Ratios of CSF Proteins Reflect Cognitive Function in ALS

Ratios of CSF Proteins Reflect Cognitive Function in ALS | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Ratios of CSF Proteins Reflect Cognitive Function in ALS Linn Öijerstedt, Sára Mravinacová, Jennie Olofsson, Louisa Azizi, and 10 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-7913800/v1 This work is licensed under a CC BY 4.0 License Status: Published Journal Publication published 31 Jan, 2026 Read the published version in Alzheimer's Research & Therapy → Version 1 posted 9 You are reading this latest preprint version Abstract Background Cognitive impairment is a recognised feature of neurodegenerative diseases, including amyotrophic lateral sclerosis (ALS). Despite advances in understanding cognitive impairment in ALS, no fluid biomarkers reliably predict these changes. Prior research in Alzheimer disease (AD) has demonstrated that CSF protein ratios enhance biomarker accuracy by mitigating inter-individual variability, improving diagnostic precision. Specifically, studies in AD have identified protein pairs reflecting key pathological processes, including synaptic dysfunction. Methods Building on findings from the AD field, we analysed 47 CSF proteins, suggested to be associated to neurodegeneration, in 66 patients with ALS and explored protein ratios to evaluate their utility in detecting cognitive impairment, hypothesising shared mechanisms between neurodegenerative diseases. Elastic net regression identified the most predictive protein pairs associated with cognitive impairment, assessed with the Edinburgh Cognitive and Behavioural ALS Screen (ECAS). Results We identified seven single proteins and eight protein pairs associated with cognitive impairment in ALS. The selected protein pairs showed stronger associations with ECAS scores compared to the individual proteins, indicating an enhanced ability to capture cognitive changes. Several of the proteins in the most predictive pairs have previously been implicated to associate to cognitive impairment in AD. Conclusion Our findings indicate that protein ratios outperform single-protein analyses in detecting associations with cognitive impairment, aligning with advancements in AD research. By extending the concept of CSF protein ratios from AD to ALS, this study highlights shared pathological mechanisms and suggests that similar proteins are linked to cognitive dysfunction in both diseases. amyotrophic lateral sclerosis cognitive impairment ECAS proteomics ratios Figures Figure 1 Figure 2 Figure 3 BACKGROUND Cognitive dysfunction is a recognised feature of numerous neurodegenerative diseases, not only dementia disorders. In amyotrophic lateral sclerosis (ALS), cognitive impairment spans a spectrum from subtle executive dysfunction to full-blown overlap with frontotemporal dementia (ALS-FTD) ( 1 , 2 ). Despite advances in understanding ALS-related cognitive impairment, efforts to translate this knowledge into clinical tools have lagged behind. In particular, biomarker research in cerebrospinal fluid (CSF) has increasingly focused on protein levels and their dynamics, with promising findings from studies of neurodegenerative diseases like Alzheimer disease (AD) ( 3 , 4 ). However, currently there are no fluid biomarkers that predict nor identify cognitive impairment in ALS. Filling this gap is critical as cognitive impairment in ALS has significant implications for disease management, affecting patients' ability to make informed decisions regarding care, treatment, and participation in clinical trials. Biomarkers that detect early cognitive dysfunction could also enable more targeted and timely interventions, potentially improving quality of life. Moreover, cognitive impairment is linked to shorter survival in ALS ( 5 ), highlighting the prognostic value of identifying at-risk individuals early in their disease course. Together, these factors underscore the need for fluid biomarkers that not only detect neurodegeneration but also track its impact on cognitive function. Emerging evidence suggests that interindividual variability in CSF protein levels can mask disease-related changes, complicating the interpretation of biomarker data ( 6 ). By adjusting for this variability, protein ratios have been shown to enhance diagnostic and prognostic accuracy in AD ( 7 , 8 ). In addition to their diagnostic utility, CSF biomarkers can provide valuable insights into disease mechanisms. For example, alterations in proteins such as neurofilaments, which are consistently elevated in ALS, reflects axonal damage while proteins like chitinases may capture central nervous system inflammation relevant to ALS pathophysiology ( 9 , 10 ). By combining proteins associated with different disease mechanisms, a synergistic or additive effect could enhance their potential as a biomarker. Building on the prior findings in AD, here we extend the concept of protein ratios to ALS, suggesting a shared utility of protein ratios as biomarkers for neurodegenerative cognitive dysfunction. We investigated a panel of CSF proteins in a well-characterized cohort of patients with ALS. Specifically, we focused on assessing the ability of these proteins to detect cognitive impairment, both individually and through protein ratios. methods Cohort and data collection Participants were recruited from the ALSrisc study, an ongoing longitudinal cohort study at Karolinska University Hospital in Stockholm, Sweden ( 11 ). We included patients with a diagnosis of ALS, an available CSF sample and assessment of cognitive function within a year from sample collection. The following clinical and phenotypic data were collected: age at sampling, sex, site of onset and C9orf72 repeat expansion. Disease progression was monitored using the ALS Functional Rating Scale–Revised (ALSFRS-R), a standardised instrument widely used to quantify functional decline in ALS ( 12 ). Cognitive function was evaluated using the Edinburgh Cognitive and Behavioural ALS Screen (ECAS), a validated screening tool specifically developed for ALS populations ( 13 ). ECAS administration takes approximately 15–20 minutes and assesses both ALS-specific (language, verbal fluency, executive function) and ALS-non-specific domains (memory, visuospatial function). The ECAS score collected closest to CSF sampling date was included. The median time between CSF sampling and ECAS was 7 weeks (range 0–35 weeks). No subject had ECAS assessed prior to sample collection. An ECAS total score below 108 will herein be referred to as “cognitive impairment” ( 14 ). Sample collection CSF samples were collected between 2019 and 2021 via lumbar puncture into polypropylene tubes and processed immediately by centrifugation at 963 ×g for 10 minutes at room temperature. The resulting supernatant was aliquoted into cryotubes and stored at − 80°C until analysis. Prior to protein analysis, the samples were stratified into 96-well PCR plates using a constrained randomisation approach based on diagnosis, age, and sex. Protein analysis Proteins (n = 47) were selected based on prior in-house neuroproteomic studies, both published ( 15 – 22 ) and unpublished. Corresponding polyclonal rabbit antibodies were sourced from the Human Protein Atlas ( www.proteinatlas.org ), except for progranulin (GRN) (AF3156-SP, R&D Systems) and apolipoprotein E4 (APOE4) (M067-3, MBL Life Science) antibodies ( Supplementary Table 1 ). To create the bead array, each antibody was individually immobilised onto uniquely color-coded, carboxylated magnetic beads (MagPlex, Luminex Corp.) as previously described ( 23 ). CSF was processed according to a previously described protocol ( 23 ). In brief, the samples were diluted 1:2 and labelled with biotin (NHS-PEG4-biotin, A39259, ThermoFisher Scientific) followed by heat-treatment at 56°C for 30 minutes. Incubation with beads was performed overnight at room temperature in 384-well plates and bound proteins detected using a streptavidin-conjugated fluorophore. Fluorescence signals, corresponding to median fluorescence intensity per bead ID and per sample, were acquired using a FlexMap 3D instrument (Luminex Corp.), providing relative quantification of protein levels. Statistical analysis All data pre-processing, analysis and illustrations were performed in R Studio version 4.3.1. Protein level data was adjusted for delayed instrument readout using robust linear regression as described previously ( 3 ). All protein levels were log2 transformed and scaled to zero mean and unit variance prior to statistical analysis. Assumptions for statistical models were assessed visually by residual plots (independence and equal variance) and normal probability plots (normality). The Wilcoxon rank-sum test was used to compare differences between sexes and the Kruskal–Wallis test was applied to assess differences across groups for site of symptom onset and genetic status. Correlation and hierarchical clustering Spearman’s rank correlation coefficients (ρ) were computed to assess the co-variation of CSF protein profiles. Protein clustering based on correlation was performed using hierarchical clustering using the Ward’s minimum variance method. Elastic net Elastic net regression is a statistical modelling technique that addresses the challenges of analyzing datasets with a large number of interrelated variables ( 24 ). It combines two forms of regularization, those used in ridge regression and Least Absolute Shrinkage and Selection Operator (LASSO), to improve model performance and interpretability. Ridge regression reduces overfitting by shrinking the size of all regression coefficients, but it does not eliminate any variables. In contrast, LASSO can shrink some coefficients to exactly zero, thus performing both shrinkage and variable selection, i.e. effectively selecting a smaller subset of variables that contribute most to the outcome. Elastic net introduces a mixing parameter, α (alpha), which controls the balance between ridge (α = 0) and LASSO (α = 1) penalties. By adjusting α, elastic net provides a flexible framework that combines the strengths of both methods. This makes elastic net especially effective when working with high-dimensional data and/or highly correlated predictors, as was the case in our study. In addition to the mixing parameter, elastic net also includes a second parameter, λ (lambda), which controls the overall strength of the penalty applied to the model. Larger values of λ impose stronger penalties, leading to greater shrinkage of the regression coefficients producing simpler, more constrained models ( Supplementary Fig. 1 ). Data setup : We partitioned the dataset into 70% for training and 30% for testing. The training data included 48 patients with a median ECAS total score of 109, the test set of 18 patients with a median ECAS total score of 108.5. Tuning procedure : A grid search over α values (step size 0.01) ranging from 0 (Ridge) to 1 (LASSO) was conducted using cross-validation to identify the optimal mixing parameter. The best α was selected based on the minimum cross-validation error. Subsequently, with the optimal α, a ten-fold cross-validation was performed to select the optimal regularisation parameter λ (λ min ​) in order to obtain the ideal trade-off between bias and variance. A cross-validation plot was used to support the tuning process ( Supplementary Fig. 2 ). Model fitting and evaluation : The elastic net model was fit on the training dataset using the optimal α and λ values. The root mean squared error (RMSE) for the training set was computed to assess model fit and potential overfitting. RMSE is a commonly used metric that quantifies the average magnitude of prediction error, with lower values indicating better model performance. Similarly, RMSE was calculated for the test set to evaluate prediction accuracy. Regression models Linear regression models were constructed to evaluate the association between candidate proteins and cognitive scores (ECAS total score). For each protein, a separate model was fitted, with ECAS total score as the outcome variable and protein levels as predictor with adjustments for age at sampling, sex, site of onset and ALSFRS-R total score. Model performance and generalisability were assessed using 10-fold cross-validation. To assess the ability for candidate proteins to predict cognitive status (cognitive impairment vs no cognitive impairment), logistic regression models were constructed. All models were adjusted for age at sampling, sex, site of onset, and ALSFRS-R total score. Model performance was evaluated by receiver operating characteristic (ROC) curve and area under the curve (AUC). Results The final cohort included 66 patients (mean age = 66 ± 11.4 years, 39% female) of whom 44% had an ECAS total score below 108 (Table 1 ). The ECAS total score was not correlated to age or disease duration. The majority of patients had a spinal onset (68%), and nine individuals (14%) carried a C9orf72 repeat expansion. Table 1 Cohort. Characteristics of the patients included in the study. N, number; SD, standard deviation; C9orf72 , C9 open reading frame 72; ECAS, Edinburgh cognitive and behavioural ALS screen; IQR, inter quartile range; ALSFRS-R, ALS functional rating scale–revised. Variable N = 66 Age at sampling, mean (SD) 66 (11.4) Sex, N females (%) 26 (39) Site of onset, N (%) Spinal 45 (68) Bulbar 19 ( 29 ) Other a 2 ( 3 ) C9orf72 repeat expansion, N (%) 9 ( 14 ) ECAS, median (IQR) 109 (21.5) ECAS < 108, N (%) 29 (44) ALSFRS-R, median (IQR) 41.5 (8.25) a respiratory Neurofilament levels do not correlate to known markers of cognitive function Pairwise correlation analysis of the 47 CSF proteins revealed two main protein clusters, one with strongly correlating proteins (n = 26, median ρ 0.81, IQR 0.14) including proteins commonly regarded as markers for dementia and cognitive function such as neurogranin (NRGN), beta-synuclein (SNCB) and neuromodulin (GAP43) (Fig. 1 ). The other protein clusters (n = 18 and n = 3) showed generally lower co-variation between proteins (median ρ 0.54, IQR 0.23, and median ρ 0.34, IQR 0.14, respectively). Neurofilament medium (NEFM), being in the small cluster of three proteins, did not exhibit strong correlations with the neuronal proteins in the first cluster (median ρ 0.20, IQR 0.18). Furthermore, chitinase 1 (CHIT1), included in the cluster together with NEFM, displayed a unique correlation profile (median ρ -0.01, range − 0.08 − 0.07, IQR 0.04). CSF proteins are associated with ECAS total score We applied elastic net as a variable selection method to identify proteins with a relevant association to ECAS total score. In our final elastic net model, the optimal hyperparameters were found to be α = 0.54 and λ = 3.63. This indicated that the final model used a nearly balanced mix of LASSO and ridge regression penalties, reflecting both the need for variable selection and stability in the presence of potential multicollinearity. The model identified 7 proteins as the most predictive of ECAS total score: NEFM, neuronal pentraxin 2 (NPTX2), GAP43, insulin like growth factor binding protein 4 (IGFBP4), insulin like growth factor binding protein 7 (IGFBP7), osteopontin (SPP1) and cadherin 8 (CDH8) (Table 2 ). The model produced an RMSE of 13.76 on the training set and 12.03 on the test set. The slightly lower RMSE on the test set suggests that the model not only fit the training data well but also generalized effectively to unseen data, with no indications of overfitting. Table 2 Candidate protein and protein pairs. Results from the elastic net and linear regression analyses. Linear regression Selected in elastic net of single proteins Selected in elastic net of protein pairs β coefficient CV R 2 Single protein IGFBP7 4.05 (0.11–7.98) 0.27 Yes No NPTX2 2.83 (-1.09–6.76) 0.29 Yes Yes CDH8 2.57 (-1.37–6.50) 0.25 Yes Yes PTPRN2 2.26 (-1.82–6.34) 0.24 No Yes IGF2 2.07 (-2.04–6.19) 0.13 No Yes CHL1 1.62 (-2.52–5.76) 0.26 No Yes CADM2 1.07 (-3.03–5.16) 0.34 No Yes BASP1 0.67 (-3.50–4.84) 0.18 No Yes IGFBP4 -0.06 (-4.37–4.24) 0.21 Yes Yes GAP43 -1.25 (-5.41–2.91) 0.23 Yes Yes NEFM -2.58 (-6.92–1.77) 0.22 Yes No SPP1 -3.45 (-7.34–0.45) 0.47 Yes Yes Protein pair PTPRN2/GAP43 7.48 (3.66–11.29) 0.34 CDH8/GAP43 6.77 (2.68–10.85) 0.32 CHL1/GAP43 6.49 (2.38–10.61) 0.30 NPTX2/SPP1 5.24 (1.49–8.99) 0.26 PTPRN2/BASP1 4.94 (0.66–9.22) 0.19 CADM2/GAP43 3.5 (-0.63–7.63) 0.22 NPTX2/IGFBP4 3.05 (-1.27–7.37) 0.28 IGF2/IGFBP4 2.92 (-1.24–7.07) 0.27 CI = confidence interval, CV R 2 = 10-fold cross-validated R 2 Protein ratios are superior to single proteins for predicting ECAS score As protein pairs have been shown to provide stronger associations with cognitive function compared to single proteins in other neurodegenerative disorders, we also assessed protein ratios for their ability to detect cognitive impairment in ALS. All 47 proteins were combined into pairs (n = 2162). Again, we used elastic net to find the pairs most predictive of ECAS total score. In the elastic net model of ratios, the optimal hyperparameters were found to be α = 0.37 and λ = 10.38. Here, 8 protein pairs were identified as the most predictive of ECAS total score (Table 2 , Fig. 2 A). The model produced an RMSE of 12.93 on the training set and 12.11 on the test set. Five of the proteins selected in the ratio elastic net model were also selected in the single protein model (NPTX2, GAP43, IGFBP4, SPP1 and CDH8). Interestingly, five additional proteins were found relevant as part of a pair, namely insulin like growth factor 2 (IGF2), cell adhesion molecule L1 like (CHL1), protein tyrosine phosphatase receptor type N2 (PTPRN2), cell adhesion molecule 2 (CADM2) and brain abundant membrane attached signal protein 1 (BASP1). As previously shown, neither of these had a strong association with ECAS total score alone. To further evaluate the association to ECAS total score and compare the performance of single proteins and protein pairs, linear regression models were created for each candidate as the predictor (Table 2 , Fig. 2 A). We assessed the predictive performance of each model using 10-fold cross-validated R ² (CV R ²). Comparing the regression models revealed that higher CV R ² values were generally observed in protein pair models (median 0.27) in contrast with those of the single protein models (median 0.24) ( Supplementary Fig. 3 ). In particular, protein pair ratios involving GAP43 exhibited larger b coefficient magnitudes, the majority of the confidence intervals excluded zero, and higher cross-validated R ² values relative to single protein metrics (Table 2 ). The PTPRN2/GAP43 ratio showed the strongest association with an estimated b coefficient of 7.48 (95% CI: 3.66–11.29) with a 10-fold cross-validated R ² of 0.34. Other prominent protein pairs included CDH8/GAP43, CHL1/GAP43 and NPTX2/SPP1. The PTPRN2/GAP43 b coefficient substantially exceeded that of PTPRN2 and GAP43 alone (2.26, 95% CI: -1.82–6.34 and − 1.25 (95% CI: -5.41–2.91, respectively), suggesting that the ratio of these highly correlating proteins is more informative than the levels alone (Fig. 2 B). This additive effect of PTPRN2 in combination with GAP43 was further explored using ROC analysis (Fig. 2 C). The performance was considerably better for PTPRN2/GAP43, with an area under the curve of 0.89 (95% CI: 0.80–0.97), compared to PTPRN2 and GAP43 alone (AUC 0.76 95% CI: 0.64–0.89 and 0.78 95% CI: 0.67–0.90, respectively). These results suggest that protein ratios potentially offer a more robust prediction of ECAS total score compared to single protein measures. The consistency of key marker selection in both elastic net and linear regression analyses further reinforces the potential of these ratios as biomarkers for cognitive impairment in ALS. We next evaluated the association of the most promising protein ratios with ECAS sub scores. The strongest associations with executive function, verbal fluency, and memory for the PTPRN2/GAP43 ratio ( Supplementary Fig. 4 ). Similar results were found for the other candidate pairs suggesting that these ratios are more likely markers of a general cognitive dysfunction and not specific to ALS frontotemporal involvement. The trajectories of CSF PTPRN2/GAP43 are different in males and females with cognitive impairment The PTPRN2/GAP43 ratio was not associated with ALSFRS-R score, nor were there any statistically significant differences in the PTPRN2/GAP43 CSF ratio between bulbar versus spinal onset or C9orf72 mutation carriers versus non-carriers ( Supplementary Fig. 5A, Supplementary Fig. 5B ). In addition, we did not find any overall differences in PTPRN2/GAP43 ratio between the sexes (p = 0.16). However, among the patients with cognitive impairment, lower ratios were found in males compared to females even though the distribution of cognitive impairment and age was similar between sexes (p = 0.002) (Fig. 3 A). Indeed, when exploring linear regression with PTPRN2/GAP43 as the outcome and including an interaction term between ECAS total score and sex, we found that the slope of PTPRN2/GAP43 ratio over ECAS total score was steeper for males than females (0.04 and 0.01 respectively, on a log2 scale) (Fig. 3 B). Discussion In this study we analysed a panel of CSF proteins and investigated their association to cognitive function in patients with ALS. No single protein strongly predicted cognitive impairment but combining them into ratios markedly improved predictive power. This resonates with findings in AD, where ratios like Aβ42/40 or Aβ42/tau better reflect disease processes than individual proteins ( 25 – 27 ). Such ratios might capture interactions between pathways, for example synaptic repair (GAP43) and metabolic alterations (PTPRN2). A possible explanation for the enhanced performance of protein pairs in relation to single proteins is that one protein in the ratio act as a reference, thereby accounting for non-disease related individual variation. Prior studies have suggested that such variability can be influenced by factors like sex, age, and ventricular volume ( 8 , 17 ). Variable selection methods like elastic net can be used to highlight proteins with important contributions to a multivariable model, but their performance may not always translate into regression models. In our case, this means that while these proteins are informative in a multivariate setting, they may not independently explain cognitive variance when clinical variables are accounted for. This could explain why several of the identified single proteins were selected in the elastic net, even though they did not demonstrate strong predictive performance in the subsequent adjusted linear models. In this study, the most promising candidate biomarker associated with cognitive impairment was the PTPRN2/GAP43 ratio, the same protein pair previously found to be among the best classifiers of cognitive decline in patients with dementia ( 3 , 6 ). While levels of GAP43 have been associated with AD in previous studies (including 15,16,19,28–30), the function and role of PTPRN2 in neurodegenerative disease pathology is poorly understood. In a clinical setting, complex multivariable models with many covariates and interaction terms can be difficult to apply or interpret, limiting their practical use. In contrast, simpler measures, like ratios of two proteins, are more feasible for clinical translation, as they balance complexity with interpretability and robustness. Our findings that protein ratios outperform single markers are in line with recent publications and these studies together suggest that the use of protein pairs might be a suitable middle way for practical biomarker implementation ( 4 , 6 ). Our cluster analysis revealed a strong co-variation between many of the CSF proteins relevant for synaptic health and function. The tight intercorrelations within the largest cluster suggest that these proteins may be co-regulated or share common biological pathways involved in neurodegeneration. Neurofilament medium, a marker of axonal integrity closely related to the well-established neurofilament light chain ( 10 ), did not strongly correlate with these cognitive markers and may be involved in only partially reflect overlapping pathological mechanisms. The clustering pattern of NEFM instead suggests stronger association with neuroinflammation or immune response rather than synaptic function. Together with the finding that protein pairs including NEFM were not selected as key features in the elastic net regression, indicate that cognitive dysfunction in ALS might be more closely driven by synaptic and cortical processes than by general axonal damage. The observed sex differences in CSF PTPRN2/GAP43 levels in relation to cognitive impairment are particularly noteworthy. Specifically, our regression analysis revealed a significant interaction between sex and ECAS total score, with a markedly steeper decline in the PTPRN2/GAP43 ratio among males. Although this ratio does not independently predict ECAS scores when modelled in the reverse direction, the interaction suggests that cognitive status is more strongly associated with biomarker levels in males. This finding not only reinforces the group differences seen in our stratified analyses but also points to a potentially meaningful biological divergence. It raises the possibility of sex-specific dynamics in biomarker expression or regulation in the context of ALS-related cognitive impairment. Further investigation is needed to determine whether these differences reflect distinct underlying mechanisms or differential susceptibility to cognitive decline between sexes. A strength of this study lies in its rigorous application of statistical methods and cross-validation, ensuring that our identified associations are robust. However, one limitation is the small sample size, especially for sex-stratified analyses, which may limit the generalisability of the results. In addition, while ECAS is a validated screening tool for cognitive impairment in ALS, our cut-offs were not adjusted for potential confounders such as age and education, which may influence cognitive performance and the classification of impairment ( 31 ). While the strong performance of CSF PTPRN2/GAP43 suggests biological relevance, future studies should aim for larger cohorts and more comprehensive cognitive assessments to confirm these findings. CONCLUSIONS Altogether, our data support the hypothesis that protein ratios, particularly PTPRN2/GAP43, may capture aspects of cognitive impairment in ALS that go undetected by single markers. They also highlight the importance of considering sex-specific interactions when interpreting biomarker data. If validated in larger cohorts, these findings could aid the development of diagnostic methods in ALS that are sensitive to cognitive impairment and argue for prioritising combinatorial or ratio-based measures over single protein readouts in future biomarker panels. Such panels could also have broader implications, as similar synaptic processes are implicated in other neurodegenerative diseases, suggesting shared targets for therapeutic intervention. Abbreviations AD alzheimer disease ALS amyotrophic lateral sclerosis ALSFRS-R ALS functional rating scale–revised AUC area under the curve BASP1 brain abundant membrane attached signal protein 1 C9orf72 C9 open reading frame 72 CADM2 cell adhesion molecule 2 CDH8 cadherin 8 CHIT1 chitinase 1 CHL1 cell adhesion molecule l1 like CI confidence interval CSF cerebraospinal fluid CV cross validation ECAS Edinburgh cognitive and behavioural ALS screen FTD frontotemporal dementia GAP43 neuromodulin IGF2 insulin like growth factor 2 IGFBP4 insulin like growth factor binding protein 4 IGFBP7 insulin like growth factor binding protein 7 IQR inter quartile range LASSO least absolute shrinkage and selection operator NEFM neurofilament medium NPTX2 neuronal pentraxin 2 NRGN neurogranin PTPRN2 protein tyrosine phosphatase receptor type n2 RMSE root mean squared error ROC receiver operating characteristics SNCB beta-synuclein SPP1 osteopontin Declarations Ethics approval and consent to participate The study was conducted in accordance with the Declaration of Helsinki. An informed consent to participate in research was obtained from each participant. The study was reviewed and approved by the Swedish Ethical Review Authority (diary numbers 2017/1895-31 and 2018/1605-31). Consent for publication Not applicable Availability of data and materials The dataset used and analysed during the current study are available from the corresponding author upon reasonable request. Competing interests The authors declare no competing interest relevant to the work conducted in the presented study. Funding This study was funded by Demensfonden, Börje Salming ALS foundation, Svenska Läkaresällskapet Björklunds fond, ALS-fonden, and Stockholm Region. Authors' contributions LÖ, SM, AM, and CI developed the study design. LÖ, AJ, UK, and CI contributed to patient data acquisition and sample collection. NdV, JO, and SB performed the experimental work. LÖ and SM did the data quality control and statistical analyses. LÖ interpreted the data with support from SM, AM and CI. CI, AM, and PN supervised the project. LÖ drafted the manuscript with input from SM and AM. All authors critically reviewed and revised the manuscript. All authors read and approved the final manuscript. Acknowledgements We would like to thank Jenny Hellqvist for all her work on data collection. We would also like to thank the entire staff of the Human Protein Atlas for their efforts. Most importantly, we would like to thank the patients who contributed to this study. References Strong MJ, Abrahams S, Goldstein LH, Woolley S, Mclaughlin P, Snowden J, et al. Amyotrophic lateral sclerosis - frontotemporal spectrum disorder (ALS-FTSD): Revised diagnostic criteria. Amyotroph Lateral Scler Frontotemporal Degener. 2017 Apr 3;18(3–4):153–74. doi:10.1080/21678421.2016.1267768 Abrahams S, Leigh PN, Goldstein LH. Cognitive change in ALS: A prospective study. Neurology. 2005 Apr 12;64(7):1222–6. doi:10.1212/01.WNL.0000156519.41681.27 Mravinacová S, Alanko V, Bergström S, Bridel C, Pijnenburg Y, Hagman G, et al. CSF protein ratios with enhanced potential to reflect Alzheimer’s disease pathology and neurodegeneration. Mol Neurodegener. 2024 Dec 1;19(1). doi:10.1186/s13024-024-00705-z Karlsson L, Vogel J, Arvidsson I, Åström K, Janelidze S, Blennow K, et al. Cerebrospinal fluid reference proteins increase accuracy and interpretability of biomarkers for brain diseases. Nat Commun. 2024 Dec 1;15(1). doi:10.1038/s41467-024-47971-5 Öijerstedt L, Foucher J, Lovik A, Yazdani S, Juto A, Kläppe U, et al. Repeated cognitive assessments show stable function over time in patients with ALS. J Neurol [Internet]. 2024 Aug 1;271(8):5267–74. doi:10.1007/s00415-024-12479-x Mravinacová S, Bergström S, Olofsson J, de San José NG, Anderl-Straub S, Diehl-Schmid J, et al. Addressing inter individual variability in CSF levels of brain derived proteins across neurodegenerative diseases. Sci Rep [Internet]. 2025 Jan 3;15(1):668. doi:10.1038/s41598-024-83281-y Oh HS-H, Urey DY, Karlsson L, Zhu Z, Shen Y, Farinas A, et al. A cerebrospinal fluid synaptic protein biomarker for prediction of cognitive resilience versus decline in Alzheimer’s disease. Nat Med [Internet]. 2025 Mar 31;31(5). doi:10.1038/S41591-025-03565-2 Karlsson L, Vogel J, Arvidsson I, Åström K, Strandberg O, Seidlitz J, et al. Machine learning prediction of tau-PET in Alzheimer’s disease using plasma, MRI, and clinical data. Alzheimers Dement. 2025 Feb 1;21(2):e14600. doi:10.1002/alz.14600 Kläppe U, Sennfält S, Lovik A, Finn A, Bofaisal U, Zetterberg H, et al. Neurodegenerative biomarkers outperform neuroinflammatory biomarkers in amyotrophic lateral sclerosis. Amyotroph Lateral Scler Frontotemporal Degener [Internet]. 2024;25(1–2):150–61. doi:10.1080/21678421.2023.2263874 Olofsson J, Bergström S, Mravinacová S, Kläppe U, Öijerstedt L, Zetterberg H, et al. Cerebrospinal fluid levels of NfM in relation to NfL and pNfH as prognostic markers in amyotrophic lateral sclerosis. Amyotroph Lateral Scler Frontotemporal Degener [Internet]. 2025 Feb;26(1–2):113–23. doi:10.1080/21678421.2024.2428930 Chourpiliadis C, Seitz C, Lovik A, Joyce EE, Pan L, Hu Y, et al. Lifestyle and medical conditions in relation to ALS risk and progression-an introduction to the Swedish ALSrisc Study. J Neurol [Internet]. 2024 Aug 1;271(8):5447–59. doi:10.1007/S00415-024-12496-W Cedarbaum JM, Stambler N, Malta E, Fuller C, Hilt D, Thurmond B, et al. The ALSFRS-R: A revised ALS functional rating scale that incorporates assessments of respiratory function. J Neurol Sci [Internet]. 1999 Oct 31;169(1–2):13–21. doi:10.1016/S0022-510X(99)00210-5 Niven E, Newton J, Foley J, Colville S, Swingler R, Chandran S, et al. Validation of the Edinburgh Cognitive and Behavioural Amyotrophic Lateral Sclerosis Screen (ECAS): A cognitive tool for motor disorders. http://dx.doi.org/103109/2167842120151030430 [Internet]. 2015 Jun 1;16(3–4):172–9. doi:10.3109/21678421.2015.1030430 Foucher J, Winroth I, Lovik A, Sennfält S, Pereira JB, Fang F, et al. Validity and reliability measures of the Swedish Karolinska version of the Edinburgh Cognitive and Behavioral ALS Screen (SK-ECAS). Amyotroph Lateral Scler Frontotemporal Degener [Internet]. 2023;24(7–8):713–8. doi:10.1080/21678421.2023.2239857 Bergström S, Remnestål J, Yousef J, Olofsson J, Markaki I, Carvalho S, et al. Multi-cohort profiling reveals elevated CSF levels of brain-enriched proteins in Alzheimer’s disease. Ann Clin Transl Neurol [Internet]. 2021 Jul 1;8(7):1456–70. doi:10.1002/ACN3.51402 Remnestål J, Bergström S, Olofsson J, Sjöstedt E, Uhlén M, Blennow K, et al. Association of CSF proteins with tau and amyloid β levels in asymptomatic 70-year-olds. Alzheimers Res Ther [Internet]. 2021 Dec 1;13(1). doi:10.1186/S13195-021-00789-5 Bergström S, Mravinacová S, Lindberg O, Zettergren A, Westman E, Wahlund L-O, et al. CSF levels of brain-derived proteins correlate with brain ventricular volume in cognitively healthy 70-year-olds. Clin Proteomics [Internet]. 2024 Dec 12;21(1):65. doi:10.1186/s12014-024-09517-1 Remnestål J, Öijerstedt L, Ullgren A, Olofsson J, Bergström S, Kultima K, et al. Altered levels of CSF proteins in patients with FTD, presymptomatic mutation carriers and non-carriers. Transl Neurodegener. 2020 Jun 23;9(1). doi:10.1186/S40035-020-00198-Y Remnestål J, Just D, Mitsios N, Fredolini C, Mulder J, Schwenk JM, et al. CSF profiling of the human brain enriched proteome reveals associations of neuromodulin and neurogranin to Alzheimer’s disease. Proteomics Clin Appl [Internet]. 2016 Dec 1;10(12):1242–53. doi:10.1002/PRCA.201500150 Ullgren A, Öijerstedt L, Olofsson J, Bergström S, Remnestål J, van Swieten JC, et al. Altered plasma protein profiles in genetic FTD – a GENFI study. Mol Neurodegener. 2023 Dec 1;18(1). doi:10.1186/s13024-023-00677-6 Bergström S, Öijerstedt L, Remnestål J, Olofsson J, Ullgren A, Seelaar H, et al. A panel of CSF proteins separates genetic frontotemporal dementia from presymptomatic mutation carriers: a GENFI study. Mol Neurodegener. 2021 Dec 1;16(1). doi:10.1186/S13024-021-00499-4 Andersson A, Remnestål J, Nellgård B, Vunk H, Kotol D, Edfors F, et al. Development of parallel reaction monitoring assays for cerebrospinal fluid proteins associated with Alzheimer’s disease. Clin Chim Acta [Internet]. 2019 Jul 1;494:79–93. doi:10.1016/J.CCA.2019.03.243 Pin E, Sjöberg R, Andersson E, Hellström C, Olofsson J, Jernbom Falk A, et al. Array-based profiling of proteins and autoantibody repertoires in CSF. In: Methods in Molecular Biology [Internet]. Methods Mol Biol; 2019. p. 303–18. doi:10.1007/978-1-4939-9706-0_19 Zou H, Hastie T. Regularization and Variable Selection Via the Elastic Net. J R Stat Soc Series B Stat Methodol [Internet]. 2005 Apr 1;67(2):301–20. doi:10.1111/J.1467-9868.2005.00503.X Shoji M, Matsubara E, Kanai M, Watanabe M, Nakamura T, Tomidokoro Y, et al. Combination assay of CSF Tau, Aβ1-40 and Aβ1-42(43) as a biochemical marker of Alzheimer’s disease. J Neurol Sci. 1998 Jun 30;158(2):134–40. doi:10.1016/S0022-510X(98)00122-1 Lewczuk P, Esselmann H, Otto M, Maler JM, Henkel AW, Henkel MK, et al. Neurochemical diagnosis of Alzheimer’s dementia by CSF Aβ42, Aβ42/Aβ40 ratio and total tau. Neurobiol Aging. 2004;25(3):273–81. doi:10.1016/S0197-4580(03)00086-1 Hansson O, Lehmann S, Otto M, Zetterberg H, Lewczuk P. Advantages and disadvantages of the use of the CSF Amyloid β (Aβ) 42/40 ratio in the diagnosis of Alzheimer’s Disease. Alzheimers Res Ther [Internet]. 2019 Apr 22;11(1):1–15. doi:10.1186/S13195-019-0485-0 Franzmeier N, Dehsarvi A, Steward A, Biel D, Dewenter A, Roemer SN, et al. Elevated CSF GAP-43 is associated with accelerated tau accumulation and spread in Alzheimer’s disease. Nat Commun [Internet]. 2024 Dec 1;15(1). doi:10.1038/S41467-023-44374-W Öhrfelt A, Benedet AL, Ashton NJ, Kvartsberg H, Vandijck M, Weiner MW, et al. Association of CSF GAP-43 With the Rate of Cognitive Decline and Progression to Dementia in Amyloid-Positive Individuals. Neurology [Internet]. 2023 Jan 17;100(3):E275–85. doi:10.1212/WNL.0000000000201417 Shi D, Ye C, Li A, Sun P, Lan G, Zhang L, et al. Presynaptic loss and axonal degeneration synergistically correlate with longitudinal neurodegeneration and cognitive decline. Alzheimers Dement [Internet]. 2025 Jun 11;21(6). doi:10.1002/ALZ.70080 Finsel J, Winroth I, Ciećwierska K, Helczyk O, Stenberg EA, Häggström AC, et al. Determining impairment in the Swedish, Polish and German ECAS: the importance of adjusting for age and education. Amyotroph Lateral Scler Frontotemporal Degener [Internet]. 2023;24(5–6):475–84. doi:10.1080/21678421.2023.2192248 Additional Declarations No competing interests reported. Supplementary Files Supplementary.pdf Cite Share Download PDF Status: Published Journal Publication published 31 Jan, 2026 Read the published version in Alzheimer's Research & Therapy → Version 1 posted Editorial decision: Revision requested 17 Dec, 2025 Reviews received at journal 17 Dec, 2025 Reviews received at journal 09 Dec, 2025 Reviewers agreed at journal 24 Nov, 2025 Reviewers agreed at journal 18 Nov, 2025 Reviewers invited by journal 03 Nov, 2025 Editor assigned by journal 27 Oct, 2025 Submission checks completed at journal 27 Oct, 2025 First submitted to journal 21 Oct, 2025 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-7913800","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":543422281,"identity":"b5a6e1ca-18b5-4a22-bb92-27c2801922f1","order_by":0,"name":"Linn Öijerstedt","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA50lEQVRIiWNgGAWjYBACPmYwJQGhGNgY5GBMnIANSQtjAwObgTFhLUhssJbEGYQcxsbOe+wBQ4UFO7/08ecPPpT9SZ/ZzsD84QNeh/GlGzCckWCW7MsxbJxxziB3NjMDmyQ+q9iYecwkGNskmA3O8DA287YZ5M5jBgkSo8X+DPvD5r9tBulyzAzMn/8QZQsPg2EzY5tBgjQwxKTxeh+kJQHoF4kzPIYze84ZG84EapTswaOFn/+MmcSHirpk/h72Bx9+lMnJS5w/fPjDD3zWgEACA0MyEhcYP8QAO6JUjYJRMApGwcgEAEAbOTsiCq2yAAAAAElFTkSuQmCC","orcid":"","institution":"Karolinska Institutet","correspondingAuthor":true,"prefix":"","firstName":"Linn","middleName":"","lastName":"Öijerstedt","suffix":""},{"id":543422282,"identity":"8267f938-9b15-488d-8182-fc460dca868b","order_by":1,"name":"Sára Mravinacová","email":"","orcid":"","institution":"KTH Royal Institute of Technology","correspondingAuthor":false,"prefix":"","firstName":"Sára","middleName":"","lastName":"Mravinacová","suffix":""},{"id":543422283,"identity":"4d3fe40e-cad8-4397-96ab-c91bae01bc6d","order_by":2,"name":"Jennie Olofsson","email":"","orcid":"","institution":"KTH Royal Institute of Technology","correspondingAuthor":false,"prefix":"","firstName":"Jennie","middleName":"","lastName":"Olofsson","suffix":""},{"id":543422284,"identity":"5668f667-b300-4248-a5f2-2947d725f281","order_by":3,"name":"Louisa Azizi","email":"","orcid":"","institution":"Karolinska Institutet","correspondingAuthor":false,"prefix":"","firstName":"Louisa","middleName":"","lastName":"Azizi","suffix":""},{"id":543422285,"identity":"f81fae49-6de6-436b-831f-918b73c9f5fe","order_by":4,"name":"Sofia Bergström","email":"","orcid":"","institution":"KTH Royal Institute of Technology","correspondingAuthor":false,"prefix":"","firstName":"Sofia","middleName":"","lastName":"Bergström","suffix":""},{"id":543422287,"identity":"f6840bc8-48a6-4ed4-8e7c-efb2672489d6","order_by":5,"name":"Solmaz Yazdani","email":"","orcid":"","institution":"Karolinska Institutet","correspondingAuthor":false,"prefix":"","firstName":"Solmaz","middleName":"","lastName":"Yazdani","suffix":""},{"id":543422288,"identity":"4e6285f0-700d-4120-817b-21b44d25db82","order_by":6,"name":"Nina De Vita","email":"","orcid":"","institution":"KTH Royal Institute of Technology","correspondingAuthor":false,"prefix":"","firstName":"Nina","middleName":"","lastName":"De Vita","suffix":""},{"id":543422289,"identity":"6d81b7e7-173f-44a4-8ecb-04cdfe18e685","order_by":7,"name":"Inci S. Aksoylu","email":"","orcid":"","institution":"KTH Royal Institute of Technology","correspondingAuthor":false,"prefix":"","firstName":"Inci","middleName":"S.","lastName":"Aksoylu","suffix":""},{"id":543422290,"identity":"2d19c936-f5cb-4dac-b8fd-fc8e9a4887ec","order_by":8,"name":"Juliette Foucher","email":"","orcid":"","institution":"Karolinska Institutet","correspondingAuthor":false,"prefix":"","firstName":"Juliette","middleName":"","lastName":"Foucher","suffix":""},{"id":543422293,"identity":"a462192e-f850-48e5-bb92-b71587c272d7","order_by":9,"name":"Alexander Juto","email":"","orcid":"","institution":"Karolinska Institutet","correspondingAuthor":false,"prefix":"","firstName":"Alexander","middleName":"","lastName":"Juto","suffix":""},{"id":543422295,"identity":"7e0ee190-0eb1-499f-88ef-3f8077b816b6","order_by":10,"name":"Ulf Kläppe","email":"","orcid":"","institution":"Karolinska Institutet","correspondingAuthor":false,"prefix":"","firstName":"Ulf","middleName":"","lastName":"Kläppe","suffix":""},{"id":543422296,"identity":"e0d61d70-33e9-4223-af9b-b0453e1c1530","order_by":11,"name":"Peter Nilsson","email":"","orcid":"","institution":"KTH Royal Institute of Technology","correspondingAuthor":false,"prefix":"","firstName":"Peter","middleName":"","lastName":"Nilsson","suffix":""},{"id":543422297,"identity":"030b3a6a-3f27-4901-acc9-cb790c380a03","order_by":12,"name":"Anna Månberg","email":"","orcid":"","institution":"KTH Royal Institute of Technology","correspondingAuthor":false,"prefix":"","firstName":"Anna","middleName":"","lastName":"Månberg","suffix":""},{"id":543422298,"identity":"891cf28d-80ce-45ee-9ca7-1d7edc976d83","order_by":13,"name":"Caroline Ingre","email":"","orcid":"","institution":"Karolinska Institutet","correspondingAuthor":false,"prefix":"","firstName":"Caroline","middleName":"","lastName":"Ingre","suffix":""}],"badges":[],"createdAt":"2025-10-21 11:08:23","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-7913800/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-7913800/v1","draftVersion":[],"editorialEvents":[{"content":"https://doi.org/10.1186/s13195-026-01976-y","type":"published","date":"2026-01-31T15:58:29+00:00"}],"editorialNote":"","failedWorkflow":false,"files":[{"id":96240463,"identity":"2dc7a5dd-7c3e-4886-9070-5774e864db9a","added_by":"auto","created_at":"2025-11-19 07:08:56","extension":"docx","order_by":0,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":1290983,"visible":true,"origin":"","legend":"","description":"","filename":"Mainmanuscript.docx","url":"https://assets-eu.researchsquare.com/files/rs-7913800/v1/550f7d2c010ea733925b1acd.docx"},{"id":96239475,"identity":"2040efca-0038-4dbd-9999-df7fe75c4c92","added_by":"auto","created_at":"2025-11-19 07:06:45","extension":"json","order_by":1,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":14066,"visible":true,"origin":"","legend":"","description":"","filename":"05231d7249a84c68b03dbbebde03d363.json","url":"https://assets-eu.researchsquare.com/files/rs-7913800/v1/a997a430ba2b24c78d87d54d.json"},{"id":96240114,"identity":"1a1f7045-731b-41a5-8e05-fae2cdfb527b","added_by":"auto","created_at":"2025-11-19 07:08:25","extension":"pdf","order_by":2,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":1264904,"visible":true,"origin":"","legend":"","description":"","filename":"Supplementary.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7913800/v1/972925bd272da8bbc7e0fbfe.pdf"},{"id":95825615,"identity":"d6a48b3d-a5ed-4901-8a56-ef38ceeca6e4","added_by":"auto","created_at":"2025-11-13 11:09:28","extension":"xml","order_by":3,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":118569,"visible":true,"origin":"","legend":"","description":"","filename":"05231d7249a84c68b03dbbebde03d3631enriched.xml","url":"https://assets-eu.researchsquare.com/files/rs-7913800/v1/f763be2da0e4b9350892ac3d.xml"},{"id":95825614,"identity":"a51f753a-3a4b-41cc-abc6-05db7bacff5f","added_by":"auto","created_at":"2025-11-13 11:09:28","extension":"png","order_by":4,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":41823,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-7913800/v1/21290f7d74ae690be9b495c1.png"},{"id":95825618,"identity":"71b6ad5e-264a-415b-a5d6-fd19509f2693","added_by":"auto","created_at":"2025-11-13 11:09:28","extension":"png","order_by":5,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":52887,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-7913800/v1/a048e1c527a7c3ff593a03d2.png"},{"id":96239507,"identity":"2978a3c1-cdf7-4d5c-8287-ffb0ac12e32f","added_by":"auto","created_at":"2025-11-19 07:06:51","extension":"png","order_by":6,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":319305,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-7913800/v1/10b1c27b5f678ded4be43029.png"},{"id":95825626,"identity":"f436edd4-40b0-49f3-a2a1-3d51808c8089","added_by":"auto","created_at":"2025-11-13 11:09:28","extension":"png","order_by":7,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":361164,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage4.png","url":"https://assets-eu.researchsquare.com/files/rs-7913800/v1/0b539f6aaba6fc983a1f06ec.png"},{"id":95825625,"identity":"3b3fa77f-7f7f-48db-8417-eead69613ddb","added_by":"auto","created_at":"2025-11-13 11:09:28","extension":"png","order_by":8,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":212117,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage5.png","url":"https://assets-eu.researchsquare.com/files/rs-7913800/v1/006e42374157eb76f7b5d0ce.png"},{"id":96239530,"identity":"d4e8c021-f99d-48d8-98d7-893e22e6c208","added_by":"auto","created_at":"2025-11-19 07:06:53","extension":"png","order_by":9,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":18693,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-7913800/v1/c3304338d82dcbafdbb99995.png"},{"id":95825623,"identity":"5cf198e3-8514-4d8e-87b7-c7ab0e5244c3","added_by":"auto","created_at":"2025-11-13 11:09:28","extension":"png","order_by":10,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":23232,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-7913800/v1/c40e2a6ee5345e21c53980fe.png"},{"id":95825621,"identity":"21190600-a87d-4d1d-8913-a787047e0a7e","added_by":"auto","created_at":"2025-11-13 11:09:28","extension":"png","order_by":11,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":57654,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-7913800/v1/43bcb6d1000b8f8c5687612a.png"},{"id":95825628,"identity":"f5425178-9a5e-4850-95d2-83b6ceddc4ef","added_by":"auto","created_at":"2025-11-13 11:09:28","extension":"png","order_by":12,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":65382,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage4.png","url":"https://assets-eu.researchsquare.com/files/rs-7913800/v1/76dc0a87d5f367ed5df64db2.png"},{"id":96239034,"identity":"0180645f-b5f8-4ae2-a96a-725b713de162","added_by":"auto","created_at":"2025-11-19 07:01:37","extension":"png","order_by":13,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":45910,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage5.png","url":"https://assets-eu.researchsquare.com/files/rs-7913800/v1/43660945f324113548c03c8e.png"},{"id":96239366,"identity":"aa147e02-3836-4236-940c-8f943ec722f9","added_by":"auto","created_at":"2025-11-19 07:06:24","extension":"xml","order_by":14,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":116337,"visible":true,"origin":"","legend":"","description":"","filename":"05231d7249a84c68b03dbbebde03d3631structuring.xml","url":"https://assets-eu.researchsquare.com/files/rs-7913800/v1/715ceeb89a9393815216f7e7.xml"},{"id":95825630,"identity":"9b52bff9-7541-4c5c-8937-79f49538d0b8","added_by":"auto","created_at":"2025-11-13 11:09:28","extension":"html","order_by":15,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":129240,"visible":true,"origin":"","legend":"","description":"","filename":"earlyproof.html","url":"https://assets-eu.researchsquare.com/files/rs-7913800/v1/f6bd5d778f16307e4e66fc10.html"},{"id":95825611,"identity":"df5c9bb9-0e0e-4fe5-b60d-c5c597fa4bda","added_by":"auto","created_at":"2025-11-13 11:09:28","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":254920,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eCorrelation matrix. \u003c/strong\u003eSpearman’s rank correlation and hierarchical clustering using the Ward’s minimum variance method. Darker colours indicate stronger correlation.\u003c/p\u003e","description":"","filename":"floatimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-7913800/v1/570d564f546ff502fac68204.png"},{"id":95825610,"identity":"578968d8-9f6f-4af4-8787-ec56698c2555","added_by":"auto","created_at":"2025-11-13 11:09:28","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":361164,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eA) Forest plot showing associations to ECAS total score for protein/protein pair candidates.\u003c/strong\u003e Each blue point represent the linear regression model b coefficient and the line the corresponding confidence interval. All models were adjusted for age at sampling, sex, site of onset and ALSFRS-R. The red dashed line correspond to a b coefficient of 0. \u003cstrong\u003eB) Relationship between PTPRN2 and GAP43 coloured by cognitive status.\u003c/strong\u003e \u003cstrong\u003eC)\u003c/strong\u003e \u003cstrong\u003eROC curve showing the predictive performance for ALS with vs without cognitive impairment\u003c/strong\u003e. Illustrating three different analyses: GAP43+PTPRN2 and GAP43 and PTPRN2 alone, all adjusted for age at sampling, sex, site of onset and ALSFRS-R total score.\u003c/p\u003e","description":"","filename":"floatimage4.png","url":"https://assets-eu.researchsquare.com/files/rs-7913800/v1/19435a832253cbfd537590c1.png"},{"id":96240177,"identity":"074c77fa-bdc0-4cef-8cd8-c8e67c232e04","added_by":"auto","created_at":"2025-11-19 07:08:32","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":212117,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eThe association between PTPRN2/GAP43 and sex. A)\u003c/strong\u003e Boxplot of PTPRN2/GAP43 for female and men with and without cognitive impairment. \u003cstrong\u003eB)\u003c/strong\u003e Scatterplot of PTPRN2/GAP43 over ECAS total score coloured by sex. Grey, shadowed areas represent the standard error. The model included ECAS total score and sex as covariates including an interaction term between these two variables, and was adjusted for age at sampling, site of onset, and ALSFRS-R.\u003c/p\u003e","description":"","filename":"floatimage5.png","url":"https://assets-eu.researchsquare.com/files/rs-7913800/v1/41a77552e1452c25fb5d839c.png"},{"id":101691244,"identity":"758304ab-3db0-4e74-a571-f852fa3092d7","added_by":"auto","created_at":"2026-02-02 16:13:11","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1904180,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7913800/v1/cc0cd38d-08c2-4487-bd0d-1df8e2cedd15.pdf"},{"id":95825612,"identity":"6573b0c9-399b-49b7-a56e-feeaee133500","added_by":"auto","created_at":"2025-11-13 11:09:28","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"supplement","size":1264904,"visible":true,"origin":"","legend":"","description":"","filename":"Supplementary.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7913800/v1/3f0976b1002ced2c32314657.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"Ratios of CSF Proteins Reflect Cognitive Function in ALS","fulltext":[{"header":"BACKGROUND","content":"\u003cp\u003eCognitive dysfunction is a recognised feature of numerous neurodegenerative diseases, not only dementia disorders. In amyotrophic lateral sclerosis (ALS), cognitive impairment spans a spectrum from subtle executive dysfunction to full-blown overlap with frontotemporal dementia (ALS-FTD) (\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e, \u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e). Despite advances in understanding ALS-related cognitive impairment, efforts to translate this knowledge into clinical tools have lagged behind. In particular, biomarker research in cerebrospinal fluid (CSF) has increasingly focused on protein levels and their dynamics, with promising findings from studies of neurodegenerative diseases like Alzheimer disease (AD) (\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e, \u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e).\u003c/p\u003e\u003cp\u003eHowever, currently there are no fluid biomarkers that predict nor identify cognitive impairment in ALS. Filling this gap is critical as cognitive impairment in ALS has significant implications for disease management, affecting patients' ability to make informed decisions regarding care, treatment, and participation in clinical trials. Biomarkers that detect early cognitive dysfunction could also enable more targeted and timely interventions, potentially improving quality of life. Moreover, cognitive impairment is linked to shorter survival in ALS (\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e), highlighting the prognostic value of identifying at-risk individuals early in their disease course. Together, these factors underscore the need for fluid biomarkers that not only detect neurodegeneration but also track its impact on cognitive function.\u003c/p\u003e\u003cp\u003eEmerging evidence suggests that interindividual variability in CSF protein levels can mask disease-related changes, complicating the interpretation of biomarker data (\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e). By adjusting for this variability, protein ratios have been shown to enhance diagnostic and prognostic accuracy in AD (\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e, \u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e). In addition to their diagnostic utility, CSF biomarkers can provide valuable insights into disease mechanisms. For example, alterations in proteins such as neurofilaments, which are consistently elevated in ALS, reflects axonal damage while proteins like chitinases may capture central nervous system inflammation relevant to ALS pathophysiology (\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e, \u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e). By combining proteins associated with different disease mechanisms, a synergistic or additive effect could enhance their potential as a biomarker.\u003c/p\u003e\u003cp\u003eBuilding on the prior findings in AD, here we extend the concept of protein ratios to ALS, suggesting a shared utility of protein ratios as biomarkers for neurodegenerative cognitive dysfunction. We investigated a panel of CSF proteins in a well-characterized cohort of patients with ALS. Specifically, we focused on assessing the ability of these proteins to detect cognitive impairment, both individually and through protein ratios.\u003c/p\u003e"},{"header":"methods","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e\u003ch2\u003eCohort and data collection\u003c/h2\u003e\u003cp\u003eParticipants were recruited from the ALSrisc study, an ongoing longitudinal cohort study at Karolinska University Hospital in Stockholm, Sweden (\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e). We included patients with a diagnosis of ALS, an available CSF sample and assessment of cognitive function within a year from sample collection. The following clinical and phenotypic data were collected: age at sampling, sex, site of onset and \u003cem\u003eC9orf72\u003c/em\u003e repeat expansion. Disease progression was monitored using the ALS Functional Rating Scale–Revised (ALSFRS-R), a standardised instrument widely used to quantify functional decline in ALS (\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e). Cognitive function was evaluated using the Edinburgh Cognitive and Behavioural ALS Screen (ECAS), a validated screening tool specifically developed for ALS populations (\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e). ECAS administration takes approximately 15–20 minutes and assesses both ALS-specific (language, verbal fluency, executive function) and ALS-non-specific domains (memory, visuospatial function). The ECAS score collected closest to CSF sampling date was included. The median time between CSF sampling and ECAS was 7 weeks (range 0–35 weeks). No subject had ECAS assessed prior to sample collection. An ECAS total score below 108 will herein be referred to as “cognitive impairment” (\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e).\u003c/p\u003e\u003c/div\u003e\n\u003ch3\u003eSample collection\u003c/h3\u003e\n\u003cp\u003eCSF samples were collected between 2019 and 2021 via lumbar puncture into polypropylene tubes and processed immediately by centrifugation at 963 ×g for 10 minutes at room temperature. The resulting supernatant was aliquoted into cryotubes and stored at − 80°C until analysis. Prior to protein analysis, the samples were stratified into 96-well PCR plates using a constrained randomisation approach based on diagnosis, age, and sex.\u003c/p\u003e\n\u003ch3\u003eProtein analysis\u003c/h3\u003e\n\u003cp\u003eProteins (n = 47) were selected based on prior in-house neuroproteomic studies, both published (\u003cspan additionalcitationids=\"CR16 CR17 CR18 CR19 CR20 CR21\" citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e–\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e) and unpublished. Corresponding polyclonal rabbit antibodies were sourced from the Human Protein Atlas (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ewww.proteinatlas.org\u003c/span\u003e\u003c/span\u003e), except for progranulin (GRN) (AF3156-SP, R\u0026amp;D Systems) and apolipoprotein E4 (APOE4) (M067-3, MBL Life Science) antibodies (\u003cb\u003eSupplementary Table\u0026nbsp;1\u003c/b\u003e). To create the bead array, each antibody was individually immobilised onto uniquely color-coded, carboxylated magnetic beads (MagPlex, Luminex Corp.) as previously described (\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e).\u003c/p\u003e\u003cp\u003eCSF was processed according to a previously described protocol (\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e). In brief, the samples were diluted 1:2 and labelled with biotin (NHS-PEG4-biotin, A39259, ThermoFisher Scientific) followed by heat-treatment at 56°C for 30 minutes. Incubation with beads was performed overnight at room temperature in 384-well plates and bound proteins detected using a streptavidin-conjugated fluorophore. Fluorescence signals, corresponding to median fluorescence intensity per bead ID and per sample, were acquired using a FlexMap 3D instrument (Luminex Corp.), providing relative quantification of protein levels.\u003c/p\u003e\u003cdiv id=\"Sec6\" class=\"Section2\"\u003e\u003ch2\u003eStatistical analysis\u003c/h2\u003e\u003cp\u003eAll data pre-processing, analysis and illustrations were performed in R Studio version 4.3.1. Protein level data was adjusted for delayed instrument readout using robust linear regression as described previously (\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e). All protein levels were log2 transformed and scaled to zero mean and unit variance prior to statistical analysis. Assumptions for statistical models were assessed visually by residual plots (independence and equal variance) and normal probability plots (normality). The Wilcoxon rank-sum test was used to compare differences between sexes and the Kruskal–Wallis test was applied to assess differences across groups for site of symptom onset and genetic status.\u003c/p\u003e\u003c/div\u003e\n\u003ch3\u003eCorrelation and hierarchical clustering\u003c/h3\u003e\n\u003cp\u003eSpearman’s rank correlation coefficients (ρ) were computed to assess the co-variation of CSF protein profiles. Protein clustering based on correlation was performed using hierarchical clustering using the Ward’s minimum variance method.\u003c/p\u003e\u003cdiv id=\"Sec8\" class=\"Section2\"\u003e\u003ch2\u003eElastic net\u003c/h2\u003e\u003cp\u003eElastic net regression is a statistical modelling technique that addresses the challenges of analyzing datasets with a large number of interrelated variables (\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e). It combines two forms of regularization, those used in ridge regression and Least Absolute Shrinkage and Selection Operator (LASSO), to improve model performance and interpretability. Ridge regression reduces overfitting by shrinking the size of all regression coefficients, but it does not eliminate any variables. In contrast, LASSO can shrink some coefficients to exactly zero, thus performing both shrinkage and variable selection, i.e. effectively selecting a smaller subset of variables that contribute most to the outcome. Elastic net introduces a mixing parameter, α (alpha), which controls the balance between ridge (α = 0) and LASSO (α = 1) penalties. By adjusting α, elastic net provides a flexible framework that combines the strengths of both methods. This makes elastic net especially effective when working with high-dimensional data and/or highly correlated predictors, as was the case in our study. In addition to the mixing parameter, elastic net also includes a second parameter, λ (lambda), which controls the overall strength of the penalty applied to the model. Larger values of λ impose stronger penalties, leading to greater shrinkage of the regression coefficients producing simpler, more constrained models (\u003cb\u003eSupplementary Fig.\u0026nbsp;1\u003c/b\u003e).\u003c/p\u003e\u003cp\u003e\u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003eData setup\u003c/span\u003e: We partitioned the dataset into 70% for training and 30% for testing. The training data included 48 patients with a median ECAS total score of 109, the test set of 18 patients with a median ECAS total score of 108.5.\u003c/p\u003e\u003cp\u003e\u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003eTuning procedure\u003c/span\u003e: A grid search over α values (step size 0.01) ranging from 0 (Ridge) to 1 (LASSO) was conducted using cross-validation to identify the optimal mixing parameter. The best α was selected based on the minimum cross-validation error. Subsequently, with the optimal α, a ten-fold cross-validation was performed to select the optimal regularisation parameter λ (λ\u003csub\u003emin\u003c/sub\u003e​) in order to obtain the ideal trade-off between bias and variance. A cross-validation plot was used to support the tuning process (\u003cb\u003eSupplementary Fig.\u0026nbsp;2\u003c/b\u003e).\u003c/p\u003e\u003cp\u003e\u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003eModel fitting and evaluation\u003c/span\u003e: The elastic net model was fit on the training dataset using the optimal α and λ values. The root mean squared error (RMSE) for the training set was computed to assess model fit and potential overfitting. RMSE is a commonly used metric that quantifies the average magnitude of prediction error, with lower values indicating better model performance. Similarly, RMSE was calculated for the test set to evaluate prediction accuracy.\u003c/p\u003e\u003c/div\u003e\n\u003ch3\u003eRegression models\u003c/h3\u003e\n\u003cp\u003eLinear regression models were constructed to evaluate the association between candidate proteins and cognitive scores (ECAS total score). For each protein, a separate model was fitted, with ECAS total score as the outcome variable and protein levels as predictor with adjustments for age at sampling, sex, site of onset and ALSFRS-R total score. Model performance and generalisability were assessed using 10-fold cross-validation. To assess the ability for candidate proteins to predict cognitive status (cognitive impairment vs no cognitive impairment), logistic regression models were constructed. All models were adjusted for age at sampling, sex, site of onset, and ALSFRS-R total score. Model performance was evaluated by receiver operating characteristic (ROC) curve and area under the curve (AUC).\u003c/p\u003e\n\u003cp\u003e\u003cimg src=\"data:image/png;base64,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\"\u003e\u003c/p\u003e"},{"header":"Results","content":"\u003cp\u003eThe final cohort included 66 patients (mean age = 66 ± 11.4 years, 39% female) of whom 44% had an ECAS total score below 108 (Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e). The ECAS total score was not correlated to age or disease duration. The majority of patients had a spinal onset (68%), and nine individuals (14%) carried a \u003cem\u003eC9orf72\u003c/em\u003e repeat expansion.\u003c/p\u003e\u003cdiv class=\"gridtable\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e\u003ccaption language=\"En\"\u003e\u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e\u003cdiv class=\"CaptionContent\"\u003e\u003cp\u003e\u003cb\u003eCohort.\u003c/b\u003e Characteristics of the patients included in the study. N, number; SD, standard deviation; \u003cem\u003eC9orf72\u003c/em\u003e, C9 open reading frame 72; ECAS, Edinburgh cognitive and behavioural ALS screen; IQR, inter quartile range; ALSFRS-R, ALS functional rating scale–revised.\u003c/p\u003e\u003c/div\u003e\u003c/caption\u003e\u003ccolgroup cols=\"2\"\u003e\u003c/colgroup\u003e\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\"\u003e\u003cp\u003eVariable\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c2\"\u003e\u003cp\u003eN = 66\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eAge at sampling, mean (SD)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e66 (11.4)\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eSex, N females (%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e26 (39)\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eSite of onset, N (%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eSpinal\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e45 (68)\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eBulbar\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e19 (\u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e)\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eOther\u003csup\u003ea\u003c/sup\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e2 (\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e)\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e\u003cem\u003eC9orf72\u003c/em\u003e repeat expansion, N (%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e9 (\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e)\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eECAS, median\u003c/p\u003e\u003cp\u003e(IQR)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e109 (21.5)\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eECAS \u0026lt; 108, N (%)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e29 (44)\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eALSFRS-R, median (IQR)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e41.5 (8.25)\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003ctfoot\u003e\u003ctr\u003e\u003ctd colspan=\"2\"\u003e\u003csup\u003ea\u003c/sup\u003e respiratory\u003c/td\u003e\u003c/tr\u003e\u003c/tfoot\u003e\u003c/table\u003e\u003c/div\u003e\u003ch3\u003eNeurofilament levels do not correlate to known markers of cognitive function\u003c/h3\u003e\u003cp\u003ePairwise correlation analysis of the 47 CSF proteins revealed two main protein clusters, one with strongly correlating proteins (n = 26, median ρ 0.81, IQR 0.14) including proteins commonly regarded as markers for dementia and cognitive function such as neurogranin (NRGN), beta-synuclein (SNCB) and neuromodulin (GAP43) (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e). The other protein clusters (n = 18 and n = 3) showed generally lower co-variation between proteins (median ρ 0.54, IQR 0.23, and median ρ 0.34, IQR 0.14, respectively). Neurofilament medium (NEFM), being in the small cluster of three proteins, did not exhibit strong correlations with the neuronal proteins in the first cluster (median ρ 0.20, IQR 0.18). Furthermore, chitinase 1 (CHIT1), included in the cluster together with NEFM, displayed a unique correlation profile (median ρ -0.01, range − 0.08 − 0.07, IQR 0.04).\u003c/p\u003e\u003ch2\u003eCSF proteins are associated with ECAS total score\u003c/h2\u003e\u003cp\u003eWe applied elastic net as a variable selection method to identify proteins with a relevant association to ECAS total score. In our final elastic net model, the optimal hyperparameters were found to be α = 0.54 and λ = 3.63. This indicated that the final model used a nearly balanced mix of LASSO and ridge regression penalties, reflecting both the need for variable selection and stability in the presence of potential multicollinearity. The model identified 7 proteins as the most predictive of ECAS total score: NEFM, neuronal pentraxin 2 (NPTX2), GAP43, insulin like growth factor binding protein 4 (IGFBP4), insulin like growth factor binding protein 7 (IGFBP7), osteopontin (SPP1) and cadherin 8 (CDH8) (Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e). The model produced an RMSE of 13.76 on the training set and 12.03 on the test set. The slightly lower RMSE on the test set suggests that the model not only fit the training data well but also generalized effectively to unseen data, with no indications of overfitting.\u003c/p\u003e\u003cdiv class=\"gridtable\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e\u003ctable float=\"Yes\" id=\"Tab2\" border=\"1\"\u003e\u003ccaption language=\"En\"\u003e\u003cdiv class=\"CaptionNumber\"\u003eTable 2\u003c/div\u003e\u003cdiv class=\"CaptionContent\"\u003e\u003cp\u003e\u003cb\u003eCandidate protein and protein pairs.\u003c/b\u003e Results from the elastic net and linear regression analyses.\u003c/p\u003e\u003c/div\u003e\u003c/caption\u003e\u003ccolgroup cols=\"5\"\u003e\u003c/colgroup\u003e\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/th\u003e\u003cth align=\"left\" colspan=\"2\" nameend=\"c3\" namest=\"c2\"\u003e\u003cp\u003eLinear regression\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c4\"\u003e\u003cp\u003eSelected in elastic net of single proteins\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c5\"\u003e\u003cp\u003eSelected in elastic net of protein pairs\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/th\u003e\u003cth align=\"left\" colname=\"c2\"\u003e\u003cp\u003eβ coefficient\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c3\"\u003e\u003cp\u003eCV \u003cem\u003eR\u003c/em\u003e\u003csup\u003e\u003cem\u003e2\u003c/em\u003e\u003c/sup\u003e\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/th\u003e\u003cth align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/th\u003e\u003c/tr\u003e\u003ctr\u003e\u003cth align=\"left\" colspan=\"5\" nameend=\"c5\" namest=\"c1\"\u003e\u003cp\u003eSingle protein\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eIGFBP7\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e4.05 (0.11–7.98)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e0.27\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003eYes\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003eNo\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eNPTX2\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e2.83 (-1.09–6.76)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e0.29\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003eYes\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003eYes\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eCDH8\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e2.57 (-1.37–6.50)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e0.25\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003eYes\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003eYes\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003ePTPRN2\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e2.26 (-1.82–6.34)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e0.24\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003eNo\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003eYes\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eIGF2\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e2.07 (-2.04–6.19)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e0.13\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003eNo\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003eYes\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eCHL1\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e1.62 (-2.52–5.76)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e0.26\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003eNo\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003eYes\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eCADM2\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e1.07 (-3.03–5.16)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e0.34\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003eNo\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003eYes\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eBASP1\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e0.67 (-3.50–4.84)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e0.18\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003eNo\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003eYes\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eIGFBP4\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e-0.06 (-4.37–4.24)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e0.21\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003eYes\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003eYes\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eGAP43\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e-1.25 (-5.41–2.91)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e0.23\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003eYes\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003eYes\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eNEFM\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e-2.58 (-6.92–1.77)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e0.22\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003eYes\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003eNo\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eSPP1\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e-3.45 (-7.34–0.45)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e0.47\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003eYes\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003eYes\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colspan=\"5\" nameend=\"c5\" namest=\"c1\"\u003e\u003cp\u003e\u003cb\u003eProtein pair\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003ePTPRN2/GAP43\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e7.48 (3.66–11.29)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e0.34\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eCDH8/GAP43\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e6.77 (2.68–10.85)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e0.32\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eCHL1/GAP43\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e6.49 (2.38–10.61)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e0.30\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eNPTX2/SPP1\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e5.24 (1.49–8.99)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e0.26\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003ePTPRN2/BASP1\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e4.94 (0.66–9.22)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e0.19\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eCADM2/GAP43\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e3.5 (-0.63–7.63)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e0.22\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eNPTX2/IGFBP4\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e3.05 (-1.27–7.37)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e0.28\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eIGF2/IGFBP4\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e2.92 (-1.24–7.07)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e0.27\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colspan=\"5\" nameend=\"c5\" namest=\"c1\"\u003e\u003cp\u003eCI = confidence interval, CV \u003cem\u003eR\u003c/em\u003e\u003csup\u003e2\u003c/sup\u003e = 10-fold cross-validated \u003cem\u003eR\u003c/em\u003e\u003csup\u003e2\u003c/sup\u003e\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/table\u003e\u003c/div\u003e\u003ch2\u003eProtein ratios are superior to single proteins for predicting ECAS score\u003c/h2\u003e\u003cp\u003eAs protein pairs have been shown to provide stronger associations with cognitive function compared to single proteins in other neurodegenerative disorders, we also assessed protein ratios for their ability to detect cognitive impairment in ALS. All 47 proteins were combined into pairs (n = 2162). Again, we used elastic net to find the pairs most predictive of ECAS total score. In the elastic net model of ratios, the optimal hyperparameters were found to be α = 0.37 and λ = 10.38. Here, 8 protein pairs were identified as the most predictive of ECAS total score (Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e, Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eA). The model produced an RMSE of 12.93 on the training set and 12.11 on the test set.\u003c/p\u003e\u003cp\u003eFive of the proteins selected in the ratio elastic net model were also selected in the single protein model (NPTX2, GAP43, IGFBP4, SPP1 and CDH8). Interestingly, five additional proteins were found relevant as part of a pair, namely insulin like growth factor 2 (IGF2), cell adhesion molecule L1 like (CHL1), protein tyrosine phosphatase receptor type N2 (PTPRN2), cell adhesion molecule 2 (CADM2) and brain abundant membrane attached signal protein 1 (BASP1). As previously shown, neither of these had a strong association with ECAS total score alone.\u003c/p\u003e\u003cp\u003eTo further evaluate the association to ECAS total score and compare the performance of single proteins and protein pairs, linear regression models were created for each candidate as the predictor (Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e, Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eA). We assessed the predictive performance of each model using 10-fold cross-validated \u003cem\u003eR\u003c/em\u003e² (CV \u003cem\u003eR\u003c/em\u003e²). Comparing the regression models revealed that higher CV \u003cem\u003eR\u003c/em\u003e² values were generally observed in protein pair models (median 0.27) in contrast with those of the single protein models (median 0.24) (\u003cb\u003eSupplementary Fig.\u0026nbsp;3\u003c/b\u003e). In particular, protein pair ratios involving GAP43 exhibited larger b coefficient magnitudes, the majority of the confidence intervals excluded zero, and higher cross-validated \u003cem\u003eR\u003c/em\u003e² values relative to single protein metrics (Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e). The PTPRN2/GAP43 ratio showed the strongest association with an estimated b coefficient of 7.48 (95% CI: 3.66–11.29) with a 10-fold cross-validated \u003cem\u003eR\u003c/em\u003e² of 0.34. Other prominent protein pairs included CDH8/GAP43, CHL1/GAP43 and NPTX2/SPP1. The PTPRN2/GAP43 b coefficient substantially exceeded that of PTPRN2 and GAP43 alone (2.26, 95% CI: -1.82–6.34 and − 1.25 (95% CI: -5.41–2.91, respectively), suggesting that the ratio of these highly correlating proteins is more informative than the levels alone (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eB). This additive effect of PTPRN2 in combination with GAP43 was further explored using ROC analysis (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eC). The performance was considerably better for PTPRN2/GAP43, with an area under the curve of 0.89 (95% CI: 0.80–0.97), compared to PTPRN2 and GAP43 alone (AUC 0.76 95% CI: 0.64–0.89 and 0.78 95% CI: 0.67–0.90, respectively). These results suggest that protein ratios potentially offer a more robust prediction of ECAS total score compared to single protein measures. The consistency of key marker selection in both elastic net and linear regression analyses further reinforces the potential of these ratios as biomarkers for cognitive impairment in ALS.\u003c/p\u003e\u003cp\u003eWe next evaluated the association of the most promising protein ratios with ECAS sub scores. The strongest associations with executive function, verbal fluency, and memory for the PTPRN2/GAP43 ratio (\u003cb\u003eSupplementary Fig.\u0026nbsp;4\u003c/b\u003e). Similar results were found for the other candidate pairs suggesting that these ratios are more likely markers of a general cognitive dysfunction and not specific to ALS frontotemporal involvement.\u003c/p\u003e\u003ch2\u003eThe trajectories of CSF PTPRN2/GAP43 are different in males and females with cognitive impairment\u003c/h2\u003e\u003cp\u003eThe PTPRN2/GAP43 ratio was not associated with ALSFRS-R score, nor were there any statistically significant differences in the PTPRN2/GAP43 CSF ratio between bulbar versus spinal onset or \u003cem\u003eC9orf72\u003c/em\u003e mutation carriers versus non-carriers (\u003cb\u003eSupplementary Fig.\u0026nbsp;5A, Supplementary Fig.\u0026nbsp;5B\u003c/b\u003e). In addition, we did not find any overall differences in PTPRN2/GAP43 ratio between the sexes (p = 0.16). However, among the patients with cognitive impairment, lower ratios were found in males compared to females even though the distribution of cognitive impairment and age was similar between sexes (p = 0.002) (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eA). Indeed, when exploring linear regression with PTPRN2/GAP43 as the outcome and including an interaction term between ECAS total score and sex, we found that the slope of PTPRN2/GAP43 ratio over ECAS total score was steeper for males than females (0.04 and 0.01 respectively, on a log2 scale) (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eB).\u003c/p\u003e"},{"header":"Discussion","content":"\u003cp\u003eIn this study we analysed a panel of CSF proteins and investigated their association to cognitive function in patients with ALS. No single protein strongly predicted cognitive impairment but combining them into ratios markedly improved predictive power. This resonates with findings in AD, where ratios like Aβ42/40 or Aβ42/tau better reflect disease processes than individual proteins (\u003cspan additionalcitationids=\"CR26\" citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e). Such ratios might capture interactions between pathways, for example synaptic repair (GAP43) and metabolic alterations (PTPRN2). A possible explanation for the enhanced performance of protein pairs in relation to single proteins is that one protein in the ratio act as a reference, thereby accounting for non-disease related individual variation. Prior studies have suggested that such variability can be influenced by factors like sex, age, and ventricular volume (\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e, \u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e).\u003c/p\u003e\u003cp\u003eVariable selection methods like elastic net can be used to highlight proteins with important contributions to a multivariable model, but their performance may not always translate into regression models. In our case, this means that while these proteins are informative in a multivariate setting, they may not independently explain cognitive variance when clinical variables are accounted for. This could explain why several of the identified single proteins were selected in the elastic net, even though they did not demonstrate strong predictive performance in the subsequent adjusted linear models.\u003c/p\u003e\u003cp\u003eIn this study, the most promising candidate biomarker associated with cognitive impairment was the PTPRN2/GAP43 ratio, the same protein pair previously found to be among the best classifiers of cognitive decline in patients with dementia (\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e, \u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e). While levels of GAP43 have been associated with AD in previous studies (including 15,16,19,28\u0026ndash;30), the function and role of PTPRN2 in neurodegenerative disease pathology is poorly understood. In a clinical setting, complex multivariable models with many covariates and interaction terms can be difficult to apply or interpret, limiting their practical use. In contrast, simpler measures, like ratios of two proteins, are more feasible for clinical translation, as they balance complexity with interpretability and robustness. Our findings that protein ratios outperform single markers are in line with recent publications and these studies together suggest that the use of protein pairs might be a suitable middle way for practical biomarker implementation (\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e, \u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e).\u003c/p\u003e\u003cp\u003eOur cluster analysis revealed a strong co-variation between many of the CSF proteins relevant for synaptic health and function. The tight intercorrelations within the largest cluster suggest that these proteins may be co-regulated or share common biological pathways involved in neurodegeneration. Neurofilament medium, a marker of axonal integrity closely related to the well-established neurofilament light chain (\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e), did not strongly correlate with these cognitive markers and may be involved in only partially reflect overlapping pathological mechanisms. The clustering pattern of NEFM instead suggests stronger association with neuroinflammation or immune response rather than synaptic function. Together with the finding that protein pairs including NEFM were not selected as key features in the elastic net regression, indicate that cognitive dysfunction in ALS might be more closely driven by synaptic and cortical processes than by general axonal damage.\u003c/p\u003e\u003cp\u003eThe observed sex differences in CSF PTPRN2/GAP43 levels in relation to cognitive impairment are particularly noteworthy. Specifically, our regression analysis revealed a significant interaction between sex and ECAS total score, with a markedly steeper decline in the PTPRN2/GAP43 ratio among males. Although this ratio does not independently predict ECAS scores when modelled in the reverse direction, the interaction suggests that cognitive status is more strongly associated with biomarker levels in males. This finding not only reinforces the group differences seen in our stratified analyses but also points to a potentially meaningful biological divergence. It raises the possibility of sex-specific dynamics in biomarker expression or regulation in the context of ALS-related cognitive impairment. Further investigation is needed to determine whether these differences reflect distinct underlying mechanisms or differential susceptibility to cognitive decline between sexes.\u003c/p\u003e\u003cp\u003eA strength of this study lies in its rigorous application of statistical methods and cross-validation, ensuring that our identified associations are robust. However, one limitation is the small sample size, especially for sex-stratified analyses, which may limit the generalisability of the results. In addition, while ECAS is a validated screening tool for cognitive impairment in ALS, our cut-offs were not adjusted for potential confounders such as age and education, which may influence cognitive performance and the classification of impairment (\u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e). While the strong performance of CSF PTPRN2/GAP43 suggests biological relevance, future studies should aim for larger cohorts and more comprehensive cognitive assessments to confirm these findings.\u003c/p\u003e"},{"header":"CONCLUSIONS","content":"\u003cp\u003eAltogether, our data support the hypothesis that protein ratios, particularly PTPRN2/GAP43, may capture aspects of cognitive impairment in ALS that go undetected by single markers. They also highlight the importance of considering sex-specific interactions when interpreting biomarker data. If validated in larger cohorts, these findings could aid the development of diagnostic methods in ALS that are sensitive to cognitive impairment and argue for prioritising combinatorial or ratio-based measures over single protein readouts in future biomarker panels. Such panels could also have broader implications, as similar synaptic processes are implicated in other neurodegenerative diseases, suggesting shared targets for therapeutic intervention.\u003c/p\u003e"},{"header":"Abbreviations","content":"\u003ctable border=\"1\" cellspacing=\"0\" cellpadding=\"0\" width=\"463\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 104px;\"\u003e\n \u003cp\u003eAD\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 359px;\"\u003e\n \u003cp\u003ealzheimer disease\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 104px;\"\u003e\n \u003cp\u003eALS\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 359px;\"\u003e\n \u003cp\u003eamyotrophic lateral sclerosis\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 104px;\"\u003e\n \u003cp\u003eALSFRS-R\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 359px;\"\u003e\n \u003cp\u003eALS functional rating scale\u0026ndash;revised\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 104px;\"\u003e\n \u003cp\u003eAUC\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 359px;\"\u003e\n \u003cp\u003earea under the curve\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 104px;\"\u003e\n \u003cp\u003eBASP1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 359px;\"\u003e\n \u003cp\u003ebrain abundant membrane attached signal protein 1\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 104px;\"\u003e\n \u003cp\u003eC9orf72\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 359px;\"\u003e\n \u003cp\u003eC9 open reading frame 72\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 104px;\"\u003e\n \u003cp\u003eCADM2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 359px;\"\u003e\n \u003cp\u003ecell adhesion molecule 2\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 104px;\"\u003e\n \u003cp\u003eCDH8\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 359px;\"\u003e\n \u003cp\u003ecadherin 8\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 104px;\"\u003e\n \u003cp\u003eCHIT1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 359px;\"\u003e\n \u003cp\u003echitinase 1\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 104px;\"\u003e\n \u003cp\u003eCHL1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 359px;\"\u003e\n \u003cp\u003ecell adhesion molecule l1 like\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 104px;\"\u003e\n \u003cp\u003eCI\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 359px;\"\u003e\n \u003cp\u003econfidence interval\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 104px;\"\u003e\n \u003cp\u003eCSF\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 359px;\"\u003e\n \u003cp\u003ecerebraospinal fluid\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 104px;\"\u003e\n \u003cp\u003eCV\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 359px;\"\u003e\n \u003cp\u003ecross validation\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 104px;\"\u003e\n \u003cp\u003eECAS\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 359px;\"\u003e\n \u003cp\u003eEdinburgh cognitive and behavioural ALS screen\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 104px;\"\u003e\n \u003cp\u003eFTD\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 359px;\"\u003e\n \u003cp\u003efrontotemporal dementia\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 104px;\"\u003e\n \u003cp\u003eGAP43\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 359px;\"\u003e\n \u003cp\u003eneuromodulin\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 104px;\"\u003e\n \u003cp\u003eIGF2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 359px;\"\u003e\n \u003cp\u003einsulin like growth factor 2\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 104px;\"\u003e\n \u003cp\u003eIGFBP4\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 359px;\"\u003e\n \u003cp\u003einsulin like growth factor binding protein 4\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 104px;\"\u003e\n \u003cp\u003eIGFBP7\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 359px;\"\u003e\n \u003cp\u003einsulin like growth factor binding protein 7\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 104px;\"\u003e\n \u003cp\u003eIQR\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 359px;\"\u003e\n \u003cp\u003einter quartile range\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 104px;\"\u003e\n \u003cp\u003eLASSO\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 359px;\"\u003e\n \u003cp\u003eleast absolute shrinkage and selection operator\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 104px;\"\u003e\n \u003cp\u003eNEFM\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 359px;\"\u003e\n \u003cp\u003eneurofilament medium\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 104px;\"\u003e\n \u003cp\u003eNPTX2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 359px;\"\u003e\n \u003cp\u003eneuronal pentraxin 2\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 104px;\"\u003e\n \u003cp\u003eNRGN\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 359px;\"\u003e\n \u003cp\u003eneurogranin\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 104px;\"\u003e\n \u003cp\u003ePTPRN2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 359px;\"\u003e\n \u003cp\u003eprotein tyrosine phosphatase receptor type n2\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 104px;\"\u003e\n \u003cp\u003eRMSE\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 359px;\"\u003e\n \u003cp\u003eroot mean squared error\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 104px;\"\u003e\n \u003cp\u003eROC\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 359px;\"\u003e\n \u003cp\u003ereceiver operating characteristics\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 104px;\"\u003e\n \u003cp\u003eSNCB\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 359px;\"\u003e\n \u003cp\u003ebeta-synuclein\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 104px;\"\u003e\n \u003cp\u003eSPP1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 359px;\"\u003e\n \u003cp\u003eosteopontin\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e"},{"header":"Declarations","content":"\u003ch3\u003e\u003cstrong\u003eEthics approval and consent to participate\u0026nbsp;\u003c/strong\u003e\u003c/h3\u003e\n\u003cp\u003eThe study was conducted in accordance with the Declaration of Helsinki. An informed consent to participate in research was obtained from each participant. The study was reviewed and approved by the Swedish Ethical Review Authority (diary numbers 2017/1895-31\u0026nbsp;and 2018/1605-31).\u0026nbsp;\u003c/p\u003e\n\u003ch3\u003e\u003cstrong\u003eConsent for publication\u0026nbsp;\u003c/strong\u003e\u003c/h3\u003e\n\u003cp\u003eNot applicable\u003c/p\u003e\n\u003ch3\u003e\u003cstrong\u003eAvailability of data and materials\u0026nbsp;\u003c/strong\u003e\u003c/h3\u003e\n\u003cp\u003eThe dataset used and analysed during the current study are available from the corresponding author upon reasonable request.\u003c/p\u003e\n\u003ch3\u003e\u003cstrong\u003eCompeting interests\u0026nbsp;\u003c/strong\u003e\u003c/h3\u003e\n\u003cp\u003eThe authors declare no competing interest relevant to the work conducted in the presented study.\u003c/p\u003e\n\u003ch3\u003e\u003cstrong\u003eFunding\u003c/strong\u003e\u003c/h3\u003e\n\u003cp\u003eThis study was funded by Demensfonden, B\u0026ouml;rje Salming ALS foundation, Svenska L\u0026auml;kares\u0026auml;llskapet Bj\u0026ouml;rklunds fond, ALS-fonden, and Stockholm Region.\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003ch3\u003e\u003cstrong\u003eAuthors\u0026apos; contributions\u0026nbsp;\u003c/strong\u003e\u003c/h3\u003e\n\u003cp\u003eL\u0026Ouml;, SM, AM, and CI developed the study design. L\u0026Ouml;, AJ, UK, and CI contributed to patient data acquisition and\u0026nbsp;sample collection. NdV, JO, and SB performed the experimental work. L\u0026Ouml; and SM did the data quality control and statistical analyses. L\u0026Ouml; interpreted the data with support from SM, AM and CI. CI, AM, and PN supervised the project. L\u0026Ouml; drafted the manuscript with input from SM and AM. All authors critically reviewed and revised the manuscript. All authors read and approved the final manuscript.\u003c/p\u003e\n\u003ch3\u003e\u003cstrong\u003eAcknowledgements\u003c/strong\u003e\u003c/h3\u003e\n\u003cp\u003eWe would like to thank Jenny Hellqvist for all her work on data collection. We would also like to thank the entire staff of the Human Protein Atlas for their efforts. Most importantly, we would like to thank the patients who contributed to this study.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eStrong MJ, Abrahams S, Goldstein LH, Woolley S, Mclaughlin P, Snowden J, et al. Amyotrophic lateral sclerosis - frontotemporal spectrum disorder (ALS-FTSD): Revised diagnostic criteria. Amyotroph Lateral Scler Frontotemporal Degener. 2017 Apr 3;18(3\u0026ndash;4):153\u0026ndash;74. doi:10.1080/21678421.2016.1267768\u003c/li\u003e\n\u003cli\u003eAbrahams S, Leigh PN, Goldstein LH. Cognitive change in ALS: A prospective study. Neurology. 2005 Apr 12;64(7):1222\u0026ndash;6. doi:10.1212/01.WNL.0000156519.41681.27\u003c/li\u003e\n\u003cli\u003eMravinacov\u0026aacute; S, Alanko V, Bergstr\u0026ouml;m S, Bridel C, Pijnenburg Y, Hagman G, et al. CSF protein ratios with enhanced potential to reflect Alzheimer\u0026rsquo;s disease pathology and neurodegeneration. Mol Neurodegener. 2024 Dec 1;19(1). doi:10.1186/s13024-024-00705-z\u003c/li\u003e\n\u003cli\u003eKarlsson L, Vogel J, Arvidsson I, \u0026Aring;str\u0026ouml;m K, Janelidze S, Blennow K, et al. Cerebrospinal fluid reference proteins increase accuracy and interpretability of biomarkers for brain diseases. Nat Commun. 2024 Dec 1;15(1). doi:10.1038/s41467-024-47971-5\u003c/li\u003e\n\u003cli\u003e\u0026Ouml;ijerstedt L, Foucher J, Lovik A, Yazdani S, Juto A, Kl\u0026auml;ppe U, et al. Repeated cognitive assessments show stable function over time in patients with ALS. J Neurol [Internet]. 2024 Aug 1;271(8):5267\u0026ndash;74. doi:10.1007/s00415-024-12479-x\u003c/li\u003e\n\u003cli\u003eMravinacov\u0026aacute; S, Bergstr\u0026ouml;m S, Olofsson J, de San Jos\u0026eacute; NG, Anderl-Straub S, Diehl-Schmid J, et al. Addressing inter individual variability in CSF levels of brain derived proteins across neurodegenerative diseases. Sci Rep [Internet]. 2025 Jan 3;15(1):668. doi:10.1038/s41598-024-83281-y\u003c/li\u003e\n\u003cli\u003eOh HS-H, Urey DY, Karlsson L, Zhu Z, Shen Y, Farinas A, et al. A cerebrospinal fluid synaptic protein biomarker for prediction of cognitive resilience versus decline in Alzheimer\u0026rsquo;s disease. Nat Med [Internet]. 2025 Mar 31;31(5). doi:10.1038/S41591-025-03565-2\u003c/li\u003e\n\u003cli\u003eKarlsson L, Vogel J, Arvidsson I, \u0026Aring;str\u0026ouml;m K, Strandberg O, Seidlitz J, et al. Machine learning prediction of tau-PET in Alzheimer\u0026rsquo;s disease using plasma, MRI, and clinical data. Alzheimers Dement. 2025 Feb 1;21(2):e14600. doi:10.1002/alz.14600\u003c/li\u003e\n\u003cli\u003eKl\u0026auml;ppe U, Sennf\u0026auml;lt S, Lovik A, Finn A, Bofaisal U, Zetterberg H, et al. Neurodegenerative biomarkers outperform neuroinflammatory biomarkers in amyotrophic lateral sclerosis. Amyotroph Lateral Scler Frontotemporal Degener [Internet]. 2024;25(1\u0026ndash;2):150\u0026ndash;61. doi:10.1080/21678421.2023.2263874\u003c/li\u003e\n\u003cli\u003eOlofsson J, Bergstr\u0026ouml;m S, Mravinacov\u0026aacute; S, Kl\u0026auml;ppe U, \u0026Ouml;ijerstedt L, Zetterberg H, et al. Cerebrospinal fluid levels of NfM in relation to NfL and pNfH as prognostic markers in amyotrophic lateral sclerosis. Amyotroph Lateral Scler Frontotemporal Degener [Internet]. 2025 Feb;26(1\u0026ndash;2):113\u0026ndash;23. doi:10.1080/21678421.2024.2428930\u003c/li\u003e\n\u003cli\u003eChourpiliadis C, Seitz C, Lovik A, Joyce EE, Pan L, Hu Y, et al. Lifestyle and medical conditions in relation to ALS risk and progression-an introduction to the Swedish ALSrisc Study. J Neurol [Internet]. 2024 Aug 1;271(8):5447\u0026ndash;59. doi:10.1007/S00415-024-12496-W\u003c/li\u003e\n\u003cli\u003eCedarbaum JM, Stambler N, Malta E, Fuller C, Hilt D, Thurmond B, et al. The ALSFRS-R: A revised ALS functional rating scale that incorporates assessments of respiratory function. J Neurol Sci [Internet]. 1999 Oct 31;169(1\u0026ndash;2):13\u0026ndash;21. doi:10.1016/S0022-510X(99)00210-5\u003c/li\u003e\n\u003cli\u003eNiven E, Newton J, Foley J, Colville S, Swingler R, Chandran S, et al. Validation of the Edinburgh Cognitive and Behavioural Amyotrophic Lateral Sclerosis Screen (ECAS): A cognitive tool for motor disorders. http://dx.doi.org/103109/2167842120151030430 [Internet]. 2015 Jun 1;16(3\u0026ndash;4):172\u0026ndash;9. doi:10.3109/21678421.2015.1030430\u003c/li\u003e\n\u003cli\u003eFoucher J, Winroth I, Lovik A, Sennf\u0026auml;lt S, Pereira JB, Fang F, et al. Validity and reliability measures of the Swedish Karolinska version of the Edinburgh Cognitive and Behavioral ALS Screen (SK-ECAS). Amyotroph Lateral Scler Frontotemporal Degener [Internet]. 2023;24(7\u0026ndash;8):713\u0026ndash;8. doi:10.1080/21678421.2023.2239857\u003c/li\u003e\n\u003cli\u003eBergstr\u0026ouml;m S, Remnest\u0026aring;l J, Yousef J, Olofsson J, Markaki I, Carvalho S, et al. Multi-cohort profiling reveals elevated CSF levels of brain-enriched proteins in Alzheimer\u0026rsquo;s disease. Ann Clin Transl Neurol [Internet]. 2021 Jul 1;8(7):1456\u0026ndash;70. doi:10.1002/ACN3.51402\u003c/li\u003e\n\u003cli\u003eRemnest\u0026aring;l J, Bergstr\u0026ouml;m S, Olofsson J, Sj\u0026ouml;stedt E, Uhl\u0026eacute;n M, Blennow K, et al. Association of CSF proteins with tau and amyloid \u0026beta; levels in asymptomatic 70-year-olds. Alzheimers Res Ther [Internet]. 2021 Dec 1;13(1). doi:10.1186/S13195-021-00789-5\u003c/li\u003e\n\u003cli\u003eBergstr\u0026ouml;m S, Mravinacov\u0026aacute; S, Lindberg O, Zettergren A, Westman E, Wahlund L-O, et al. CSF levels of brain-derived proteins correlate with brain ventricular volume in cognitively healthy 70-year-olds. Clin Proteomics [Internet]. 2024 Dec 12;21(1):65. doi:10.1186/s12014-024-09517-1\u003c/li\u003e\n\u003cli\u003eRemnest\u0026aring;l J, \u0026Ouml;ijerstedt L, Ullgren A, Olofsson J, Bergstr\u0026ouml;m S, Kultima K, et al. Altered levels of CSF proteins in patients with FTD, presymptomatic mutation carriers and non-carriers. Transl Neurodegener. 2020 Jun 23;9(1). doi:10.1186/S40035-020-00198-Y\u003c/li\u003e\n\u003cli\u003eRemnest\u0026aring;l J, Just D, Mitsios N, Fredolini C, Mulder J, Schwenk JM, et al. CSF profiling of the human brain enriched proteome reveals associations of neuromodulin and neurogranin to Alzheimer\u0026rsquo;s disease. Proteomics Clin Appl [Internet]. 2016 Dec 1;10(12):1242\u0026ndash;53. doi:10.1002/PRCA.201500150\u003c/li\u003e\n\u003cli\u003eUllgren A, \u0026Ouml;ijerstedt L, Olofsson J, Bergstr\u0026ouml;m S, Remnest\u0026aring;l J, van Swieten JC, et al. Altered plasma protein profiles in genetic FTD \u0026ndash; a GENFI study. Mol Neurodegener. 2023 Dec 1;18(1). doi:10.1186/s13024-023-00677-6\u003c/li\u003e\n\u003cli\u003eBergstr\u0026ouml;m S, \u0026Ouml;ijerstedt L, Remnest\u0026aring;l J, Olofsson J, Ullgren A, Seelaar H, et al. A panel of CSF proteins separates genetic frontotemporal dementia from presymptomatic mutation carriers: a GENFI study. Mol Neurodegener. 2021 Dec 1;16(1). doi:10.1186/S13024-021-00499-4\u003c/li\u003e\n\u003cli\u003eAndersson A, Remnest\u0026aring;l J, Nellg\u0026aring;rd B, Vunk H, Kotol D, Edfors F, et al. Development of parallel reaction monitoring assays for cerebrospinal fluid proteins associated with Alzheimer\u0026rsquo;s disease. Clin Chim Acta [Internet]. 2019 Jul 1;494:79\u0026ndash;93. doi:10.1016/J.CCA.2019.03.243\u003c/li\u003e\n\u003cli\u003ePin E, Sj\u0026ouml;berg R, Andersson E, Hellstr\u0026ouml;m C, Olofsson J, Jernbom Falk A, et al. Array-based profiling of proteins and autoantibody repertoires in CSF. In: Methods in Molecular Biology [Internet]. Methods Mol Biol; 2019. p. 303\u0026ndash;18. doi:10.1007/978-1-4939-9706-0_19\u003c/li\u003e\n\u003cli\u003eZou H, Hastie T. Regularization and Variable Selection Via the Elastic Net. J R Stat Soc Series B Stat Methodol [Internet]. 2005 Apr 1;67(2):301\u0026ndash;20. doi:10.1111/J.1467-9868.2005.00503.X\u003c/li\u003e\n\u003cli\u003eShoji M, Matsubara E, Kanai M, Watanabe M, Nakamura T, Tomidokoro Y, et al. Combination assay of CSF Tau, A\u0026beta;1-40 and A\u0026beta;1-42(43) as a biochemical marker of Alzheimer\u0026rsquo;s disease. J Neurol Sci. 1998 Jun 30;158(2):134\u0026ndash;40. doi:10.1016/S0022-510X(98)00122-1\u003c/li\u003e\n\u003cli\u003eLewczuk P, Esselmann H, Otto M, Maler JM, Henkel AW, Henkel MK, et al. Neurochemical diagnosis of Alzheimer\u0026rsquo;s dementia by CSF A\u0026beta;42, A\u0026beta;42/A\u0026beta;40 ratio and total tau. Neurobiol Aging. 2004;25(3):273\u0026ndash;81. doi:10.1016/S0197-4580(03)00086-1\u003c/li\u003e\n\u003cli\u003eHansson O, Lehmann S, Otto M, Zetterberg H, Lewczuk P. Advantages and disadvantages of the use of the CSF Amyloid \u0026beta; (A\u0026beta;) 42/40 ratio in the diagnosis of Alzheimer\u0026rsquo;s Disease. Alzheimers Res Ther [Internet]. 2019 Apr 22;11(1):1\u0026ndash;15. doi:10.1186/S13195-019-0485-0\u003c/li\u003e\n\u003cli\u003eFranzmeier N, Dehsarvi A, Steward A, Biel D, Dewenter A, Roemer SN, et al. Elevated CSF GAP-43 is associated with accelerated tau accumulation and spread in Alzheimer\u0026rsquo;s disease. Nat Commun [Internet]. 2024 Dec 1;15(1). doi:10.1038/S41467-023-44374-W\u003c/li\u003e\n\u003cli\u003e\u0026Ouml;hrfelt A, Benedet AL, Ashton NJ, Kvartsberg H, Vandijck M, Weiner MW, et al. Association of CSF GAP-43 With the Rate of Cognitive Decline and Progression to Dementia in Amyloid-Positive Individuals. Neurology [Internet]. 2023 Jan 17;100(3):E275\u0026ndash;85. doi:10.1212/WNL.0000000000201417\u003c/li\u003e\n\u003cli\u003eShi D, Ye C, Li A, Sun P, Lan G, Zhang L, et al. Presynaptic loss and axonal degeneration synergistically correlate with longitudinal neurodegeneration and cognitive decline. Alzheimers Dement [Internet]. 2025 Jun 11;21(6). doi:10.1002/ALZ.70080\u003c/li\u003e\n\u003cli\u003eFinsel J, Winroth I, Ciećwierska K, Helczyk O, Stenberg EA, H\u0026auml;ggstr\u0026ouml;m AC, et al. Determining impairment in the Swedish, Polish and German ECAS: the importance of adjusting for age and education. Amyotroph Lateral Scler Frontotemporal Degener [Internet]. 2023;24(5\u0026ndash;6):475\u0026ndash;84. doi:10.1080/21678421.2023.2192248\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":true,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"alzheimers-research-and-therapy","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"azrt","sideBox":"Learn more about [Alzheimer's Research and Therapy](http://alzres.biomedcentral.com/)","snPcode":"13195","submissionUrl":"https://submission.nature.com/new-submission/13195/3","title":"Alzheimer's Research \u0026 Therapy","twitterHandle":"@AlzheimersRes","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"em","reportingPortfolio":"BMC/SO AJ","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"amyotrophic lateral sclerosis, cognitive impairment, ECAS, proteomics, ratios","lastPublishedDoi":"10.21203/rs.3.rs-7913800/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-7913800/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003e\u003cstrong\u003eBackground\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eCognitive impairment is a recognised feature of neurodegenerative diseases, including amyotrophic lateral sclerosis (ALS). Despite advances in understanding cognitive impairment in ALS, no fluid biomarkers reliably predict these changes. Prior research in Alzheimer disease (AD) has demonstrated that CSF protein ratios enhance biomarker accuracy by mitigating inter-individual variability, improving diagnostic precision. Specifically, studies in AD have identified protein pairs reflecting key pathological processes, including synaptic dysfunction.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eMethods\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eBuilding on findings from the AD field, we analysed 47 CSF proteins, suggested to be associated to neurodegeneration, in 66 patients with ALS and explored protein ratios to evaluate their utility in detecting cognitive impairment, hypothesising shared mechanisms between neurodegenerative diseases. Elastic net regression identified the most predictive protein pairs associated with cognitive impairment, assessed with the Edinburgh Cognitive and Behavioural ALS Screen (ECAS).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eResults\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eWe identified seven single proteins and eight protein pairs associated with cognitive impairment in ALS. The selected protein pairs showed stronger associations with ECAS scores compared to the individual proteins, indicating an enhanced ability to capture cognitive changes. Several of the proteins in the most predictive pairs have previously been implicated to associate to cognitive impairment in AD.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConclusion\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eOur findings indicate that protein ratios outperform single-protein analyses in detecting associations with cognitive impairment, aligning with advancements in AD research. By extending the concept of CSF protein ratios from AD to ALS, this study highlights shared pathological mechanisms and suggests that similar proteins are linked to cognitive dysfunction in both diseases.\u0026nbsp;\u003c/p\u003e","manuscriptTitle":"Ratios of CSF Proteins Reflect Cognitive Function in ALS","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-11-13 11:09:23","doi":"10.21203/rs.3.rs-7913800/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"decision","content":"Revision requested","date":"2025-12-17T23:09:24+00:00","index":"","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-12-17T18:53:26+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-12-10T02:46:17+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"163787857024678805070670807468577692427","date":"2025-11-24T08:56:39+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"44444672086646140615679924323266253583","date":"2025-11-18T06:07:59+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2025-11-03T21:55:11+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2025-10-27T11:06:03+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2025-10-27T11:03:36+00:00","index":"","fulltext":""},{"type":"submitted","content":"Alzheimer's Research \u0026 Therapy","date":"2025-10-21T11:03:00+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"alzheimers-research-and-therapy","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"azrt","sideBox":"Learn more about [Alzheimer's Research and Therapy](http://alzres.biomedcentral.com/)","snPcode":"13195","submissionUrl":"https://submission.nature.com/new-submission/13195/3","title":"Alzheimer's Research \u0026 Therapy","twitterHandle":"@AlzheimersRes","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"em","reportingPortfolio":"BMC/SO AJ","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"e6cd59f3-7b23-44d4-bafb-b535056fd4e6","owner":[],"postedDate":"November 13th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"published-in-journal","subjectAreas":[],"tags":[],"updatedAt":"2026-02-02T16:07:02+00:00","versionOfRecord":{"articleIdentity":"rs-7913800","link":"https://doi.org/10.1186/s13195-026-01976-y","journal":{"identity":"alzheimers-research-and-therapy","isVorOnly":false,"title":"Alzheimer's Research \u0026 Therapy"},"publishedOn":"2026-01-31 15:58:29","publishedOnDateReadable":"January 31st, 2026"},"versionCreatedAt":"2025-11-13 11:09:23","video":"","vorDoi":"10.1186/s13195-026-01976-y","vorDoiUrl":"https://doi.org/10.1186/s13195-026-01976-y","workflowStages":[]},"version":"v1","identity":"rs-7913800","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-7913800","identity":"rs-7913800","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}