Biomarker and method for determination, monitoring, stratification, and prediction and determination of disease-modifying drug therapeutic effect in amyotrophic lateral sclerosis

Abstract

Provided is a biomarker for determination, monitoring, stratification, and prediction and determination of disease-modifying drug therapeutic effect in amyotrophic lateral sclerosis (ALS), the biomarker being (1) one or more first proteins selected from a first protein group having increased expression in ALS when compared to a control group, and / or (2) one or more second proteins selected from a second protein group having decreased expression in ALS when compared to a control group.

Description

Biomarkers and methods for assessing, predicting and assessing the efficacy of disease-modifying drugs, monitoring, and stratifying in amyotrophic lateral sclerosis

[0001] The present invention relates to biomarkers and methods for the diagnosis, prediction and assessment of disease-modifying drug treatment efficacy, monitoring, and stratification in amyotrophic lateral sclerosis.

[0002] Amyotrophic lateral sclerosis (ALS) is a fatal neurodegenerative disease characterized by progressive motor neuron (MN) degeneration and progressive muscle weakness. 90% of cases are sporadic, making diagnosis difficult. ALS diagnosis is based on pathological findings, such as abnormal accumulation of TDP-43 in motor neurons in the brain and / or spinal cord, combined with medical interviews, visual examination, palpation, electromyography, and imaging tests such as MRI. Early diagnostic methods have yet to be developed.

[0003] Previously, attempts have been made to identify protein biomarkers for diagnosing amyotrophic lateral sclerosis (ALS) by studying extracellular vesicles (EVs) (see Non-Patent Documents 1 and 2).

[0004] Barbo M, Ravnik-Glavac M. Genes (Basel) 2023;14. https: / / doi.org / 10.3390 / genes14020325.Gagliardi D et al., Cell Mol Life Sci 2021;78:561-72. https: / / doi.org / 10.1007 / s00018-020-03619-3.

[0005] However, longitudinal and comprehensive analysis of extracellular vesicle (EV) proteins, along with biomarkers with high diagnostic accuracy for ALS, are still needed.

[0006] The present invention aims to solve the above-mentioned conventional problems and achieve the following objectives: Namely, the present invention aims to provide biomarkers and methods for diagnosing amyotrophic lateral sclerosis, predicting and diagnosing the therapeutic effect of disease-modifying drugs, monitoring, and stratifying amyotrophic lateral sclerosis.

[0007] The means for solving the above problems are as follows: <1> Biomarkers for diagnosis, prediction and diagnosis of disease-modifying drug treatment effect, monitoring, and stratification in amyotrophic lateral sclerosis, including: (1) B4GALT1, C1RL, C4A, C4B, C8B, C9, CD14, CFH, COL6A3, FBLN1, PROS1, SDCBP, SERPINA1, SERPINA3, SERPINA5, SERPINA7, SERPINC1, SERPIND1, SERPINF2, SERPING1, APOL1, CNDP1, EFEMP1, ISLR, MINPP1, OAF, PGLYRP2, QSOX1, SVEP1, TGFBR3, and TTR; OGN, FETUB, VASN, SELENOP, and CNDP1; and FRMPD1, COL4A2, TRAP1, PDGFRA, and TGM3, and / or (2) one or more first proteins selected from a first protein group consisting of ACTN4, ARPC2, ARPC4, CFL1, HSP90AA1, HSP90AB1, HSPA1A, MAP2K2, PFN1, ACTN1, SPTBN1, TXN, CALR, DNAJC3, GANAB, HSPA5, HSPA8, HSPD1, LMAN2, NECTIN2, P4HB, PDIA3, UGGT1, AARS1, ARHGDIA, CBLN4, CETP, ENO1, GNB1, GSTP1, IGFALS, PCYOX1, PGAM1, PKM, RAB14, SCPEP1, SLITRK4, and YWHAZ; RPS4X, CCDC144A, BLMH, LCN2, and S100A9; and RAB14, A biomarker characterized by being one or more second proteins selected from a second protein group consisting of ATP6V1B2, ACLY, USP14, and TIMP2. <2> A method for diagnosing amyotrophic lateral sclerosis, predicting and diagnosing the therapeutic effect of a disease-modifying drug, monitoring, and stratifying the disease, comprising: a detection amount D of the biomarker according to <1> in a sample obtained from a subject; ｓand obtaining a detected amount D of the biomarker in a sample obtained from a control group. ｃ (1) the biomarker is the first protein, and D ｓ >D ｃ and / or (2) the biomarker is the second protein, and D ｓ <D ｃ and determining that the patient is at risk of amyotrophic lateral sclerosis when the level of the detected amount D of the biomarker is 0. <3> The method according to <2> above, wherein the sample is at least one of blood, serum, plasma, cerebrospinal fluid, saliva, and urine. <4> The method according to <2> above, wherein the detected amount D of the biomarker is 0. ｓ and D ｃ <5> The method according to <3>, wherein the detected amount D of the biomarker in the sample obtained from the subject is ｓ０ and the detected amount D of the biomarker in the sample obtained from the subject after a certain time has elapsed. ｓ１ (1) the biomarker is the first protein; and (2) comparing ｓ０ >D ｓ１ When the above condition is met, the patient falls into one of the following stratification groups: high response to disease-modifying drug treatment, high effect of disease-modifying drug treatment, no progression of disease or slow progression of disease; D ｓ０ =D ｓ１ If the condition is as above, there is no change in the course of the disease, or ｓ０ <D ｓ１ or the patient is in any of the stratified groups of rapid progression of the disease, low response to disease-modifying drug treatment, low effect of disease-modifying drug treatment, and progression of the disease; (2) the biomarker is the second protein; ｓ０ <D ｓ１ When the above condition is met, the patient falls into one of the following stratification groups: high response to disease-modifying drug treatment, high effect of disease-modifying drug treatment, no progression of disease or slow progression of disease; D ｓ０ =D ｓ１ If the condition is as above, there is no change in the course of the disease, or ｓ０ >Dｓ１ and determining, when the result is positive, that the subject belongs to any of the stratification groups of low responsiveness to disease-modifying drug treatment, low disease-modifying drug treatment effect, advanced disease state, or rapid disease progression. <6> The method according to <5>, further comprising the step of administering a medication to the subject during the certain period of time. <7> The method according to <6>, wherein the medication comprises administering one or more medications selected from the group consisting of ropinirole hydrochloride, riluzole, edaravone, methylcobalamin, bosutinib, hydrochlorothiazide, cortisone acetate, aldioxa, sibutramine hydrochloride, alendronate sodium hydrate, ceftibuten, etropibate, and ketorolac tromethamine. <8> A device for assessment, disease-modifying drug treatment effect prediction and assessment, monitoring, and stratification in amyotrophic lateral sclerosis, comprising: a detection amount D of the biomarker according to <1> in a sample obtained from a subject ｓ a data acquisition unit that acquires the detected amount D of the biomarker in the sample obtained from the control group; ｃ (1) the biomarker is the first protein, and D ｓ >D ｃ and / or (2) the biomarker is the second protein, and D ｓ <D ｃ <9> A program for diagnosis of amyotrophic lateral sclerosis, prediction and diagnosis of disease modifying drug treatment effect, monitoring, and stratification, comprising: a determination unit that determines that the subject is at risk of amyotrophic lateral sclerosis when the subject is at risk of amyotrophic lateral sclerosis, and a display unit that displays the subject's risk of amyotrophic lateral sclerosis. ｓ a data acquisition unit that acquires the detected amount D of the biomarker in the sample obtained from the control group; ｃ (1) the biomarker is the first protein, and D ｓ >D ｃ and / or (2) the biomarker is the second protein, and D ｓ <Dｃ and a display unit that displays the risk of amyotrophic lateral sclerosis for the subject. <10> A method for the diagnosis of amyotrophic lateral sclerosis, prediction and diagnosis of disease-modifying drug treatment effect, monitoring, and stratification, comprising: a step of acquiring extracellular vesicle protein analysis data of a sample obtained from the subject; a step of inputting the extracellular vesicle protein analysis data of the subject into a determination model that determines the risk of amyotrophic lateral sclerosis for a subject group from the extracellular vesicle protein analysis data of the subject, and outputting the risk of amyotrophic lateral sclerosis for the subject; and a step of displaying the risk of amyotrophic lateral sclerosis for the subject. <11> An apparatus for diagnosing amyotrophic lateral sclerosis, predicting and diagnosing the therapeutic effect of disease-modifying drugs, monitoring, and stratifying the disease, comprising: a data acquisition unit that acquires extracellular vesicle protein analysis data of a sample obtained from a subject; a determination unit that inputs the extracellular vesicle protein analysis data of the subject into a determination model that diagnosing the risk of amyotrophic lateral sclerosis for a subject group from the extracellular vesicle protein analysis data of the subject, and outputs the risk of the subject having amyotrophic lateral sclerosis; and a display unit that displays the risk of the subject having amyotrophic lateral sclerosis. <12> A program for the assessment of amyotrophic lateral sclerosis, prediction and assessment of the therapeutic effect of disease-modifying drugs, monitoring, and stratification, which causes an apparatus to function as: a data acquisition unit that acquires extracellular vesicle protein analysis data of a sample obtained from a subject; a determination unit that inputs the extracellular vesicle protein analysis data of a subject into a determination model that assesses the risk of a subject group having amyotrophic lateral sclerosis from the extracellular vesicle protein analysis data of the subject, and outputs the risk of the subject having amyotrophic lateral sclerosis; and a display unit that displays the risk of the subject having amyotrophic lateral sclerosis.

[0008] According to the present invention, the above-mentioned conventional problems can be solved and the above-mentioned object can be achieved, and biomarkers and methods for the diagnosis of amyotrophic lateral sclerosis, prediction and diagnosis of the therapeutic effect of disease-modifying drugs, monitoring, and stratification can be provided.

[0009] Figure 1A shows an overview of the ROPALS study and the timing of serum and CSF collection. Figure 1B shows the study design. Figure 1C shows a comparison of protein profiles between sEVs and cEVs derived from controls and SALS patients. Figure 1D shows the results of UMAP analysis of all EVs, sEVs, and cEVs. Figure 1E shows a correlation plot and Pearson correlation analysis of proteins detected in both sEVs and cEVs. Figure 1F shows the results of correlation analysis of protein profiles detected in both sEVs and cEVs collected from the same subject at the same time. Figure 2A shows a volcano diagram comparing EV proteins in serum from a control group and a ROPI-untreated SALS patient. Figure 2B shows the results of GO and KEGG pathway analysis of DAPs (BP, CC, MF) in sEVs. Figure 2C shows a volcano diagram comparing EV proteins in CSF from a control group and a ROPI-untreated SALS patient. Figure 2D shows the results of GO and KEGG pathway analysis of DAPs (BP, CC, MF) in cEVs. Figure 2E shows a comparison of DAPs between sEVs and cEVs, a list of proteins commonly decreased / increased in sEVs and cEVs of SALS patients compared with controls, and the results of GO (BP) and KEGG pathway clustering analysis. Figure 2F shows a schematic diagram of the disease progression modeling method and aALSFRS-R calculation. Figure 2G shows a Pearson correlation analysis between the mean values ​​of proteins in the complement coagulation regulation group, actin polymerization regulation group, and unfolded protein processing group and the aALSFRS-R. Figure 3A shows a Cohen's d rank plot comparing the log2 (fold change) of each sEV protein between the ROPI and placebo groups for each patient at weeks 0 and 24. Figure 3B shows the results of GO (BP, CC, MF) and KEGG pathway analysis of es-DAPs in sEVs (placebo 0-24w vs. ROPI 0-24w). Figure 3C is a Venn diagram comparing es-DAP (placebo 0-24w vs. ROPI 0-24w) and DAP (control samples vs. SALS patient samples) in sEV.Figure 3D shows the results of GO (BP, CC, MF) and KEGG pathway analysis for the proteins in SALS patient samples and sEVs that were increased / decreased by ROPI treatment. Figure 3E shows Cohen's d rank plots comparing the log2 (fold change) between the ROPI and placebo groups for each patient and each protein in cEVs at weeks 0 and 24. Figure 3F shows the results of GO (BP, CC, MF) and KEGG pathway analysis for es-DAPs in cEVs (placebo 0-24w vs. ROPI 0-24w). Figure 3G shows a Venn diagram comparing es-DAPs in cEVs (placebo 0-24w vs. ROPI 0-24w) with DAPs (control samples vs. SALS patient samples). Figure 3H shows the results of GO (BP, CC, MF) and KEGG pathway analysis for the proteins in SALS patient samples and cEVs that were increased / decreased by ROPI treatment. Figure 4A shows a schematic diagram of a time-series analysis using a clustering method to examine the temporal changes in sEV and cEV proteins in SALS patients who participated in the ROPALS study and whose samples were available up to 48 weeks after treatment. Figure 4B shows plots of the log2 (fold change) changes of each protein in sEV from 0 weeks to each sampling time in patients in the placebo and ROPI groups. Figure 4C shows GO (BP, CC, MF) and KEGG pathway analysis results for increased and decreased proteins based on the log2 (fold change) clustering results of sEV proteins in the placebo group. Figure 4D shows plots of the log2 (fold change) changes of each protein in cEV from 0 weeks to each sampling time in patients in the placebo and ROPI groups. Figure 4E shows GO (BP, CC, MF) and KEGG pathway analysis results for increased and decreased proteins based on the log2 (fold change) clustering results of cEV proteins in the placebo group.Figure 5A is a heat map showing the results of cosine similarity analysis of proteins in sEVs (DAP (controls vs. SALS patients), es-DAP (placebo 0-24w vs. ROPI 024w), and proteins that changed over time. Figure 5B is a bar graph showing the GO (BP, CC, MF) and KEGG pathway results of proteins that increased / decreased or increased / decreased over time in sEVs of SALS patients and decreased / increased by ROPI. Figure 5C is a heat map showing the results of cosine similarity analysis of proteins in cEVs (controls vs. SALS patients), es-DAP (placebo 0-24w vs. ROPI 024w), and proteins that changed over time. Figure 5D is a bar graph showing the GO (BP, CC, MF) and KEGG pathway results of proteins that increased / decreased or increased / decreased over time in cEVs of SALS patients and decreased / increased by ROPI. Figure 5E is a scatter plot showing the difference from 0w in protein levels that increased or increased over time in SALS patients on the vertical axis and the difference from 0w in protein levels that decreased or decreased over time in SALS patients on the horizontal axis. Figure 5F is a schematic diagram of the protein composition in control EVs and SALS EVs and the changes due to ROPI. Figure 6A is a diagram showing the results of PCA analysis of the iPast transcriptome with and without ROPI treatment. Figure 6B is a diagram showing dopamine receptor expression in iPast with and without ROPI treatment. Figure 6C is a volcano diagram showing changes in iPast with and without ROPI treatment. Figure 6D is a heat map showing the expression levels of DEGs identified in the comparative analysis of iPast and iPSC-MN with and without ROPI treatment. Figure 6E is a diagram showing the results of PCA analysis of the iPSC-MN transcriptome with and without ROPI treatment. Figure 6F is a diagram showing dopamine receptor expression in iPSC-MN with and without ROPI treatment. Figure 6G is a volcano diagram showing changes in iPSC-MN with and without ROPI treatment. Figure 6H shows a heatmap of the expression levels of DEGs identified by comparative analysis of iPast and iPSC-MN treated with or without ROPI. Figure 6I shows a schematic diagram of the changes in DRD2 expression and ROPI-induced expression in iPast and iPSC-MN.Figure 7A shows the aALSFRS-R calculated by applying the approximation formula to the trends in the ALSFRS-R scores of 20 SALS patients enrolled in the ROPALS study. Figure 7B shows a rank plot showing the results of a correlation analysis between the abundance of each protein in sEV at 0w and the subsequent rate of adaptive progression (normalized a-value). Figure 7C shows a table of the top five positive and negative proteins in sEV. Figure 7D shows a scatter plot showing the proteins with the highest predictive accuracy in sEV compared with different indices. Figure 7E shows a rank plot showing the results of a correlation analysis between the abundance of each protein in cEV at 0w and the subsequent rate of adaptive progression (normalized a-value). Figure 7F shows a table of the top five positive and negative proteins in cEV. Figure 7G shows a scatter plot showing the proteins with the highest predictive accuracy in cEV compared with different indices. Figure 8A shows a schematic diagram of the learning framework for a limited sample size and an unbalanced dataset. Figure 8B shows a table and heatmap showing the accuracy of the trained model according to the protein composition in sEV. FIG. 8C is a table and heat map showing the accuracy of the trained model according to the protein composition in cEV. FIG. 9 is a diagram showing the overall configuration of the device of the first embodiment. FIG. 10 is a functional block diagram of the device of the first embodiment. FIG. 11 is a flowchart of one aspect of the method of the first embodiment. FIG. 12 is a hardware configuration diagram of the device of the first embodiment. FIG. 13 is a diagram showing the overall configuration of the device of the second embodiment. FIG. 14 is a functional block diagram of a learning device of the device of the second embodiment. FIG. 15 is a functional block diagram of a determination device of the device of the second embodiment. FIG. 16 is a flowchart of the learning process of the method of the second embodiment. FIG. 17 is a flowchart of the determination process of the method of the second embodiment.

[0010] (Biomarkers) The biomarkers of this embodiment are biomarkers for diagnosis of amyotrophic lateral sclerosis, prediction and diagnosis of the therapeutic effect of disease-modifying drugs, monitoring, and stratification, and include: (1) B4GALT1, C1RL, C4A, C4B, C8B, C9, CD14, CFH, COL6A3, FBLN1, PROS1, SDCBP, SERPINA1, SERPINA3, SERPINA5, SERPINA7, SERPINC1, SERPIND1, SERPINF2, SERPING1, APOL1, CNDP1, EFEMP1, ISLR, MINPP1, OAF, PGLYRP2, QSOX1, SVEP1, TGFBR3, and TTR; OGN, FETUB, VASN, SELENOP, and CNDP1; and FRMPD1, COL4A2, TRAP1, PDGFRA, and TGM3, and / or (2) one or more first proteins selected from a first protein group consisting of ACTN4, ARPC2, ARPC4, CFL1, HSP90AA1, HSP90AB1, HSPA1A, MAP2K2, PFN1, ACTN1, SPTBN1, TXN, CALR, DNAJC3, GANAB, HSPA5, HSPA8, HSPD1, LMAN2, NECTIN2, P4HB, PDIA3, UGGT1, AARS1, ARHGDIA, CBLN4, CETP, ENO1, GNB1, GSTP1, IGFALS, PCYOX1, PGAM1, PKM, RAB14, SCPEP1, SLITRK4, and YWHAZ; RPS4X, CCDC144A, BLMH, LCN2, and S100A9; and RAB14, The second protein is one or more selected from a second protein group consisting of ATP6V1B2, ACLY, USP14, and TIMP2.

[0011] (Method) The method of the first embodiment is a method for determining amyotrophic lateral sclerosis, predicting and determining the therapeutic effect of a disease-modifying drug, monitoring, and stratifying the disease, and includes detecting a detected amount D of the biomarker of the present embodiment in a sample obtained from a subject. ｓand obtaining a detected amount D of the biomarker in a sample obtained from a control group. ｃ (1) the biomarker is the first protein, and D ｓ >D ｃ and / or (2) the biomarker is the second protein, and D ｓ <D ｃ and determining that there is a risk of amyotrophic lateral sclerosis if

[0012] Amyotrophic lateral sclerosis (ALS) is a fatal neurodegenerative disease characterized by the degeneration and subsequent loss of function of upper and lower motor neurons. The disease progresses rapidly, with a median survival time from onset to death of 20 to 48 months. Approximately 90% of ALS patients present sporadically (sporadic ALS; SALS), highlighting the great heterogeneity of its etiology.

[0013] A key pathological feature of ALS is the nuclear loss of transactive response DNA-binding protein 43 (TDP-43; an RNA-binding protein encoded by TARDBP) and its aggregation in the cytoplasm of motor neurons. This pathological feature is observed in 97% of SALS patients and in FALS cases, including those with TARDBP or C9orf72 (GGGGCC repeat expansion) mutations, indicating a common disease mechanism. Despite these unifying features, the detailed molecular mechanisms of motor neuron degeneration have yet to be fully defined. Furthermore, the lack of reliable biomarkers that can reflect disease pathophysiology and monitor treatment response has hindered the development of effective diagnostic and therapeutic methods. In particular, the inability to assess early molecular changes or track disease progression in vivo poses a major barrier to successful clinical application.

[0014] Extracellular Vesicles (EVs) are ubiquitously secreted by most cell types and have been identified in many body fluids, including blood and cerebrospinal fluid (CSF). EVs encapsulate proteins, nucleic acids, and lipids and function as conduits for cellular communication. Given their lipid bilayer composition and stability in circulation, the potential contribution of EVs to the pathogenesis of diseases such as cancer and neurodegenerative disorders, as well as their potential as biomarkers, has attracted considerable interest. Many studies have attempted to identify ALS biomarkers by studying EVs. However, in this short period of time, there remains a notable lack of longitudinal analysis in these ALS-related EV studies. While the importance of longitudinal analysis for understanding this disease cannot be overemphasized, longitudinal analysis of EV protein composition in ALS has been significantly underexplored.

[0015] Ropinirole hydrochloride (ROPI), a dopamine D2 receptor agonist, was approved for the treatment of Parkinson's disease over 15 years ago and is widely used. Subsequently, ROPI was identified as a potential new drug candidate for the treatment of ALS through drug screening using induced pluripotent stem cell-derived spinal motor neurons (iPSC-MNs) derived from ALS patients. A single-center, randomized, double-blind, placebo-controlled trial, the Ropinirole Hydrochloride Remedy for Amyotrophic Lateral Sclerosis (ROPALS), was conducted and suggested that ROPI may slow disease progression in patients with ALS.

[0016] As a result of extensive investigations to achieve the above object, the present inventors have made the following findings, as demonstrated in the examples below, and have completed the present invention based on these findings.

[0017] To address these unmet needs, we focused on comprehensive analysis of the protein profiles of extracellular vesicles (EVs) isolated from cerebrospinal fluid (CSF) and blood of healthy controls and ALS patients. EVs are lipid bilayer-enclosed nanovesicles secreted by almost all cell types, including neurons, and circulate in biological fluids such as CSF and blood. By transporting proteins, lipids, and nucleic acids, including miRNAs, EVs mediate intercellular communication and have emerged as stable and accessible carriers of disease-related molecular information.

[0018] To facilitate this research, we utilized clinical and cellular resources obtained from the ROPALS trial, an investigator-initiated phase I / IIa clinical trial evaluating the therapeutic potential of ropinirole hydrochloride (ROPI) in patients with amyotrophic lateral sclerosis (ALS). Extracellular vesicles (EVs) hold the potential to elucidate the pathogenesis of amyotrophic lateral sclerosis (ALS) and are useful as biomarkers. In particular, the comparative and time-dependent changes in protein profiles of serum-derived EVs (sEVs) and cerebrospinal fluid (CSF)-derived EVs (cEVs) from sporadic ALS (SALS) patients remain unclear. Ropinirole hydrochloride (ROPI; a dopamine D2 receptor agonist) is a novel anti-ALS drug candidate identified through induced pluripotent stem cell (iPSC)-based drug discovery and has been shown to inhibit ALS disease progression in clinical trials. However, the mechanism of action of ROPI is not fully understood. Therefore, we sought to clarify the time-dependent changes associated with disease progression and the effects of ROPI on EV protein profiles.

[0019] In the Examples described below, serum and CSF samples were collected at regular intervals from 10 control subjects and 20 patients with amyotrophic lateral sclerosis (ROPALS) participating in a ropinirole hydrochloride treatment trial. Comprehensive and quantitative proteomic analysis of EVs extracted from these samples was performed using liquid chromatography / mass spectrometry (LC / MS).

[0020] We found that the protein profiles of sEVs and cEVs were highly consistent, despite significant differences in disease state, time, and ROPI administration. We also identified a first group of proteins elevated in ALS patients and / or a second group of proteins decreased in ALS patients as biomarkers. In SALS, both sEVs and cEVs showed elevated levels of inflammation-related proteins but decreased levels associated with the unfolded protein response (UPR). These results reflected longitudinal changes after disease onset and correlated with the ALS Functional Rating Scale-Revised (ALSFRS-R) at the time of sampling, suggesting a link between SALS onset and progression. ROPI counteracted these changes, attenuating inflammation-related protein levels and increasing UPR-related protein levels in SALS, suggesting an anti-ALS effect on EV protein profiles. Reverse translation studies using iPSC-derived astrocytes indicated that these changes may partly reflect the DRD2-dependent neuroinflammatory inhibitory effect of ROPI. We also identified biomarkers that predict diagnosis and disease progression through machine learning-driven biomarker discovery.

[0021] Despite the limited sample size, this study reports time-series proteomic changes in serum and CSF EVs of SALS patients, providing comprehensive insights into SALS pathogenesis, ROPI-induced changes, and potential prognostic and diagnostic biomarkers.

[0022] According to the biomarker and method of this embodiment, it is possible to provide a biomarker and method for the diagnosis of amyotrophic lateral sclerosis, prediction and diagnosis of the therapeutic effect of disease-modifying drugs, monitoring, and stratification.

[0023] Each step in the biomarker and method of this embodiment will be described below. <Data Acquisition Step> The data acquisition step is a step of acquiring the detected amount D of the biomarker according to claim 1 in a sample obtained from a subject. ｓ This is the process of obtaining the above.

[0024] The biomarkers are selected from the group consisting of: (1) one or more first proteins selected from a first group of proteins consisting of B4GALT1, C1RL, C4A, C4B, C8B, C9, CD14, CFH, COL6A3, FBLN1, PROS1, SDCBP, SERPINA1, SERPINA3, SERPINA5, SERPINA7, SERPINC1, SERPIND1, SERPINF2, SERPING1, APOL1, CNDP1, EFEMP1, ISLR, MINPP1, OAF, PGLYRP2, QSOX1, SVEP1, TGFBR3, and TTR; OGN, FETUB, VASN, SELENOP, and CNDP1; and FRMPD1, COL4A2, TRAP1, PDGFRA, and TGM3; and / or (2) ACTN4, ARPC2, ARPC4, CFL1, The second protein is one or more selected from a second group of proteins consisting of HSP90AA1, HSP90AB1, HSPA1A, MAP2K2, PFN1, ACTN1, SPTBN1, TXN, CALR, DNAJC3, GANAB, HSPA5, HSPA8, HSPD1, LMAN2, NECTIN2, P4HB, PDIA3, UGGT1, AARS1, ARHGDIA, CBLN4, CETP, ENO1, GNB1, GSTP1, IGFALS, PCYOX1, PGAM1, PKM, RAB14, SCPEP1, SLITRK4, and YWHAZ; RPS4X, CCDC144A, BLMH, LCN2, and S100A9; and RAB14, ATP6V1B2, ACLY, USP14, and TIMP2.

[0025] The first proteins classified into the first protein group are proteins whose expression is elevated in ALS patients compared to controls. The first proteins can be classified into the following three groups: (1-1) B4GALT1, C1RL, C4A, C4B, C8B, C9, CD14, CFH, COL6A3, FBLN1, PROS1, SDCBP, SERPINA1, SERPINA3, SERPINA5, SERPINA7, SERPINC1, SERPIND1, SERPINF2, SERPING1, APOL1, CNDP1, EFEMP1, ISLR, MINPP1, OAF, PGLYRP2, QSOX1, SVEP1, TGFBR3, and TTR are proteins whose expression is elevated in ALS patients in both serum and CSF-derived EV proteomes (Figure 2E). (1-2) OGN, FETUB, VASN, SELENOP, and CNDP1 are serum-derived EV proteins associated with rapid disease progression, as measured by the ALS Functional Rating Scale (ALSFRS-R) (Fig. 7B). (1-3) FRMPD1, COL4A2, TRAP1, PDGFRA, and TGM3 are CSF-derived EV proteins associated with rapid disease progression, as measured by the ALS Functional Rating Scale (ALSFRS-R) (Fig. 7E).

[0026] The second protein classified into the second protein group is a protein whose expression is reduced in ALS compared to a control group. The second set of proteins can be classified into the following categories (2-1) to (2-3): (2-1) ACTN4, ARPC2, ARPC4, CFL1, HSP90AA1, HSP90AB1, HSPA1A, MAP2K2, PFN1, ACTN1, SPTBN1, TXN, CALR, DNAJC3, GANAB, HSPA5, HSPA8, HSPD1, LMAN2, NECTIN2, P4HB, PDIA3, UGGT1, AARS1, ARHGDIA, CBLN4, CETP, ENO1, GNB1, GSTP1, IGFALS, PCYOX1, PGAM1, PKM, RAB14, SCPEP1, SLITRK4, and YWHAZ are proteins commonly down-regulated in serum and CSF-derived EVs in ALS (Fig. 2E). (2-2) RPS4X, CCDC144A, BLMH, LCN2, and S100A9 are serum-derived EV proteins associated with slower disease progression, as measured by the ALS Functional Rating Scale (ALSFRS-R) (Fig. 7B). (3-3) RAB14, ATP6V1B2, ACLY, USP14, and TIMP2 are cerebrospinal fluid-derived EV proteins associated with slower disease progression, as measured by the ALS Functional Rating Scale (ALSFRS-R) (Fig. 7E).

[0027] The first protein classified into the first protein group and the second protein classified into the second protein group are both publicly known proteins that are published in protein databases such as "SwissProt" (URL: https: / / www.expasy.org / resources / uniprotkb-swiss-prot) and "UniProt" (URL: https: / / www.uniprot.org / ).

[0028] Among these biomarkers, DNAJC3 and HSPA5 are preferred because there is a large and / or significant difference in the expression levels of the biomarkers between control and ALS subjects. Furthermore, it is preferred to use multiple biomarkers to improve accuracy, with 10 or more biomarkers being preferred, 20 or more being more preferred, and 50 or more being even more preferred.

[0029] The subject is generally a mammal, such as a human, non-human primate, dog, cat, mouse, rat, cow, horse, or pig. Among these, a human is preferred. The subject may be any of adults, infants, children, and elderly subjects, and is preferably a subject suspected of having amyotrophic lateral sclerosis (ALS), a subject diagnosed or confirmed to have a disease associated with ALS, or a subject undergoing therapeutic intervention for a disease associated with ALS.

[0030] The specimen is not particularly limited and can be appropriately selected depending on the purpose, but is preferably a biological fluid, and more preferably at least one of blood, serum, plasma, cerebrospinal fluid, saliva, and urine.

[0031] <<Detection Amount of Biomarker>> The detection amount of the biomarker is preferably the detection amount of the biomarker in proteins derived from extracellular vesicles (EVs) in the biological fluid. The method for detecting the detection amount of the biomarker is not particularly limited, and any known method can be appropriately selected depending on the purpose. Examples include a method for quantifying a target biomarker and a relative quantification method using proteome analysis.

[0032] Specifically, the isolation and extraction of proteins from extracellular vesicles (EVs) can be performed as follows. First, the sample's extracellular vesicles (EVs) are collected. Proteins in the collected EVs are reduced with 10 mM TCEP at 100°C for 10 minutes, alkylated with 50 mM iodoacetamide at room temperature for 45 minutes, and then digested on beads with Trypsin / Lys-C Mix (Promega) at 37°C for 12 hours. The resulting peptides are analyzed using an Orbitrap Fusion Lumos mass spectrometer (Thermo Fisher Scientific) and an UltiMate 3000 RSLC nanoflow HPLC system (Thermo Fisher Scientific). Peptides were concentrated using a μ-precolumn (0.3 mm i.d. × 5 mm, 5 μm, Thermo Fisher Scientific) and separated on an AURORA column (0.075 mm i.d. × 250 mm, 1.6 μm, Ion Opticks Pty Ltd) using a two-step gradient: 2%–40% acetonitrile for 110 min, followed by 40%–95% acetonitrile with 0.1% formic acid for 5 min.

[0033] Examples of the method for quantifying the biomarkers include Western blotting, immunohistochemistry (IHC), immunofluorescence (IF), immunochromatography, enzyme-linked immunosorbent assay (ELISA), chemiluminescent enzyme immunoassay (CLEIA), chemiluminescent immunoassay (CLIA), and fluorescent enzyme immunoassay (e.g., Luminex) using extracellular vesicle-derived proteins. TM Examples of methods include the xMAP™ system (Medical and Biological Laboratories Co., Ltd.), aptamer method, immunoturbidimetric method, dye colorimetric method, immunonephelometric method, latex agglutination method, Jaffe method, and gold colloid colorimetric method.

[0034] Specifically, the relative quantification method using proteome analysis of the biomarkers involves separating and extracting proteins from extracellular vesicles (EVs), obtaining mass spectrometry (MS) / mass spectrometry (MS) data of the EV proteins, and performing label-free relative quantitative analysis of the proteins, thereby obtaining the quantitative analysis results of the target biomarker as the detected amount.

[0035] Specifically, mass spectrometry (MS) / mass spectrometry (MS) data of EV proteins can be acquired by analyzing the obtained EV proteins using an MS / MS device (e.g., Orbitrap Fusion Lumos mass spectrometer, Thermo Fisher Science) under the following conditions: Analysis parameters of the Orbitrap Fusion Lumos mass spectrometer include, for example, full scan resolution of 50,000; scan range (m / z) of 350-1500; full scan maximum injection time of 50 ms; and full scan AGC (Auto Gain Control: ion quantity control function) target of 4 × 10. 5 Dynamic exclusion time, 30 seconds; Data-dependent MS / MS acquisition cycle time, 2 seconds; Activation type, HCD; MS / MS detector, ion trap; Maximum MS / MS injection time, 35 ms; MS / MS AGC target, 1 x 10 4 can be set to.

[0036] Next, relative proteome analysis of EV proteins can be performed using the following procedure to obtain the detected amount of the target biomarker. The obtained MS / MS spectra, which represent the protein analysis data of extracellular vesicles (EVs), are searched against the SwissProt Homo sapiens protein sequence database using Proteome Discoverer 2.5 software (Thermo Fisher Scientific), with the peptide identification filter set to a false discovery rate of <1%. Label-free relative quantitative analysis of proteins can be performed using the default parameters of the Minora Feature Detector, Feature Mapper, and Precursor Ions Quantifier nodes in Proteome Discoverer 2.5. The quantitative analysis results can then be used to obtain the detected amount of the target biomarker.

[0037] <Determination step> The determination step is performed by determining the detected amount D of the biomarker in the sample obtained from the control group. ｃ (1) the biomarker is the first protein, and Dｓ >D ｃ and / or (2) the biomarker is the second protein, and D ｓ <D ｃ and determining that there is a risk of amyotrophic lateral sclerosis if

[0038] The control group refers to a plurality of healthy subjects who are not affected by ALS and serve as a comparison control.

[0039] The above D ｃ More than the above D ｓ When is small, compare comparable measurements and ｃ >D ｓ As long as the above relationship is satisfied, there are no particular limitations and the method can be appropriately selected depending on the purpose. For example, the method can be determined when a statistically significant difference is recognized between the proteins.

[0040] The above D ｃ More than the above D ｓ When D is large, compare comparable measurements and ｃ <D ｓ As long as the above relationship is satisfied, there are no particular limitations and the method can be appropriately selected depending on the purpose. For example, the method can be determined when a statistically significant difference is recognized between the proteins.

[0041] In addition, D determined in the examples described below ｃ Based on this, D was calculated individually. ｓ It can also be determined from the absolute value of

[0042] [Method for predicting progression] In one embodiment of the assessment method, the progression of a specific subject can be predicted. Examples of the progression include the progression of ALS, prognosis, and evaluation of the therapeutic effect of a therapeutic drug or a candidate drug in a subject administered with the drug. The method for predicting progression can be used to evaluate the progression of ALS, prognosis, and therapeutic effect, as well as to screen therapeutic drugs and candidate drugs. In this embodiment, the assessment method further includes a comparison step and a prediction step in addition to the data acquisition step and assessment step.

[0043] <Comparing Step> The comparing step is performed by comparing the detected amount D of the biomarker in the sample obtained from the subject. ｓ０ and the detected amount D of the biomarker in the sample obtained from the subject after a certain period of time has elapsed. ｓ１ This is a process of comparing the above.

[0044] <Prediction Step> The prediction step includes the steps of: (1) determining, by the comparison, that the biomarker is the first protein; ｓ０ >D ｓ１ When the above condition is met, the patient falls into one of the following stratification groups: high response to disease-modifying drug treatment, high effect of disease-modifying drug treatment, no progression of disease or slow progression of disease; D ｓ０ =D ｓ１ If the condition is as above, there is no change in the course of the disease, or ｓ０ <D ｓ１ or the patient is in any of the stratified groups of rapid progression of the disease, low response to disease-modifying drug treatment, low effect of disease-modifying drug treatment, and progression of the disease; (2) the biomarker is the second protein; ｓ０ <D ｓ１ When the above condition is met, the patient falls into one of the following stratification groups: high response to disease-modifying drug treatment, high effect of disease-modifying drug treatment, no progression of disease or slow progression of disease; D ｓ０ =D ｓ１ If the condition is as above, there is no change in the course of the disease, or ｓ０ >D ｓ１ and predicting that the patient belongs to any of the stratification groups of low response to disease-modifying drug treatment, low effect of disease-modifying drug treatment, progression of the disease state, or rapid progression of the disease state, if the patient is in the stratification group of low response to disease-modifying drug treatment, low effect of disease-modifying drug treatment, progression of the disease state, or rapid progression of the disease state.

[0045] This makes it possible to evaluate, for example, the progression, prognosis, and therapeutic effects of ALS, as well as to screen therapeutic drugs and candidate drugs.

[0046] Here, "disease-modifying drug response" refers to the prediction of whether a patient will respond to a treatment that can improve the ALS condition (i.e., slow the progression of the disease). "Disease-modifying drug efficacy" refers to the degree to which the treatment improves the ALS condition. "Stratification" refers to the classification of patients into groups based on different therapeutic responses or efficacy.

[0047] When the method for predicting the course of disease is a method for predicting the therapeutic effect on amyotrophic lateral sclerosis, the method further comprises a step of administering medication to the subject during the certain period of time (administration step).

[0048] The timing of administration is not particularly limited and can be appropriately selected depending on the purpose. ｓ１ and the aforementioned E ｓ１ The period until the detection of the above-mentioned drug may be any period sufficient for evaluating the therapeutic effect of the administered drug, and may be appropriately selected from the above-mentioned certain periods.

[0049] The drug used for the administration is not particularly limited and can be appropriately selected depending on the purpose, and may be a therapeutic agent or a candidate drug. These may be used alone or in combination of two or more.

[0050] Examples of therapeutic agents include ropinirole hydrochloride, riluzole, edaravone, methylcobalamin, bosutinib, hydrochlorothiazide, cortisone acetate, aldioxa, sibutramine hydrochloride, alendronate sodium, ceftibuten, estropipate, ketorolac tromethamine, etc. Among these, ropinirole hydrochloride is preferred.

[0051] The above-mentioned drug may be used in combination with other drugs, and may be formulated with a pharmaceutically acceptable carrier appropriate for the route of administration.

[0052] "Pharmaceutically acceptable" means a non-toxic ingredient, composition, etc. that is physiologically tolerable and does not normally cause gastrointestinal upset, allergic reactions such as dizziness, or similar reactions when administered to humans. Examples of the carrier include solvents, dispersion media, oil-in-water or water-in-oil emulsions, aqueous compositions, liposomes, microbeads and microsomes, biodegradable nanoparticles, etc.

[0053] The administration route of the drug is not particularly limited and can be appropriately selected depending on the purpose, and may be, for example, orally or parenterally. Parenteral administration routes include, for example, transdermal, nasal, abdominal, intramuscular, subcutaneous, or intravenous administration.

[0054] When the drug is administered orally, the pharmaceutical composition containing the drug may be formulated together with a suitable carrier for oral administration into the form of, but is not limited to, powder, granules, tablets, pills, sugar-coated tablets, capsules, liquids, gels, syrups, suspensions, wafers, and the like, by a method known in the art.

[0055] The medicaments may be formulated using methods known in the art so as to provide quick, sustained or delayed release of the active ingredient after administration to a mammal.

[0056] The drug formulated in the above manner may be administered in an effective amount via various routes, including oral, transdermal, subcutaneous, intravenous, or intramuscular administration. In the above, the term "effective amount" refers to the amount of a substance that allows the diagnostic or therapeutic effect to be tracked when administered to a subject.

[0057] The dosage of the drug can be appropriately selected depending on the route of administration, the subject, the target disease and its severity, age, sex, body weight, individual differences, and disease state. The drug can vary in the content of the active ingredient depending on the severity of the disease, but typically, based on an adult, the effective amount per administration is 0.1 mg to 50 mg (e.g., 5 mg), and may be administered several times a day.

[0058] When the drug is used in combination with another drug, the drugs may be contained in a single formulation or in separate formulations. If the drugs are in separate formulations, the formulations may be administered simultaneously or separately at intervals.

[0059] (Device) The device of the first embodiment is a device for determining amyotrophic lateral sclerosis, predicting and determining the therapeutic effect of a disease-modifying drug, monitoring, and stratifying the disease, and includes a detection amount D of the biomarker of this embodiment in a sample obtained from a subject. ｓ a data acquisition unit that acquires the detected amount D of the biomarker in the sample obtained from the control group; ｃ (1) the biomarker is the first protein, and D ｓ >D ｃ and / or (2) the biomarker is the second protein, and D ｓ <D ｃ The device comprises a determination unit that determines that the subject is at risk of amyotrophic lateral sclerosis if the subject is at risk of amyotrophic lateral sclerosis, and a display unit that displays the subject's risk of amyotrophic lateral sclerosis, and further comprises other components as necessary.

[0060] (Program) The program of the first embodiment is a program for diagnosis of amyotrophic lateral sclerosis, prediction and diagnosis of therapeutic effect of disease-modifying drugs, monitoring, and stratification, and includes an apparatus for detecting a detected amount D of the biomarker of the present embodiment in a sample obtained from a subject. ｓ a data acquisition unit that acquires the detected amount D of the biomarker in the sample obtained from the control group; ｃ (1) the biomarker is the first protein, and D ｓ >D ｃ and / or (2) the biomarker is the second protein, and D ｓ <D ｃ and a display unit that displays the subject's risk of amyotrophic lateral sclerosis.

[0061] [System Configuration] Fig. 9 is a diagram showing the overall configuration of the device of the first embodiment. In the ALS determination system of the first embodiment, an analyzer 10 determines the risk of ALS in a subject 30, and in the second embodiment, a data management device 11 (e.g., a computer such as a personal computer or a server) capable of acquiring analytical data determines the ALS risk of the subject 30. Note that the subject 30 may be a person who is being examined by a doctor at a medical institution, a person who is undergoing a health check, or the like.

[0062] [Functional Blocks] The functional configuration will be described below. Fig. 10 is a functional block diagram of the data management device 11 according to the first embodiment (in the case of embodiment 2). The data management device 11 can include a data acquisition unit 101, a determination unit 102, and a display unit 103. The data management device 11 can function as the data acquisition unit 101, the determination unit 102, and the display unit 103.

[0063] The data acquisition unit 101 acquires the detected amount D of the biomarker in the sample obtained from the subject 30 as analysis data. ｓ Get.

[0064] The determination unit 102 stores the detected amount D of the biomarker in the sample obtained from the control group as comparison data. ｃ are stored in advance. Alternatively, newly acquired comparison data may be integrated with the stored comparison data. The determination unit 102 compares the analysis data of the subject 30 with the comparison data, and determines the ALS risk of the subject 30 based on predetermined criteria appropriately set from the items described in the determination step. The determination unit 102 transmits the determination result to the display unit 103.

[0065] The display unit 103 displays the ALS risk of the subject 30 determined by the determination unit 102 (for example, by displaying the analysis data of the subject 30 together with a comparison with the comparison data). The display unit 103 may display on the data management device 11 or on another terminal.

[0066] [Flowchart] FIG. 11 is a flowchart of one aspect of the determination method of the first embodiment.

[0067] In step 101 (S101), the detected amount D of the biomarker of the subject 30 is used as analytical data. ｓ The analytical data may be obtained by measurements made at any medical institution or analytical institution.

[0068] In step 102 (S102), the determination unit 102 compares the analysis data of the target 30 acquired in S101 with the stored comparison data.

[0069] In step 103 (S103), the determination unit 102 determines the ALS risk of the subject 30 based on the comparison and predetermined criteria.

[0070] In step 104 (S104), the display unit 103 displays the ALS risk of the subject 30 determined in S103 (for example, displaying the analysis data of the subject 30 together with a comparison with the comparison data).

[0071] Physicians such as general internists and neurologists can refer to the ALS risk determined by the determination method of this embodiment using analytical data obtained from medical examinations and health checkups, and, if necessary, combine it with conventional diagnostic tests to efficiently make a diagnosis.

[0072] [Hardware Configuration] The hardware configuration will be described below.

[0073] 12 is a hardware configuration diagram of the data management device 11 according to the first embodiment. The data management device 11 can include a control unit 1001, a main memory unit 1002, an auxiliary memory unit 1003, an input unit 1004, an output unit 1005, and an interface unit 1006. Each of these units will be described below.

[0074] The control unit 1001 is a processor (for example, a CPU (Central Processing Unit), a GPU (Graphics Processing Unit), etc.) that executes various programs installed in the auxiliary storage unit 1003 .

[0075] The main memory unit 1002 includes a non-volatile memory (Read Only Memory (ROM)) and a volatile memory (Random Access Memory (RAM)). The ROM stores various programs, data, etc. required for the control unit 1001 to execute various programs installed in the auxiliary memory unit 1003. The RAM provides a working area into which the various programs installed in the auxiliary memory unit 1003 are expanded when executed by the control unit 1001.

[0076] The auxiliary storage unit 1003 is an auxiliary storage device that stores various programs and information used when the various programs are executed.

[0077] The input unit 1004 is an input device through which the operator of the data management device 11 inputs various instructions to the data management device 11 .

[0078] The output unit 1005 is an output device that outputs the internal state of the data management device 11 and the like.

[0079] The interface unit 1006 is a communication device for connecting to a network and communicating with other devices.

[0080] [Method, device, and program of second embodiment] The method of the second embodiment is a method for assessing amyotrophic lateral sclerosis, predicting and assessing the therapeutic effect of disease-modifying drugs, monitoring, and stratifying the disease, and includes the steps of: acquiring extracellular vesicle protein analysis data of a sample obtained from a subject (data acquisition step); inputting the extracellular vesicle protein analysis data of the subject into a assessment model that assesses the risk of amyotrophic lateral sclerosis for a subject group from the extracellular vesicle protein analysis data of the subject, and outputting the risk of the subject having amyotrophic lateral sclerosis (assessment step); and displaying the risk of the subject having amyotrophic lateral sclerosis (display step).

[0081] The device of the second embodiment is a device for assessing amyotrophic lateral sclerosis, predicting and assessing the therapeutic effect of disease-modifying drugs, monitoring, and stratification, and is equipped with: a data acquisition unit that acquires extracellular vesicle protein analysis data of a sample obtained from a subject; a determination unit that inputs the extracellular vesicle protein analysis data of a subject into a determination model that assesses the risk of a subject group having amyotrophic lateral sclerosis from the extracellular vesicle protein analysis data of the subject, and outputs the risk of the subject having amyotrophic lateral sclerosis; and a display unit that displays the risk of the subject having amyotrophic lateral sclerosis.

[0082] The program of the second embodiment is a program for making a diagnosis of amyotrophic lateral sclerosis, predicting and determining the therapeutic effect of disease-modifying drugs, monitoring, and stratification, and causes an apparatus to function as: a data acquisition unit that acquires extracellular vesicle protein analysis data of a sample obtained from a subject; a determination unit that inputs the extracellular vesicle protein analysis data of a subject into a determination model that determines the risk of a subject group having amyotrophic lateral sclerosis from the extracellular vesicle protein analysis data of the subject, and outputs the risk of the subject having amyotrophic lateral sclerosis; and a display unit that displays the risk of the subject having amyotrophic lateral sclerosis.

[0083] According to the method, device, and program of the second embodiment, a judgment model is obtained by having artificial intelligence (AI) comprehensively learn analytical data of extracellular vesicle (EV) proteins, and judgment is made based on the judgment model, thereby providing a method, device, and program that can accurately perform judgment in amyotrophic lateral sclerosis, prediction and judgment of the therapeutic effect of disease-modifying drugs, monitoring, and stratification.

[0084] <Data Acquisition Step and Data Acquisition Unit> The data acquisition step is a step of acquiring analysis data of extracellular vesicle proteins of interest, and can be suitably carried out using a data acquisition unit that acquires analysis data of extracellular vesicle proteins of interest. The data acquisition unit may be an analyzer such as a mass spectrometry (MS) / mass spectrometry (MS) device, or may be a data management device that acquires and manages data (detection amount) measured by an external device.

[0085] The extracellular vesicle protein analysis data is preferably either mass spectrometry (MS) / mass spectrometry (MS) data of EV proteins or label-free relative quantitative analysis data of EV proteins, which can be obtained by the method described in the first embodiment.

[0086] <Determination step and determination unit> The determination step is a step of inputting extracellular vesicle protein analysis data of a subject into a determination model that determines the risk of a subject group having amyotrophic lateral sclerosis from the extracellular vesicle protein analysis data of the subject, and outputting the risk of the subject having amyotrophic lateral sclerosis, and can be suitably performed using a determination unit.

[0087] The determination model can be generated by a learning process described below. The method of the second embodiment may further include a determination model generation step, and the generated determination model may be used. The determination model may be generated by unsupervised learning or supervised learning. However, unsupervised learning is preferred because EV proteins comprehensively contain a large number of proteins, enabling determination based on novel indicators.

[0088] The risk of ALS is preferably assessed based on indicators from known ALS testing and diagnostic methods, which have high accuracy in the assessment model, and more preferably based on the ALSFRS-R, an important clinical indicator of ALS symptoms. The accuracy of the assessment model can be evaluated by internal validity and / or external validity. The assessment model then inputs extracellular vesicle protein analysis data of the subject to be assessed and outputs the risk of the subject having amyotrophic lateral sclerosis.

[0089] <Display Step and Display Unit> The display step is a step of displaying the risk that the subject has amyotrophic lateral sclerosis, and can be suitably performed using a display unit.

[0090] [System Configuration] Fig. 13 is a diagram showing the overall configuration of the device of the second embodiment. The system configuration of the device of the second embodiment can be appropriately selected in the same manner as the device of the first embodiment, except that it further includes a learning device 20. In embodiment 1, the analyzer 10, which is the determination device, has a determination model generated by the learning device 20. In embodiment 2, the data management device 11, which is the determination device, has a determination model generated by the learning device 20.

[0091] [Functional Blocks] -Learning Device- Fig. 14 is a functional block diagram of a learning device 20 according to the second embodiment. The learning device 20 can include a learning data acquisition unit 201, a learning data storage unit 202, a machine learning unit 203, and a determination model storage unit 204. By executing a program, the learning device 20 can function as the learning data acquisition unit 201 and the machine learning unit 203. Each of these units will be described below.

[0092] The training data acquisition unit 201 acquires training data used to generate a determination model. Specifically, the training data acquisition unit 201 acquires extracellular vesicle protein analysis data of a subject group, and may acquire information indicating whether or not the subject group has ALS (an indicator of ALS) in the case of supervised learning or for evaluating the accuracy of the determination model.

[0093] The learning data storage unit 202 stores extracellular vesicle protein analysis data of a subject group and, if necessary, information indicating whether the subject group has ALS.

[0094] The machine learning unit 203 generates a determination model through machine learning. Specifically, the machine learning unit 203 learns the target extracellular vesicle protein analysis data.

[0095] The judgment model storage unit 204 stores the judgment model generated by the machine learning unit 203 .

[0096] In one embodiment, it is preferable to perform regression analysis and / or classification analysis in supervised learning, and suitable examples include logistic regression, perceptron, random forest classifier, extra trees classifier, gradient boosting classifier, decision tree classifier, AdaBoost classifier, XGBoost classifier, XGBoost classifier, etc. Model accuracy verification is performed using a test dataset and / or a validation dataset, and a model with high accuracy can be preferably adopted.

[0097] The extracellular vesicle protein analysis data can be, for example, mass spectrometry (MS) / mass spectrometry (MS) data of EV proteins or label-free relative quantitative analysis data of EV proteins. The test data used for learning can be extracellular vesicle protein analysis data of healthy control subjects and SALS patient subjects before or after administration of the therapeutic RPOI.

[0098] Fig. 8A is a diagram for explaining a decision model for the apparatus of the second embodiment. Fig. 8A is a schematic diagram of a learning framework for limited sample size and imbalanced dataset. This framework can be composed of three steps: oversampling, parameter search, and feature selection.

[0099] As shown in Figure 8A, an imbalanced dataset with a limited sample size is expanded into a balanced dataset using an imbalanced learning algorithm. The balanced dataset is then randomly assigned to a training dataset and a test dataset. Separately measured data can be used as validation data. To confirm the general performance of the model, it is preferable to perform parameter tuning using k-fold cross validation. Recursive feature elimination (RFE) can also be used to select the features most relevant to predicting the target variable by removing features with the lowest importance.

[0100] As a result of the training, as shown in Figure 8B and Figure 8C, 155 proteins were selected as features for serum EV-derived proteins, with an accuracy of 90.4% (Figure 8B), and 19 proteins were selected as features for cerebrospinal fluid EV-derived proteins, with an accuracy of 80.3% (Figure 8C).

[0101] 15 is a functional block diagram of a data management device 11 (determination device) according to the second embodiment of the second embodiment. The data management device 11 can include a data acquisition unit 301, a determination unit 302, and a display unit 303. The data management device 11 can function as the data acquisition unit 301, the determination unit 302, and the display unit 303.

[0102] The data acquisition unit 301 acquires EV protein analysis data of the subject 30.

[0103] The determination unit 302 inputs the EV protein analysis data acquired from the specimen of the subject 30 by the data acquisition unit 301 into the determination model stored in the determination model storage unit 304 of the determination device (or the determination model stored in the determination model storage unit 204 of the determination device 20), and outputs the ALS risk of the subject 30. Specifically, the determination unit 302 inputs the EV protein analysis data into the determination model, and outputs the ALS risk of the subject 30.

[0104] The determination model storage unit 304 stores a determination model for determining a subject's risk of ALS from the subject's EV protein analysis data.

[0105] [Processing Method] Hereinafter, the learning processing method will be described with reference to FIG. 16, and the determination processing method will be described with reference to FIG.

[0106] -Learning Process- FIG. 16 is a flowchart of the learning process according to the determination method of this embodiment.

[0107] In step 201 (S201), the training data acquisition unit 201 acquires EV protein analysis data for a training group and, if necessary, information indicating whether the group has ALS. The training data may be mass spectrometry (MS) / mass spectrometry (MS) data of EV proteins or label-free relative quantitative analysis data of EV proteins.

[0108] In step 202 (S202), the machine learning unit 203 generates a judgment model through machine learning. Specifically, the machine learning unit 203 performs regression analysis and / or classification analysis in supervised learning using the EV protein analysis data of the learning subject group acquired in S201, and also performs model accuracy verification using a test dataset and / or a validation dataset to obtain a highly accurate judgment model.

[0109] [Determination Process] FIG. 17 is a flowchart of the determination process according to the determination method of this embodiment.

[0110] First, in an optional step not shown, EV protein analysis data is measured for the subject 30. The EV protein analysis data can be measured at any medical institution or analytical institution.

[0111] In step 301 (S301), the data acquisition unit 301 acquires EV protein analysis data of the measured subject 30. The data acquisition unit 301 may acquire EV protein analysis data of the subject 30 measured by the analyzer 10 in real time, or may acquire EV protein analysis data of the subject 30 measured by the analyzer 10 in the past.

[0112] In step 302 (S302), the judgment unit 302 inputs the EV protein analysis data of the subject 30 obtained in S301 into the judgment model (i.e., the judgment model generated in S202 of Figure 16) and outputs the ALS risk of the subject 30.

[0113] In step 303 (S303), the display unit 303 displays the ALS risk of the subject 30 determined in S302 (for example, displaying the analysis data of the subject 30 together with a comparison with the comparison data).

[0114] Physicians such as general internists and neurologists can refer to the ALS risk determined by the determination method of this embodiment using analytical data obtained from medical examinations and health checkups, and, if necessary, combine it with conventional diagnostic tests to efficiently make a diagnosis.

[0115] [Hardware Configuration] The hardware configuration of the device of the second embodiment can be appropriately selected from the same hardware configuration as the hardware configuration of the device of the first embodiment shown in FIG.

[0116] The present invention will be described in more detail below based on examples, but the present invention is not limited to the following examples.

[0117] <Methods> <<Study Participants>> This study involved 20 SALS patients enrolled in the ROPALS trial conducted at Keio University Hospital. Details of the inclusion and exclusion criteria are described in the ROPALS trial protocol (Reference 1: Morimoto, S. et al. Cell Stem Cell 30, 766-780.e9 (2023)). In addition, serum and cerebrospinal fluid samples from 10 non-neurodegenerative Japanese adults (controls) were obtained from the NCNP Biobank, a member of the National Center Biobank Network.

[0118] <<ROPALS Study and Blood / CSF Sampling>> The ROPALS study enrolled 20 SALS patients without reported ALS mutations. The study consisted of a 12-week observation period, a 24-week double-blind period, and a 24-week open-label extension period. Patients with SALS were assigned to either the ROPI treatment (ROPI group) or placebo group (13 and 7 patients, respectively). Thirteen patients in the ROPI group received ROPI for the entire 48-week intervention period, while seven patients in the placebo group received placebo for the first 24 weeks (double-blind period) and ROPI for the second 24 weeks (open-label extension period) of the 48-week intervention period (Figure 1A).

[0119] The first week of ROPI / placebo administration was designated week 0 (week 0). Blood samples were collected from patients at week 0 (n = 20), week 13 (n = 20), week 24 (n = 18), week 39 (n = 15), and week 48 (n = 11). Similarly, CSF samples were collected from patients at week 0 (n = 20), week 24 (n = 17), and week 48 (n = 6). Sample collection and storage methods followed clinical trial protocols for SALS patient samples and NCNP Biobank methods for control samples.

[0120] EV isolation and mass spectrometry. Extracellular vesicles (EVs) released into serum, CSF, and iPSC-derived astrocyte (iPast) culture media were collected using an EV Second L70 column (GL Sciences) according to the manufacturer's instructions. Proteins in EVs were reduced with 10 mM TCEP at 100°C for 10 min, alkylated with 50 mM iodoacetamide at room temperature for 45 min, and then digested on-bead with Trypsin / Lys-C Mix (Promega) at 37°C for 12 h. The resulting peptides were analyzed using an Orbitrap Fusion Lumos mass spectrometer (Thermo Fisher Scientific) and an UltiMate 3000 RSLC nanoflow HPLC system (Thermo Fisher Scientific). Peptides were concentrated using a μ-precolumn (0.3 mm i.d. × 5 mm, 5 μm, Thermo Fisher Scientific) and separated on an AURORA column (0.075 mm i.d. × 250 mm, 1.6 μm, Ion Opticks Pty Ltd) using a two-step gradient: 2%–40% acetonitrile for 110 min, followed by 40%–95% acetonitrile with 0.1% formic acid for 5 min. The analytical parameters of the Orbitrap Fusion Lumos mass spectrometer were: full scan resolution: 50,000; scan range (m / z): 350–1500; full scan maximum injection time: 50 ms; full scan AGC (Auto Gain Control) target: 4 × 10 5 Dynamic exclusion time, 30 seconds; Data-dependent MS / MS acquisition cycle time, 2 seconds; Activation type, HCD; MS / MS detector, ion trap; Maximum MS / MS injection time, 35 ms; MS / MS AGC target, 1 x 10 4MS / MS spectra were searched against the SwissProt Homo sapiens protein sequence database using Proteome Discoverer 2.5 software (Thermo Fisher Scientific), with a peptide identification filter set to a false discovery rate <1%. Label-free relative quantification of proteins was performed using the default parameters of the Minora Feature Detector, Feature Mapper, and Precursor Ions Quantifier nodes in Proteome Discoverer 2.5 (Figure 1B).

[0121] <<Comprehensive quantification of EV proteins>> The first batch measured longitudinal samples from 20 SALS patients, and the second batch measured samples from 10 controls and three SALS patients (0w samples, sEV ID; RPR-01-31 / 32 / 33, cEV ID; RPR-01-24 / 25 / 33). To eliminate the possibility of batch effects due to processing samples from different batches, three identical samples were measured in two batches each for sEVs and cEVs. The measurements from the second batch were corrected using the following formula:

[0122]

[0123] For downstream analysis, all protein measurements were log-corrected after batch-to-batch correction to ensure data followed a normal distribution. 10 Converted.

[0124] All machine learning analyses were performed in Python (version 3.8) using scikit-learn (version 1.2.1) and xgboost (version 1.7.5). Taking into account the limited sample size and the presence of imbalanced classes, a learning framework was constructed (Figure 8A). In the oversampling stage, the imbalanced dataset of 13 samples (controls: n = 10, SALS patients: n = 3) measured in the first batch was expanded to a balanced dataset of 19 samples (controls: n = 10, SALS patients: n = 9) using the SMOTE algorithm in imbalanced learning (version 0.0). The data from all 19 samples were randomly assigned to a training dataset containing 13 samples (68.4%) and a testing dataset containing 6 samples (31.6%), taking into account the control and SALS patient samples. Validation was performed using 84 serum samples and 43 CSF samples measured in the second batch. Of these samples, 33 serum samples and 13 CSF samples were from ROPI-naïve patients, and the remaining samples were from ROPI-exposed patients. We used scikit-learn's GridSearchCV for parameter tuning by two-fold cross-validation and RFECV for recursive feature elimination in the feature selection phase, with a random seed of 111.

[0125] <iPSC-derived astrocytes and spinal motor neuron research> <<iPSC-derived astrocyte research>> - Cell lines and culture - Human iPSCs derived from a healthy control (WD39; CVCL_Y528) were used to differentiate into human iPlasts according to the method previously reported by the inventors (Reference: Leventoux N et al., Cells 2020;9.).

[0126] - Ribonucleic acid sequencing (RNA-seq) - RNA quantity and quality were confirmed using an Agilent TapeStation (Agilent Technologies). RNA was amplified using the SMART-Seq v4 Ultra Low Input RNA Kit for Sequencing (Takara Bio Inc.), and DNA libraries were prepared using the Nextera XT DNA Library Prep Kit (Illumina). RNA-seq was performed using the NovaSeq system (Illumina). All procedures were performed by Takara Bio Inc.

[0127] -PSC-MN analysis- RNA-seq data for iPSC-MN (t84h) derived from healthy individuals was obtained from data deposited in the NCBI Gene Expression Omnibus (GEO) database (GEO accession number GSE209696).

[0128] <<Analysis Tools>> All analyses except for the machine learning portion were performed using R (version 4.1.2). Effect size calculations were performed using Effsize (version 0.8.1). Dunnett's tests were performed using the aov function and multcomp (version 1.4.0). PCA was performed using the prcomp function, with the argument "scale" set to TRUE, as recommended in the official documentation.

[0129] For RNA-seq analysis, fastp (version 0.23.2, all arguments set to default) was used for quality control of fastq files. Read counts were quantified using Salmon (version 1.6, all arguments set to default) for samples that passed quality control. The Salmon reference was gencode, release 42 (GRCh38.p13). Salmon output files were converted to count files using the R package tximport (version 1.22.0). Group comparisons were performed using the R package DESeq2 (version 1.34.0).

[0130] Gene Ontology (GO) (biological pathways BP; cellular component CC; molecular function MF) enrichment and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis were performed using g:Profiler. The adjusted P values ​​(adj. P) calculated by statistical analysis were entered in ascending order, and the "Ordered query" option was checked. According to the reference (Reimond et al. Nat Protoc 2019;14:482-517), the "term size" was set to "5-350." Cosine similarity analysis was performed using lsa (version 0.73.3). The nls function was used to approximate changes in the ALS Functional Rating Scale-Revised (ALSFRS-R), an important clinical indicator of ALS symptoms in the ROPALS study. The following model equation was used in the ROPALS study:

[0131]

[0132] For visualization of GO (BP) and KEGG pathway analyses, the output files in GEM files were clustered and annotated by g:profiler using Cytoscape (version 3.9.1), EnrichmentMap (version 3.3.4), and AutoAnnotate (version 1.4.0).

[0133] <<Statistics>> Demographic data were analyzed using Student's t-test (age) and chi-square test (gender and race). Log 10Student's t-tests were used to compare transformed protein measurements. Storey's method was used to correct for multiple testing. Longitudinal, two-group analyses of the effects of treatment (placebo vs. ROPI) and period on the response variable (protein abundance) were performed using two-way analysis of variance followed by Bonferroni's multiple comparison test using the ez package, and estimated marginal means were calculated using the emmeans package. Pairwise comparisons were performed using these estimated marginal means, and P values ​​were adjusted for family-wise error rates using the Bonferroni method. Correlation analyses used Pearson's product-moment correlation coefficients unless otherwise noted. P values ​​or adj. P values ​​were calculated using two-tailed tests, and values ​​less than 0.05 indicated statistical significance. This cutoff value was also used to identify significantly differentially abundant proteins (DAPs) and differentially expressed genes (DEGs). As described in the manual of the effsize (version 0.8.1) package, Cohen's d was considered negligible, small, medium, or large if its absolute value was <0.2, <0.5, <0.8, or ≥0.8, respectively.

[0134] <Results> <<Research participants>> The subjects were 10 cases in the control group and 20 cases in the SALS group. Gender (χ ２ (1)=0.1563, p=0.6926), race (χ ２ (1) = NA, p > 0.9999) showed no significant difference between the two groups, but the SALS group was significantly older than the control group (t (27.14) = 3.8354, P < 0.001).

[0135] <<The protein composition of sEVs and cEVs differs between the control and SALS groups>> The protein composition of EVs in bodily fluids was compared between the control and SALS groups. Serum and cerebrospinal fluid were collected over time from 20 SALS patients enrolled in the ROPALS trial (Fig. 1A). EV fractions were extracted and comprehensive quantitative proteomic analysis was performed (Fig. 1B). Typical EV markers, such as CD9 and Alix, were detected in all samples, indicating that highly pure EVs had been extracted (not shown). Furthermore, 3,719 and 2,754 proteins derived from sEVs and cEVs, respectively, were detected.

[0136] To investigate differences between serum-derived extracellular vesicles (sEVs) and CSF-derived extracellular vesicles (cEVs), we compared the protein types detected in all sEV samples or cEV samples (from both controls and SALS patients). We identified 802 and 884 proteins in sEVs and cEVs, respectively, of which 387 proteins were shared (Fig. 1C, Venn diagram). sEV-specific proteins were primarily associated with platelet activation (KEGG pathway; KP, adj. P value < 0.001), while cEV-specific proteins were primarily associated with ECM-receptor interactions (KP, adj. P value < 0.001) (Fig. 1C, bar graph). To clarify differences in the composition of sEVs and cEVs, we performed uniform manifold approximation and projection (UMAP) analysis of proteins detected in all samples from both groups. sEVs and cEVs were clearly separated into two groups, suggesting that the protein compositions of sEVs and cEVs differ significantly (Fig. 1D). Furthermore, UMAP analysis of sEVs and cEVs showed that their distribution in both groups differed from that in controls and SALS patients, although the changes over time were smaller than the differences due to the presence of disease.

[0137] Next, we compared the levels of proteins detected in both sEVs and cEVs. The levels of each protein present in both groups showed a significant positive correlation (R 2 = 0.22, adj. P value < 0.001) (Fig. 1E). However, some proteins were heterogeneously present in both sEVs and cEVs. Furthermore, we performed a correlation analysis of the different protein levels in sEVs and cEVs from each SALS patient at each time point. The coefficients of determination for the protein content of sEVs and cEVs at 0 wk, 24 wk, and 36 wk were significantly higher in SALS patients than in controls (Dunnett's test, adj. P values ​​= 0.0497, 0.0129, and 0.0129, respectively), indicating that the protein compositions of sEVs and cEVs were more similar in SALS patients (Fig. 1F). Furthermore, the correlation coefficient tended to increase over time in SALS patients, suggesting that the compositions of sEVs and cEVs became more similar over time.

[0138] Figures 1A–F show that the protein profiles of sEVs and cEVs differ between controls and SALS patients. Figure 1A shows an overview of the ROPALS study and the timing of serum and CSF collection. Figure 1B shows the study design. EVs were extracted from serum and CSF collected from controls (n = 10, point-in-time collection, NCNP Biobank) and SALS patients (n = 20, longitudinal collection) participating in the ROPALS study. Comprehensive quantitative proteomic analysis was performed using an Orbitrap Fusion Lumos mass spectrometer. Figure 1C shows a comparison of the protein profiles of sEVs and cEVs from controls and SALS patients. The Venn diagram in the figure shows a comparison of proteins within sEVs and cEVs detected in all samples (both control and SALS patient samples). The bar graph in the figure shows the results of KEGG pathway analysis of proteins specifically identified in sEVs and cEVs. Figure 1D shows the results of UMAP analysis of all EVs, sEVs, and cEVs. Figure 1E shows a correlation plot and Pearson correlation analysis of proteins detected in both sEVs and cEVs. Each point represents one protein. Figure 1F shows the correlation analysis of protein profiles detected in both sEVs and cEVs collected from the same subject at the same time. Each point represents the correlation coefficient comparing the protein content of each protein in sEVs and cEVs collected from the same subject at the same time (Dunnett's test, controls-0w: adj. P = 0.0497, controls-24w: adj. P = 0.0129, control group: adj. P = 0.0129, mean ± SD).

[0139] Comparative analysis of the protein composition of EVs derived from controls and SALS patients. First, to characterize the protein profiles of sEVs and cEVs derived from ROPI-untreated SALS patients, we defined sEV- and cEV-derived SALS-specific proteins as those identified in ≥90% of sEVs and cEVs derived from SALS patients and ≤10% of control sEVs and cEVs. We identified 365 EV-derived SALS-specific proteins from sEVs and 398 EV-derived SALS-specific proteins from cEVs (not shown).

[0140] sEV-derived SALS-specific proteins were suggested to be involved in lipase activity (MF, adj. P = 0.002) (data not shown). Furthermore, cEV-derived SALS-specific proteins were associated with immunoglobulin complexes (CC, adj. P < 0.001) and antigen binding (MF, adj. P < 0.001). Furthermore, cEV-derived SALS-specific proteins included C-reactive protein, suggesting that they may induce inflammation in the CSF of SALS patients (data not shown).

[0141] Next, we compared the protein composition of sEVs and cEVs from control and ROPI-untreated SALS patients. To improve reproducibility, proteins detected in all samples from control and ROPI-untreated SALS patients were used for comparative analysis. As a result, 723 significantly differentially abundant proteins (DAPs) were identified in sEVs (Fig. 2A) and 246 in cEVs (Fig. 2C).

[0142] Increased DAP in sEVs (n = 159) was associated with complement and coagulation cascade (KP, adj. P < 0.001), whereas decreased DAP (n = 564) was associated with platelet activation (KP, adj. P < 0.001) (Fig. 2B). Furthermore, increased DAP in cEVs (n = 148) and sEVs was associated with complement and coagulation cascade (KP, adj. P < 0.001), whereas decreased DAP (n = 98) was significantly associated with extracellular vesicle protein processing (KP, adj. P < 0.001) (Fig. 2D).

[0143] Finally, we investigated TDP-43, which forms aggregates in motor neurons of ALS patients. Previous studies have shown that TDP-43 is elevated in plasma, CSF, and cultured cell-derived EVs of SALS patients compared with controls and over time. The detection rates of TDP-43 were 0% (0 / 10) in controls, 10.7% (9 / 84) in SALS patient sEVs, 0% (0 / 10) in SALS patient cEVs, and 18.6% (8 / 43) in SALS patient cEVs (not shown). There was no significant difference in the detection rates between controls and SALS patients (sEVs; χ ２ (1)=0.9800, p=0.3222, cEV;χ ２ (1) = 0.2705, p = 0.6030). Furthermore, TDP-43-positive serum and CSF samples from the same SALS patient were not consistent over time (not shown).

[0144] <<Comparison of DAPs between sEVs and cEVs>> Next, we compared DAPs between sEVs and cEVs. In SALS patients, 31 DAPs were elevated and 38 DAPs were decreased in both sEVs and cEVs (Figure 2E, left). Functional analysis of these DAPs revealed that DAPs elevated in SALS patients were involved in the regulation of complement coagulation, while DAPs decreased in SALS patients were involved in the regulation of actin polymerization and unfolded protein processing (Figure 2E, right; the cluster name assigned by AutoAnnotate was "unfolded protein processing," but will be referred to below as the more general term "unfolded protein response (UPR)").

[0145] Next, to determine the relationship between the mean EV protein levels (regulation of complement coagulation, polymerization regulation of actin, and UPR) in each category and the ALSFRS-R score at the time of sampling, we calculated the approximate ALSFRS-R (aALSFRS-R) at the time of sampling. The aALSFRS-R score was calculated by approximating each patient's pathological progression using the following formula and substituting the number of weeks at the time of sampling (Figure 2F):

[0146] Correlation analysis of the mean EV content of each protein category with the aALSFRS-R score at the time of collection revealed a significant inverse correlation between proteins related to the regulation of complement coagulation and UPR-related proteins in both sEVs (R = -0.82, p < 0.001) and cEVs (R = -0.62, p < 0.001) (Fig. 2G). On the other hand, the concentration of UPR-related proteins in cEVs was significantly correlated with the aALSFRS-R score (R = 0.52, p < 0.001), and the concentration of UPR-related proteins in cEVs decreased with disease progression.

[0147] These results suggest that SALS disease progression is characterized by increased regulation of complement coagulation-related proteins and decreased regulation of EV UPR-related proteins.

[0148] Figures 2A-G show the results of a comparative analysis of EV protein profiles between the control group and SALS patients. Figures 2A and 2C are volcano diagrams showing a comparison of EV proteins in serum and cerebrospinal fluid between the control group and ROPI-untreated SALS patients (sEV: Figure 2A, cEV: Figure 2C). Proteins detected in all samples from the control group and ROPI-untreated SALS patients were compared. The comparison between the control group and SALS patients is log 10A Student's t-test was performed on transformed data, and P values ​​were corrected using Storey's method. Figures 2B and 2D show the results of GO and KEGG pathway analysis of DAPs (BP, CC, MF) in sEVs and cEVs (sEV: Figure 2B, cEV: Figure 2D). Figure 2E shows the left panel: a comparison of DAPs in sEVs and cEVs (control samples vs. SALS patient samples), and the right panel: a list of proteins commonly decreased / increased in sEVs and cEVs from SALS patients compared to controls, along with the results of GO (BP) and KEGG pathway clustering analysis. "Regulation of complement coagulation" refers to the complement coagulation regulation group, "polymerization regulation of actin" refers to the actin polymerization regulation group, "unfolded protein processing" refers to the unfolded protein processing group, and "others" refers to other groups. Figure 2F shows a schematic diagram of the disease progression modeling method and aALSFRS-R calculation. Figure 2G shows the Pearson correlation analysis between the average values ​​of each protein in the complement coagulation regulation group, actin polymerization regulation group, and unfolded protein processing group and the aALSFRS-R. Circles indicate correlation coefficients.

[0149] ROPI Decreases Complement- and Coagulation-Related Proteins in EVs and Increases UPR-Related Proteins. We calculated the log2 (fold change) of sEVs and cEVs from week 0 to week 24 for 20 SALS patients in the ROPALS study (13 in the ROPI group and 7 in the placebo group) and compared their values ​​between the ROPI and placebo groups. Three patients lacking CSF samples at week 24 and two patients lacking serum samples at week 24 were excluded (Figure 1A). A Student's t-test with Storey correction for P values ​​revealed no statistically significant differences for any proteins. This was primarily due to the small sample size relative to the data variability. Therefore, according to the effsize manual, an R package for calculating effect sizes, we set the effect size cutoff value for Student's t-test (Cohen's d) at ±0.5, and defined proteins with a moderate or greater effect of ROPI as effect-size-based differentially abundant proteins (es-DAPs).

[0150] Comparing the changes in sEV and cEV protein content between 0 and 24 weeks of age between the ROPI and placebo groups, 287 and 712 es-DAPs were identified, respectively (Fig. 3A, 3E). Among sEVs, es-DAPs increased by ROPI (n = 182) were associated with proteasomes (KP, adj. P = 0.0034), whereas es-DAPs decreased by ROPI (n = 105) were associated with complement and coagulation cascades (KP, adj. P < 0.001) (Fig. 3B). In cEVs, increased es-DAP (n = 99) by ROPI administration was associated with the pentose phosphate pathway (KP, adj. P = 0.0017), whereas decreased es-DAP (n = 613) by ROPI administration was associated with the complement and coagulation cascade (KP, adj. P < 0.001) (Fig. 3F).

[0151] Furthermore, we compared DAP (controls vs. SALS patients) and es-DAP with or without ROPI treatment (ROPI 0w-24w group vs. placebo 0w-24w group) (Fig. 3C, G). Proteins that were decreased in SALS patients and increased with ROPI (sEVs: n = 99, cEVs: n = 15) were associated with Parkinson's disease (KP, adj. P < 0.001) and proteasome (KP, adj. P < 0.001) in sEVs (Fig. 3D), and with extracellular vesicle protein processing (KP, adj. P = 0.0366) in cEVs (Fig. 3H). Conversely, proteins that were increased in SALS patients and decreased by ROPI (sEVs: n = 47, cEVs: n = 76) were associated with the complement and coagulation cascades in both sEVs and cEVs (KP, adj. P values ​​< 0.001 and < 0.001, respectively) (Fig. 3D, Fig. 3H).

[0152] Figures 3A–H show changes in EV protein profiles following ROPI administration. Figures 3A and 3E show Cohen's d rank plots comparing log2 (fold change) and each protein in sEV or cEV between the ROPI and placebo groups for each patient at weeks 0 and 24 (sEV: Figure 3A, cEV: Figure 3E). es-DAPs were identified using Cohen's d with a Student's t-test at a cutoff value of ±0.5. Figures 3B and 3F show the results of GO (BP, CC, MF) and KEGG pathway analysis of es-DAPs in sEV and cEV (placebo 0–24w vs. ROPI 0–24w) (sEV: Figure 3B, cEV: Figure 3B). Figures 3C and 3G show Venn diagrams comparing es-DAP (placebo 0-24w vs. ROPI 0-24w) and DAP (control samples vs. SALS patient samples) in sEVs and cEVs (sEV: Fig. 3C, cEV: Fig. 3G). Figures 3D and 3H show GO (BP, CC, MF) and KEGG pathway analysis results for the protein groups increased / decreased by ROPI treatment in SALS patient samples and sEVs and cEVs (sEV: Fig. 3D, cEV: Fig. 3H).

[0153] <<Changes in sEV and cEV Proteins Over Time>> Of the 20 SALS patients who participated in the ROPALS study, we selected patients with samples up to the final week (48 weeks) (CSF: ROPI group, n = 4, placebo group, n = 2; serum: ROPI group, n = 8, placebo group, n = 3). Next, we performed clustering analysis using Ward's method on each sEV and cEV in the placebo group to determine the difference (log2 (fold change)) in EV protein levels between 0 weeks and each sampling time point (Fig. 4A).

[0154] Clustering analysis was performed on the differential changes in EV proteins at each sampling time point between the 0w and placebo groups, and the changes in each protein were classified into three groups: increase (sEV: n = 298, Figure 4B: solid gray line in the left graph; cEV: n = 260, Figure 4D: solid gray line in the left graph), relatively unchanged (sEV: n = 297, Figure 4B: dashed gray line in the left graph; cEV: n = 1073, Figure 4D: dashed gray line in the left graph), and decrease (sEV: n = 466, Figure 4B: solid black line in the left graph; cEV: n = 51, Figure 4D: solid black line in the left graph).

[0155] Next, we examined the changes in these protein groups in the ROPI group (Fig. 4B and Fig. 4D, right graphs). The time-dependent increase in EV proteins in the placebo group was significantly suppressed by ROPI administration (all P<0.001, Fig. 4B and Fig. 4D, right graphs, gray solid lines). Furthermore, these protein groups were associated with the complement and coagulation cascade in both sEVs (Fig. 4C, left graph) (KP, adj. P<0.001) and cEVs (KP, adj. P<0.001) (Fig. 4E, left graph).

[0156] However, the proteins whose EV levels decreased over time in the placebo group were significantly restored by ROPI treatment (all P<0.001, solid black lines in Figures 4B and 4D). These proteins were involved in platelet aggregation (KP, adj. P<0.001) in sEVs (Figure 4C, right graph) and antioxidant activity (MF, adj. P<0.001) in cEVs (Figure 4E, right graph).

[0157] Figures 4A–E show the time course of EV protein changes in SALS patients. Figure 4A shows the scheme of time series analysis using clustering techniques to examine the time course of sEV and cEV protein changes in SALS patients who participated in the ROPALS study and whose samples were available up to 48 weeks. Figures 4B and 4D show the log2 (fold change) changes of each protein from 0 weeks to each sampling time in patients in the placebo and ROPI groups (sEV: Figure 4B, cEV: Figure 4D). Clustering analysis (Ward's method, k = 3) of differential changes in EV proteins from 0w to each sampling time in the placebo group classified proteins into three groups based on their time course: increased (sEV: n = 298, Figure 4B left, gray solid line; cEV: n = 260, Figure 4D left, gray solid line), relatively unchanged (sEV: n = 297, Figure 4B left, gray dashed line; cEV: n = 1073, Figure 4D left, gray dashed line), and decreased (sEV: n = 466, Figure 4B left, black solid line; cEV: n = 51, Figure 4D left, black solid line). After clustering, the time course of the ROPI group for proteins belonging to each clustering group was plotted (sEV: Figure 4B right; cEV: Figure 4D right). Statistical analysis was performed using two-way ANOVA followed by Bonferroni's multiple comparison test. Figures 4C and 4E show the results of GO (BP, CC, MF) and KEGG pathway analysis of increased and decreased proteins in EVs of the placebo group based on log2 (fold change) clustering (sEV: Fig. 4C, cEV: Fig. 4E). *** indicates statistical significance at P<0.001.

[0158] Comparative Analysis of DAP, es-DAP, and Time-Varying Proteins Next, we calculated the cosine similarity between DAP (controls vs. SALS patients), es-DAP (placebo 0-24w vs. ROPI 0-24w), and time-varying protein clusters. The increased / decreased protein clusters identified in sEVs and cEVs were clearly similar, a finding supported by cluster analysis (Figures 5A and 5C). Interestingly, the protein clusters that increased / decreased or increased / decreased over time in SALS patients were highly similar to the protein clusters that decreased / increased with ROPI administration (Figures 5A and 5C). These results suggest that ROPI has an anti-ALS effect on the protein profile of EVs.

[0159] Proteins that increased / decreased in SALS patients or increased over time and decreased / increased with ROPI administration were subjected to GO (MF, BP, CC) and KEGG pathway analysis. Proteins that increased in SALS patients or increased and decreased over time with ROPI administration were involved in the complement and coagulation cascades in both sEVs and cEVs (KP, adj. P values ​​<0.001 and <0.001, respectively). Conversely, proteins that were decreased or decreased over time in SALS patients but increased with ROPI administration were enriched in Parkinson's disease (KP, adj. P < 0.001) and proteasome (KP, adj. P = 0.0001) in sEVs (Fig. 5B), and in protein folding shaperone (MF, adj. P = 0.0016), unfolded protein binding (MF, adj. P = 0.0113), and protein processing in the endoscopic reticulum (KP, adj. P = 0.0344) in cEVs (Fig. 5D).

[0160] Next, we calculated the mean values ​​of the proteins that increased or decreased over time in SALS patients and examined the time-dependent changes in the placebo and ROPI groups up to 48 weeks (Fig. 5E). Both the placebo and ROPI groups deviated from the control group distribution over time, but the ROPI group distribution was suppressed relative to the deviation from the control group distribution.

[0161] These results suggest that SALS induces inflammation, typified by the complement and coagulation cascades, and reduces the EV content of proteins involved in the UPR, but that ROPI administration ameliorates inflammation and increases UPR-related proteins (Fig. 5F).

[0162] Figures 5A–F show a comparison of DAP, es-DAP, and protein changes over time. Figures 5A and 5C are heat maps showing the results of cosine similarity analysis of DAP (controls vs. SALS patients), es-DAP (placebo 0-24w vs. ROPI 0-24w), and proteins that changed over time (sEV: Figure 5A, cEV: Figure 5C). Figures 5B and 5D are bar graphs showing the GO (BP, CC, MF) and KEGG pathway results for proteins that increased / decreased or increased / decreased over time in SALS patients and decreased / increased by ROPI (sEV: Figure 5B, cEV: Figure 5D). Figure 5E is a scatter plot showing the difference in protein levels from 0w on the vertical axis (increased or increased over time in SALS patients) and the difference in protein levels from 0w on the horizontal axis (decreased or decreased over time in SALS patients) (sEV: left, cEV: right). Light-colored dots indicate individual samples, and arrows indicate trends over time. The black boxes indicate the mean values ​​for each group. The control plots show the difference between the mean protein content at 0w and the mean value for each patient. Figure 5F is a schematic diagram of the protein composition in control and SALS EVs and its changes with ROPI.

[0163] <<Reverse Translational Research: ROPI-Induced Changes in the Transcriptomes of iPSC-Derived Astrocytes and iPSC-MNs>> We investigated the possibility that ROPI acts on astrocytes and motor neurons to suppress neuroinflammation. First, we generated iPSC-derived astrocytes (iPast) from healthy individuals and performed RNA-seq on astrocytes with and without ROPI treatment. Furthermore, we obtained RNA-seq data for induced pluripotent stem cell-derived spinal motor neurons (iPSC-MNs) from healthy individuals with and without ROPI treatment from the NCBI GEO database. PCA revealed that the transcriptomes of iPasts and iPSC-MNs under ROPI treatment and those without treatment were clearly separated into two groups, suggesting that ROPI treatment induces changes in the transcriptomes of iPasts and iPSC-MNs (Figures 6A and 6E). Next, we confirmed sufficient expression of DRD2 (Figures 6B and 6F), suggesting that both iPasts and iPSC-MNs may be affected by the D2R-dependent effects of ROPI treatment.

[0164] Furthermore, we examined the transcriptome changes induced by ROPI treatment and identified nine and eight genes whose expression levels were significantly altered in iPasts and iPSC-MNs, respectively (Figures 6C and 6G). In iPasts, ROPI significantly upregulated the expression of CRYAB and downregulated the expression of the chemokines CXCL14 and CCN2 (Figure 6D, left panel). However, the expression changes induced by ROPI treatment in iPasts were not consistent with those observed in iPSC-MNs (Figure 6H, left panel). Genes involved in cholesterol synthesis, such as MVD and ACAT2, were significantly decreased in iPSC-MNs by ROPI treatment, and a similar trend was observed in iPasts (Figures 6D and 6H, right panel). The inhibitory effect of ROPI on the expression of genes involved in cholesterol synthesis was consistent with a previous report examining iPSC-MNs from SALS patients.

[0165] Figures 6A-H show the results of reverse translation studies: ROPI delivery to iPast and iPSC-MN. Figures 6A and 6E show the PCA results of the transcriptomes of iPast and iPSC-MN with and without ROPI treatment (circle dots: (+) ROPI, diamond dots: (-) ROPI). Figures 6B and 6F show dopamine receptor expression in iPast and iPSC-MN with and without ROPI treatment. Figures 6C and 6G show volcano diagrams showing changes in iPast and iPSC-MN with and without ROPI treatment. Figures 6D and 6H show heat maps showing the expression levels of DEGs identified in the comparative analysis of iPast and iPSC-MN with and without ROPI treatment. Figure 6I shows a schematic diagram of DRD2 expression and changes in expression in iPast and iPSC-MN with ROPI treatment.

[0166] <<Searching for biomarkers using proteins in EVs as clinical indicators of ALS>> Based on the detailed patient data collected in the ROPALS study, we searched for biomarkers using proteins in EVs as clinical indicators of ALS. First, we approximated the changes in the ALSFRS-R, an important clinical indicator of ALS symptoms, during the ROPALS study using the following model formula. The z-score-transformed a value (adj. a value) for each ROPI group and placebo group was calculated using the following formula: -log 10 a, where a represents the disease progression rate for each patient (Figure 7A). A higher adj. a value indicates slower disease progression. aALSFRS-R was calculated from the coefficients a and b at each sampling time. The sentinel disease progression rate was calculated by differentiating the model equation at each sampling time.

[0167]

[0168] A correlation analysis was performed between the content of each protein in EVs at 0w and the adj. a value (sEVs: Figure 7B; cEVs: Figure 7E). Osteoglycin (OGN) and FERM and PDZ domain-containing protein 1 (FRMPD1) were identified as the markers that best predicted the adj. a value, i.e., the prognostic value of sEVs (R = 0.809, P < 0.001, Figures 7C and 7D) and cEVs (R = 0.641, P = 0.0029, Figures 7F and 7G).

[0169] Furthermore, correlation analyses were performed between each protein in EVs at each sampling time, aALSFRS-R (not shown), and the progression rate of aALSFRS-R at that time point (not shown).

[0170] Figures 7A–G show the results of biomarker discovery using EV proteins to identify clinical indicators of SALS. Figure 7A shows the aALSFRS-R calculated by applying an approximation formula to the trends in ALSFRS-R scores of 20 SALS patients enrolled in the ROPALS study. The center line in the left panel represents the measured and predicted ALSFRS-R (aALSFRS-R), and the table on the right shows the predicted results. Figures 7B and 7E show rank plots showing the correlation analysis results between the abundance of each protein in sEVs or cEVs at 0w and the subsequent rate of adaptive progression (normalized a-value) (sEV: Figure 7B, cEV: Figure 7E). Colors indicate the top five positive and negative proteins. Figures 7C and 7F show tables of the top five positive and negative proteins (sEV: Figure 7C, cEV: Figure 7F). Figures 7D and 7G are scatter plots showing the proteins with the highest prediction accuracy compared to different indices (sEV: Figure 7D, cEV: Figure 7G).

[0171] <<Diagnostic biomarker discovery using machine learning and machine learning-based classifier generation>> Machine learning methods were used to build a classifier to distinguish between control and SALS patient samples based on EV protein composition.

[0172] The accuracy of each model on the sEV test dataset was 100% for the Random Forest Classifier, Gradient Boosting Classifier, and AdaBoost Classifier. Furthermore, the Gradient Boosting Classifier achieved the highest average accuracy for validation dataset 1 (ROPI-naive) and validation dataset 2 (ROPI-exposed), which contained batches different from the training and test datasets, with 155 proteins selected as features and an accuracy of 90.4% (Figure 8B).

[0173] The accuracy of each model on the cEV test dataset was 100% for Perceptron, Random Forest, Extra Trees, Decision Tree, and XBG Classifier. Furthermore, the Random Forest Classifier had the highest average accuracy of each model on the validation dataset, selecting 19 proteins as features with an accuracy of 80.3% (Figure 8C).

[0174] Figures 8A–C illustrate the use of machine learning techniques to discover diagnostic biomarkers and generate machine learning-based classifiers. Figure 8A is a schematic diagram of the training framework for limited sample sizes and imbalanced datasets. This framework consists of three steps: oversampling, parameter search, and feature selection. Figures 8B and 8C are tables and heat maps showing the accuracy of the trained models according to protein composition (sEV: Figure 8B and cEV: Figure 8C). The test datasets contain equal numbers of samples from controls and SALS patients (ROPI-naïve). The ROPI-naïve dataset consists of specimens collected from SALS patients before ROPI administration, and the ROPI-exposed dataset consists of specimens collected from SALS patients after ROPI administration. Accuracy is expressed as a percentage, and the average accuracy of the ROPI-naïve and ROPI-exposed datasets is shown in the heat map.

[0175] Discussion: A key finding of this study is the characterization of the protein profiles of sEVs and cEVs derived from SALS patients. Compared with controls, these EVs were characterized by an upregulation of inflammation-related proteins, such as proteins involved in the complement and coagulation cascades, and a downregulation of proteins associated with the UPR. These changes manifested over time and in a pathological progression-dependent manner (Fig. 2A–E). Specifically, the detection of SALS-specific proteins, such as immunoglobulin complexes and C-reactive protein, in cEVs derived from SALS patients suggests inflammation in the CSF of these patients. These proteins are known activators of the complement pathway, particularly the classical pathway, and this finding is consistent with the results of this study. Previous studies have similarly demonstrated an increase in inflammatory complement proteins in astrocyte-derived EVs from Alzheimer's disease patients and in animal models of Gulf War syndrome. These findings suggest that neuroinflammation associated with neurodegeneration may enhance complement-related proteins in EVs, exacerbating neurological disorders.

[0176] In contrast, levels of UPR-related proteins were reduced in sEVs and cEVs from SALS patients (Figure 2E). Previous studies have shown reduced sEV HSP90 levels in ALS patients (without distinguishing between familial ALS and SALS), which is consistent with the results of this study. In vitro studies suggest that UPR-related proteins, including HSP family proteins, alleviate TDP-43 aggregation. Of note, UPR-related proteins, such as HSP70 and HSP90 family proteins, transported by EVs enhance the protein folding environment in recipient cells. Astrocytes also release HSP70-rich EVs under stress conditions. These observations suggest that EVs transporting HSPs and other UPR proteins exert neuroprotective effects. The reduction of UPR-related proteins in EVs from SALS patients may indirectly promote TDP-43 aggregation in motor neurons, exacerbating SALS pathology. Interestingly, UPR-associated protein levels were decreased in both sEVs and cEVs, suggesting that the reduction of these proteins in systemic circulating EVs as well as the brain and spinal cord may be involved in the pathogenesis of SALS.

[0177] On the other hand, the increase in complement and coagulation-related proteins and decrease in UPR-related proteins in EVs observed in this study may be common to some extent in neurodegenerative diseases. As mentioned above, an increase in inflammatory proteins in EVs is associated with neuroinflammation. In neurodegenerative diseases, inflammatory responses are triggered by neurodegeneration, so the increase in complement and coagulation-related proteins in EVs may not be specific to SALS but may be a common phenomenon in neurodegenerative diseases to some extent. Similarly, the decrease in UPR-related proteins may be involved in the disruption of neuronal proteostasis. It cannot be ruled out that the decrease in UPR-related proteins may be a common phenomenon in neurodegenerative diseases, where the accumulation of degenerated proteins is common. Future comprehensive studies of the protein composition in EVs of not only SALS but also other neurodegenerative diseases, such as Alzheimer's disease and Parkinson's disease, may advance our understanding of neurodegenerative diseases as a whole.

[0178] Furthermore, comparison of the protein composition of sEVs and cEVs revealed similar trends in mutations in disease-associated proteins (Figure 1F). The protein content of sEVs and cEVs showed remarkable similarities between SALS patients and controls, suggesting, at least in part, increased blood-brain barrier permeability in SALS patients. This observation is consistent with previous reports of increased blood-brain barrier permeability in ALS patients and suggests that peripheral blood-derived EVs may be useful for the diagnosis and treatment of central nervous system lesions.

[0179] Another important finding of this study is that ROPI administration regulated the pathological changes in EV protein profiles toward control or early pathology. ROPI, a D2R agonist, may be involved in the observed reduction of inflammation-related proteins. Many studies support the important role of D2R in reducing neuroinflammation by microglia and astrocytes. D2R activation is thought to induce increased expression of CRYAB and inhibit its nuclear translocation by promoting its binding to STAT3 and NF-κB in the cytoplasm.

[0180] In this study, we confirmed sufficient expression of DRD2 in iPasts from healthy individuals (Figure 6B), and ROPI administration significantly increased CRYAB expression and significantly decreased CCN2 and CXCL14 expression (Figures 6C and 6D). As previously described, D2R stimulation in astrocytes regulates innate immunity through CRYAB and suppresses neuroinflammation. Conversely, CCN2, as part of the CTGF / CCN2 complex, increases the expression of inflammatory genes, including chemokines and cytokines, and enhances inflammatory responses through astrocyte activation. Furthermore, CTGF was significantly upregulated in reactive astrocytes in the spinal cord anterior horn and white matter of patients with SALS and familial ALS. Although the neuroinflammatory role of CXCL14 is not fully understood, overexpression of CXCL14 exacerbates collagen-induced arthritis, and its addition to dendritic cells increases NF-κB activity and its activation, suggesting its involvement in inflammatory responses. However, ROPI-induced expression changes in iPSC-MNs were not clearly consistent with those in iPasts (Figures 6C-D, 6G-H), suggesting that the reduction in inflammation-related proteins in EVs by ROPI administration may be due to the effects of ROPI on cells other than motor neurons, such as astrocytes. These results and previous studies strongly support the possibility that ROPI suppresses neuroinflammation through D2R stimulation and suppresses complement and coagulation-related protein levels in EVs. Furthermore, suppressing inflammation and reducing the levels of proteins involved in the complement and coagulation cascades in ALS patients may be potential therapeutic targets for ALS.

[0181] We also investigated biomarkers for various clinical indicators of ALS. Among all candidate biomarkers in sEVs and cEVs, OGN (also known as mimecan) in sEVs was the best marker for predicting disease progression. OGN is a tandem leucine-rich repeat-containing protein that plays an important role in extracellular matrix assembly and influences bone formation and fibrosis. Interestingly, OGN significantly increased IGF-2 / IGFb-induced neurite outgrowth. In this study, sEVs with higher baseline OGN concentrations were associated with slower disease progression and a better prognosis, suggesting that OGN in peripheral blood EVs may have anti-ALS effects on neurons. However, given the small sample size and the lack of comprehensive protein discovery in EVs, these results may be merely correlational. Furthermore, it remains unclear whether these candidate biomarker proteins are transported to MNs by EVs from other cell types, and if so, whether they play a functional role within MNs. To investigate this, one possible approach would be to generate non-MN central nervous system cells, such as astrocytes, oligodendrocytes, and microglia, from iPSCs and co-culture them with MNs. This method would facilitate research into the influence of non-MN cells in ALS, particularly focusing on EV-mediated protein transport. Therefore, further large-scale studies are needed to determine whether these biomarkers are useful and have biological significance.

[0182] Furthermore, this study evaluated the performance of trained machine learning diagnostic models on the protein profiles of sEV and cEV datasets. The Gradient Boosting Classifier performed well on the sEV dataset (155 selected features; 100% accuracy on the test dataset; 90.4% accuracy on the validation dataset), while the Random Forest Classifier performed best on the cEV dataset (19 selected features; 100% accuracy on the test dataset; 80.3% accuracy on the validation dataset). Because the test dataset contained approximately half of the control samples, these trained models are estimated to be able to distinguish between control and SALS patient samples with high accuracy. Furthermore, SALS data measured in different batches were also able to be distinguished with high accuracy, suggesting the versatility of these trained models. Interestingly, the machine learning model classification was more accurate for sEVs than for cEVs, suggesting that blood, which can be collected less invasively, can be used with sufficient accuracy for diagnosing SALS. However, several factors need to be considered in this study. First, the sample size was small, which may affect the robustness of the results. Second, differences between batches may affect the accuracy of the results. Third, because this study was based on a 60-week clinical trial, long-term prognosis could not be included in the analysis. Finally, the selected features do not necessarily guarantee biological interpretability. Therefore, further validation through larger, longer-term studies is essential for the development of diagnostic methods based on EV protein composition and their clinical application.

[0183] Most previous ALS studies of fluid-derived EV proteins have focused on EV TDP-43. Because EVs function as "garbage cans" that release abnormal proteins accumulated within cells, detecting TDP-43 in EVs is important for diagnosing ALS and understanding its pathophysiology. However, by identifying the comprehensive protein composition in EVs, we revealed that various functional proteins are altered in ALS, strongly suggesting that these proteins may be involved in the pathogenesis of the disease. Furthermore, the protein composition significantly changed with disease progression over time and was suppressed by ROPI administration.

[0184] Conclusions: Our comprehensive EV protein composition analysis revealed changes in various functional proteins in ALS, which may be involved in the pathogenesis. Furthermore, this composition significantly changed over time and was suppressed by ROPI administration. Our approach is highly useful for understanding the pathogenesis and drug mechanisms of ALS and awaits validation in future large-scale studies.

[0185] Although the examples of the present invention have been described in detail above, the present invention is not limited to the specific embodiments described above, and various modifications and changes are possible within the scope of the gist of the present invention as set forth in the claims.

[0186] This international application claims priority to U.S. patent application Ser. No. 63 / 661,114, filed Jun. 18, 2024, the entire contents of which are incorporated herein by reference.

[0187] 1 ALS determination system 10 Analyzer 11 Data management device 20 Learning device 30 Target 101 Data acquisition unit 102 Determination unit 103 Display unit 201 Learning data acquisition unit 202 Learning data storage unit 203 Machine learning unit 204 Determination model storage unit 301 Data acquisition unit 302 Determination unit 303 Display unit 304 Determination model storage unit 1001 Control unit 1002 Main storage unit 1003 Auxiliary storage unit 1004 Input unit 1005 Output unit 1006 Interface unit

Claims

1. Biomarkers for diagnosis, disease-modifying drug treatment effect prediction and assessment, monitoring, and stratification in amyotrophic lateral sclerosis, including: (1) B4GALT1, C1RL, C4A, C4B, C8B, C9, CD14, CFH, COL6A3, FBLN1, PROS1, SDCBP, SERPINA1, SERPINA3, SERPINA5, SERPINA7, SERPINC1, SERPIND1, SERPINF2, SERPING1, APOL1, CNDP1, EFEMP1, ISLR, MINPP1, OAF, PGLYRP2, QSOX1, SVEP1, TGFBR3, and TTR; OGN, FETUB, VASN, SELENOP, and CNDP1; and FRMPD1, COL4A2, TRAP1, PDGFRA, and TGM3, and / or (2) one or more first proteins selected from a first protein group consisting of ACTN4, ARPC2, ARPC4, CFL1, HSP90AA1, HSP90AB1, HSPA1A, MAP2K2, PFN1, ACTN1, SPTBN1, TXN, CALR, DNAJC3, GANAB, HSPA5, HSPA8, HSPD1, LMAN2, NECTIN2, P4HB, PDIA3, UGGT1, AARS1, ARHGDIA, CBLN4, CETP, ENO1, GNB1, GSTP1, IGFALS, PCYOX1, PGAM1, PKM, RAB14, SCPEP1, SLITRK4, and YWHAZ; RPS4X, CCDC144A, BLMH, LCN2, and S100A9; and RAB14, A biomarker characterized by being one or more second proteins selected from a second protein group consisting of ATP6V1B2, ACLY, USP14, and TIMP2.

2. A method for diagnosing amyotrophic lateral sclerosis, predicting and diagnosing the therapeutic effect of a disease-modifying drug, monitoring, and stratifying the disease, comprising detecting a detected amount D of the biomarker according to claim 1 in a sample obtained from a subject. ｓ and obtaining a detected amount D of the biomarker in a sample obtained from a control group. ｃ (1) the biomarker is the first protein, and D ｓ >D ｃ and / or (2) the biomarker is the second protein, and D ｓ <D ｃ and determining that the patient is at risk of amyotrophic lateral sclerosis if the patient is at risk of amyotrophic lateral sclerosis.

3. The method according to claim 2, wherein the sample is at least one of blood, serum, plasma, cerebrospinal fluid, saliva, and urine.

4. The detected amount D of the biomarker ｓ and D ｃ The method of claim 3, wherein the detected amount of the biomarker in extracellular vesicle proteins is .

5. The detected amount D of the biomarker in the sample obtained from the subject ｓ０ and the detected amount D of the biomarker in the sample obtained from the subject after a certain time has elapsed. ｓ１ (1) the biomarker is the first protein; and (2) comparing ｓ０ >D ｓ１ When the above condition is met, the patient falls into one of the following stratification groups: high response to disease-modifying drug treatment, high effect of disease-modifying drug treatment, no progression of disease or slow progression of disease; D ｓ０ =D ｓ１ If the condition is as above, there is no change in the course of the disease, or ｓ０ <D ｓ１ or the patient is in any of the stratified groups of rapid progression of the disease, low response to disease-modifying drug treatment, low effect of disease-modifying drug treatment, and progression of the disease; (2) the biomarker is the second protein; ｓ０ <D ｓ１ When the above condition is met, the patient falls into one of the following stratification groups: high response to disease-modifying drug treatment, high effect of disease-modifying drug treatment, no progression of disease or slow progression of disease; D ｓ０ =D ｓ１ If the condition is as above, there is no change in the course of the disease, or ｓ０ >D ｓ１ and determining, when the condition is as follows, that the patient belongs to any of the stratification groups of low response to disease-modifying drug treatment, low effect of disease-modifying drug treatment, progressive disease, or rapid disease progression.

6. The method of claim 5, further comprising administering a medication to said subject during said period of time.

7. The method of claim 6, wherein the medication comprises one or more medications selected from the group consisting of ropinirole hydrochloride, riluzole, edaravone, methylcobalamin, bosutinib, hydrochlorothiazide, cortisone acetate, aldioxa, sibutramine hydrochloride, alendronate sodium hydrate, ceftibuten, etropibate, and ketorolac tromethamine.

8. A device for diagnosing amyotrophic lateral sclerosis, predicting and diagnosing the therapeutic effect of disease-modifying drugs, monitoring, and stratifying the disease, comprising: a detection amount D of the biomarker according to claim 1 in a sample obtained from a subject; ｓ a data acquisition unit that acquires the detected amount D of the biomarker in the sample obtained from the control group; ｃ (1) the biomarker is the first protein, and D ｓ >D ｃ and / or (2) the biomarker is the second protein, and D ｓ <D ｃ and a display unit that displays the subject's risk of amyotrophic lateral sclerosis.

9. A program for diagnosis of amyotrophic lateral sclerosis, prediction and diagnosis of the therapeutic effect of a disease-modifying drug, monitoring, and stratification, comprising: a device; a detection amount D of the biomarker according to claim 1 in a sample obtained from a subject; ｓ a data acquisition unit that acquires the detected amount D of the biomarker in the sample obtained from the control group; ｃ (1) the biomarker is the first protein, and D ｓ >D ｃ and / or (2) the biomarker is the second protein, and D ｓ <D ｃ a determination unit that determines that the subject is at risk of amyotrophic lateral sclerosis when the subject is at risk of amyotrophic lateral sclerosis; and a display unit that displays the subject's risk of amyotrophic lateral sclerosis.

10. A method for determining amyotrophic lateral sclerosis, predicting and determining the therapeutic effect of disease-modifying drugs, monitoring, and stratifying, comprising the steps of: acquiring extracellular vesicle protein analysis data of a sample obtained from a subject; and inputting the detected amount D of the subject into a determination model for determining the risk of amyotrophic lateral sclerosis for a subject group based on the extracellular vesicle protein analysis data of the subject group. ｓ and outputting a risk of the subject having amyotrophic lateral sclerosis; and displaying the risk of the subject having amyotrophic lateral sclerosis.

Images