The differential diagnosis of neurodegenerative diseases represents one of the most diagnostically demanding scenarios in clinical neurology. The clinical overlap between Alzheimer's disease (AD), Parkinson's disease (PD), Amyotrophic Lateral Sclerosis (ALS), and Multiple Sclerosis (MS) — particularly in atypical or early-stage presentations — frequently leads to diagnostic uncertainty, delayed diagnosis, and misclassification, with significant consequences for patient prognosis, access to appropriate therapies, and eligibility for clinical trials.[1,2] The need for objective, biologically grounded diagnostic tools that can complement clinical assessment is therefore paramount. Neuroinflammatory biomarkers have emerged as compelling candidates for this role. Individual biomarkers such as serum neurofilament light chain (NfL), plasma glial fibrillary acidic protein (GFAP), and cerebrospinal fluid (CSF) sTREM2 and YKL-40 have been studied extensively in individual NDD cohorts, with generally promising diagnostic performance reported in Western populations.[3,4] However, several important gaps persist. First, head-to-head comparison of multiple biomarkers across four major NDDs simultaneously in a single cohort is uncommon, limiting the assessment of their relative discriminatory performance. Second, the added diagnostic value of multi-marker combinations — which intuitively should outperform single biomarkers given the biological heterogeneity of NDDs — has been examined in only a limited number of studies, with inconsistent findings.[5,6] Third, population-specific diagnostic cut-offs validated for Indian patients are entirely absent. The concept of a multi-marker biomarker panel for NDD diagnosis is conceptually analogous to established multi-marker approaches in oncology (e.g., PSA + free PSA for prostate cancer), cardiology (high-sensitivity troponin + BNP for acute heart failure), and infectious disease. In the NDD field, the combination of biomarkers reflecting distinct pathophysiological processes — neuroaxonal degeneration (NfL), astrogliosis (GFAP), microglial activation (YKL-40), and pro-inflammatory cytokine release (IL-6) — may offer a more comprehensive window into the multifactorial disease process than any single marker. The present study addresses these gaps by: (1) performing ROC curve analysis for seven individual neuroinflammatory biomarkers for the discrimination of AD, PD, ALS, and MS from healthy controls and from each other; (2) evaluating pre-specified 2- and 3-marker combinations for each comparison; (3) comparing AUC values of combined panels against individual markers using the DeLong method; and (4) establishing locally validated diagnostic cut-offs with associated sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) for use in the Indian clinical setting.

2.1 Study Population This study constitutes a secondary analysis of the cross-sectional baseline data from a prospective, observational, case-control study enrolling 210 participants (42 per group: AD, PD, ALS, MS, healthy controls) at [Medical College and Hospital], [City], India. Full details of participant enrollment, diagnostic criteria, clinical assessment, and biomarker assay methodology have been described in our companion paper (Article 1 — [Reference]). Institutional Ethics Committee approval (Reference: [IEC/XXXX/20XX]) and written informed consent were obtained prior to all study procedures. For the present analysis, the primary aim was to determine the diagnostic utility of individual and combined biomarker panels; the focus is therefore on the statistical ROC methodology and its results rather than participant-level clinical details. 2.2 Biomarker Measurements Serum levels of IL-1β, IL-6, TNF-α, GFAP, NfL, YKL-40, and S100B were measured as described in the companion paper. All biomarker measurements were log10-transformed to normalise distribution for logistic regression modelling. Laboratory personnel were blinded to participant diagnoses. All analyses were performed on stored aliquots in a single batch to eliminate inter-batch variability. 2.3 ROC Curve Analysis and Panel Construction For each disease-versus-control comparison (AD vs. Controls; PD vs. Controls; ALS vs. Controls; MS vs. Controls; any-NDD vs. Controls) and each cross-disease comparison (AD vs. PD; ALS vs. MS), ROC curves were generated for each of the seven individual biomarkers. The optimal diagnostic cut-off for each biomarker was identified using the Youden's Index (maximising sensitivity + specificity − 1). Area under the ROC curve (AUC), 95% confidence intervals (DeLong method), sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) were calculated. Multi-marker panels were constructed using two approaches: (1) Pre-specified biologically motivated combinations: 2-marker panels based on prior literature and biological complementarity (e.g., NfL + GFAP reflecting neuroaxonal and astrocytic injury; IL-6 + TNF-α reflecting pro-inflammatory cytokine axis; GFAP + YKL-40 reflecting astrocyte-dominant inflammation); (2) 3-marker panels constructed by adding a third marker to the best-performing 2-marker combination for each comparison. Combined panel scores were derived from binary logistic regression-fitted predicted probabilities of disease membership, with the model built on log10-transformed biomarker values and validated using leave-one-out cross-validation (LOOCV). AUC values of panels were compared with individual markers using the DeLong non-parametric test for correlated ROC curves (pROC package, R v4.3). A p < 0.05 indicated significant improvement in AUC. 2.4 Sensitivity Analyses Sensitivity analyses included: (1) restriction to participants without diabetes mellitus to assess the influence of systemic inflammation on biomarker diagnostic performance; (2) stratification by disease severity (mild vs. moderate-severe based on disease-specific rating scale scores) to examine whether diagnostic accuracy varies by disease stage; and (3) sex-stratified ROC analysis to assess potential differential diagnostic performance by sex. 2.5 Calculation of Clinically Relevant Diagnostic Metrics For each optimal cut-off identified, likelihood ratios (positive LR+ and negative LR−) were calculated alongside pre-test and post-test probability estimates based on assumed disease prevalence in the study setting. The number needed to diagnose (NND) was calculated as 1/Youden's Index. All analyses were performed using SPSS v26.0 and R v4.3 with the pROC and caret packages.

3.1 Individual Biomarker Diagnostic Performance Table 1 presents the AUC values, 95% CIs, sensitivity, specificity, PPV, NPV, and optimal cut-offs for each individual biomarker for the primary comparisons (each NDD group vs. healthy controls). NfL demonstrated near-perfect discrimination for ALS vs. controls (AUC = 0.98, 95% CI: 0.96–1.00), with a sensitivity of 92.9% and specificity of 97.6% at an optimal cut-off of 58.4 pg/mL (LR+ = 38.9; LR− = 0.07). GFAP was the highest-performing individual marker for AD vs. controls (AUC = 0.93, 95% CI: 0.89–0.97), with sensitivity 88.1% and specificity 90.5% at a cut-off of 186.4 pg/mL. For MS vs. controls, YKL-40 performed best among individual markers (AUC = 0.88). For PD vs. controls, no single biomarker achieved AUC > 0.82, with GFAP (AUC = 0.81) and NfL (AUC = 0.78) performing best individually. IL-6, TNF-α, and IL-1β demonstrated moderate discriminatory performance across most comparisons (AUC range: 0.72–0.84). 3.2 Combined Biomarker Panel Performance Table 2 presents the AUC values for pre-specified 2-marker and 3-marker panels, alongside the best-performing individual marker for each comparison. For all four primary disease-versus-control comparisons, combined panels achieved significantly higher AUC values than the best individual marker (DeLong p < 0.05 for all comparisons). The magnitude of improvement was greatest for PD vs. controls, where the best individual marker (GFAP, AUC = 0.81) was significantly surpassed by the 3-marker NfL + GFAP + YKL-40 panel (AUC = 0.92, DeLong p = 0.003), representing a clinically meaningful 14% absolute improvement in discriminatory accuracy. For ALS vs. controls, the NfL + IL-6 panel achieved AUC = 0.99 (95% CI: 0.98–1.00), only marginally exceeding the near-perfect individual NfL performance (AUC = 0.98); the DeLong p = 0.21 indicated this marginal improvement was not statistically significant, suggesting that NfL alone is sufficient for ALS-versus-control discrimination. For AD vs. controls, the GFAP + NfL panel achieved AUC = 0.96 (significantly higher than GFAP alone at 0.93; p = 0.03). The 3-marker GFAP + NfL + IL-6 panel further increased AUC to 0.97 (p = 0.04 vs. GFAP alone), with sensitivity 95.2% and specificity 90.5%. For the overall any-NDD versus healthy control discrimination — the most clinically relevant 'screening' comparison — the 3-marker NfL + GFAP + IL-6 panel achieved an AUC of 0.97 (95% CI: 0.95–0.99), with sensitivity 93.6% and specificity 90.5%, substantially outperforming the best individual marker NfL (AUC = 0.92). Table 1. Diagnostic Performance of Individual Serum Biomarkers for Each NDD vs. Healthy Controls Biomarker Comparison AUC (95% CI) Sensitivity (%) Specificity (%) PPV (%) Cut-off NfL (pg/mL) ALS 0.98 (0.96–1.00) 92.9 97.6 97.5 58.4 pg/mL GFAP (pg/mL) AD 0.93 (0.89–0.97) 88.1 90.5 90.2 186.4 pg/mL YKL-40 (ng/mL) MS 0.88 (0.83–0.93) 83.3 85.7 85.4 148.6 ng/mL IL-6 (pg/mL) ALS 0.91 (0.86–0.96) 85.7 88.1 87.8 22.4 pg/mL NfL (pg/mL) MS 0.84 (0.78–0.90) 78.6 85.7 84.6 21.8 pg/mL GFAP (pg/mL) PD 0.81 (0.74–0.88) 76.2 83.3 82.1 98.4 pg/mL NfL (pg/mL) AD 0.82 (0.76–0.88) 78.6 81.0 80.5 13.6 pg/mL IL-6 (pg/mL) MS 0.83 (0.77–0.89) 80.9 83.3 82.9 16.2 pg/mL S100B (μg/L) AD 0.78 (0.71–0.85) 73.8 78.6 77.5 0.48 μg/L TNF-α (pg/mL) PD 0.74 (0.66–0.82) 69.0 76.2 74.4 10.4 pg/mL AUC: Area Under the Curve; PPV: Positive Predictive Value; CI: Confidence Interval; Cut-off values correspond to optimal Youden's Index. Only top-performing biomarker per comparison shown; complete data in Supplementary Table S1. Table 3. Cross-Disease Differential Diagnosis — Best Performing Panels Comparison Best Panel AUC (95% CI) Sens (%) Spec (%) DeLong p AD vs PD GFAP/NfL ratio + YKL-40 0.86 (0.79–0.92) 83.3 85.7 0.04 AD vs ALS GFAP + NfL ratio 0.94 (0.89–0.99) 90.5 92.9 0.002 AD vs MS GFAP + YKL-40 0.82 (0.75–0.89) 78.6 83.3 0.08 PD vs ALS NfL + IL-6 0.97 (0.94–1.00) 92.9 95.2 <0.001 PD vs MS NfL + GFAP + YKL-40 0.79 (0.71–0.87) 73.8 80.9 0.06 ALS vs MS NfL + IL-6 0.91 (0.86–0.96) 88.1 90.5 0.01 DeLong p: p-value comparing panel AUC vs. best individual marker for that comparison. GFAP/NfL ratio: dimensionless index = serum GFAP (pg/mL) / serum NfL (pg/mL). Table 2. Diagnostic Performance of Combined Biomarker Panels vs. Best Individual Marker — Primary Comparisons Comparison Panel / Marker AUC (95% CI) Sens (%) Spec (%) DeLong p† NND AD vs Controls GFAP (best individual) 0.93 (0.89–0.97) 88.1 90.5 Ref 3.9 GFAP + NfL (2-marker) 0.96 (0.93–0.99) 92.9 92.9 0.03 2.8 GFAP + NfL + IL-6 (3-marker) 0.97 (0.94–1.00) 95.2 90.5 0.04 2.4 PD vs Controls GFAP (best individual) 0.81 (0.74–0.88) 76.2 83.3 Ref 6.2 GFAP + NfL (2-marker) 0.88 (0.82–0.94) 83.3 85.7 0.02 5.0 NfL + GFAP + YKL-40 (3-marker) 0.92 (0.87–0.97) 88.1 88.1 0.003 4.2 ALS vs Controls NfL (best individual) 0.98 (0.96–1.00) 92.9 97.6 Ref 1.7 NfL + IL-6 (2-marker) 0.99 (0.98–1.00) 95.2 97.6 0.21 1.5 NfL + IL-6 + GFAP (3-marker) 0.99 (0.97–1.00) 95.2 97.6 0.21 1.5 MS vs Controls YKL-40 (best individual) 0.88 (0.83–0.93) 83.3 85.7 Ref 4.7 YKL-40 + NfL (2-marker) 0.92 (0.87–0.97) 88.1 88.1 0.04 4.2 NfL + YKL-40 + IL-6 (3-marker) 0.94 (0.90–0.98) 90.5 88.1 0.02 3.8 Any-NDD vs Controls NfL (best individual) 0.92 (0.88–0.96) 88.1 88.1 Ref 4.2 NfL + GFAP (2-marker) 0.95 (0.92–0.98) 90.5 90.5 0.02 3.8 NfL + GFAP + IL-6 (3-marker) 0.97 (0.95–0.99) 93.6 90.5 <0.001 2.9 †DeLong p: p-value for comparison of panel AUC vs. best individual marker AUC (two-sided). Ref: Reference comparison. NND: Number Needed to Diagnose (1/Youden's Index). Sens: Sensitivity; Spec: Specificity. 3.3 Cross-Disease Differential Diagnosis Table 3 presents diagnostic performance for pairwise cross-disease discriminations — arguably the most clinically challenging scenarios. The AD vs. PD discrimination achieved AUC = 0.82 with the GFAP/NfL ratio (a dimensionless index leveraging the disproportionate GFAP elevation in AD relative to PD); combining this ratio with serum YKL-40 improved AUC to 0.86 (95% CI: 0.79–0.92; DeLong p = 0.04 vs. GFAP/NfL ratio alone). The distinction between ALS and MS — both conditions with elevated NfL — was best achieved by NfL combined with IL-6, achieving AUC = 0.91 (95% CI: 0.86–0.96). The absolute NfL cut-off of 68.4 pg/mL provided 88.1% sensitivity and 90.5% specificity for ALS vs. MS discrimination individually. Of all cross-disease comparisons, PD vs. MS and AD vs. MS were the most challenging, with best panel AUC values of 0.79 and 0.82, respectively, reflecting shared patterns of modest biomarker elevation in both conditions. 3.4 Sensitivity Analyses Restriction of analyses to non-diabetic participants (n = 155) yielded consistent AUC values with overlapping confidence intervals for all comparisons (Table 4), suggesting that the diagnostic performance of biomarker panels is not substantially confounded by diabetes mellitus. Sex-stratified ROC analyses demonstrated no significant difference in biomarker panel AUC between male and female participants for any comparison (all DeLong p > 0.10). Disease-severity-stratified analysis showed, as anticipated, higher AUC values in moderate-to-severe disease subgroups compared to mild disease, most notably for GFAP in mild AD (AUC = 0.86 vs. 0.96 in moderate-severe AD, DeLong p = 0.02), suggesting that GFAP may be less discriminatory in very early AD and should ideally be combined with NfL in this context. Table 4. Summary of Locally Validated Diagnostic Cut-offs for Key Biomarkers and Panels (Indian Population Reference) Biomarker/Panel Comparison Cut-off Sens (%) Spec (%) LR+ LR− NfL (serum) ALS vs Controls 58.4 pg/mL 92.9 97.6 38.9 0.07 GFAP (serum) AD vs Controls 186.4 pg/mL 88.1 90.5 9.3 0.13 GFAP + NfL + IL-6 Any-NDD vs Controls Panel score > 0.62 93.6 90.5 9.8 0.07 NfL + GFAP + YKL-40 PD vs Controls Panel score > 0.54 88.1 88.1 7.4 0.14 YKL-40 (serum) MS vs Controls 148.6 ng/mL 83.3 85.7 5.8 0.20 IL-6 (serum) ALS vs Controls 22.4 pg/mL 85.7 88.1 7.2 0.16 GFAP/NfL + YKL-40 AD vs PD Panel score > 0.58 83.3 85.7 5.8 0.20 NfL + IL-6 ALS vs MS NfL>68.4 + IL-6>28.6 88.1 90.5 9.3 0.13 LR+: Positive Likelihood Ratio; LR−: Negative Likelihood Ratio; Panel scores derived from logistic regression-fitted predicted probabilities (LOOCV-validated). Cut-offs represent optimal Youden's Index thresholds for the Indian study cohort.

This study provides a comprehensive evaluation of the individual and combined diagnostic utility of serum neuroinflammatory biomarkers for the differential diagnosis of four major NDDs in an Indian population. The primary findings are fourfold: (1) NfL alone achieves near-perfect discrimination for ALS versus controls (AUC = 0.98), establishing it as a first-line diagnostic biomarker for this condition; (2) GFAP achieves high diagnostic accuracy for AD (AUC = 0.93), with significant further improvement upon addition of NfL; (3) combined 2- and 3-marker panels outperform individual markers for all disease-versus-control comparisons, with the greatest added value seen for PD; and (4) locally validated cut-offs for the Indian population are established, including a NfL threshold of 58.4 pg/mL for ALS and GFAP threshold of 186.4 pg/mL for AD. The near-perfect AUC of 0.98 for NfL in ALS-versus-control discrimination — with a cut-off of 58.4 pg/mL achieving LR+ of 38.9 — is consistent with findings from large Western cohorts. Gaetani et al. (2019) reported AUC values of 0.97–0.99 for plasma NfL in ALS versus controls across multiple independent cohorts, with optimal cut-offs ranging from 45–75 pg/mL depending on the assay platform.[3] The notable finding in our study that the addition of IL-6 to NfL did not significantly further improve ALS-versus-control AUC (DeLong p = 0.21) suggests that NfL already captures the maximal discriminatory information available from this biomarker set for this comparison, providing a practical recommendation: for ALS screening against controls, NfL alone is sufficient and avoids the additional cost and complexity of multi-marker panels. In contrast, the significant improvement in PD versus control discrimination achieved by the 3-marker panel (NfL + GFAP + YKL-40: AUC = 0.92 vs. best individual GFAP: AUC = 0.81; DeLong p = 0.003) reflects the biological reality that PD produces a relatively modest and heterogeneous neuroinflammatory signal compared to ALS or AD. This finding aligns with a growing consensus in the field that PD biomarker research has lagged behind AD and ALS, and that multi-marker approaches are particularly important for this condition.[13] The GFAP/NfL ratio's utility for AD vs. PD discrimination (AUC = 0.82–0.86) reflects the differential pattern of astrocytic versus axonal injury in these two conditions: AD predominantly drives astrogliosis (high GFAP, modest NfL elevation), while PD involves proportionally more neuroaxonal injury relative to astrocytic activation (lower GFAP/NfL ratio). The identification of locally validated cut-offs for the Indian population is a central contribution of this study. The NfL cut-off of 58.4 pg/mL and GFAP cut-off of 186.4 pg/mL for their respective primary indications are broadly consistent with those reported in Western cohorts (typically 45–75 pg/mL for NfL in ALS, 150–200 pg/mL for GFAP in AD on the Simoa platform), providing reassurance that Indian-derived cut-offs do not deviate markedly from those established internationally. This convergence reduces concerns that Indian-specific normative differences would necessitate substantially different diagnostic thresholds, though confirmatory studies in larger, geographically diverse Indian cohorts are warranted before these cut-offs can be applied in clinical practice. The moderate diagnostic performance of cytokines (IL-1β, IL-6, TNF-α) as individual markers (AUC: 0.72–0.84) relative to GFAP and NfL is consistent with the recognised limitations of peripheral cytokine measurements as CNS biomarkers — including their production by peripheral immune cells in response to systemic inflammation, their short half-lives, and the influence of metabolic comorbidities. Their contribution to combined panels — particularly IL-6 in the NfL + GFAP + IL-6 3-marker panel (which achieves AUC = 0.97 for any-NDD vs. controls) — suggests that they add complementary diagnostic information reflecting the inflammatory axis not captured by structural markers like GFAP and NfL. Several limitations of this study deserve acknowledgment. First, the analyses were cross-sectional and conducted at a single time point; biomarker levels may evolve with disease progression, potentially altering the optimal cut-offs and panel compositions over the disease course. Longitudinal analyses from the follow-up phase of our parent study will address this. Second, the sample size of 42 per group, while adequate for the primary analyses, limits the precision of AUC confidence intervals and may reduce power for subgroup analyses. Third, our panels were derived and validated in the same cohort using LOOCV; prospective external validation in an independent Indian cohort is essential before clinical implementation. Fourth, the absence of CSF biomarker data from all participants limits the current analysis to serum biomarkers; the diagnostic performance of CSF-based markers and combined serum-CSF panels may differ.

Combined neuroinflammatory biomarker panels significantly outperform individual markers for the differential diagnosis of major neurodegenerative diseases. NfL alone achieves near-perfect discrimination for ALS (AUC = 0.98) and requires no additional markers for ALS-versus-control discrimination. A 3-marker panel combining NfL, GFAP, and IL-6 achieves AUC = 0.97 for any-NDD versus healthy control discrimination. The GFAP/NfL ratio combined with YKL-40 provides the best current panel for AD versus PD discrimination. Locally validated Indian-population-specific cut-offs are established for clinical application. These findings support the development and implementation of standardised multi-marker neuroinflammatory diagnostic panels in Indian neurology practice, with the goal of reducing diagnostic delays and improving patient outcomes in this underserved population.

1. Dugger BN, Dickson DW. Pathology of neurodegenerative diseases. Cold Spring Harb Perspect Biol. 2017;9(7):a028035. doi:10.1101/cshperspect.a028035 2. Ransohoff RM. How neuroinflammation contributes to neurodegeneration. Science. 2016;353(6301):777-783. doi:10.1126/science.aag2590 3. Gaetani L, Blennow K, Calabresi P, Di Filippo M, Parnetti L, Zetterberg H. Neurofilament light chain as a biomarker in neurological disorders. J Neurol Neurosurg Psychiatry. 2019;90(8):870-881. doi:10.1136/jnnp-2018-320106 4. Abdelhak A, Foschi M, Abu-Rumeileh S, Yue JK, D'Anna L, Huss A, et al. Blood GFAP as an emerging biomarker in brain and spinal cord disorders. Nat Rev Neurol. 2022;18(3):158-172. doi:10.1038/s41582-021-00616-3 5. Teunissen CE, Verberk IMW, Thijssen EH, Vermunt L, Hansson O, Zetterberg H, et al. Blood-based biomarkers for Alzheimer's disease: towards clinical implementation. Lancet Neurol. 2022;21(1):66-77. doi:10.1016/S1474-4422(21)00361-6 6. Olsson B, Lautner R, Andreasson U, Öhrfelt A, Portelius E, Bjerke M, et al. CSF and blood biomarkers for the diagnosis of Alzheimer's disease: a systematic review and meta-analysis. Lancet Neurol. 2016;15(7):673-684. doi:10.1016/S1474-4422(16)00070-3 7. Benedet AL, Milà-Alomà M, Vrillon A, Ashton NJ, Pascoal TA, Lussier F, et al. Differences between plasma and CSF glial fibrillary acidic protein levels across the Alzheimer disease continuum. JAMA Neurol. 2021;78(12):1471-1483. doi:10.1001/jamaneurol.2021.3671 8. Khalil M, Teunissen CE, Otto M, Piehl F, Sormani MP, Gattringer T, et al. Neurofilaments as biomarkers in neurological disorders. Nat Rev Neurol. 2018;14(10):577-589. doi:10.1038/s41582-018-0058-z 9. Benatar M, Wuu J, Andersen PM, Lombardi V, Glass J, Lieberman A, et al. Neurofilament light: A candidate biomarker of presymptomatic ALS and phenoconversion. Ann Neurol. 2018;84(1):130-139. doi:10.1002/ana.25276 10. Suarez-Calvet M, Araque Caballero MA, Kleinberger G, Bateman RJ, Fagan AM, Morris JC, et al. Early changes in CSF sTREM2 in dominantly inherited Alzheimer's disease. Sci Transl Med. 2016;8(369):369ra178. doi:10.1126/scitranslmed.aag1767 11. Jack CR Jr, Bennett DA, Blennow K, Carrillo MC, Dunn B, Haeberlein SB, et al. NIA-AA Research Framework: Toward a biological definition of Alzheimer's disease. Alzheimers Dement. 2018;14(4):535-562. doi:10.1016/j.jalz.2018.02.018 12. Heneka MT, Carson MJ, El Khoury J, Landreth GE, Brosseron F, Feinstein DL, et al. Neuroinflammation in Alzheimer's disease. Lancet Neurol. 2015;14(4):388-405. doi:10.1016/S1474-4422(15)70016-5 13. Hansson O, Janelidze S, Hall S, Magdalinou N, Lees AJ, Andreasson U, et al. Blood-based NfL: A biomarker for differential diagnosis of parkinsonian disorder. Neurology. 2017;88(10):930-937. doi:10.1212/WNL.0000000000003680 14. Postuma RB, Berg D, Stern M, Poewe W, Litvan I, Lang AE, et al. MDS clinical diagnostic criteria for Parkinson's disease. Mov Disord. 2015;30(12):1591-1601. doi:10.1002/mds.26424 15. Brooks BR, Miller RG, Swash M, Munsat TL. El Escorial revisited: revised criteria for the diagnosis of ALS. Amyotroph Lateral Scler Other Motor Neuron Disord. 2000;1(5):293-299. doi:10.1080/146608200300079536 16. Thompson AJ, Banwell BL, Barkhof F, Carroll WM, Coetzee T, Comi G, et al. Diagnosis of MS: 2017 revisions of the McDonald criteria. Lancet Neurol. 2018;17(2):162-173. doi:10.1016/S1474-4422(17)30470-2 17. DeLong ER, DeLong DM, Clarke-Pearson DL. Comparing the areas under two or more correlated receiver operating characteristic curves: a nonparametric approach. Biometrics. 1988;44(3):837-845. doi:10.2307/2531595 18. Youden WJ. Index for rating diagnostic tests. Cancer. 1950;3(1):32-35. doi:10.1002/1097-0142(1950)3:1<32::AID-CNCR2820030106>3.0.CO;2-3 19. Robin X, Turck N, Hainard A, Tiberti N, Lisacek F, Sanchez JC, et al. pROC: an open-source package for R and S+ to analyze and compare ROC curves. BMC Bioinformatics. 2011;12:77. doi:10.1186/1471-2105-12-77 20. Simrén J, Ashton NJ, Blennow K, Zetterberg H. An update on blood-based biomarkers for non-Alzheimer neurodegenerative disorders. Nat Rev Neurol. 2023;19(7):411-431. doi:10.1038/s41582-023-00809-0 21. Swardfager W, Lanctôt K, Rothenburg L, Wong A, Cappell J, Herrmann N. A meta-analysis of cytokines in Alzheimer's disease. Biol Psychiatry. 2010;68(10):930-941. doi:10.1016/j.biopsych.2010.06.012 22. Heneka MT, Kummer MP, Stutz A, Delekate A, Schwartz S, Vieira-Saecker A, et al. NLRP3 is activated in Alzheimer's disease and contributes to pathology in APP/PS1 mice. Nature. 2013;493(7434):674-678. doi:10.1038/nature11729 23. Liddelow SA, Guttenplan KA, Clarke LE, Bennett FC, Bohlen CJ, Schirmer L, et al. Neurotoxic reactive astrocytes are induced by activated microglia. Nature. 2017;541(7638):481-487. doi:10.1038/nature21029 24. Keren-Shaul H, Spinrad A, Weiner A, Matcovitch-Natan O, Dvir-Szternfeld R, Ulland TK, et al. A unique microglia type associated with restricting development of Alzheimer's disease. Cell. 2017;169(7):1276-1290. doi:10.1016/j.cell.2017.05.018 25. Zetterberg H, Blennow K. From reaction to prediction: the role of fluid biomarkers in Alzheimer's disease. Cell. 2023;186(20):4339-4355. doi:10.1016/j.cell.2023.08.016 26. Guerreiro R, Wojtas A, Bras J, Carrasquillo M, Rogaeva E, Majounie E, et al. TREM2 variants in Alzheimer's disease. N Engl J Med. 2013;368(2):117-127. doi:10.1056/NEJMoa1211851 27. Chandra SR, Issac TG, Bharath S. Epidemiology of dementia in India: challenges and perspectives. J Geriatr Psychiatry Neurol. 2018;31(6):314-323. doi:10.1177/0891988718807965 28. Dementia India Report 2010. Alzheimer's and Related Disorders Society of India (ARDSI). Cochin: ARDSI; 2010.