Assay for TDP-43 determination

Abstract

The present invention is related to the field of medical diagnosis. In particular, we disclose the development and validation of a novel assay for TDP-43 determination, which is highly sensitive, can quantify TDP-43 levels in human lymphocytes and is useful in the diagnosis of ALS and other related TDP-43 proteinopathies.

Description

[0001] Assay for TDP-43 determination

[0002] TECHNICAL FIELD

[0003] The present invention is related to the field of medical diagnosis. In particular, we disclose the development and validation of a novel assay for TDP-43 determination, which is highly sensitive, can quantify TDP-43 levels in human lymphocytes and is useful in the diagnosis of ALS and other related TDP-43 proteinopathies.

[0004] BACKGROUND ART TAR DNA-binding protein 43 (TDP-43) is a 414-amino acid nuclear protein belonging to the heterogeneous ribonucleoprotein (hnRNP) family and has been identified as a major component of ubiquitin-positive inclusions found in degenerating neurons and glial cells of patients with frontotemporal lobar degeneration (FTLD) and amyotrophic lateral sclerosis (ALS). Since the systemic degeneration of motor neurons in ALS causes progressive muscle atrophy, paralysis, and eventual death within 2-5 years, early and accurate diagnosis would be valuable for patients, their families, and doctors. However, the early diagnosis of ALS is challenging because differentiating it from other diseases that mimic ALS based only on clinical symptoms and signs is difficult. To address this problem, molecular biomarkers, especially fluid biomarkers that are safe, inexpensive, and widely accessible, are necessary. Thus far, neurofilament light chain (NfL) is the most studied fluid biomarker for ALS, and its levels in the cerebrospinal fluid (CSF) and serum / plasma have been reported to increase in patients with ALS compared with those in controls. Moreover, they were associated with poor outcomes. Despite these promising results, changes in NfL levels in the CSF and blood are not specific to ALS; therefore, other fluid biomarkers that reflect pathognomonic pathologies of ALS, such as TDP-43 accumulation in neurons, are still needed.

[0005] Thus far, some studies have used conventional enzyme-linked immunosorbent assay (ELISA) techniques that reported the potential usefulness of TDP-43 levels in the CSF and plasma as a biochemical biomarker to support ALS diagnosis. Although these studies have identified high TDP-43 levels in the CSF or plasma from patients with ALS compared with controls, the absolute concentrations of TDP-43 immunoassays are inconsistent for measuring this protein in biofluids. This is primarily caused by those studies that used ELISA techniques that did not have sufficient sensitivity to correctly measure small amounts of TDP-43 in human biofluids. To overcome this insufficientsensitivity, authors have reported using different techniques, but none have disclosed a useful method to identified high TDP-43 levels in non-immortalized lymphocytes isolated from subjects at risk of suffering from

[0006] In this invention, we report the development and validation of a novel assay for TDP-43 determination, which is highly sensitive, can quantify TDP-43 levels in human lymphocytes and is useful in the diagnosis of ALS and other related TDP-43 proteinopathies.

[0007] BRIEF DESCRIPTION OF THE FIGURES

[0008] Figure 1- Flow cytometry analysis of TDP-43 and pTDP-43 in ALS patients lymphoblasts and healthy controls. (A) Single patient analysis of TDP-43 using QD605 and QD655 secondary antibody conjugate and QD655 streptavidin conjugates. (B) TDP-43 quantification using QD655 and Alexa488 in healthy controls (red) and ALS patients (green). (*p<0.05, **p<0.01)

[0009] Figure 2- Flow cytometry quantification of ALS lymphocytes. (A) Single-patient quantification comparing controls and ALS patients representing only QD MFI. (B) Quantification comparing controls and ALS patients using the QD / Alexa TDP-43 MFI ratio.

[0010] Figure 3- Flow cytometry quantification of FTLD lymphocytes. Quantification comparing controls and FTLD patients using the QD / Alexa TDP-43 MFI ratio.

[0011] Figure 4- Summary table reporting the raw counts for nuclear, cytoplasmic, and total signal in mononuclear cells from controls, pre-ALS subjects, and ALS patients, together with the calculated nucleus-to-cytoplasm (N / C) and cytoplasm-to-total (C / T) ratios for each group.

[0012] Figure 5- Bar plots showing the mean nucleus-to-cytoplasm (N / C) and cytoplasm-to-total (C / T) ratios across groups (CONTROL, preALS, ALS)

[0013] Figure 6- One-way ANOVA analysis of the cytoplasm-to-total ratio (C / T) across groups.

[0014] Figure 7- One-way ANOVA analysis of the nucleus-to-cytoplasm ratio (N / C) across groups.

[0015] DETAILED DESCRIPTION OF THE INVENTIONIn the present invention, we herein confront the problem of developing a novel assay for TDP-43 determination, which is highly sensitive, can quantify TDP-43 levels in human non-immortalized lymphocytes and is useful in the diagnosis of ALS and other related TDP-43 proteinopathies.

[0016] In example 1, we observed that in an established cell line (immortalized lymphocytes) derived from healthy individuals and ALS patients, the selective cytosolic labelling by using quantum dots, resulted in significant differences in the TDP-43 levels between ALS patients and healthy individuals or treated ALS patients’ samples (Figure 1A). However, if instead of using a selective cytosolic labelling such as by using quantum dots, TDP-43 levels are determined by using traditional fluorophores without the ability of selective cytosolic labelling (such as Alexa), healthy individuals and ALS patients present similar values (Figure 1B).

[0017] Based on the above, we decided to reproduce the results from example 1 but in fresh lymphocytes. However, when working with lymphocytes directly obtained from a blood draw, we observed highly variable results with both QD and Alexa staining values, it was thus not possible to correctly strategy ALS patients from healthy individuals (Figure 2A). To resolve this problem, instead of using absolute values we used the ratio QD / Alexa as a measure of TDP-43 pathology. Remarkably, by using said ratio, ALS patients and healthy controls were correctly grouped (Figure 2B).

[0018] In conclusion, this novel assay for TDP-43 determination could be used not only for correctly classifying ALS patients from healthy controls but specifically to measure TDP-43 pathology burden and monitor pharmacological activity at the molecular level from a blood draw.

[0019] Therefore, a first aspect of the invention refers to a method for calculating a ratio of the cytoplasmic concentration to the total cellular concentration of TDP-43 in an isolated biological sample comprising mononuclear immunological cells, wherein the method comprises;

[0020] a. determining the cytoplasmic concentration of TDP-43 in the mononuclear immunological cells, wherein the cytoplasmic concentration of the TDP- 43 protein is determined by detecting and quantifying the signals from Quantum Dots (QDs) by preferably using flow cytometry;

[0021] b. determining the total cellular concentration of TDP-43 in the mononuclear immunological cells by using flow cytometry; and

[0022] c. calculating a ratio of the cytoplasmic concentration to the total cellular concentration of TDP-43;and wherein the mononuclear immunological cells are primary, whole, non-immortalized cells, obtained directly from peripheral blood from a subject, excluding any artificially immortalized or cultured cell lines.

[0023] The term “whole cell”, as used herein, refers to a cell that has not undergone physical separation or lysis of any of its compartments, such as the nucleus or the cytoplasm, thereby maintaining its membrane and organelles intact.

[0024] It is noted that the methods describe herein (be it the method described in the first, second or second alternative aspect) include standard flow cytometry permeabilization steps to allow the binding moieties to access the cells intracellular compartments while maintaining overall cellular integrity. Such permeabilization steps are known to the skilled person. Permeabilization can be performed under conditions to avoid complete lysis or fractionation of the cells. Preferably, for permeabilizing the cells reagents can be used, such as those selected from a group consisting of: Tween 20 detergents, Triton X-100 detergents, saponin-based buffers, NP-40 buffers, digitonin buffers, and methanol or mixtures thereof in PBS, applied at concentrations and incubating times effectives to permeabilize the cell without disrupting the overall cell integrity.

[0025] That is, in a preferred embodiment (also of the second aspect of the invention, including its alternative second aspect), permeabilization is performed under conditions to avoid complete lysis or fractionation of the cells to allow the binding moieties of steps a) and b) to access the cells intracellular compartments while maintaining overall cellular integrity. Preferably, for permeabilizing the cells reagents can be used, such as those selected from a group consisting of: Tween 20 detergents, Triton X-100 detergents, saponin-based buffers, NP-40 buffers, digitonin buffers, and methanol or mixtures thereof in PBS, applied at concentrations and incubating times effectives to permeabilize the cell without disrupting the overall cell integrity.

[0026] In a preferred embodiment of the first aspect of the invention (from herein referred to as “first method to determine the cytoplasmic concentration of the TDP-43 protein”), the cytoplasmic concentration of the TDP-43 protein is determined by incubating the biological sample with a first binding moiety selected from the group consisting of an antibody or an antibody fragment such as, but not limited to, nanobodies, wherein said first binding moiety specifically binds the TDP-43 protein, and incubating the biological sample with a second binding moiety selected from the group consisting of an antibody or an antibody fragment such as, but not limited to, nanobodies, conjugated to one or more Quantum Dots (QD) under conditions sufficient to form a complex with the firstbinding moiety; and analysing the biological sample to detect signals from QDs, thereby determining the cytoplasmic concentration of the TDP-43 protein.

[0027] The term 'antibody fragment', as indicated throughout the present invention, refers to a portion of an intact immunoglobulin molecule that comprises sufficient regions to confer specific binding to a target antigen. This term includes, but is not limited to Fab, F(ab')2, Fv, single-chain variable fragments (scFv), and single-domain antibodies (sdAb / VHH). In another preferred embodiment of the first aspect of the invention (from hereinafter “second method to determine the cytoplasmic concentration of the TDP-43 protein”), optionally in combination with any previous or subsequent preferred embodiments of the first aspect of the invention, the cytoplasmic concentration of the TDP-43 protein is determined by incubating the biological sample with a first binding moiety selected from the group consisting an antibody or an antibody fragment that specifically binds the TDP-43 protein, and incubating the biological sample with a second binding moiety selected from the group consisting an antibody or an antibody fragment and one or more quantum dots, wherein the second binding moiety specifically binds to the one or more QDs under conditions sufficient to first form a complex between the first and second binding moiety and second through a high-affinity binding interaction between the second binding moiety and the QDs, and analysing the biological sample to detect signals from QDs, thereby determining the cytoplasmic concentration of the TDP-43 protein.

[0028] In another preferred embodiment of the first aspect of the invention (from hereinafter referred to as “first method to determine the total concentration of the TDP-43 protein”), optionally in combination with any previous or subsequent preferred embodiments of the first aspect of the invention, the total cellular concentration of the TDP-43 protein is determined by incubating the biological sample with a first binding moiety selected from the group consisting of antibodies or antibody fragments that specifically binds the TDP-43 protein and incubating the biological sample with a second binding moiety selected from the group consisting of antibodies or antibody fragments conjugated to a fluorophore (F), distinct from a QD and capable of labelling the nucleus and the cytoplasm of the cells, under conditions sufficient to form a complex with the first binding moiety and analysing the biological sample to detect signals from F, thereby determining the total cellular concentration of the TDP-43 protein.

[0029] In another preferred embodiment of the first aspect of the invention (from hereinafter referred to as “second method to determine the total concentration of the TDP-43 protein”), optionally in combination with any previous or subsequent preferred embodiments of the first aspect of the invention, the total cellular concentration of theTDP-43 protein is determined by incubating the biological sample with a first binding moiety selected from the group consisting of antibodies and antibody fragments that specifically bind the TDP-43 protein and incubating the biological sample with a second binding moiety selected from the group consisting of antibodies and antibody fragments and one or more fluorophores (F), distinct from a QD and capable of labelling the nucleus and the cytoplasm of the cells, under conditions sufficient to first form a complex between the first and second binding moiety and second through a high-affinity binding interaction between the second binding moiety and the Fs, and analysing the biological sample to detect signals from Fs, thereby determining the total cellular concentration of the TDP-43 protein.

[0030] In another preferred embodiment of the first aspect of the invention, optionally in combination with any previous or subsequent preferred embodiments of the first aspect of the invention, the cytoplasmic concentration of the TDP-43 protein is determined according to the first or second method to determine the cytoplasmic concentration of the TDP-43 protein and the total cellular concentration of the TDP-43 protein is determined according to the first or second method to determine the cytoplasmic concentration of the TDP-43 protein.

[0031] In another preferred embodiment of the first aspect of the invention, optionally in combination with any previous or subsequent preferred embodiments of the first aspect of the invention, the QD and the fluorophore (F) are selected from different emission spectra suitable for multicolor flow cytometry, allowing simultaneous detection of multiple signals, preferably wherein the QD is Qdot 655 and the fluorophore (F) is Alexa fluor 488. In another preferred embodiment of the first aspect of the invention, optionally in combination with any previous or subsequent preferred embodiments of the first aspect of the invention, calculating the ratio from step (c) is performed by a computer-implemented algorithm that processes input data obtained from the measurements indicated in steps (a) and (b).

[0032] A second aspect of the invention refers to a method for screening or identifying subjects at risk of suffering or having a TDP-43 proteinopathy, the method comprising calculating a ratio of the cytoplasmic concentration to the total cellular concentration of TDP-43 in mononuclear immunological cells (from herein after “cell sample”) isolated from a blood sample of the subject; comparing the calculated ratio to one or more predefined decision thresholds to classify the subject into at least one category indicative of a higher or lower likelihood of disease presence; and generating a risk stratification output that specifieswhether the subject falls within a category associated with an increased or decreased likelihood of having the disease (TDP-43 proteinopathy).

[0033] TDP-43 proteinopathies consist of a group of neurodegenerative diseases defined by the pathological presence of misfolded proteins and insoluble deposits of the transactive response DNA-binding protein of 43 kDa (TDP-43) in the central nervous system (CNS), in association with progressive neuronal loss and gliosis. Preferred TDP-43 proteinopathies are selected from the list consisting of amyotrophic lateral sclerosis, frontotemporal lobar degeneration or limbic-predominant age-related TDP-43 encephalopathy. Preferably, amyotrophic lateral sclerosis.

[0034] The term "cytoplasmic concentration," as used herein, refers to the measurable quantity of TDP-43 localized within the cytoplasmic compartment of a mononuclear immunological cell in the cell sample, expressed per unit volume or mass of the cytoplasm.

[0035] The term "total cellular concentration," as used herein, refers to the aggregate measurable quantity of TDP-43 present throughout the entirety of the mononuclear immunological cell in the cell sample, encompassing all cellular compartments, expressed per unit volume or mass of the cell.

[0036] The term "subject," as used herein, refers to a living mammal, including but not limited to humans, from whom a biological sample, such as a blood sample, is obtained for the purpose of evaluating biomarker concentrations, determining the likelihood of disease presence, or performing risk stratification. Preferably, the subject is a human being, encompassing individuals of any age, gender, or health status.

[0037] The term "predefined decision thresholds," as used herein, refers to one or more predetermined quantitative or qualitative criteria established based on empirical data, statistical models, or expert consensus, against which the calculated ratio of cytoplasmic concentration to total cellular concentration of the biomarker is compared to categorize the subject’s likelihood of disease presence.

[0038] The term "category," as used herein, refers to a predefined classification group into which a subject is placed based on the comparison of biomarker concentration ratios to the predefined decision thresholds, such categories being indicative of distinct levels of likelihood of disease presence.

[0039] The term "higher or lower likelihood of disease presence," as used herein, refers to the relative probability that a subject either currently has, or is at increased or decreased risk of developing, a specific disease, condition, or pathological state. This probability isquantitatively or qualitatively determined by comparing a calculated biomarker value or ratio, such as the cytoplasmic concentration-to-total cellular concentration ratio of a biomarker, to one or more predefined decision thresholds.

[0040] The term "risk stratification output," as used herein, refers to a generated result or report, derived from the classification of the subject based on biomarker analysis, which specifies whether the subject belongs to a category associated with an increased or decreased likelihood of disease presence, potentially guiding clinical decisions or further diagnostic evaluation.

[0041] Mononuclear immunological cells, as used herein, are a subclass of immune cells characterized by the presence of a singular, non-lobulated nucleus, distinct from polymorphonuclear cells. These cells encompass monocytes and lymphocytes, which include subsets such as T lymphocytes, B lymphocytes, and natural killer (NK) cells. Mononuclear immunological cells are integral to immune system function, encompassing roles in antigen presentation, cytokine production, phagocytosis, antibody generation, and cellular immunity. In a specified embodiment, mononuclear immunological cells include monocytes, which are precursors to macrophages and dendritic cells, functioning in the innate immune response and antigen presentation to adaptive immune system components. Alternatively, these cells may include lymphocytes, which mediate adaptive immunity through mechanisms such as cell-mediated cytotoxicity, humoral immunity via antibody secretion, and immune regulation. In the context of the present invention, mononuclear immunological cells isolated from a blood sample of the subject are preferably lymphocytes. The lymphocytes are primary, non-immortalized cells obtained directly from the subject's blood via standard blood collection techniques, such as venipuncture. These lymphocytes are naturally occurring cells that have not undergone any artificial immortalization processes or genetic modifications to render them capable of indefinite proliferation. Instead, they represent the subject's authentic immune profile at the time of collection. Importantly, the lymphocytes are not derived from established or artificially maintained cell lines but are freshly isolated from the subject's peripheral blood following standard separation techniques, such as density gradient centrifugation, to specifically enrich for mononuclear cell populations.

[0042] A primary antibody that specifically binds the target protein refers to an antibody that directly recognizes and binds to a specific epitope on the target protein with high affinity. Nowadays, antibodies exist to detect TDP-43 and phosphorylated TDP-43. In the present invention, the primary antibody can bind TDP-43 and / or phosphorylated TDP-43. Moreover, more than one primary antibody can be used.It is noted that in the context of the present invention, the terms “subject” and “individual” are interchangeable.

[0043] In a preferred embodiment, the method of the second aspect of the invention comprises the steps of:

[0044] a. determining the cytoplasmic concentration and the total cellular concentration of TDP-43 in the mononuclear immunological cells (cell sample) isolated from the subject;

[0045] b. calculating a ratio of the cytoplasmic concentration to the total cellular concentration of TDP-43;

[0046] c. comparing the calculated ratio to one or more predefined decision thresholds to classify the individual into at least one category indicative of a higher or lower likelihood of disease presence; and

[0047] d. generating a risk stratification output that specifies whether the individual falls within a category associated with an increased or decreased likelihood of having the disease.

[0048] In another preferred embodiment of the second aspect of the invention, optionally in combination with any previous or subsequent preferred embodiment of the invention, the TDP-43 proteinopathy is selected from the list consisting of amyotrophic lateral sclerosis, frontotemporal lobar degeneration or limbic-predominant age-related TDP-43 encephalopathy. Preferably, the TDP-43 proteinopathy is amyotrophic lateral sclerosis. In another preferred embodiment of the second aspect of the invention, optionally in combination with any previous or subsequent preferred embodiment of the invention, the total cellular concentration and the cytoplasmic concentration of the TDP-43 protein are determined by an immunoassay selected from the group consisting of:

[0049] a. Western blot,

[0050] b. enzyme-linked immunosorbent assay (ELISA),

[0051] c. immunofluorescence, and

[0052] d. flow cytometry or imaging flow cytometry.

[0053] In another preferred embodiment of the second aspect of the invention, optionally in combination with any previous or subsequent preferred embodiments of the invention, the cytoplasmic concentration of the TDP-43 protein is determined by incubating the cell sample with a primary antibody that specifically binds the TDP-43 protein and incubating the cell sample with a secondary antibody conjugated to one or more Quantum Dots (QD), under conditions sufficient to form a complex with the primary antibody; andanalyzing the cell sample to detect signals from QDs, thereby determining the cytoplasmic concentration of the TDP-43 protein. Preferably, but not limited to, the analysis is performed using flow cytometry or any other similar technique capable of detecting and quantifying the signals from QDs.

[0054] In another preferred embodiment of the second aspect of the invention, optionally in combination with any previous or subsequent preferred embodiment of the invention, the cytoplasmic concentration of the biomarker is determined by incubating the cell sample with a primary antibody that specifically binds the TDP-43 protein and incubating the cell sample with a secondary antibody and one or more quantum dots, wherein the secondary antibody specifically binds to the one or more QDs under conditions sufficient to first form a complex between the primary and secondary antibodies and second through a high-affinity binding interaction between the secondary antibody and the QDs; and analyzing the cell sample to detect signals from QDs, thereby determining the cytoplasmic concentration of the TDP-43 protein.

[0055] High-affinity binding interactions between a secondary antibody and a fluorophore or a QD, can occur through specific molecular recognition and binding mechanisms under conditions sufficient to facilitate such interactions. For instance, in one embodiment, the secondary antibody is conjugated to a binding moiety (e.g., biotin) capable of engaging in a high-affinity interaction with a complementary molecule (e.g., streptavidin).

[0056] The high-affinity interaction between biotin and streptavidin is exemplary of such a mechanism. Biotin, a small molecule structurally designed to fit with high specificity into the binding pocket of streptavidin, forms a stable non-covalent complex characterized by a dissociation constant (Kd) in the range of ~10“15M. This interaction is stabilized by multiple molecular forces, including hydrogen bonding, van der Waals forces, and electrostatic interactions. In certain embodiments, a biotin-conjugated secondary antibody first forms a complex with a primary antibody bound to the target molecule. Subsequently, this biotin-labeled secondary antibody binds to streptavidin through the high-affinity interaction, thereby forming a stable and specific multi-component complex. The resulting complex remains intact under stringent conditions, facilitating its use in applications requiring robust molecular interactions, such as immunoassays, purification, or detection methods.

[0057] In another preferred embodiment of the second aspect of the invention, optionally in combination with any previous or subsequent preferred embodiment of the invention, the total cellular concentration of the TDP-43 protein is determined by incubating the cell sample with a primary antibody that specifically binds the TDP-43 protein and incubatingthe cell sample with a secondary antibody conjugated to a fluorophore (F), distinct from a QD and capable of labelling the nucleus and the cytoplasm of the cells, under conditions sufficient to form a complex with the primary antibody; and analyzing the cell sample to detect signals from F, thereby determining the total cellular concentration of the TDP-43 protein.

[0058] In another preferred embodiment of the second aspect of the invention, optionally in combination with any previous or subsequent preferred embodiment of the invention, the total cellular concentration of the TDP-43 protein is determined by incubating the cell sample with a primary antibody that specifically binds the TDP-43 protein and incubating the cell sample with a secondary antibody and one or more fluorophores (F), distinct from a QD and capable of labelling the nucleus and the cytoplasm of the cells, under conditions sufficient to first form a complex between the primary and secondary antibodies and second through a high-affinity binding interaction between the secondary antibody and the Fs; and analyzing the cell sample to detect signals from Fs, thereby determining the total cellular concentration of the TDP-43 protein.

[0059] In another preferred embodiment of the second aspect of the invention, optionally in combination with any previous or subsequent preferred embodiment of the invention, the cytoplasmic concentration of the TDP-43 protein is determined:

[0060] by incubating the cell sample with a primary antibody that specifically binds the TDP-43 protein and incubating the cell sample with a secondary antibody conjugated to one or more Quantum Dots (QD), under conditions sufficient to form a complex with the primary antibody; and analysing the cell sample to detect signals from QDs, thereby determining the cytoplasmic concentration of the TDP- 43 protein. Preferably, but not limited to, the analysis is performed using flow cytometry or any other similar technique capable of detecting and quantifying the signals from QDs; and / or

[0061] by incubating the cell sample with a primary antibody that specifically binds the TDP-43 protein and incubating the cell sample with a secondary antibody and simultaneously or subsequently with one or more quantum dots, wherein the secondary antibody specifically binds to the one or more QDs under conditions sufficient to first form a complex between the primary and secondary antibodies and second through a high-affinity binding interaction between the secondary antibody and the QDs; and analysing the cell sample to detect signals from QDs, thereby determining the cytoplasmic concentration of the TDP-43 protein. Preferably, but not limited to, the analysis is performed using flow cytometry orany other similar technique capable of detecting and quantifying the signals from QDs;

[0062] and the total cellular concentration of the TDP-43 protein is determined:

[0063] by incubating the cell sample with a primary antibody that specifically binds the TDP-43 protein and incubating the cell sample with a secondary antibody conjugated to a fluorophore (F), distinct from a QD and capable of labelling the nucleus and the cytoplasm of the cells, under conditions sufficient to form a complex with the primary antibody; and analysing the cell sample to detect signals from F, thereby determining the total cellular concentration of the TDP-43 protein. Preferably, but not limited to, the analysis is performed using flow cytometry or any other similar technique capable of detecting and quantifying the signals from F; and / or

[0064] by incubating the cell sample with a primary antibody that specifically binds the TDP-43 protein and incubating the cell sample with a secondary antibody and simultaneously or subsequently with one or more fluorophores (F), distinct from a QD and capable of labelling the nucleus and the cytoplasm of the cells, under conditions sufficient to first form a complex between the primary and secondary antibodies and second through a high-affinity binding interaction between the secondary antibody and the Fs; and analysing the cell sample to detect signals from Fs, thereby determining the total cellular concentration of the TDP-43 protein. Preferably, but not limited to, the analysis is performed using flow cytometry or any other similar technique capable of detecting and quantifying the signals from F.

[0065] An alternative aspect of the second aspect of the invention (from hereinafter referred to as “the second alternative aspect”) refers to a method for determining or assessing the likelihood that a subject is suffering from amyotrophic lateral sclerosis (ALS), wherein the method comprises:

[0066] a) calculating a ratio of the cytoplasmic concentration to the total cellular concentration of TDP-43 in mononuclear immunological cells isolated from a biological sample of the subject, preferably wherein the cytoplasmic concentration is determined according to the first or second method to determine the cytoplasmic concentration of the TDP-43 protein and the total cellular concentration of the TDP-43 protein is determined according to the first or second method to determine the cytoplasmic concentration of the TDP-43 protein, andb) comparing the calculated ratio to one or more predefined decision thresholds to classify the subject into at least one category indicative of a higher or lower likelihood of disease presence.

[0067] In a preferred embodiment of the second alternative aspect, the predefined decision threshold is selected based on statistical analysis of TDP-43 ratios obtained from a reference or control population, wherein if in the comparing step the ratio result is higher than the decision threshold the subject shall be categorized as an individual into at least the higher-risk group of suffering from a TDP-43 proteinopathy otherwise the subject shall be categorized as a lower-risk group.

[0068] In another preferred embodiment of the second alternative aspect, optionally in combination with any previous or subsequent embodiment of this aspect of the invention, the cytoplasmic concentration of the TDP-43 protein is determined according to the first or second method to determine the cytoplasmic concentration of the TDP-43 protein and the total cellular concentration of the TDP-43 protein is determined according to the first or second method to determine the total concentration of the TDP-43 protein, wherein the predetermined threshold is selected based on statistical analysis of TDP-43 ratios obtained from a reference or control population, wherein if in the comparing step the ratio result is higher than the decision threshold the subject shall be categorize as a higher-risk group of suffering from a TDP-43 proteinopathy otherwise the subject shall be categorized as a lower-risk group, and wherein the TDP-43 proteinopathy is amyotrophic lateral sclerosis (ALS). Preferaby, wherein the QD and the fluorophore (F) are selected from different emission spectra suitable for multicolor flow cytometry, allowing simultaneous detection of multiple signals, preferably wherein the QD is Qdot 655 and the fluorophore (F) is Alexa fluor 488.

[0069] In another preferred embodiment of the second aspect of the invention (including its alternative second aspect), optionally in combination with any previous or subsequent preferred embodiments of the invention, the predetermined threshold is selected based on statistical analysis of TDP-43 ratios obtained from a reference or healthy control population, wherein if in the comparing step the ratio result is higher than the decision threshold the subject shall be categorize as an individual into at least the higher-risk group of suffering from a TDP-43 proteinopathy (an increased likelihood of having the disease) otherwise the subject shall be categorized as a lower-risk group (a decrease likelihood of having the disease).

[0070] In another preferred embodiment of the second aspect of the invention (including its alternative second aspect),, optionally in combination with any previous or subsequentpreferred embodiment of the invention, the predetermined threshold is selected based on statistical analysis of TDP-43 ratios obtained from a reference or control population, wherein if in the comparing step the ratio result is higher than the decision threshold the subject shall be categorize as a higher-risk group (an increased likelihood of having the disease) of suffering from a TDP-43 proteinopathy otherwise the subject shall be categorized as a lower-risk group (a decrease likelihood of having the disease); and the TDP-43 proteinopathy is preferably amyotrophic lateral sclerosis.

[0071] Quantum dots (QDs), as used throughout the present invention, are nanocrystals of semiconducting material possessing quantum mechanical characteristics with capability to get conjugated with drug moieties. The particle size of QDs varies from 2 to 10 nm and can radiate a wide range of colours depending upon their size. Their wide and diverse usage of QDs across the world is due to their adaptable properties like large quantum yield, photostability, and adjustable emission spectrum. QDs are nanomaterials with inherent electrical characteristics that can be used as drug carrier vehicle and as a diagnostic in the field of nanomedicine. Typically, commercially available quantum dots are named for their peak emission wavelength (i.e. , 525, 565, 585, 605, and 655 nm). In one embodiment, any quantum dot such as QD525, 565, 585, 605, or 655 nm, can be used in the present invention.

[0072] In another preferred embodiment of the second aspect of the invention (including its alternative second aspect),, optionally in combination with any previous or subsequent preferred embodiment of the invention, the QD and the fluorophore (F) are selected from different emission spectra suitable for multicolor flow cytometry, allowing simultaneous detection of multiple signals.

[0073] In another preferred embodiment of the second aspect of the invention (including its alternative second aspect), optionally in combination with any previous or subsequent preferred embodiment of the invention, the QD has a peak emission wavelength of 525, 565, 585, 605, or 655 nm such as Qdot 655 and the fluorophore (F) is preferably Alexa fluor 488.

[0074] In another preferred embodiment of the second aspect of the invention (including its alternative second aspect), optionally in combination with any previous or subsequent preferred embodiment of the invention, the method further comprises selecting or adjusting a therapeutic regimen for the subject based on the classification, wherein subjects classified above a certain ratio threshold receive a different therapy than subjects classified below said threshold.In another preferred embodiment of the second aspect of the invention (including its alternative second aspect), optionally in combination with any previous or subsequent preferred embodiment of the invention, calculating the ratio and classifying the subject is performed by a computer-implemented algorithm that processes input data from the measurements in steps (b) and (c) and outputs an indication of the subject’s classification.

[0075] A third aspect of the invention refers to a computer implemented method for determining or assessing the likelihood that, or determining if, a subject is suffering from amyotrophic lateral sclerosis (ALS), wherein the method comprises:

[0076] a. determining or receiving the cytoplasmic concentration of TDP-43 in the mononuclear immunological cells, preferably wherein the cytoplasmic concentration of the TDP-43 protein is determined by detecting and quantifying the signals from Quantum Dots (QDs) by using flow cytometry; b. determining or receiving the total cellular concentration of TDP-43 in the mononuclear immunological cells, preferably determining said cellular concentration by using flow cytometry;

[0077] c. integrating both the cytoplasmatic concentration and the total cellular concentration of TDP-43 in the mononuclear immunological cells into a diagnostic model configured to generate, based on said cytoplasmatic and total cellular concentrations of TDP-43, an output indicating the likelihood that, or indicating if, a subject is suffering from amyotrophic lateral sclerosis (ALS).

[0078] In a preferred embodiment of the third aspect of the invention, generating an output indicating the likelihood that a subject is suffering from amyotrophic lateral sclerosis (ALS) comprises classifying the subject into at least one category indicative of a higher or lower likelihood of suffering from ALS, and wherein generating an output indicating if a subject is suffering from amyotrophic lateral sclerosis (ALS) comprises classifying the subject into at least one category indicative of suffering or not suffering from ALS.

[0079] In another preferred embodiment of the third aspect of the invention, optionally in combination with any previous or subsequent embodiment of the third aspect of the invention, the diagnostic model is a machine learning model; a rule-based decision algorithm; or a mathematical model wherein the cytoplasmatic and total cellular concentrations of TDP-43 are parameters or variables used to fit said mathematical model.In another preferred embodiment of the third aspect of the invention, optionally in combination with any previous or subsequent embodiment of the third aspect of the invention, the machine learning model is a logistic regression model, a linear or polynomial model, a discriminant analysis model, a quadratic discriminant analysis model, a Bayesian model, a decision tree model, a random forest model, a gradient boosting model, a support vector machine, a k-nearest neighbour model, an artificial neural network, a transformer-based model, a feed-forward neural network or any ensemble model thereof.

[0080] In another preferred embodiment of the third aspect of the invention, optionally in combination with any previous or subsequent embodiment of the third aspect of the invention, the machine learning model is a binary classifier configured to map an input vector x = (c_cyto, c_total) to a score s in [0,1] indicative of ALS, preferably wherein the classifier has been trained on labelled measurement data of said concentrations, more preferably wherein the output class is determined by thresholding s at a threshold T selected to optimize diagnostic performance on a validation set.

[0081] In another preferred embodiment of the third aspect of the invention, optionally in combination with any previous or subsequent embodiment of the third aspect of the invention, the diagnostic model is generated according to the following steps:

[0082] i. Receiving the cytoplasmatic concentration and the total cellular concentration TDP-43 in mononuclear immunological cells from a biological sample obtained from a cohort of subjects confirmed to have ALS and from a cohort of subjects confirmed to not have ALS, together with corresponding diagnostic labels indicating whether each subject has ALS or not; and ii. generating the diagnostic model by using at least a training set or subset of the cytoplasmatic and total cellular concentrations and corresponding diagnostic labels received in step (i) to train a machine learning model, to generate a rule-based decision algorithm or to fit the parameters of a mathematical model, thus generating the diagnostic model.

[0083] In another preferred embodiment of the third aspect of the invention, optionally in combination with any previous or subsequent embodiment of the third aspect of the invention, the generation of the diagnostic model further comprises (iii) validating the diagnostic model generated in step (ii) by applying it to at least a validation set or subset of the transcript expression data and corresponding diagnostic labels received in step (i).In yet another preferred embodiment of the third aspect of the invention, it is noted that the cytoplasmic concentration of the TDP-43 protein can be determined according to the first or second method to determine the cytoplasmic concentration of the TDP-43 protein and the total cellular concentration of the TDP-43 protein can be determined according to the first or second method to determine the total concentration of the TDP-43 protein. It is noted that the expression “determining or assessing the likelihood that, or determining if,” encompasses both probabilistic and categorical diagnostic outputs. In some embodiments the output is preferably a likelihood value mapped to a continuous or quasi-continuous scale, such as, but not limited to, a probability between 0 and 1, a percentage between 0 and 100 percent, a log-odds score, a likelihood ratio, or a calibrated risk index on an arbitrary scale, for example 0 to 10. It is further noted that in some embodiments determining if a subject is suffering from amyotrophic lateral sclerosis (ALS) comprises generating a binary or multi-category decision derived from, for example, one or more thresholds applied to such continuous outputs, for example reporting positive or negative for ALS, or reporting low, intermediate, or high likelihood categories using two thresholds that define indeterminate zones.

[0084] It is further noted that step (a) may comprise either calculating the cytoplasmic concentration of TDP-43, for example from instrument signals or receiving said concentration as an input, for example from a laboratory information system. In some embodiments the concentration is obtained in mononuclear immunological cells such as, but not limited to, T lymphocytes, B lymphocytes, natural killer cells, monocytes, and dendritic cells, preferably measured on peripheral blood mononuclear cells. Preferably, the concentration is derived from flow cytometry data, for example acquired with Quantum Dot labelled anti-TDP-43 reagents, while other labels may be used, such as, but not limited to, organic fluorophores, tandem dyes, metal-tag labels for mass cytometry, or enzymatic colorimetric probes.

[0085] It is further noted that step (b) may comprise either calculating the total cellular concentration of TDP-43, for example from instrument signals or receiving said concentration, for example as a digital input. Preferably, the total cellular concentration is determined by flow cytometry, for example with permeabilisation conditions configured to expose all cellular compartments, while alternative techniques may be used, such as, but not limited to, imaging flow cytometry, high-content microscopy, immunoassays, capillary electrophoresis immunoassay, or single-cell proteomics. In some embodiments the concentrations in steps (a) and (b) are expressed as molecules of equivalent solublefluorophore, median fluorescence intensity normalised to calibration beads, or copies per cell, for example using regression fits to standard curves.

[0086] In some embodiments the diagnostic model of step (c) integrates as inputs the cytoplasmic concentration and the total cellular concentration of TDP-43 at the single-cell level, at the population level, or both. Population level summaries may comprise, but are not limited to, mean, median, geometric mean, trimmed mean, interquartile range, 5th to 95th percentiles, or robust z-scores across the mononuclear immunological cell gate or within specific subpopulations. Single-cell inputs may be aggregated by the model through embedded statistics or pooling layers, for example by computing per-cell decision scores and summarising them via quantiles.

[0087] It is noted that derived variables may be computed from the two concentrations and provided to the diagnostic model, in addition to or instead of the raw concentrations. Such derived variables may comprise, but are not limited to, the ratio of cytoplasmic to total concentration, the cytoplasmic fraction defined as cytoplasmic divided by total, the difference between cytoplasmic and total after scale alignment, a log-ratio, or a piecewise transformation that is preferably configured for stabilising variance, for example log or arcsine-square-root transforms. In some embodiments the model is preferably configured for capturing non-linear dependencies between the cytoplasmic concentration and the total cellular concentration, such that, for example, the ratio provides high discriminative power at low cytoplasmic values, while the absolute cytoplasmic concentration predominates at higher values.

[0088] In some embodiments the diagnostic model comprises, but is not limited to, logistic regression, penalised regression, linear or quadratic discriminant analysis, decision trees, random forests, gradient boosting machines, support vector machines with linear or radial kernels, k-nearest neighbours, artificial neural networks including multilayer perceptrons, or ensemble combinations thereof. Preferably the model outputs a calibrated score between 0 and 1 that is interpreted as a likelihood, or I some embodimetns at least one threshold is applied to determine a categorical decision. Thresholds may be selected, for example, by maximising Youden’s J on validation data, fixing specificity at a predefined level such as 95 percent, fixing sensitivity, or optimising a cost function reflecting clinical utility.

[0089] It is further noted that the likelihood or risk may be expressed in alternative formats tailored to clinical workflows. Examples include: a posterior probability conditioned on pre-test probability using Bayes updating, a likelihood ratio positive or negative, a decile rank relative to a reference cohort, a traffic-light colour code mapped from score intervals,or a numerical grade, for example from 1 to 5. In some embodiments two thresholds define three zones, such as <0.2 negative, 0.2 to 0.8 indeterminate, and >0.8 positive, among other options.

[0090] In some embodiments the diagnostic model receives exclusively the two variables of interest, namely the cytoplasmic concentration and the total cellular concentration of TDP-43, optionally together with derived terms computed from these two inputs. In other embodiments the diagnostic model receives a superset of inputs that may comprise, but is not limited to, age, sex, clinical site, sample type, batch identifiers, additional biomarkers such as phosphorylated TDP-43 signals, neurofilament light chain levels, cell-subset-specific features, viability indicators, or instrument metadata, while retaining the two TDP-43 concentrations, preferably as primary features. Preferably, when additional inputs are present, the model is structured so that inference remains valid when only the two concentrations are available, for example by employing feature-wise missing handling or model variants with reduced feature sets.

[0091] It is noted that preprocessing may comprise normalisation and calibration steps configured for cross-instrument and cross-site harmonisation. Examples include, but are not limited to, bead-based fluorescence normalisation to a reference scale, compensation of spectral spillover, logicle or biexponential transform of intensities, batch effect mitigation such as location-scale alignment or empirical Bayes adjustment, outlier trimming, and imputation of missing values using, for example, k-nearest neighbours or model-based imputation. Quality controls may comprise replicate control samples, instrument performance tracking, and lot-to-lot bridging calibrators to constrain drift. In some embodiments the computational pipeline comprises receiving digital inputs for step (a) and step (b), applying deterministic preprocessing transforms, computing any derived variables, evaluating the diagnostic model parameters stored on a non-transitory computer-readable medium, and generating both a continuous likelihood and, where applicable, a categorical decision by comparing the likelihood to at least one threshold. The method may further comprise reporting confidence intervals around the likelihood, for example 95 percent bootstrap intervals, reporting the distance to the decision boundary, and logging model versioning and data provenance.

[0092] In some embodiments the model is trained using cross-validation or external validation strategies, for example k-fold cross-validation, leave-one-site-out validation, or temporal validation. Performance metrics may comprise, but are not limited to, area under the receiver operating characteristic curve, sensitivity, specificity, positive and negative predictive values, Brier score, calibration slope and intercept, and decision curveanalysis. The trained model may be locked and distributed to multiple centres with harmonised thresholds, for example with a single threshold applicable across instruments after normalisation, or with site-specific offsets determined from a small panel of reference samples.

[0093] It is further noted that the method may operate on per-cell inputs or on aggregated summaries. For per-cell operation, an interpretable decision tree may first branch on cytoplasmic concentration at a threshold, for example 80th percentile of controls, and then apply a ratio rule within the low cytoplasmic branch and an absolute cytoplasmic rule within the high branch. For parametric models, a logistic model with interaction and spline terms may capture such non-linearities, for example by including a restricted cubic spline of cytoplasmic concentration and an interaction term with the ratio. Neural networks may approximate similar behaviour without explicit feature engineering.

[0094] In some embodiments the output is stored in an electronic health record and trended longitudinally. Temporal aggregation may comprise, but is not limited to, exponential smoothing, mixed-effects modelling with subject-specific intercepts, or change-from-baseline analyses, thereby enabling monitoring of disease trajectories using the same two inputs over time.

[0095] Advantageously, integrating the cytoplasmic concentration with the total cellular concentration of TDP-43 in a diagnostic model provides a compartment-aware measure of TDP-43 biology that is preferably configured for capturing mislocalisation phenomena relevant to ALS. The approach yields an objective likelihood or a clear positive or negative decision using minimal inputs, supports non-linear relationships between features, enables harmonised thresholds across centres through standardised preprocessing, and is compatible with high-throughput workflows, thereby improving diagnostic performance and facilitating reproducible clinical deployment.

[0096] According to a preferred embodiment of the invention, the method of the third aspect of the invention further comprises generating an output indicating the likelihood that a subject is suffering from amyotrophic lateral sclerosis (ALS) comprises classifying the subject into at least one category indicative of a higher or lower likelihood of suffering from ALS, and generating an output indicating if a subject is suffering from amyotrophic lateral sclerosis (ALS) comprises classifying the subject into at least one category indicative of suffering or not suffering from ALS.

[0097] It is noted that the categorisation preferably comprises an ordinal or nominal labelling scheme produced from a calibrated score, with the labelling rule selected to optimise clinical utility under predefined cost ratios. It is further noted that the categorisation mayemploy a hierarchical process, in which a first stage assigns likelihood strata and a second stage, preferably configured for binary decisioning, maps strata to a positive or negative determination with an optional abstention category when uncertainty exceeds a preset bound. In some embodiments the categorisation may be set-valued, for example via conformal prediction, thereby returning one or more categories when evidence is ambiguous. Alternatively the categorisation may be dynamic, wherein thresholds are adapted to covariates, for example age or sample type, using monotone transformations that preserve calibration. In some embodiments the categorisation may be derived from utility-aware decision functions, such as, but not limited to, rule-out, rule-in, or refer for further testing. Alternatively ties or borderline cases may be resolved by secondary rules comprising consensus across model ensembles, replicate consistency checks, or temporal aggregation over serial samples. In some embodiments the output comprises both the assigned category and supporting indicators, for example confidence level or distance to decision boundary, together with machine-interpretable flags suitable for clinical pathways integration.

[0098] Advantageously, such categorisation provides operational decisions while explicitly accommodating uncertainty, variable pre-test probabilities, and heterogeneous clinical workflows.

[0099] According to another preferred embodiment of the invention, the diagnostic model is a machine learning model; a rule-based decision algorithm; or a mathematical model wherein the cytoplasmatic and total cellular concentrations of TDP-43 are parameters or variables used to fit said mathematical model.

[0100] It is noted that, in some embodiments, a machine-learning implementation comprises training exclusively on two input variables, namely the cytoplasmic concentration of TDP-43 and the total cellular concentration of TDP-43, preferably after predefined scaling. It is further noted that variables may be represented as raw magnitudes, normalised intensities, or monotone transforms, for example log or arcsine-square-root, while parameters may comprise coefficients, intercepts, margins, kernel scales, or regularisation strengths determined during fitting. In some embodiments a rule-based decision algorithm comprises threshold parameters on one or both variables, optionally including a ratio or difference computed internally without introducing additional fitted variables, and decision rules that are preferably configured for two-zone or three-zone categorisation. Alternatively a mathematical model may comprise a linear, polynomial, or spline function y = f(x1, x2) where x1 is the cytoplasmic concentration and x2 is the total concentration, with fitted parameters comprising, among other options, slopes,interaction terms, knot locations, or piecewise breakpoints. It is further noted that fitting may comprise least squares, maximum likelihood, or Bayesian estimation using priors on parameters, while validation may comprise cross-validation with metric-driven model selection under constraints that the only explanatory variables are said two concentrations. In some embodiments calibration parameters, such as Platt scaling or isotonic mapping, are estimated solely from the model score derived from the same two inputs.

[0101] Advantageously, limiting fitting to two biologically grounded variables simplifies estimation, improves robustness to confounders, and facilitates transfer across laboratories.

[0102] According to another preferred embodiment of the invention, the machine learning model is a logistic regression model, a linear or polynomial model, a discriminant analysis model, a quadratic discriminant analysis model, a Bayesian model, a decision tree model, a random forest model, a gradient boosting model, a support vector machine, a k-nearest neighbour model, an artificial neural network, a transformer-based model, a feed-forward neural network or any ensemble model thereof.

[0103] It is noted that transformer-based implementations may comprise tabular transformers or sequence transformers that ingest per-cell feature vectors and attention-pool them into sample-level representations, preferably with heads configured for binary or ordinal outputs and with regularisation that comprises dropout, L2 penalties, or early stopping. It is further noted that feed-forward neural networks may comprise 1 to 4 hidden layers with, for example, 4 to 64 units per layer, ReLLI or tanh activations, and an output sigmoid, while weights may be constrained by sparsity priors in Bayesian variants. In some embodiments ensemble models may comprise stacking with a meta-learner, soft-voting over heterogeneous base learners, or bagging with bootstrap resampling, and may include calibration layers fitted post-hoc. Alternatively Bayesian models may comprise priors on coefficients or margins and yield posterior credible intervals used to define abstention zones. In some embodiments hyperparameters may be selected by nested cross-validation with target metrics comprising ROC-AUC or Brier score, and class imbalance may be addressed by class weights or probability-based thresholds. As a small example, the method comprises a logistic regression trained on the cytoplasmic and total TDP-43 concentrations, yielding a calibrated score; a clinical user inputs the two concentrations, receives a likelihood, and applies a predefined threshold to report negative, indeterminate, or positive. Alternatively the example may be implemented with a shallow decision tree or a gradient boosting model.Advantageously, such model diversity and ensembling improves robustness, supports calibration, and adapts to varying data structures and clinical operating points.

[0104] According to another preferred embodiment of the invention, the machine learning model is a binary classifier configured to map an input vector x = (c_cyto, c_total) to a score s in [0,1] indicative of ALS, preferably wherein the classifier has been trained on labelled measurement data of said concentrations, more preferably wherein the output class is determined by thresholding s at a threshold T selected to optimize diagnostic performance on a validation set.

[0105] It is noted that the binary classification pipeline preferably comprises a scoring function that outputs s confined to [0,1], for example by applying a sigmoid or by post-hoc isotonic calibration to an unbounded margin. It is further noted that training may comprise labelled measurement data where labels derive from clinical diagnosis or adjudicated consensus, with procedures to mitigate label noise such as robust losses or sample reweighting. In some embodiments validation is stratified by site and class to ensure that the selection of T reflects deployment conditions, and stability of T may be assessed across folds or temporal splits. Alternatively T may be selected on receiver-operating characteristic or precision-recall curves according to predefined clinical utility, and ties at s equal to T may be resolved by abstention or by secondary rules using, for example, replicate concordance. It is further noted that monotonicity constraints may be imposed so that s is preferably non-decreasing with cytoplasmic concentration or with derived ratios, while still allowing non-linear effects of c_total. In some embodiments decision margins s minus T are reported to quantify proximity to the boundary, and operating bands may trigger confirmatory testing. Alternatively post-deployment recalibration and periodic T review may be implemented under drift monitoring using reference controls.

[0106] Advantageously, selecting T on validated data yields reproducible operating points, facilitates clinical alignment, and supports transparent and auditable decision criteria. According to another preferred embodiment of the third aspect of the invention, the diagnostic model is generated according to the following steps:

[0107] i. Receiving the cytoplasmatic concentration and the total cellular concentration TDP-43 in mononuclear immunological cells from a biological sample obtained from a cohort of subjects confirmed to have ALS and from a cohort of subjects confirmed to not have ALS, together with corresponding diagnostic labels indicating whether each subject has ALS or not; andii. generating the diagnostic model by using at least a training set or subset of the cytoplasmatic and total cellular concentrations and corresponding diagnostic labels received in step (i) to train a machine learning model, to generate a rule-based decision algorithm or to fit the parameters of a mathematical model, thus generating the diagnostic model.

[0108] It is noted that step (i) preferably comprises supervised acquisition of paired inputs and labels, preferably with case and control cohorts curated under inclusion and exclusion criteria, and with labels adjudicated by consensus or reference standards to minimise noise. It is further noted that data partitioning may be stratified by site and class, for example 60, 20, 20 percent for training, validation, and test, while ensuring no subject overlap across splits and blinding of the test set. In some embodiments preprocessing parameters, such as scaling or transformation coefficients, are fit on the training partition and applied to validation and test to avoid leakage. Alternatively k-fold or nested cross-validation may be applied within step (ii) for hyperparameter tuning and unbiased performance estimation. In some embodiments supervised methods comprise empirical risk minimisation with loss functions aligned to target metrics, such as, but not limited to, log-loss, hinge loss, or focal loss. It is further noted that class imbalance may be handled by class weights, calibrated resampling, or threshold adjustment learned on validation data. Alternatively rule-based algorithms may be induced from labelled data using programme-synthesis or greedy splitting with monotonicity constraints, and mathematical models may be fit by penalised least squares with cross-validated regularisation. In some embodiments label uncertainty is modelled by soft labels or robust losses and performance is bootstrapped for confidence intervals.

[0109] Advantageously, the supervised protocol yields reproducible models with controlled generalisation, calibrated thresholds, and auditable data provenance across cohorts and sites.

[0110] According to another preferred embodiment of the invention, the generation of the diagnostic model further comprises:

[0111] iii. iii validating the diagnostic model generated in step (ii) by applying it to at least a validation set or subset of the transcript expression data and corresponding diagnostic labels received in step (i).

[0112] It is noted that validation preferably comprises applying the trained parameters to a held-out subset and computing prespecified performance, calibration, and utility metrics. It is further noted that, when the subset comprises transcript expression data, harmonisation to the model inputs may comprise cross-modal mapping, for example bylearned or predefined transforms that relate transcript levels to the two concentration features, with coefficients fixed prior to validation. In some embodiments the validation protocol may include site-wise or time-wise splits, permutation testing to assess chance-level performance, and confidence intervals computed by stratified bootstrap or DeLong methods. Alternatively early stopping may be governed by the validation loss or Brier score monitored on mini-batches derived from the subset. In some embodiments recalibration on the validation subset is prohibited, while post-hoc calibration may be evaluated in a separate tuning fold nested within the validation to avoid leakage. It is further noted that subgroup analyses may be conducted across age strata, cell-subset composition, or assay batch to verify stability. In some embodiments decision-analytic evaluations may comprise decision curves, net benefit, and cost-weighted utility at candidate thresholds pre-registered before analysis. Alternatively robustness checks may include stress tests under simulated noise, missingness, and covariate shift, and fairness audits on protected attributes with parity constraints recorded.

[0113] Advantageously, such validation quantifies generalisation, detects modality mismatch, and supports threshold selection under clinically relevant operating conditions.

[0114] The term "diagnostic model", in the context of the invention, refers to a computational mapping that receives one or more quantitative inputs, preferably cytoplasmatic concentrations and total cellular concentration of TDP-43 in mononuclear immunological cells, more preferably receives as input only said concentrations, and outputs a value or class indicative of a diagnostic state, for example a score in [0,1] or a categorical label. The term encompasses supervised, semi-supervised or unsupervised approaches and algorithmic or mathematical formulations such as, but not limited to, logistic regression, linear or polynomial models, generalized linear models, discriminant analysis, Bayesian classifiers, decision trees, random forests, gradient boosting, support vector machines, k-nearest neighbours, artificial neural networks, convolutional or transformer-based networks, ensembles, rule-based decision systems, and parametric or non-parametric statistical models. Parameters may be fixed, hand-set or learned from data, and outputs may include binary, multiclass, ordinal, or continuous risk estimates. The diagnostic model may be implemented in software, firmware or hardware.

[0115] The term "machine learning model", in the context of the invention, refers to a parameterized computational construct that maps one or more input variables to one or more outputs by learning patterns from data through an optimization or training procedure, without being explicitly programmed for the specific mapping. It is noted that the input variables preferably comprise or consist of cytoplasmatic concentrations andtotal cellular concentration of TDP-43 in mononuclear immunological cells. The term encompasses supervised, unsupervised, semi-supervised and reinforcement learning approaches, such as, but not limited to, logistic regression, linear and polynomial regression, ridge or lasso models, linear or quadratic discriminant analysis, naive Bayes and Bayesian networks, decision trees, random forests, gradient boosting machines, XGBoost, LightGBM, CatBoost, support vector machines, k-nearest neighbours, Gaussian processes, hidden Markov models, artificial neural networks, feed-forward, convolutional and recurrent networks, LSTM, GRU, transformer-based architectures, autoencoders, variational autoencoders, probabilistic graphical models and ensembles thereof. The model may comprise learnable parameters, hyperparameters, loss functions, regularization terms, and training algorithms, for example stochastic gradient descent, Adam, coordinate descent or expectation-maximization. The inputs and outputs may be scalars, vectors, matrices or tensors, may be continuous or discrete, and may be post-processed by calibration, thresholding or other decision functions. The model may be trained, validated and deployed on any computing platform, may operate online or offline, and may incorporate feature preprocessing, dimensionality reduction or feature selection, such as, but not limited to, PCA, ICA or mutual information filters.

[0116] The term "binary classifier", in the context of the invention, refers to a computational model or decision function that maps an input vector comprising one or more features to a discrete output representing one of two classes, for example positive versus negative, presence versus absence, or higher versus lower likelihood. It is noted that the input variables preferably comprise or consist of cytoplasmatic concentrations and total cellular concentration of TDP-43 in mononuclear immunological cells. The term encompasses probabilistic and deterministic formulations that output either a class label or a scalar score in [0,1] that is thresholded to assign a class. The term encompasses algorithms such as, but not limited to, logistic regression, linear or quadratic discriminant analysis, naive Bayes, decision trees, random forests, gradient boosting, support vector machines with linear or nonlinear kernels, k-nearest neighbours, perceptrons, feed-forward neural networks, convolutional or transformer architectures constrained to two outputs, and ensemble methods. The classifier may be trained in supervised, semi-supervised or weakly supervised regimes, may incorporate regularisation, calibration, and class imbalance handling, and may operate on raw, normalised or engineered features. The output may be a hard label or a calibrated probability, and decision thresholds may be fixed, adaptive or selected to meet performance criteria such as sensitivity, specificity, ROG-AUG or Youden’s J.The term "input vector (c_cyto, c_total)", in the context of the invention, refers to a two-dimensional ordered pair comprising numerical values representing, respectively, the cytoplasmic concentration of TDP-43 and the total cellular concentration of TDP-43 measured in mononuclear immunological cells. The term encompasses raw intensities, background-subtracted signals, calibrated concentrations, normalized or z-scored values, log-transformed values, or ratios mapped to equivalent two-component representations, among other options. Values may be expressed in arbitrary fluorescence units, molecules of equivalent soluble fluorophore, concentration units such as ng per 10A6 cells, median fluorescence intensity, or probabilistic estimates derived from signal models. In some embodiments, c_cyto and c_total are derived from flow cytometry, imaging cytometry, immunoassays or other quantification methods, such as, but not limited to, Quantum Dot fluorescence, enzyme-linked detection, bead-based assays or mass cytometry. The vector may further incorporate scaling or harmonization factors applied uniformly to both components, for example min-max scaling to [0,1], robust scaling using medians and interquartile ranges, or instrument-specific calibration to reference standards.

[0117] The term "score s", in the context of the invention, refers to a scalar output generated by a diagnostic model from an input comprising the cytoplasmic and total cellular concentrations of TDP-43, which quantifies the likelihood that a subject has ALS. The term encompasses continuous values in a bounded or unbounded range, for example s in [0.1]. [-1 ]. or ,andmay represent a probability, a log-odds, a margin, a confidence value, or a normalized index. The score s may be derived from models such as, but not limited to, logistic regression, discriminant analysis, support vector machines, decision trees, random forests, gradient boosting, k-nearest neighbours, neural networks, rule-based systems or mathematical formulas. The score s may be calibrated or transformed, for example via Platt scaling, isotonic regression, min-max normalization, z-score standardization, or sigmoid mapping, and may be thresholded at one or more predetermined values to assign diagnostic categories. The score s may be accompanied by or computed with auxiliary quantities, such as, but not limited to, confidence intervals, posterior probabilities, variance estimates, or cross-validated predictions, among other options.

[0118] The term "threshold T", in the context of the invention, refers to a decision boundary value applied to the classifier output score s in [0,1] for assigning a diagnostic class, encompassing fixed, adaptive or probabilistic cutoffs used to distinguish ALS-positive from ALS-negative outputs. The threshold T may be a single scalar, a set of class-specific thresholds, or a function of ancillary variables such as, but not limited to, age, instrumentbatch, site, or measurement uncertainty, T may be determined by optimization criteria such as, but not limited to, maximizing Youden’s J, minimizing expected cost, fixing sensitivity or specificity at predefined levels, equal error rate, ROC-based selection, PR-curve optimization, cross-validated risk minimization, or Bayesian decision rules, T may be derived from calibration procedures such as, but not limited to, Platt scaling, isotonic regression, temperature scaling, or empirical quantile mapping. In some embodiments, T may vary over time or across centers, may comprise an interval [TL, TLI] defining indeterminate or gray zones, or may be applied after score normalization or drift correction. Examples include T = 0.5 for symmetric 0-1 loss, T = 0.23 to achieve 90 percent sensitivity, T chosen to maximize F1, or patient-specific T computed from posterior class probabilities.

[0119] A fourth aspect of the invention refers to a computer program comprising processor readable instructions which, when the program is executed by a computer, cause the computer to carry out steps of the method of the second (including its alternative aspect) or third aspect of the invention, including any of its preferred embodiments.

[0120] A fifth aspect of the invention refers to a computer-readable storage medium comprising processor readable instructions which, when executed by a computer, cause the computer to carry out steps of the method of the second (including its alternative aspect) or third aspect of the invention, including any of its preferred embodiments.

[0121] A sixth aspect of the invention refers to a system comprising a processor and a memory, wherein the memory comprises instructions which, when executed by the processor, cause the system to carry out the method of the second or third aspect of the invention, including any of its preferred embodiments.

[0122] A seventh aspect of the invention refers to the use of a kit or system comprising:

[0123] a. optionally a first binding moiety as defined in the first or second aspects of the invention (including the second alternative aspect);

[0124] b. optionally a second binding moiety as defined in the first or second aspects of the invention (including the second alternative aspect); c. instructions for performing the method of any of the first to second aspect of the invention (including the second alternative aspect); and

[0125] d. a data processing application or software configured to receive input data from the measurements in steps (a) and (b) of the methods of any one of the first or second aspects of the invention and outputs an indication of the subject’s classification.The following example is merely for illustrative purposes and does not limit the present invention.

[0126] EXAMPLE

[0127] Example 1: Characterization of TDP-43 pathology in lymphoblasts from ALS patients using QDs, both secondary antibody-conjugated QDs and streptavidin-conjugated QDs. In this case, we are using an established cell line (immortalized lymphocytes) derived from healthy individuals and ALS patients. Due to their selective cytosolic labelling, there are significant differences between Controls, ALS patients, and treated ALS patients’ samples (Figure 1 A). We observe that traditional fluorophores without ability for selective cytosolic labelling (such as Alexa) present similar values in high contrast to QDs that classifies correctly the samples (Figure 1B).

[0128] Example 2: Characterization of TDP-43 pathology in lymphocytes from ALS patients using streptavidin-conjugated QDs, and Alexa Fluor 488.

[0129] In this case, the original idea was to reproduce the results from example 1 but in fresh lymphocytes. However, when working with lymphocytes directly obtained from a blood draw, we observed highly variable results with both QD and Alexa staining values, and a correct classification cannot be done in a similar way to Example 1 (Figure 2A). This challenge has been solved using the ratio QD / Alexa as a measure of TDP-43 pathology, that is performed in each patient data. ALS patients and healthy controls are correctly grouped with this correction (Figure 2B). We believe this methodology could be used not only for correctly classifying ALS patients from healthy controls but specifically to measure TDP-43 pathology burden and monitor pharmacological activity at the molecular level from a blood draw. It could be useful for companies developing TDP-43 targeting therapies to show in vivo target engagement and in vivo efficacy (PoC).

[0130] Example 3: Characterization of TDP-43 pathology in lymphocytes from FTLD patients using streptavidin-conjugated QDs, and Alexa Fluor 488.

[0131] In this case, we have used the method to measure TDP-43 pathology in lymphocytes from FTLD patients. While the ratio has not given significant differences from healthy controls, given that FTLD pathology is roughly 50% based on tau pathology and 50% on TDP-43 pathology, it could help to classify patients according to their molecular profile that could be extremely useful to select effective drug for these patients.Example 4: Comparative analysis of nucleus / cytoplasm versus cytoplasm / total ratios in primary

[0132] In this example, two approaches for calculating a relative index based on TDP-43 measurements in different cellular compartments were compared: the nucleus / cytoplasm (N / C) ratio and the cytoplasm / total (C / T) ratio, where the total (T) is defined as the sum of the nuclear and cytoplasmic signals (T = Nucleus + Cytoplasm). Representative values for three groups (healthy controls, pre-ALS subjects and ALS patients) are shown in Figure 4. For each sample, nuclear and cytoplasmic signals were measured, the total cellular value was calculated, and both N / C and C / T ratios were derived.

[0133] Overall, controls exhibited high N / C ratios and low C / T ratios, ALS patients showed the opposite pattern (very low N / C and high C / T), and pre-ALS subjects fell in between. These values were further analysed and presented as bar plots with error bars in Figure 5. The C / T ratio demonstrated lower dispersion and stronger statistical significance (p = 0.0006) compared with the N / C ratio (p = 0.0012), indicating improved group separation when using the cytoplasm-to-total normalisation. This improved separation is consistent with normalising the cytoplasmic signal against the total cellular TDP-43 content, which may reduce variability arising from differences in cell size and sample heterogeneity.

[0134] Example 5: Statistical analysis of nucleus / cytoplasm ratio

[0135] In this example, a one-way ANOVA followed by Tukey’s multiple comparisons test was performed to evaluate the discriminatory performance of the nucleus / cytoplasm (N / C) ratio across three groups: healthy controls, pre-ALS subjects and ALS patients. As shown in Figure 6, all pairwise comparisons were statistically significant (p < 0.05), with Control vs ALS showing the largest mean difference (2.273, adjusted p = 0.0002). These results confirm that the N / C ratio can separate groups, however, the confidence intervals are wide and variability within groups is high, indicating that this metric is sensitive to sample heterogeneity and may lack robustness for clinical classification. Furthermore, the N / C ratio depends on fractionation and loading controls, which introduces additional sources of variability and reduces reproducibility.

[0136] Example 6: Statistical comparison of nucleus / cytoplasmIn this example, the same statistical approach as Example 5 was applied to the cytoplasm / total (C / T) ratio. As shown in Figure 7, all pairwise comparisons were highly significant, with Control vs ALS showing the largest mean difference (0.5701, adjusted p < 0.0001) and preALS vs ALS showing p = 0.0001. Compared to the N / C ratio, the C / T ratio demonstrated narrower confidence intervals, lower p-values and reduced dispersion, confirming that this metric provides a more robust and reproducible measure for distinguishing ALS patients from controls and pre-ALS subjects. This improvement is attributed to the intra-cellular normalization achieved by dividing the cytoplasmic signal by the total cellular TDP-43 content, mitigating variability due to cell size and heterogeneity. Both signals used to compute the C / T ratio are acquired in the same cell population by multiplex flow cytometry, eliminating the need for fractionation, loading controls and densitometric analysis required by conventional approaches.

[0137] MATERIALSAND METHODS

[0138] 1. Flow cytometry protocol

[0139] 1.1. Fresh lymphocytes are extracted from the blood of the patients and obtained by Ficoll-Paque.

[0140] Cells are fixed with 4% PFA in PBS1X for 15 min at 37°C. Subsequently, they are permeabilized with 0.5% Tween 20 in PBS1X for 25 min, after which the blocking step is performed by adding 5% NGS for 30 min.

[0141] Cells are incubated with primary antibody (c-terminal TDP-43) for 1h at 37 °C used at 1:100 dilution in 5% NGS and the cells are incubated overnight at 4°. For the negative control, IgG isotype primary antibody is employed.

[0142] 1.2. Streptavidin Quantum Dots (QD-sav)

[0143] Subsequent to primary antibody incubation, cells are incubated with biotinylated antimouse or anti-rabbit secondary antibody at a 1:500 dilution in 5% NGS for 1 h at room temperature. The samples are then washed twice with PBS1X and incubated with QD655-sav at 1 :100 dilution in 5% NGS at room temperature for 1 h followed by a wash with PBS1X.

[0144] 1.3. Quantum Dots Secondary Antibody (QD-Ab2)Subsequent to primary antibody incubation, cells are incubated with QD655 conjugated with secondary antibody at a 1:100 dilution in 5% NGS at room temperature for 1 h followed by a PBS1X wash.

[0145] 1.4. Alexa Fluor 488 (A488) Secondary Antibody

[0146] Subsequent to primary antibody incubation, cells are incubated with Alexa Fluor 488 conjugated with secondary antibody at a 1:500 dilution in 5% NGS at room temperature for 1 h and washed with PBS1X.

[0147] 2. Data Acquisition and analysis

[0148] Cell suspension is analysed using a Cytoflex flow cytometer (Beckman-Coulter) and 405 nm laser as excitation source for QDs and 488 nm laser for A488 were used. A minimum of 1 million cells are acquired unless limited by the total number of isolated cells. Sample gating is performed by taking the lymphoblasts population based on the FSC (cell size) and SSC (complexity). Cytexpert (Beckman-Coulter) software is used to analyze all data. Immortalized lymphoblast data is analysed by calculating the relative fluorescence unit (RFU) against control’s mean for each day of measurement. Mean fluorescence intensity (MFI) ratio among QD intensity and Alexa488 values of each condition is done for Patients’ lymphocytes data. GraphPad Prism 8 software is used for data analysis. Firstly, normality test is done for each data and, as the data followed a normal distribution, a one-way ANOVA analysis was employed. Statistically differences were analysed using Bonferroni’s analysis. Data is represented by the mean±SEM.

[0149] 3. Patient data:

[0150] Table 1. Lymphoblasts

[0151] Clinical

[0152] Subject IDs

[0153] presentatio

[0154]

[0155] Control NA 4

[0156] ALS Bulbar,

[0157]

[0158] respiratory

[0159]

[0160] Table 2. Lymphocytes

[0161]

[0162] Table 3. Reagents table

[0163]

[0164]

[0165] CLAUSES

[0166] 1. A method for screening or identifying subjects at risk of suffering or having a TDP- 43 proteinopathy, the method comprising calculating a ratio of the cytoplasmic concentration to the total cellular concentration of TDP-43 in mononuclear immunological cells (from herein after “cell sample”) isolated from a blood sample of the subject; comparing the calculated ratio to one or more predefined decision thresholds to classify the subject into at least one category indicative of a higher or lower likelihood of disease presence; and generating a risk stratification output that specifies whether the subject falls within a category associated with an increased or decreased likelihood of having the disease; wherein the mononuclear immunological cells are primary, non-immortalized cells, obtained directly from the subject's peripheral blood, excluding any artificially immortalized or cultured cell lines.

[0167] 2. The method of clause 1, wherein the method comprises:

[0168] a. determining the cytoplasmic concentration and the total cellular concentration of TDP-43 in the mononuclear immunological cells (cell sample) isolated from the subject;

[0169] b. calculating a ratio of the cytoplasmic concentration to the total cellular concentration of TDP-43;

[0170] c. comparing the calculated ratio to one or more predefined decision thresholds to classify the individual into at least one category indicative of a higher or lower likelihood of disease presence; and

[0171] d. generating a risk stratification output that specifies whether the subject falls within a category associated with an increased or decreased likelihood of having the disease.

[0172] 3. The method of clause 1 or 2, wherein the TDP-43 proteinopathy is selected from the list consisting of amyotrophic lateral sclerosis, frontotemporal lobar degeneration and limbic-predominant age-related TDP-43 encephalopathy.

[0173] 4. The method of any one of clauses 1 to 3, wherein the total cellular concentration and the cytoplasmic concentration of the TDP-43 protein are determined by an immunoassay selected from the group consisting of:

[0174] a. Western blot,b. enzyme-linked immunosorbent assay (ELISA),

[0175] c. immunofluorescence, and

[0176] d. flow cytometry or imaging flow cytometry.

[0177] 5. The method of clause 1 to 3, wherein the cytoplasmic concentration of the TDP- 43 protein is determined by incubating the cell sample with a primary antibody that specifically binds the TDP-43 protein (TAR DNA-binding protein 43 (TDP-43) and / or phosphorylated TDP-43 (pTDP-43)) and incubating the cell sample with a secondary antibody conjugated to one or more Quantum Dots (QD) under conditions sufficient to form a complex with the primary antibody; and analysing the cell sample to detect signals from QDs, thereby determining the cytoplasmic concentration of the TDP-43 protein.

[0178] 6. The method of clause 1 to 3, wherein the cytoplasmic concentration of the biomarker is determined by incubating the cell sample with a primary antibody that specifically binds the TDP-43 protein (TAR DNA-binding protein 43 (TDP-43) and / or phosphorylated TDP-43 (pTDP-43)) and incubating the cell sample with a secondary antibody and simultaneously or subsequently with one or more quantum dots, wherein the secondary antibody specifically binds to the one or more QDs under conditions sufficient to first form a complex between the primary and secondary antibodies and second through a high-affinity binding interaction between the secondary antibody and the QDs; and analysing the cell sample to detect signals from QDs, thereby determining the cytoplasmic concentration of the TDP-43 protein.

[0179] 7. The method of clause 1 to 6, wherein the total cellular concentration of the TDP- 43 protein is determined by incubating the cell sample with a primary antibody that specifically binds the TDP-43 protein (TAR DNA-binding protein 43 (TDP-43) and / or phosphorylated TDP-43 (pTDP-43)) and incubating the cell sample with a secondary antibody conjugated to a fluorophore (F), distinct from a QD and capable of labelling the nucleus and the cytoplasm of the cells, under conditions sufficient to form a complex with the primary antibody; and analysing the cell sample to detect signals from F, thereby determining the total cellular concentration of the TDP-43 protein.

[0180] 8. The method of clause 1 to 6, wherein the total cellular concentration of the TDP- 43 protein is determined by incubating the cell sample with a primary antibodythat specifically binds the TDP-43 protein (TAR DNA-binding protein 43 (TDP-43) and / or phosphorylated TDP-43 (pTDP-43)) and incubating the cell sample with a secondary antibody and one or more fluorophores (F), distinct from a QD and capable of labelling the nucleus and the cytoplasm of the cells, under conditions sufficient to first form a complex between the primary and secondary antibodies and second through a high-affinity binding interaction between the secondary antibody and the Fs; and analysing the cell sample to detect signals from Fs, thereby determining the total cellular concentration of the TDP-43 protein.

[0181] 9. The method according to clauses 5 to 8, wherein the cytoplasmic concentration of the TDP-43 protein is determined according to clause 5 or 6 and the total cellular concentration of the TDP-43 protein is determined according to clause 7 or 8.

[0182] 10. The method of any one of clauses 5 to 7, wherein the QD and the fluorophore (F) are selected from different emission spectra suitable for multicolor flow cytometry, allowing simultaneous detection of multiple signals.

[0183] 11. The method of any one of clauses 1 to 8, wherein the predetermined threshold is selected based on statistical analysis of TDP-43 ratios obtained from a reference or control population, wherein if in the comparing step the ratio result is higher than the decision threshold the subject shall be categorize as an individual into at least the higher-risk group of suffering from a TDP-43 proteinopathy otherwise the subject shall be categorized as a lower-risk group.

[0184] 12. The method according to clause 2, wherein the cytoplasmic concentration of the TDP-43 protein is determined according to claim 5 or 6 and the total cellular concentration of the TDP-43 protein is determined according to clause 7 or 8; wherein the predetermined threshold is selected based on statistical analysis of TDP-43 ratios obtained from a reference or control population, wherein if in the comparing step the ratio result is higher than the decision threshold the subject shall be categorize as a higher-risk group of suffering from a TDP-43 proteinopathy otherwise the subject shall be categorized as a lower-risk group; and wherein the TDP-43 proteinopathy is amyotrophic lateral sclerosis.

[0185] 13. The method according to clause 12, wherein the QD is Qdot 655 and the fluorophore (F) is Alexa fluor 488.The method of any one of clauses 1 to 13, further comprising selecting or adjusting a therapeutic regimen for the subject based on the classification, wherein subjects classified above a certain ratio threshold receive a different therapy than subjects classified below said threshold.

[0186] The method of any one of clauses 1 to 14, wherein calculating the ratio and classifying the subject is performed by a computer-implemented algorithm that processes input data from the measurements in steps (b) and (c) and outputs an indication of the subject’s classification.

[0187] The method of any one of clauses 1 to 15, further comprising utilizing a kit or system configured to:

[0188] a. isolate cytoplasmic fractions from the sample,

[0189] b. detect or quantify the biomarker in both the cytoplasmic fraction and the whole-cell fraction, and

[0190] c. provide instructions for comparing the ratio to the predetermined threshold to classify the subject.

Claims

CLAIMS1. A method for calculating a ratio of the cytoplasmic concentration to the total cellular concentration of TDP-43 in an isolated biological sample comprising mononuclear immunological cells, wherein the method comprises;a. determining the cytoplasmic concentration of TDP-43 in the mononuclear immunological cells, wherein the cytoplasmic concentration of the TDP- 43 protein is determined by detecting and quantifying the signals from Quantum Dots (QDs) by using flow cytometry;b. determining the total cellular concentration of TDP-43 in the mononuclear immunological cells by using flow cytometry; andc. calculating a ratio of the cytoplasmic concentration to the total cellular concentration of TDP-43;and wherein the mononuclear immunological cells are primary, whole, non-immortalized cells, obtained directly from peripheral blood from a subject, excluding any artificially immortalized or cultured cell lines.

2. The method of claim 1, wherein the cytoplasmic concentration of the TDP-43 protein is determined by incubating the biological sample with a first binding moiety, preferably selected from the group consisting of antibodies or antibody fragments, that specifically binds the TDP-43 protein and incubating the biological sample with a second binding moiety capable of forming a complex with the first binding moiety, preferably selected from the group consisting of antibodies or antibody fragments, conjugated to one or more Quantum Dots (QD), under conditions sufficient to form a complex with the first binding moiety; and analysing the biological sample to detect signals from QDs, thereby determining the cytoplasmic concentration of the TDP-43 protein.

3. The method of claim 1 , wherein the cytoplasmic concentration of the TDP-43 protein is determined by incubating the biological sample with a first binding moiety, preferably selected from the group consisting of antibodies or antibody fragments, that specifically binds the TDP-43 protein and incubating the biological sample with a second binding moiety capable of forming a complex with the first binding moiety, preferably selected from the group consisting of antibodies or antibody fragments, and one or more quantum dots, wherein the second binding moiety specifically binds to the one or more QDs under conditions sufficient tofirst form a complex between the first and second binding moiety and second through a high-affinity binding interaction between the second binding moiety and the QDs, and analysing the biological sample to detect signals from QDs, thereby determining the cytoplasmic concentration of the TDP-43 protein.

4. The method of any one of claims 1 to 3, wherein the total cellular concentration of the TDP-43 protein is determined by incubating the biological sample with a first binding moiety, preferably selected from the group consisting of antibodies or antibody fragments, that specifically binds the TDP-43 protein and incubating the biological sample with a second binding moiety capable of forming a complex with the first binding moiety, preferably selected from the group consisting of antibodies or antibody fragments, conjugated to a fluorophore (F), distinct from a QD and capable of labelling the nucleus and the cytoplasm of the cells, under conditions sufficient to form a complex with the first binding moiety , and analysing the biological sample to detect signals from F, thereby determining the total cellular concentration of the TDP-43 protein.

5. The method of any one of claims 1 to 3, wherein the total cellular concentration of the TDP-43 protein is determined by incubating the biological sample with a first binding moiety, preferably selected from the group consisting of antibodies or antibody fragments, that specifically binds the TDP-43 protein and incubating the biological sample with a second binding moiety capable of forming a complex with the first binding moiety, preferably selected from the group consisting of antibodies or antibody fragments, and one or more fluorophores (F), distinct from a QD and capable of labelling the nucleus and the cytoplasm of the cells, under conditions sufficient to first form a complex between the first and second binding moiety and second through a high-affinity binding interaction between the second binding moiety and the Fs, and analysing the biological sample to detect signals from Fs, thereby determining the total cellular concentration of the TDP-43 protein.

6. The method according to claims 2 to 5, wherein the cytoplasmic concentration of the TDP-43 protein is determined according to claim 2 or 3 and the total cellular concentration of the TDP-43 protein is determined according to claim 4 or 5.

7. The method according to claims 2 to 5, wherein the QD and the fluorophore (F) are selected from different emission spectra suitable for multicolor flow cytometry,allowing simultaneous detection of multiple signals, preferably wherein the QD is Qdot 655 and the fluorophore (F) is Alexa fluor 488.

8. The method according to any one of claims 1 to 7, wherein calculating the ratio from step (c) of claim 1 is performed by a computer-implemented algorithm that processes input data from the measurements in step (b) of claim 1.

9. A method for determining or assessing the likelihood that a subject is suffering from amyotrophic lateral sclerosis (ALS), wherein the method comprises:a. calculating a ratio of the cytoplasmic concentration to the total cellular concentration of TDP-43 in mononuclear immunological cells isolated from a biological sample of the subject, wherein the cytoplasmic concentration is determined according to claims 2 or 3 and wherein the total cellular concentration is determined according to claims 4 or 5, and b. comparing the calculated ratio to one or more predefined decision thresholds to classify the subject into at least one category indicative of a higher or lower likelihood of disease presence.

10. The method according to claim 9, wherein the predefined decision threshold is selected based on statistical analysis of TDP-43 ratios obtained from a reference or control population, wherein if in the comparing step the ratio result is higher than the decision threshold the subject shall be categorized as an individual into at least the higher-risk group of suffering from a TDP-43 proteinopathy otherwise the subject shall be categorized as a lower-risk group.

11. The method according to claims 9 and 10, wherein the cytoplasmic concentration of the TDP-43 protein is determined according to claim 2 or 3 and the total cellular concentration of the TDP-43 protein is determined according to claim 4 or 5, wherein the predetermined threshold is selected based on statistical analysis of TDP-43 ratios obtained from a reference or control population, wherein if in the comparing step the ratio result is higher than the decision threshold the subject shall be categorize as a higher-risk group of suffering from a TDP-43 proteinopathy otherwise the subject shall be categorized as a lower-risk group, and wherein the TDP-43 proteinopathy is amyotrophic lateral sclerosis (ALS).

12. The method according to claim 10, wherein the QD and the fluorophore (F) are selected according to claim 7.

13. A computer implemented method for determining or assessing the likelihood that, or determining if, a subject is suffering from amyotrophic lateral sclerosis (ALS), wherein the method comprises:a. determining or receiving the cytoplasmic concentration of TDP-43 in the mononuclear immunological cells, preferably wherein the cytoplasmic concentration of the TDP-43 protein is determined by detecting and quantifying the signals from Quantum Dots (QDs) by using flow cytometry; b. determining or receiving the total cellular concentration of TDP-43 in the mononuclear immunological cells, preferably determining said cellular concentration by using flow cytometry;c. integrating both the cytoplasmatic concentration and the total cellular concentration of TDP-43 in the mononuclear immunological cells into a diagnostic model configured to generate, based on said cytoplasmatic and total cellular concentrations of TDP-43, an output indicating the likelihood that, or indicating if, a subject is suffering from amyotrophic lateral sclerosis (ALS).

14. The computer implemented method according to claim 13, wherein generating an output indicating the likelihood that a subject is suffering from amyotrophic lateral sclerosis (ALS) comprises classifying the subject into at least one category indicative of a higher or lower likelihood of suffering from ALS, and wherein generating an output indicating if a subject is suffering from amyotrophic lateral sclerosis (ALS) comprises classifying the subject into at least one category indicative of suffering or not suffering from ALS.

15. The computer implemented method according to any one of claims 13 or 14, wherein the diagnostic model is a machine learning model; a rule-based decision algorithm; or a mathematical model wherein the cytoplasmatic and total cellular concentrations of TDP-43 are parameters or variables used to fit said mathematical model.

16. The method according to claim 15, wherein the machine learning model is a logistic regression model, a linear or polynomial model, a discriminant analysis model, a quadratic discriminant analysis model, a Bayesian model, a decision tree model, a random forest model, a gradient boosting model, a support vectormachine, a k-nearest neighbour model, an artificial neural network, a transformerbased model, a feed-forward neural network or any ensemble model thereof.

17. The method according to claim 15, wherein the machine learning model is a binary classifier configured to map an input vector x = (c_cyto, c_total) to a score s in [0,1] indicative of ALS, preferably wherein the classifier has been trained on labelled measurement data of said concentrations, more preferably wherein the output class is determined by thresholding s at a threshold T selected to optimize diagnostic performance on a validation set.

18. The method according to any one of claims 13 to 17, wherein the diagnostic model is generated according to the following steps:iii. Receiving the cytoplasmatic concentration and the total cellular concentration TDP-43 in mononuclear immunological cells from a biological sample obtained from a cohort of subjects confirmed to have ALS and from a cohort of subjects confirmed to not have ALS, together with corresponding diagnostic labels indicating whether each subject has ALS or not; and iv. generating the diagnostic model by using at least a training set or subset of the cytoplasmatic and total cellular concentrations and corresponding diagnostic labels received in step (i) to train a machine learning model, to generate a rule-based decision algorithm or to fit the parameters of a mathematical model, thus generating the diagnostic model.

19. The method according to claim 18, wherein the generation of the diagnostic model further comprises (iii) validating the diagnostic model generated in step (ii) by applying it to at least a validation set or subset of the transcript expression data and corresponding diagnostic labels received in step (i).

20. A computer program comprising processor readable instructions which, when the program is executed by a computer, cause the computer to carry out steps of the method of any of the previous claims.

21. A computer-readable storage medium comprising processor readable instructions which, when executed by a computer, cause the computer to carry out steps of the method of any one of the previous claims.

22. A system comprising a processor and a memory, wherein the memory comprises instructions which, when executed by the processor, cause the system to carry out the method of any one of the previous claims.

Images