, the degree of MS disability, or the risk of developing MS.

[0649] 32. The method according to embodiment 31, wherein the general disease activity is the presence or absence of one or more gadolinium-enhanced MRI lesions, and wherein the mild disease activity is the presence of one gadolinium-enhanced MRI lesion.

[0650] 33. The method according to embodiment 31, wherein the severity of MS corresponds to the number of gadolinium-enhanced MRI lesions.

[0651] 34. The method according to embodiment 31, wherein the MS status is a deteriorating or quiescent state of multiple sclerosis.

[0652] 35. A method for determining multiple sclerosis activity in a subject, the method comprising:

[0653] Obtaining or having obtained a data set comprising expression levels of a plurality of biomarkers, wherein the plurality of biomarkers comprises two or more of the following: NEFL, MOG, CD6, CXCL9, OPG, OPN, CXCL13, GFAP, CDCP1, CCL20, IL-12B, APLP1, TNFRSF10A, COL4A1, SERPINA9, FLRT2, TNFSF13B, GH, VCAN, PRTG, and CNTN2; and

[0654] Generating a prediction of multiple sclerosis disease activity by applying a prediction model to the expression levels of the plurality of biomarkers.

[0655] 36. A method for preparing a biomarker fraction from a sample obtained from a subject, the method comprising:

[0656] (a) Extracting biomarkers from the sample;

[0657] (b) Generating a fraction of the extracted biomarkers after (a),

[0658] wherein the fraction of the extracted biomarkers after (b) comprises a plurality of biomarkers, wherein the plurality of biomarkers comprises two or more of the following: NEFL, MOG, CD6, CXCL9, OPG, OPN, CXCL13, GFAP, CDCP1, CCL20, IL-12B, APLP1, TNFRSF10A, COL4A1, SERPINA9, FLRT2, TNFSF13B, GH, VCAN, PRTG, and CNTN2;

[0659] (c) Analyzing the plurality of biomarkers in the fraction of the extracted biomarkers generated in (b).

[0660] 37. The method according to embodiment 35 or 36, wherein the two or more biomarkers comprise NEFL and at least one other biomarker.

[0661] 38. The method according to embodiment 37, wherein the two or more biomarkers are biomarker pairs selected from any one of the following: A) NEFL and CD6, B) NEFL and MOG, C) NEFL and CXCL9, and D) NEFL and TNFRSF10A.

[0662] 39. The method according to embodiment 37, wherein the two or more biomarkers are a biomarker triplet selected from any one of the following: A) NEFL, CD6, and CXCL9, and B) NEFL, TNFRSF10A, and COL4A1.

[0663] 40. The method according to embodiment 37, wherein the two or more biomarkers are a biomarker quadruplet selected from any one of the following: A) NEFL, MOG, CD6, and CXCL9, B) NEFL, CXCL9, TNFRSF10A, and COL4A1, C) NEFL, CD6, CXCL9, and CXCL13, and D) NEFL, TNFRSF10A, COL4A1, and CCL20.

[0664] 41. The method according to embodiment 35 or 36, wherein the two or more biomarkers do not include NEFL.

[0665] 42. The method according to embodiment 41, wherein the two or more biomarkers are a biomarker pair selected from any one of the following: A) MOG and IL-12B, B) CXCL9 and CD6, C) MOG and CXCL9, D) MOG and CD6, E) CXCL9 and COL4A1, and F) CD6 and VCAN.

[0666] 43. The method according to embodiment 41, wherein the two or more biomarkers are a biomarker triplet selected from any one of the following: A) MOG, IL-12B, and APLP1, B) MOG, CD6, and CXCL9, C) CXCL9, COL4A1, and VCAN, D) MOG, IL-12B, and CNTN2, and E) CD6, CCL20, and VCAN.

[0667] 44. The method according to embodiment 41, wherein the two or more biomarkers are a biomarker quadruplet selected from any one of the following: A) MOG, CXCL9, IL-12B, APLP1, B) CXCL9, COL4A1, OPG, and VCAN, C) CXCL9, OPG, APLP1, and OPN, D) MOG, IL-12B, OPN, and CNTN2, and E) CD6, COL4A1, CCL20, and VCAN.

[0668] 45. The method according to any one of embodiments 37-44, wherein the multiple sclerosis activity is a deviation of disease activity.

[0669] 46. The method according to embodiment 37, wherein the two or more biomarkers are biomarker pairs selected from any one of the following: A) NEFL and TNFSF13B, B) NEFL and CNTN2, and C) NEFL and CXCL9.

[0670] 47. The method according to embodiment 37, wherein the two or more biomarkers are biomarker triples selected from any one of the following: A) NEFL, CNTN2 and TNFSF13B, B) NEFL, APLP1 and TNFSF13B, and C) NEFL, TNFRSF10A and TNFSF13B.

[0671] 48. The method according to embodiment 37, wherein the two or more biomarkers are biomarker quadruples selected from any one of the following: A) NEFL, TNFRSF10A, CNTN2 and TNFSF13B, B) NEFL, COL4A1, CNTN2 and TNFSF13B, and C) NEFL, TNFRSF10A, APLP1 and TNFSF13B.

[0672] 49. The method according to embodiment 41, wherein the two or more biomarkers are biomarker pairs selected from any one of the following: A) MOG and CDCP1, B) MOG and TNFSF13B, and C) MOG and CXCL9.

[0673] 50. The method according to embodiment 41, wherein the two or more biomarkers are biomarker triples, and wherein one biomarker in the biomarker triple is MOG.

[0674] 51. The method according to embodiment 50, wherein the biomarker triple is selected from any one of the following: A) MOG, CXCL9 and TNFSF13B, B) MOG, OPG and TNFSF13B, and C) MOG, CCL20 and TNFSF13B.

[0675] 52. The method according to embodiment 41, wherein the two or more biomarkers are biomarker quadruples, and wherein one biomarker in the biomarker quadruple is MOG.

[0676] 53. The method according to embodiment 52, wherein the biomarker quadruple is selected from any one of the following: A) MOG, CXCL9, APLP1 and TNFSF13B, B) MOG, CXCL9, OPG and TNFSF13B, and C) MOG, CXCL9, OPG and CNTN2.

[0677] 54. The method according to any one of embodiments 46 - 53, wherein the multiple sclerosis activity is the presence or absence of multiple sclerosis.

[0678] 55. The method according to embodiment 37, wherein the two or more biomarkers are a pair of biomarkers selected from any one of the following: A) NEFL and TNFSF13B, B) NEFL and SERPINA9, and C) NEFL and GH.

[0679] 56. The method according to embodiment 37, wherein the two or more biomarkers are a triple of biomarkers selected from any one of the following: A) NEFL, SERPINA9, and TNFSF13B, B) NEFL, CNTN2, and TNFSF13B, and C) NEFL, APLP1, and TNFSF13B.

[0680] 57. The method according to embodiment 37, wherein the two or more biomarkers are a quadruple of biomarkers selected from any one of the following: A) NEFL, CCL20, SERPINA9, and TNFSF13B, B) NEFL, APLP1, SERPINA9, and TNFSF13B, and C) NEFL, CCL20, APLP1, and TNFSF13B.

[0681] 58. The method according to embodiment 41, wherein the two or more biomarkers are a pair of biomarkers selected from any one of the following: A) MOG and TNFSF13B, B) MOG and CXCL9, and C) MOG and IL - 12B.

[0682] 59. The method according to embodiment 41, wherein the two or more biomarkers are a triple of biomarkers selected from any one of the following: A) MOG, CXCL9, and TNFSF13B, B) MOG, SERPINA9, and TNFSF13B, and C) MOG, OPG, and TNFSF13B.

[0683] 60. The method according to embodiment 41, wherein the two or more biomarkers are a quadruple of biomarkers selected from any one of the following: A) MOG, CXCL9, OPG, and TNFSF13B, B) MOG, CXCL9, OPG, SERPINA9, and TNFSF13B, C) MOG, CXCL9, SERPINA9, and TNFSF13B.

[0684] 61. The method according to any one of embodiments 55 - 60, wherein the multiple sclerosis activity is the disease severity based on the predicted number of gadolinium - enhancing lesions.

[0685] 62. The method according to embodiment 35 or 36, wherein the two or more biomarkers are biomarker pairs selected from any one of the following: A) GFAP and MOG, B) GFAP and NEFL, C) APLP1 and GFAP, D) NEFL and MOG, or E) CXCL9 and OPG.

[0686] 63. The method according to embodiment 35 or 36, wherein the two or more biomarkers are biomarker triples selected from any one of the following: A) GFAP, NEFL and MOG, B) NEFL, MOG and GH, C) NEFL, MOG and SERPINA9, D) CXCL9, OPG and SERPINA9, or E) CXCL9, OPG and TNFRSF10A.

[0687] 64. The method according to embodiment 35 or 36, wherein the two or more biomarkers are biomarker triples selected from any one of the following: A) GFAP, NEFL, MOG and IL12B, B) GFAP, NEFL, MOG and PRTG, C) GFAP, NEFL, MOG and APLP1, D) NEFL, MOG, GH and SERPINA9, E) NEFL, MOG, GH and TNFRSF10A, F) MOG, CXCL9, OPG and SERPINA9, or G) CD6, IL12B, APLP1 and CCL20.

[0688] 65. The method according to any one of embodiments 62 - 64, wherein the multiple sclerosis activity is disease progression.

[0689] 66. The method according to any one of embodiments 1 - 19 and 26 - 65, further comprising classifying the subject based on the prediction.

[0690] 67. The method according to any one of embodiments 1 - 19 and 26 - 65, wherein generating the prediction of multiple sclerosis disease activity comprises comparing the score output by the prediction model with a reference score.

[0691] 68. The method according to embodiment 67, wherein the reference score corresponds to any one of the following:

[0692] A) a patient at a baseline time point, at which the patient does not exhibit disease activity,

[0693] B) patients clinically diagnosed as having no disease activity, or

[0694] C) healthy patients.

[0695] 69. The method according to any one of embodiments 1-19 and 26-68, wherein the expression levels of the plurality of biomarkers are determined from a test sample obtained from the subject.

[0696] 70. The method according to embodiment 69, wherein the test sample is a blood or serum sample.

[0697] 71. The method according to embodiment 69 or 70, wherein the subject has multiple sclerosis, is suspected of having multiple sclerosis, or has previously been diagnosed with multiple sclerosis.

[0698] 72. The method according to any one of embodiments 1-19 and 26-71, wherein obtaining or having obtained the data set comprises performing an immunoassay to determine the expression levels of the plurality of biomarkers.

[0699] 73. The method according to embodiment 72, wherein the immunoassay is a proximity extension assay (PEA) or a LUMINEX xMAP multiplex assay.

[0700] 74. The method according to embodiment 72 or 73, wherein performing the immunoassay comprises contacting the test sample with a plurality of reagents comprising antibodies.

[0701] 75. The method according to embodiment 74, wherein the antibodies comprise one of monoclonal antibodies and polyclonal antibodies.

[0702] 76. The method according to embodiment 74, wherein the antibodies comprise both monoclonal antibodies and polyclonal antibodies.

[0703] 77. The method according to any one of embodiments 1-76, further comprising:

[0704] selecting a therapy for administration to the subject based on the prediction of multiple sclerosis disease activity.

[0705] 78. The method according to any one of embodiments 1-76, further comprising:

[0706] determining the therapeutic efficacy of a therapy previously administered to the subject based on the prediction of multiple sclerosis disease activity.

[0707] 79. The method according to embodiment 78, wherein determining the therapeutic efficacy of the therapy comprises comparing the prediction with a previous prediction determined for the subject at a previous time point.

[0708] 80. The method according to embodiment 79, wherein determining the therapeutic efficacy of the therapy comprises determining that the therapy exhibits efficacy in response to a difference between the prediction and the previous prediction.

[0709] 81. The method according to embodiment 79, wherein determining the therapeutic efficacy of the therapy comprises determining that the therapy lacks efficacy in response to a lack of difference between the prediction and the previous prediction.

[0710] 82. The method according to any one of embodiments 1-76, further comprising:

[0711] Determining a differential diagnosis of multiple sclerosis based on the prediction of multiple sclerosis disease activity.

[0712] 83. The method according to embodiment 82, wherein the differential diagnosis of multiple sclerosis comprises at least one of the following: relapsing-remitting multiple sclerosis (RRMS), secondary progressive multiple sclerosis (SPMS), primary progressive multiple sclerosis (PPMS), progressive relapsing multiple sclerosis (PRMS), and clinically isolated syndrome (CIS).

[0713] 84. A non-transitory computer-readable medium for determining multiple sclerosis activity in a subject, the non-transitory computer-readable medium comprising instructions that, when executed by a processor, cause the processor to perform the following steps:

[0714] Obtaining a data set comprising expression levels of a plurality of biomarkers, wherein the plurality of biomarkers comprises each biomarker selected from at least one of group 1, group 2, and group 3,

[0715] wherein group 1 comprises biomarker 1, biomarker 2, biomarker 3, biomarker 4, biomarker 5, biomarker 6, biomarker 7, and biomarker 8,

[0716] wherein biomarker 1 is NEFL, MOG, CADM3, or GFAP,

[0717] wherein biomarker 2 is MOG, CADM3, KLK6, BCAN, OMG, or GFAP,

[0718] wherein biomarker 3 is CD6, CD5, CRTAM, CD244, or TNFRSF9,

[0719] wherein biomarker 4 is CXCL9, CXCL10, IL-12B, CXCL11, or GFAP,

[0720] where biomarker 5 is OPG, TFF3, or ENPP2,

[0721] where biomarker 6 is OPN, OMD, MEPE, or GFAP,

[0722] where biomarker 7 is CXCL13, NOS3, or MMP-2, and

[0723] where biomarker 8 is GFAP, NEFL, OPN, CXCL9, MOG, or CHI3L1, and

[0724] where Group 2 includes biomarker 9, biomarker 10, biomarker 11, biomarker 12, biomarker 13, biomarker 14, biomarker 15, biomarker 16, and biomarker 17,

[0725] where biomarker 9 is CDCP1, IL-18BP, IL-18, GFAP, or MSR1,

[0726] where biomarker 10 is CCL20, CCL3, or TWEAK,

[0727] where biomarker 11 is IL-12B, IL12A, or CXCL9,

[0728] where biomarker 12 is APLP1, SEZ6L, BCAN, DPP6, NCAN, or KLK6,

[0729] where biomarker 13 is TNFRSF10A, TNFRSF11A, SPON2, CHI3L1, or IFI30,

[0730] where biomarker 14 is COL4A1, IL6, NOTCH3, or PCDH17,

[0731] where biomarker 15 is SERPINA9, TNFRSF9, or CNTN4,

[0732] where biomarker 16 is FLRT2, DDR1, NTRK2, CDH6, MMP-2, and

[0733] where biomarker 17 is TNFSF13B, CXCL16, ALCAM, or IL-18,

[0734] where Group 3 includes biomarker 18, biomarker 19, biomarker 20, and biomarker 21,

[0735] where biomarker 18 is GH, GH2, or IGFBP-1,

[0736] wherein biomarker 19 is VCAN, TINAGL1, CANT1, NECTIN2, MMP-9, or NPDC1,

[0737] wherein biomarker 20 is PRTG, NTRK2, NTRK3, or CNTN4, and

[0738] wherein biomarker 21 is CNTN2, DPP6, GDNFR-α-3, or SCARF2; and

[0739] a prediction of multiple sclerosis disease activity is generated by applying the prediction model to the expression levels of the plurality of biomarkers.

[0740] 85. The non-transitory computer-readable medium according to embodiment 84, wherein the plurality of biomarkers includes each biomarker in group 1, wherein biomarker 1 is NEFL, wherein biomarker 2 is MOG, wherein biomarker 3 is CD6, wherein biomarker 4 is CXCL9, wherein biomarker 5 is OPG, wherein biomarker 6 is OPN, wherein biomarker 7 is CXCL13, and wherein biomarker 8 is GFAP.

[0741] 86. The non-transitory computer-readable medium according to embodiment 84 or 85, wherein the performance of the prediction model is characterized by an area under the curve (AUC) in the range of 0.667 to 0.869.

[0742] 87. The non-transitory computer-readable medium according to embodiment 84 or 85, wherein the performance of the prediction model is characterized by a positive predictive value in the range of 0.617 to 0.861.

[0743] 88. The non-transitory computer-readable medium according to embodiment 84, wherein the plurality of biomarkers includes each biomarker in group 2, wherein biomarker 9 is CDCP1, wherein biomarker 10 is CCL20, wherein biomarker 11 is IL-12B, wherein biomarker 12 is APLP1, wherein biomarker 13 is TNFRSF10A, wherein biomarker 14 is COL4A1, wherein biomarker 15 is SERPINA9, wherein biomarker 16 is FLRT2, and wherein biomarker 17 is TNFSF13B.

[0744] 89. The non-transitory computer-readable medium according to embodiment 88, wherein the performance of the prediction model is characterized by an area under the curve (AUC) in the range of 0.595 to 0.761.

[0745] 90. The non-transitory computer-readable medium as described in embodiment 88, wherein the performance of the prediction model is characterized in that the positive predictive value ranges from 0.523 to 0.769.

[0746] 91. The non-transitory computer-readable medium as described in embodiment 85, wherein the plurality of biomarkers further includes each biomarker in group 2, wherein biomarker 9 is CDCP1, wherein biomarker 10 is CCL20, wherein biomarker 11 is IL-12B, wherein biomarker 12 is APLP1, wherein biomarker 13 is TNFRSF10A, wherein biomarker 14 is COL4A1, wherein biomarker 15 is SERPINA9, wherein biomarker 16 is FLRT2, and wherein biomarker 17 is TNFSF13B.

[0747] 92. The non-transitory computer-readable medium as described in embodiment 91, wherein the performance of the prediction model is characterized in that the area under the curve (AUC) ranges from 0.693 to 0.892.

[0748] 93. The non-transitory computer-readable medium as described in embodiment 91, wherein the performance of the prediction model is characterized in that the positive predictive value ranges from 0.613 to 0.843.

[0749] 94. The non-transitory computer-readable medium as described in embodiment 84, wherein the plurality of biomarkers includes each biomarker in group 3, wherein biomarker 18 is GH, wherein biomarker 19 is VCAN, wherein biomarker 20 is PRTG, and wherein biomarker 21 is CNTN2.

[0750] 95. The non-transitory computer-readable medium as described in embodiment 94, wherein the performance of the prediction model is characterized in that the area under the curve (AUC) ranges from 0.566 to 0.644.

[0751] 96. The non-transitory computer-readable medium as described in embodiment 94, wherein the performance of the prediction...

Claims

1. A method for determining multiple sclerosis activity in a subject, the method comprising: obtaining or having obtained a data set comprising expression levels of a plurality of biomarkers, wherein each biomarker of the plurality of biomarkers is selected from at least one of Group 1, Group 2, and Group 3, wherein Group 1 comprises Biomarker 1, Biomarker 2, Biomarker 3, Biomarker 4, Biomarker 5, Biomarker 6, Biomarker 7, and Biomarker 8, wherein Biomarker 1 is NEFL, MOG, CADM3, or GFAP, wherein Biomarker 2 is MOG, CADM3, KLK6, BCAN, OMG, or GFAP, wherein Biomarker 3 is CD6, CD5, CRTAM, CD244, or TNFRSF9, wherein Biomarker 4 is CXCL9, CXCL10, IL-12B, CXCL11, or GFAP, wherein Biomarker 5 is OPG, TFF3, or ENPP2, wherein Biomarker 6 is OPN, OMD, MEPE, or GFAP, wherein Biomarker 7 is CXCL13, NOS3, or MMP-2, and wherein Biomarker 8 is GFAP, NEFL, OPN, CXCL9, MOG, or CHI3L1, and wherein Group 2 comprises Biomarker 9, Biomarker 10, Biomarker 11, Biomarker 12, Biomarker 13, Biomarker 14, Biomarker 15, Biomarker 16, and Biomarker 17, wherein Biomarker 9 is CDCP1, IL-18BP, IL-18, GFAP, or MSR1, wherein Biomarker 10 is CCL20, CCL3, or TWEAK, wherein Biomarker 11 is IL-12B, IL12A, or CXCL9, wherein Biomarker 12 is APLP1, SEZ6L, BCAN, DPP6, NCAN, or KLK6, wherein Biomarker 13 is TNFRSF10A, TNFRSF11A, SPON2, CHI3L1, or IFI30, wherein Biomarker 14 is COL4A1, IL6, NOTCH3, or PCDH17, wherein Biomarker 15 is SERPINA9, TNFRSF9, or CNTN4, wherein Biomarker 16 is FLRT2, DDR1, NTRK2, CDH6, MMP-2, and wherein Biomarker 17 is TNFSF13B, CXCL16, ALCAM, or IL-18, wherein Group 3 comprises Biomarker 18, Biomarker 19, Biomarker 20, and Biomarker 21, wherein Biomarker 18 is GH, GH2, or IGFBP-1, wherein Biomarker 19 is VCAN, TINAGL1, CANT1, NECTIN2, MMP-9, or NPDC1, wherein Biomarker 20 is PRTG, NTRK2, NTRK3, or CNTN4, and wherein biomarker 21 is CNTN2, DPP6, GDNFR-α-3, or SCARF2; and generating a prediction of multiple sclerosis disease activity by applying the prediction model to the expression levels of the plurality of biomarkers.

2. The method according to claim 1, wherein the plurality of biomarkers includes each biomarker in Group 1, wherein biomarker 1 is NEFL, wherein biomarker 2 is MOG, wherein biomarker 3 is CD6, wherein biomarker 4 is CXCL9, wherein biomarker 5 is OPG, wherein biomarker 6 is OPN, wherein biomarker 7 is CXCL13, and wherein biomarker 8 is GFAP.

3. The method according to any one of claims 1-2, wherein the performance of the prediction model is characterized by an area under the curve (AUC) in the range of 0.667 to 0.

869.

4. The method according to any one of claims 1-2, wherein the performance of the prediction model is characterized by a positive predictive value in the range of 0.617 to 0.

861.

5. The method according to claim 1, wherein the plurality of biomarkers includes each biomarker in Group 2, wherein biomarker 9 is CDCP1, wherein biomarker 10 is CCL20, wherein biomarker 11 is IL-12B, wherein biomarker 12 is APLP1, wherein biomarker 13 is TNFRSF10A, wherein biomarker 14 is COL4A1, wherein biomarker 15 is SERPINA9, wherein biomarker 16 is FLRT2, and wherein biomarker 17 is TNFSF13B.

6. The method according to claim 5, wherein the performance of the prediction model is characterized by an area under the curve (AUC) in the range of 0.595 to 0.

761.

7. The method according to claim 5, wherein the performance of the prediction model is characterized by a positive predictive value in the range of 0.523 to 0.

769.

8. The method according to claim 2, wherein the plurality of biomarkers further includes each biomarker in Group 2, wherein biomarker 9 is CDCP1, wherein biomarker 10 is CCL20, wherein biomarker 11 is IL-12B, wherein biomarker 12 is APLP1, wherein biomarker 13 is TNFRSF10A, wherein biomarker 14 is COL4A1, wherein biomarker 15 is SERPINA9, wherein biomarker 16 is FLRT2, and wherein biomarker 17 is TNFSF13B.

9. The method according to claim 8, wherein the performance of the prediction model is characterized by an area under the curve (AUC) in the range of 0.693 to 0.

892.

10. The method according to claim 8, wherein the performance of the prediction model is characterized by a positive predictive value in the range of 0.613 to 0.843.

Images