Cervantes González et al. Molecular Neurodegeneration (2022) 17:29 https://doi.org/10.1186/s13024-022-00534-y RESEARCH ARTICLE Open Access Multimarker synaptic protein cerebrospinal fluid panels reflect TDP‑43 pathology and cognitive performance in a pathological cohort of frontotemporal lobar degeneration Alba Cervantes González1,2,3, David J. Irwin4, Daniel Alcolea1,2,3, Corey T. McMillan4, Alice Chen‑Plotkin5, David Wolk5, Sònia Sirisi1,2,3, Oriol Dols‑Icardo1,2,3, Marta Querol‑Vilaseca1,2,3, Ignacio Illán‑Gala1,2,3, Miguel Angel Santos‑Santos1,2,3, Juan Fortea1,2,3, Edward B. Lee6, John Q. Trojanowski6, Murray Grossman4, Alberto Lleó1,2,3 and Olivia Belbin1,2,3* Abstract Background: Synapse degeneration is an early event in pathological frontotemporal lobar degeneration (FTLD). Consequently, a surrogate marker of synapse loss could be used to monitor early pathologic changes in patients with underlying FTLD. The aim of this study was to evaluate the relationship of antemortem cerebrospinal fluid (CSF) levels of 8 synaptic proteins with postmortem global tau and TDP-43 burden and cognitive performance and to assess their diagnostic capacity in a neuropathological FTLD cohort. Methods: We included patients with a neuropathological confirmation of FTLD-Tau (n = 24, mean age-at-CSF 67 years ± 11), FTLD-TDP (n = 25, 66 years ± 9) or AD (n = 25, 73 years ± 6) as well as cognitively normal controls (n = 35, 69 years ± 7) from the Penn FTD Center and ADRC. We used a semi-quantitative measure of tau and TDP-43 inclusions to quantify pathological burden across 16 brain regions. Statistical methods included Spearman rank corre‑ lations, one-way analysis of covariance, ordinal regression, step-wise multiple linear regression and receiver-operating characteristic curves. Result: CSF calsyntenin-1 and neurexin-2a were correlated in all patient groups (rs = .55 to .88). In FTLD-TDP, we observed low antemortem CSF levels of calsyntenin-1 and neurexin-2a compared to AD (.72-fold, p = .001, .77-fold, p = .04, respectively) and controls (.80-fold, p = .02, .78-fold, p = .02, respectively), which were inversely associated with post-mortem global TDP-43 burden (regression r2 = .56, p = .007 and r2 = .57, p = .006, respectively). A multimarker panel including calsyntenin-1 was associated with TDP-43 burden (r2 = .69, p = .003) and MMSE score (r2 = .19, p = .03) in FTLD. A second multimarker synaptic panel, also including calsyntenin-1, was associated with MMSE score in FTLD- tau (r2 = .49, p = .04) and improved diagnostic performance to discriminate FTLD-Tau and FTLD-TDP neuropathologic subtypes (AUC = .83). *Correspondence: obelbin@santpau.cat 3 Memory Unit and Biomedical Research Institute, IIB Sant Pau, c/Sant Quintí 77, 08041 Barcelona, Spain Full list of author information is available at the end of the article John Q Trojanowski is deceased © The Author(s) 2022. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://​creat​iveco​mmons.​org/​licen​ses/​by/4.​0/. The Creative Commons Public Domain Dedication waiver (http://​creat​iveco​ mmons.​org/​publi​cdoma​in/​zero/1.​0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data. Cervantes González et al. Molecular Neurodegeneration (2022) 17:29 Page 2 of 12 Conclusion: These synaptic panels have potential in the differential diagnosis of FTLD neuropathologic subtypes and as surrogate markers of cognitive performance in future clinical trials targeting TDP-43 or tau. Keywords: Calsyntenin-1, Frontotemporal lobar degeneration, Frontotemporal dementia, Cerebrospinal fluid, Biomarker, TDP-43, Tau Background trafficking [16, 17], synaptic vesicle exocytosis [18–21] A prominent feature of frontotemporal lobar degen- and synaptic plasticity [22–25]. eration (FTLD) is a region-specific degeneration and loss of synapses [1–4], which correlates with cogni- Methods tive impairment [1]. Characterization of the changes in Study design quantity and function of synaptic proteins and the rela- This is a single-center cross-sectional study. Antemortem tionship to pathologic abnormalities and clinical symp- CSF samples obtained between 1992 and 2015 were retro- toms is essential to improve our understanding of early spectively selected from samples at the Penn FTD Center FTLD pathogenesis. An objective marker of synapse and Alzheimer’s Disease Research Center (ADRC) at the degeneration that is directly related to FTLD pathol- University of Pennsylvania (Philadelphia, USA) [8, 9]. The ogy would be invaluable for monitoring the progres- study includes patients with FTLD-related syndromes sion of these early pathological changes, which could followed to autopsy with a neuropathological diagnosis aid patient management, clinical trial recruitment and of FTLD-Tau (n = 24) or FTLD-TDP (n = 25) and patients has potential as a measure of drug response in future with a clinical and neuropathological diagnosis of AD clinical trials. Moreover, the clinical diagnosis of FTLD (n = 25). Patients included for study had prospective CSF syndromes is challenged by the variable correspond- collection as part of ongoing observational research pro- ence between the clinical syndrome and underlying grams at The Penn FTD Center aimed at developing bio- neuropathological changes and there is a need for a logical markers that reflect pathophysiological changes in biomarker that can distinguish neuropathologic sub- FTLD proteinopathies to improve antemortem diagnosis types. FTLD can be classified postmortem into four and prognostication. Patients underwent a Mini-Mental major neuropathological subtypes; FTLD-Tau (hyper- State Examination (MMSE) by neurologists masked to phosphorylated tau inclusions), FTLD-TDP (trans- synaptic biomarker data within 5 months of the lumbar active response DNA-binding protein-43; TDP-43), puncture. We also analyzed CSF samples from cogni- FTLD-FET (fused in sarcoma, Ewing’s sarcoma, TATA- tively normal controls (n = 35) that were recruited from binding protein associated factor 15) and FTLD-UPS the community and screened by revising their medical (ubiquitin and p62) [2–4]. Here, we have focused on and medication history. Controls self-reported a nega- FTLD-Tau and FTLD-TDP, which represent 90–95% of tive neurological and psychiatric history and were within FTLD cases. the normal range of cognitive performance (Mini-Mental The aim of this study is to a) compare the cerebro- State Examination > 27). CSF samples from patients and spinal fluid (CSF) profile of a panel of synaptic proteins controls were processed in the same way. Pre-analytical in a neuropathological cohort of confirmed FTLD-Tau processing details for these CSF samples can be found and FTLD-TDP, b) evaluate the relationship of ante- elsewhere [8]. mortem CSF levels with postmortem Tau and TDP-43 burden and cognitive performance and c) determine Neuropathological classification and quantification their diagnostic accuracy. As reference groups we of neuropathological burden include cognitively normal controls where synapse Neuropathological diagnosis was established follow- degeneration is absent and Alzheimer’s disease (AD) ing previously described methods and international where synapse degeneration is widespread. The pro- published criteria [9–12]. Patients with a primary neu- teins included in the study (calsyntenin-1, glua2, glua4, ropathological diagnosis of Pick’s disease (n = 3), corti- Neurexin-2a, Neurexin-3a, neuroligin-2, syntaxin-1b, cobasal degeneration (n = 7), progressive supranuclear thy-1 and VAMP-2) were brought forward from our palsy (n = 9), argyrophilic grain disease (n = 4) or non- previous study in clinical AD cohorts [5]. These pro- classifiable non-AD tauopathies (n = 1) were classi- teins are enriched at the synapse [9] and are involved fied as FTLD-Tau (4-repeat tauopathy; n = 20, 3-repeat in a range of synaptic functions including pre-synaptic tauopathy; n = 3, unclassified tauopathy; n = 1). Patients differentiation [6], dendritic spine assembly [11–14], with TDP-43 inclusions were classified as FTLD-TDP postsynaptic Ca2 + signalling [7], AMPA receptor (subtype A; n = 1, subtype B; n = 11, subtype C; n = 8, Cervantes González et al. Molecular Neurodegeneration (2022) 17:29 Page 3 of 12 non-specified; n = 5). The majority of patients with labeled with 13C6 15N4 (Arg) or 13C6 15N2 (Lys) isotopes FTLD-Tau (n = 20/24) and FTLD-TDP (n = 18/25) had a (Pepotech SRM custom peptides, grade 2 > 70% purity, neurofibrillary tangle score of B0 or B1 in the National Thermo Fisher Scientific) corresponding to the synaptic Institute on Aging-Alzheimer’s Association (NIA-AA) proteins and selected from our previously reported shot- classification [26] and therefore had no evidence of sig- gun mass spectromic analysis of CSF [9]). We spiked the nificant AD comorbidity. All patients with a neuro- isotopically-labeled peptides into each sample as internal pathological diagnosis of AD had scores of B2 or B3 in standards. The peptide sequence of the internal stand- the NIA-AA classification. The majority of AD patients ards is reported in [13]. Samples (5 µl) were analyzed showed no TDP-43 pathology beyond the amygdala in a randomized order over a 120-min gradient (0–35% and therefore had no evidence of significant TDP-43 ACN + 0.1% FA) in SRM mode using a triple quadrupole- comorbidity. Qtrap mass spectrometer (5500 QTrap, Sciex, Masachus- Tau and TDP-43 pathology was detected using well- setts) coupled to a nano-LC chromatography column characterized antibodies (i.e. anti-phosphorylated tau, (300 µl/min, 25-cm C18 column, 75 µm I.d., 2 µm par- mouse monoclonal PHF1, gift of Professor Peter Davies, ticle size). We visualized and analyzed transitions using Albert Einstein College of Medicine, New York, NY, USA Skyline 3.5 as previously described [9]. To evaluate the and anti-phosphorylated 409/410 TDP-43, rat monoclo- precision of the internal standards and the stability of the nal TAR5P-1D3, gift of Manuela Neumann and Elisa- peptides over the course of the experiment, we injected beth Kremmer) and established immunohistochemical a pool of all the samples over the duration of the mass methods and neuropathological criteria for scoring as spectrometric measurements and monitored the peak described previously [10]. Postmortem tau and TDP-43 area of the labeled and endogenous peptides. The % coef- burden were rated on a standard ordinal score (0 = none, ficient of variation in light/heavy ratio for each protein 0.5 = rare, 1 = mild, 2 = moderate, 3 = severe) in 16 brain was as follows: calsyntenin-1; 4%, gluA2; 60%, gluA4; regions (amygdala, hippocampus (CA1), entorhinal cor- 9%, neuroligin-2; 12%, neurexin-2a; 7%, neurexin-3a; 6%, tex, mid-frontal gyrus, angular gyrus, superior-middle syntaxin-1B; 25%, thy-1; 10%, VAMP-2; 26%. Except for temporal gyrus, cingulate gyrus, caudate/putamen, glo- GluA2, all proteins gave %CV < 25%. GluA2 was removed bus pallidus, thalamus, midbrain, substantia nigra, locus from further analyses. coeruleus, pons, cerebellum, medulla) as previously Peak areas for the endogenous (light) and isotopically- described according to established neuropathological labeled (heavy) transitions were extracted and used as criteria [13]. Cumulative burden across regions (Global input for processing. Using the dataProcess function of tau and Global TDP) was obtained as the sum of these MSstats v3.5 package in R [14], light and heavy transi- values. tions were log-2 transformed and normalized using the Targeted liquid chromatography mass spectrometry. EqualizeMedians function. The light/heavy intensity was We monitored a set of 22 proteotypic peptides corre- calculated for all transitions and summarized using the sponding to 9 proteins (Calsyntenin-1, GluA2, GluA4, Tukey’s median polish function. We removed transitions Neurexin-2a, Neurexin-3a, Neuroligin-2, Syntaxin-1B, with between-run interference (between RunInterferenc- Thy-1 and VAMP-2) using the previously described eScore < 0.8). All remaining transitions passed the mini- Selected Reaction Monitoring (SRM) method [9]. Briefly, mum log2 intensity cut-off designated by the MSstats we precipitated the individual CSF samples (100 μl) with package (7.724). We did not impute missing transition cold acetone (-20 ºC overnight) and resuspended in 6 M data. One of the three calsyntenin-1 peptides was not Urea/ 200 mM ammonium bicarbonate. We reduced detected in any samples and was excluded from further the samples with 10 mM dithiothreitol followed by in- analyses. The final number of peptides corresponding to solution digestion with LysC (1:10 ratio enzyme:protein, each protein was as follows: calsyntenin-1 (2), gluA4 (3), w:w) overnight at 37 °C (650 rpm), followed by an 8 h neuroligin-2 (2), neurexin-2A (3), neurexin-2A (3), syn- incubation with trypsin (1:10 ratio enzyme:protein, w:w). taxin-1B (1), thy-1 (3), vamp-2 (1). The digestion reaction was stopped by adding formic acid (10% total volume). We digested 10 µg of E. Coli Statistical analysis protein extract in triplicate in parallel with the sam- Statistical analyses were performed in R version 4.1.2 ples to evaluate digestion efficiency; 78.2% of the pep- [15]. Outliers were excluded using the 3 × IQR rule. tide groups showed complete cleavage, 18.7% showed 1 Where residuals deviated from a Gaussian distribution missed cleavage and 3.1% showed more than 1% missed (Shapiro–Wilk p < 0.05), we used square root transformed cleavage. Missed cleavages were consistent across trip- values, which did not deviate from a Gaussian distribu- licates. CSF samples were purified using off-line reverse tion. Group differences were compared using χ2 test for phase purification. We synthesized custom peptides categorical variables or one-way analysis of covariance Cervantes González et al. Molecular Neurodegeneration (2022) 17:29 Page 4 of 12 (ANCOVA) for continuous variables including sex and analysis and death was longer in the AD group compared age-at-CSF as covariates with post hoc Tukey method. A to FTLD-Tau (1.4 year difference, post-hoc p = 0.04) and Kruskal–Wallis test was used to compare Global Tau and FTLD-TDP (2.4 year difference, post-hoc p = 0.01). Based TDP-43 burden across male and female participants. We on these differences, we included sex and age-at-CSF col- used Spearman rank correlations to determine the rela- lection as a covariate in statistical analysis for group com- tionship between antemortem synaptic protein levels in parisons. Eighty-three percent of the FTLD-Tau and 76% the CSF and demographic variables. We performed ordi- of FTLD-TDP showed little to no AD pathology and 72% nal regression to test for association of synaptic proteins of the AD patients showed no TDP-43 pathology beyond with Tau and TDP-43 scores in individual brain regions, the amygdala, a proportion that was comparable across including age-at-death and time from CSF to autopsy as groups (p = 0.63). The global pathological burden of tau covariates. We performed a mixed entry backward step- (p = 0.66) and TDP-43 (p = 0.31) pathology was com- wise linear regression (stepAIC function) to identify the parable between male and female participants. Global best possible predictors of postmortem Global tau and pathological burden was not associated with the time Global TDP-43 (sum of scores across all regions). Age-at- interval from CSF collection to death (tau; p = 0.12, TDP- death and time from CSF to death were forced into the 43; p = 0.21) or disease duration (tau; p = 0.08, TDP-43; model and test variables included all 8 synaptic proteins, p = 0.19) but was inversely related to age-at-death (tau; biological sex, and AD comorbidity. Akaike Informa- p = 0.04, TDP-43; p = 0.02). The mean MMSE score and tion criteria and variant inflation factors were used to set time interval between MMSE and CSF collection were a limit on the total number of variables included in the comparable between FTLD-Tau and FTLD-TDP (both final model. The same approach was taken for regressions p > 0.29). using MMSE as the outcome measure, with the modifica- tion that age-at-CSF and years of education were forced Correlation between antemortem CSF levels into the model. All reported r2 values were adjusted for of the synaptic proteins the number of predictors in the model. All ANCOVA, We first sought to compare the pair-wise correlation Spearman and regression p-values were adjusted for between antemortem CSF levels of the synaptic pro- multiple testing of 8 synaptic proteins using the Ben- teins in each neuropathologic group (Fig. 1). A set of 3 jamini–Hochberg method. We assessed the diagnostic proteins (calsyntenin-1, neurexin-2a and thy-1) showed utility (area-under-the-curve; AUC) of CSF levels of the pair-wise correlation in all groups (FTLD-Tau; rs = 0.79- synaptic proteins and multimarker panels using receiver 0.87, p < 0.0001, FTLD-TDP; rs = 0.70-0.88, p < 0.009, operating characteristic (ROC) curves in the pROC pack- AD; rs = 0.55-0.78, p < 0.007, Controls; rs = 0.68-0.81, age implemented in R. We used the ‘DeLong’ method to p < 0.0001), while all other protein combinations were compare ROC curves [16]. group-dependent. Results Antemortem CSF profile of the synaptic proteins Demographics and clinical data across neuropathologic groups Table 1 shows the demographic, clinical and neuro- We next compared the CSF profile of the synaptic pro- pathological data for the samples included in the study teins across neuropathologic groups, including sex and grouped according to their final neuropathological diag- age-at-CSF as covariates (Fig. 2). Of the 8 synaptic pro- nosis. Forty-five percent of the study participants were teins, calsyntenin-1 showed the strongest association female. The FTLD-Tau group had a lower proportion with neuropathologic group (F = 6.36, p = 0.004). In of females (25%) compared to controls (57%, p = 0.03), FTLD-TDP, we observed low CSF calsyntenin-1 com- whereas the proportion was comparable across all other pared to AD (0.72-fold, p = 0.001) and controls (0.80- groups (p > 0.05). There was no difference in years of edu- fold, p = 0.02). In FTLD-Tau, we observed low CSF cation across groups (p = 0.40). The mean age-at-disease calsyntenin-1 compared to AD (0.79-fold, p = 0.03) but onset (6 year difference, post-hoc p = 0.02) and mean not controls (0.88-fold, p = 0.24). Calsyntenin-1 was age-at-CSF analysis (7 year difference, post-hoc p = 0.04) comparable between FTLD neuropathological subtypes were higher in the AD group compared to FTLD-TDP. (p = 0.79) and between AD and controls (p = 0.60). The mean age-at-death was higher in AD compared to Neurexin-2a (F = 4.60, p = 0.01), thy-1 (F = 3.79, FTLD-Tau (8 year difference, post-hoc p = 0.03) and p = 0.03), neurexin-3a (F = 3.53, p = 0.03) and VAMP-2 FTLD-TDP (9 year difference, post-hoc p = 0.006). There (F = 4.58, p = 0.01) were also associated with neuro- was no difference in mean disease duration (p = 0.07) or pathologic group. Neurexin-2a was lower in FTLD- time interval between disease-onset and CSF analysis TDP compared to AD (0.77-fold, p = 0.04) and controls (p = 0.79), whereas the mean time interval between CSF (0.78-fold, p = 0.02). Thy-1 was lower in FTLD-TDP Cervantes González et al. Molecular Neurodegeneration (2022) 17:29 Page 5 of 12 Fig. 1 Pair-wise correlation of antemortem CSF levels of 8 synaptic proteins in FTLD, AD and controls. Pair-wise r­ s coefficients resulting from statistical tests performed in (a) FTLD-Tau, (b) FTLD-TDP, (c) AD and (d) controls are shown. Degree of shading is relative to size of Spearman r­ s coefficients, which are shown in bold where p < .05. A set of 3 proteins that correlated in all groups are highlighted by a black box compared to AD (0.73-fold, p = 0.04). Neurexin-3a or duration from CSF analysis to autopsy (all p > 0.06) was lower in FTLD-Tau compared to AD (0.66-fold, in any neuropathologic group. p = 0.01). VAMP-2 was elevated in AD compared to controls (1.6-fold, p = 0.005) and FTLD-Tau (1.43- Association of antemortem CSF levels of the synaptic fold, p = 0.03). The association of CSF calsyntenin-1, proteins with post‑mortem pathological burden in FTLD neurexin-2a and VAMP-2 with patient group remained In FTLD-Tau, we observed no association between when sex and age-at-CSF were excluded from the antemortem CSF levels of the synaptic proteins with model (p = 0.009, p = 0.02, p = 0.02, respectively). age-at-death (all p > 0.11). In FTLD-TDP, neurexin-3a To determine the influence of AD comorbidity, we was inversely associated with age-at-death (rs = -0.55, repeated these analyses in the FTLD subgroups without p = 0.04). We observed no association between ante- significant AD pathology (FTLD-Tau; n = 20, FTLD- mortem CSF levels of the 8 synaptic proteins with TDP; n = 19). The principal associations we observed postmortem global tau burden (Additional File 2; all across groups in the total dataset held in this subset p > 0.17). Calsyntenin-1, neurexin-2a, thy-1 and neu- (Additional file 1). We observed no difference in any of roligin-2 were inversely associated with postmortem the synaptic proteins between male and female partici- global TDP-43 burden (Fig. 3). With the exception of pants (all p > 0.21) and no association with age-at-CSF neuroligin-2 and thy-1, biological sex also significantly analysis (all p > 0.10) in any neuropathologic group. The contributed to the associations (all p = 0.03). Moreo- synaptic proteins were not associated with age-at-dis- ver, the inclusion of covariates explained global post- ease onset (all p > 0.07), duration from disease onset to mortem TDP-43 burden better than simple linear CSF analysis (all p > 0.15), disease duration (all p > 0.15) regression models for each synaptic protein alone (all ANOVA p < 0.003). In a secondary analysis, we Cervantes González et al. Molecular Neurodegeneration (2022) 17:29 Page 6 of 12 Fig. 2 Antemortem CSF SRM intensities of the synaptic panel across patient groups. Violin plots show the distribution of SRM intensities for each synaptic protein quantified in CSF for patients with neuropathological confirmation of FTLD-Tau (Tau), FTLD-TDP (TDP) or AD and cognitively normal subjects (Control). Solid horizontal lines show the mean value for each and boxes represent the standard error of the mean. The horizontal dotted line represents the mean value in the control group for each protein. Summary statistics for ANCOVA including sex and age-at-CSF performed on square root transformed data are shown at the top of each plot. ANCOVA p-values are adjusted for multiple testing (9 proteins). *p < .05, **p < .01 for Dunnet’s post-hoc tests observed a stronger association of calsyntenin-1 when p = 0.01). This may be due to a ceiling effect in regions the outcome measure was restricted to regions with with high burden. low to medium TDP-43 burden (brainstem + cerebel- To determine whether a combination of synap- lum + thalamus + basal ganglia; r2 = 0.61; p = 0.03). tic proteins could better explain postmortem tau and We also observed an association for neurexin-3a in TDP-43 burden, we performed stepwise mixed entry regions with low to medium TDP-43 burden (r2 = 0.69; linear regression starting with the synaptic proteins Cervantes González et al. Molecular Neurodegeneration (2022) 17:29 Page 7 of 12 Fig. 3 Association of antemortem CSF levels of synaptic proteins with postmortem TDP-43 burden. Scatter plots show the antemortem SRM intensities for selected synaptic proteins quantified in CSF and post-mortem TDP-43 burden for patients with neuropathological confirmation of FTLD. Linear regression lines and standard error (shaded region) are shown for each plot. Adj.r2, p-values and t-statistics for the synaptic protein (syn t), time from CSF to autopsy (aut t), age-at-death (age t) and biological sex (sex t) are shown for each linear regression. P-values were adjusted for multiple testing (9 proteins) and covariates. No multimarker model was associ- the significant contributors to this model (henceforth ated with global Tau (p > 0.13). On the other hand, a 6 named Panel A[CLSTN1-VAMP2]). When controlling variable model including calsyntenin-1, neurexin-3a, for the other variables, lower age-at-death (p = 0.01), VAMP-2 and all 3 covariates was associated with global lower CSF calsyntenin-1 (p = 0.0006), elevated CSF TDP-43 burden (Table 2a; r2 = 0.69, p = 0.003). Age- VAMP-2(p = 0.02) and being female (p = 0.01) were at-death, biological sex, calsyntenin-1, VAMP-2 were associated with elevated global TDP-43 protein. Cervantes González et al. Molecular Neurodegeneration (2022) 17:29 Page 8 of 12 Table 1 Demographic, clinical and neuropathological data for the samples included in the study N FTLD-Tau FTLD-TDP AD Control p-value 24 25 25 35 Female, n (%) 6 (25) 13 (52) 10 (40) 20 (57) .08 Education (years) 15.6 (3.5) 15.1 (2.7) 14.8 (3.1) 15.7 (3.4) .40 Age-at-disease onset (years) 64 (9.3) 63 (9.0) 69 (5.6) NA .02a Age-at-CSF analysis (years) 67 (11.3) 66 (8.7) 73 (5.7) 69 (7.3) .04b Age-at-death (years) 71 (12.6) 70 (9.5) 79 (6.0) NA .005c Time interval, onset–CSF (years) 3.7 (2.7) 3.7 (2.7) 3.6 (1.6) NA .79 Time interval, CSF–death (years) 4.6 (3.6) 3.6 (3.0) 6.0 (2.3) NA .01d Total disease duration, onset–death (years) 8.4 (4.5) 7.2 (3.4) 9.6 (3.0) NA .07 Participants without comorbidity, n (%) 20 (83) 19 (76) 18 (72) NA .63 Global Tau pathology 29.3 (12.0) 10.0 (8.0) NA NA < .0001 Global TDP pathology 9.7 (10.1) 26.1 (6.8) NA NA .0002 MMSE score 24.0 (5.2) 20.9 (6.8) NA NA .29 Time interval, CSF–MMSE (months) 0.7 (1.2) 0.7 (0.8) NA NA .84 Unless otherwise specified, the results are expressed as mean (SD). Global tau and TDP pathology refers to sum of ordinal values across 16 brain regions. Comorbidity refers to FTLD without incidental AD pathology (NIA-AA stage ≤ B1) and AD without TDP-43 pathology beyond the amygdala. P-values were obtained from χ2 test, one-way ANOVA with post-hoc Tukey or Kruskal–Wallis rank-sum tests with post-hoc Dunn’s test. Post-hoc tests where adj.p < .05 are as follows a FTLD-TDP vs AD; p = .02 b FTLD-TDP vs AD; p = .04 c FTLD-TDP vs AD; p = .03, FTLD-TDP vs AD; p = .006 d FTLD-TDP vs AD; p = .04, FTLD-TDP vs AD; p = .01 MMSE Mini-mental state examination Additional File 3 plots the relative regression line of stronger association with MMSE in FTLD-Tau (Table 2b; each variable. r2 = 0.49, p = 0.04). Age-at-CSF, calsyntenin-1, VAMP-2 and gluA4 were the significant contributors to the model Association of antemortem CSF levels of the synaptic (henceforth termed Panel B[CLSTN1-VAMP2-GLUA4]). proteins with cognitive performance in FTLD When controlling for the other variables, elevated age We first tested the association of the synaptic proteins (p = 0.03), elevated CSF calsyntenin-1 (p = 0.005), lower from Panel A[CLSTN1-VAMP2] with MMSE score in CSF VAMP-2 (p = 0.01) and lower CSF gluA4 (p = 0.03) FTLD patients including sex, age-at-CSF, and years of were associated with worse MMSE score. Additional education as covariates. The panel was associated with file 5 plots the relative regression line for each variable. MMSE in the total FTLD group (Table 2b; r2 = 0.19, An alternative 6 variable model, including gluA4, neu- p = 0.03) but not in FTLD-Tau (r2 = 0.33, p = 0.09) or roligin-2, neurexin-3a, syntaxin-1b, age-at-CSF and edu- FTLD-TDP (r2 = 0.17, p = 0.13). For comparison, no indi- cation, showed a stronger association with MMSE in vidual synaptic protein was associated with MMSE in FTLD-TDP (Table 2b; r2 = 0.38, p = 0.02) and in all FTLD FTLD (all p > 0.48) or in either neuropathologic subtype (Table 2b; r2 = 0.30, p = 0.002). GluA4, neurexin-3a and (FTLD-Tau; p = 0.35, FTLD-TDP; p > 0.52). Neurexin-3a syntaxin-1b, age-at-CSF and years of education were the and VAMP-2 were the significant contributors to this significant contributors to this model (henceforth termed model. When controlling for the other variables, lower Panel C[GLUA4-NRX3A-STX1B]). When controlling for CSF VAMP-2 (p = 0.02) and elevated CSF neurexin- the other variables, less education (p = 0.02), elevated 3a (p = 0.02) were associated with worse MMSE score. age (p = 0.005), lower CSF gluA4 (p = 0.002), elevated Additional file 4 plots the relative regression line of each neurexin-3a (p = 0.003) and elevated CSF syntaxin-1b variable. (p = 0.03) were associated with worse MMSE score. To determine whether other combinations of synaptic Additional file 6 plots the relative regression line for each proteins could improve the association of Panel A, we variable. performed stepwise linear regression starting with the Thus, several multimarker panels of synaptic proteins synaptic proteins and covariates. Compared to Panel A, are associated with cognitive performance in FTLD. Pan- a 6 variable model including calsyntenin-1, gluA4, syn- els B and C were not associated with postmortem global taxin-1b, VAMP-2, age-at-CSF and education showed a TDP-43 or global Tau burden (all p > 0.1). Table 2 Association of step-wise linear regression models with postmortem TDP-43 burden and MMSE outcome measures and diagnostic accuracy Summary Statistics Forced Entry Test variables 2 Panel Study AIC adj.r p CSF- Age-at-death AD comorbidity Sex Calsyn‑ VAMP-2 Neurexin-3a GluA4 Syntaxin-1b Neuroligin-2 Neurexin-2a Thy-1 group autopsy (a) / Age-at- tenin-1 (a) / CSF (b) Educa‑ tion (b) a) Outcome measure: Global TDP-43 A[CLSTN1-VAMP2] FTLD 119 0.69 0.003 -0.91 -2.91 3.1 -4.8 2.71 -1.28 Cervantes González et al. Molecular Neurodegeneration b) Outcome measure: MMSE A[CLSTN1-VAMP2] FTLD 273 0.19 0.03 1.21 -1.9 -0.36 2.41 -2.43 B[CLSTN1-VAMP2- FTLD-Tau 98 0.49 0.04 -1.79 -2.45 -3.60 3.15 2.57 1.90 GLUA4] FTLD-TDP 174 0.02 0.40 0.93 -1.46 -0.23 0.80 0.91 -0.68 FTLD 279 0.09 0.16 1.46 -1.29 -0.65 1.16 0.99 -1.29 (2022) 17:29 C[GLUA4-NRX3A- FTLD-Tau 131 0.02 0.45 1.49 -0.55 0.22 -1.59 1.28 -0.57 0.65 STX1B] FTLD-TDP 162 0.40 0.02 2.54 -3.21 1.33 -3.42 3.71 -2.4 1.69 FTLD 291 0.30 0.004 2.11 -2.53 1.12 -3.5 3.74 -2.22 1.22 c) Outcome measure: AUC​ FTLD-Tau vs Controls FTLD-Tau vs AD FTLD-TDP vs Controls FTLD-TDP vs AD FTLD-TDP vs FTLD-Tau Model AUC​ 95%CI AUC​ 95%CI AUC​ 95%CI AUC​ 95%CI AUC​ 95%CI A[CLSTN1-VAMP2] 71.1 56–85 80.1 66–92 76.9 64–88 77.2 63–89 72.6 57–86 B[CLSTN1-VAMP2- 71.3 56–85 75.8 60–89 82.3 71–93 81.3 68–92 83.0 70–94 GLUA4] C[GLUA4-NRX3A- 64.6 49–79 79.7 66–91 65.8 52–79 66.7 51–82 70.8 55–84 STX1B] Summary statistics and composition of each step-wise regression model is shown for models associated with postmortem TDP-43 burden and MMSE outcome measures (a-b) and their diagnostic performance in FTLD and FTLD neuropathologic subtypes (c) The model statistics and effect size for each variable included in the final regression model resulting from backward step-wise regression as predictors for the outcome variable are given for models where p < 0.05 Variables that significantly contributed to the model (p < 0.05) are shown in bold AIC Akaike Information criteria Page 9 of 12 Cervantes González et al. Molecular Neurodegeneration (2022) 17:29 Page 10 of 12 Diagnostic value. performance in FTLD-Tau. Thus, when measured in CSF, The synaptic proteins showed modest diagnostic accu- combinations of multiple synaptic proteins were better racy to distinguish FTLD-Tau from controls (highest predictors of global TDP burden and cognitive perfor- AUC = 0.67 for neurexin-2a). The best multimarker panel mance and showed better diagnostic accuracy compared was Panel B[CLSTN1-VAMP2-GLUA4], which showed to individual synaptic proteins. To our knowledge, this is only a nominal improvement (Table 2c; AUC = 0.71) the first study to relate CSF levels of synaptic proteins to compared to neurexin-2a (p = 0.67). This panel was sta- FTLD neuropathological subtypes. tistically comparable to tau CSF biomarkers, pTau, tTau Calsyntenin-1 is abundant in most neurons of the CNS and pTau/tTau ratio (AUC = 0.65–73, all p > 0.53 vs Panel [36] where it has been shown to modulate postsynaptic B). When distinguishing FTLD-Tau from AD, neurexin- calcium signaling and promote dendritic spine assem- 3a gave nominally the highest AUC (0.75, 95% CI 0.60- bly [15]. VAMP-2 and syntaxin-1b are synaptic vesicle 0.88). Panel A[CLSTN1-VAMP2] showed a nominal recycling proteins predominantly found at glutamater- improvement (Table 2c; AUC = 0.80) compared to neu- gic synapses [17] as part of the synaptic exocytosis core rexin-3a (p = 0.61) but still performed worse than pTau vesicular complex [18] where they regulate the releasable and the ptau/Aβ42 ratio (AUC = 0.95–98, all p < 0.04 vs pool of glutamate at the pre-synapse [19]. GluA4 is the Panel A). predominant AMPA receptor subunit in PV interneu- Neurexin-2a gave nominally the highest AUC (0.75, rons [44]. The neurexins are a complex family of proteins 95% CI 0.62-0.87) for distinguishing FTLD-TDP from that are generated from three different genes (NRXN1, controls. Panel B[CLSTN1-VAMP2-GLUA4] showed NRXN2, NRXN3) with alternative promoters (α, β, γ) only a nominal improvement (Table 2c; AUC = 0.82) and extensive alternative splicing resulting in over 1000 compared to neurexin-2a (p = 0.41) but performed bet- distinct neurexin isoforms, each associated with spe- ter than CSF tau markers (AUC = 0.62–68, all p < 0.04 cific neuronal cell types [20]. The formation of a trans- vs Panel B). When distinguishing FTLD-TDP from AD, synaptic complex with their post-synaptic partners, the calsyntenin-1 gave nominally the highest AUC (0.76, 95% neuroligins, allows the structural assembly of excitatory CI 0.62-0.89). Panel B showed only a nominal improve- synapses, by triggering the recruitment of postsynap- ment (Table 2c; AUC = 0.81) compared to calsyntenin-1 tic NMDA and AMPA glutamate receptors [21, 22]. The (p = 0.56) and was significantly worse than the pTau/ composition of these panels thus support the growing Aβ42 ratio (AUC = 0.96, p < 0.03 vs Panel B). evidence for involvement of the glutamatergic system in None of the synaptic proteins alone could discriminate FTLD syndromes [23]. FTLD-TDP from FTLD-Tau subtypes (all AUC < 0.67). A previous study has reported an association between However, Panel B performed significantly better low CSF calsyntenin-1 levels in patients with behavioral (Table 2c; AUC = 0.83) than tau markers (0.51–62, all variant FTD (bvFTD) compared to cognitively normal p < 0.05 vs Panel B) and the ptau/aβ42 ratio (AUC = 0.56, controls and compared to presymptomatic carriers of p = 0.009). C9orf72, GRN or VCP mutations [24]. As the pathology underlying clinically defined FTLD syndromes is het- Discussion erogeneous, the results of the previous study cannot be Here we report a comprehensive evaluation of the directly compared to the current study; clinical diagno- antemortem CSF levels of 8 synaptic proteins in a neu- ses of bvFTD is split between FTLD-TDP and FTLD-Tau ropathological cohort of FTLD (FTLD-Tau and FTLD- subtypes and the small number of cases with a bvFTD TDP), AD and cognitively normal controls. CSF levels diagnosis in the current study (n = 10) prohibits mean- of calsyntenin-1 and neurexin-2a were correlated in all ingful comparison across syndromes. Nevertheless, both patient groups and were lower in the FTLD-TDP neuro- studies point towards altered abundance of calsyntenin-1 pathological subtype compared to neuropathologically in CSF from patients clinically diagnosed with FTLD syn- confirmed AD and cognitively normal controls. Further- dromes and those with neuropathologically confirmed more, antemortem CSF levels of calsyntenin-1 and neu- FTLD compared to cognitively normal controls. Neu- rexin-2a inversely correlated with global TDP-43 burden rexins have also been shown to facilitate the calcium- post-mortem. Combining calsyntenin-1 with VAMP-2 dependent release of synaptic vesicles [25, 27, 28]. and neurexin-3a improved this association with TDP-43 Another important finding of this study is that while burden and was also correlated with cognitive perfor- AD neurofibrillary tangle pathology (Braak score < 2 mance in FTLD. Combining calsyntenin-1 with VAMP-2, [31]) was present in 17% of the FTLD-Tau and 24% of the gluA4 and syntaxin-1b showed high accuracy in the dif- FTLD-TDP patients, stepwise regression showed that AD ferential diagnosis of FTLD-TDP and FTLD-Tau neuro- comorbidity was not a good predictor of TDP-43 burden pathologic subtypes and was associated with cognitive and was therefore excluded from the final model. On the Cervantes González et al. Molecular Neurodegeneration (2022) 17:29 Page 11 of 12 other hand, AD comorbidity was a predictor of MMSE Supplementary Information and was included in Panel C. Thus, AD comorbidity may The online version contains supplementary material available at https://​doi.​ influence cognitive performance in FTLD. This is consist- org/​10.​1186/​s13024-​022-​00534-y. ent with a recent study showing that neurodegenerative comorbidities contribute to transition from mild cogni- Additional file 1. tive impairment to dementia [29]. Additional file 2. Additional file 3. Additional file 4. Study limitations Additional file 5. A limitation of this work is the relatively small study size Additional file 6. compared to other CSF biomarker studies. However, CSF samples from well-annotated autopsy cases of these Acknowledgements uncommon conditions are scarce and the neuropatho- We would like to thank all the participants who agreed to take part in this logical confirmation reduces the substantial heteroge- study and all the members of the clinical teams involved in the selection and assessment. We also thank Dr. P. Davies for providing the PHF1 antibody; and neity inherent in clinical FTD cohorts. It should also be Dr. M. Neumann and Dr. E. Kremmer for providing the p409-410 antibody noted that the mean time from CSF to death was 5 years (TAR5P-1D3). in FTLD-Tau and 4 years in the FTLD-TDP subtypes, Authors’ contributions reaching up to 10 years in some cases. This long interval OB and AL contributed to the design and concept of the study, AC-G, DJI, and variability could influence the relationship between DA, CTM, AC-P, DW, SS, OD-I, M Q–V, II-G, MAS-S, JF, EBL, JQT, MG and OB participants’ CSF biochemical signature and their final contributed to the acquisition and analysis of data. AC-G, DA, JF, AL and OB drafted a significant portion of the manuscript and figures. All authors read neuropathological findings. Finally, our control partici- and approved the final manuscript. pants lack neuropathological confirmation. However, complete clinical and neuropsychological evaluations Funding This work was supported by research grants from Institute of Health Carlos were performed to exclude significant medical (and spe- III (ISCIII), Spain (PI18/00327 to OB, AC19/00103 to AL, PI20/01330 to AL and cifically neurological) conditions in these participants. CIBERNED Program 1, Alzheimer Disease to AL, PI17/01019 and PI20/01473 to JF). This work has also been supported by the Department of Health Generali‑ tat de Catalunya the Department of Health, Generalitat de Catalunya AGAUR and PERIS programs (AGAUR 2019 PROD 00088 to OB, SLT002/16/00408 to Conclusions AL) and partly funded by Fondo Europeo de Desarrollo Regional (FEDER), A biomarker that can distinguish FTLD neuropatho- Unión Europea, “Una manera de hacer Europa”. This work was also sup‑ ported by grants from the NIH (P01AG066597,P30AG072979, U19 AG062418, logic subtypes in clinical cohorts would be a valuable 1R01AG056850-01A1, R21AG056974, R01AG061566). OB is contractually addition to the biomarker arsenal. In this regard, Panel supported by the Miguel Servet Type II fellowship (CP18/00011; ISCIII, FEDER). B[CLSTN1-VAMP2-GLUA4] identified in this study The funders of this study played no role in the design of the study, collection, analysis or interpretation of data or writing the manuscript. provides proof-of-concept that these synaptic proteins are differentially altered and can distinguish between the Availability of data and materials two most common neuropathologic subtypes. Moreover, The data analyzed during the current study are available from the correspond‑ ing author on reasonable request. all panels described here were associated with MMSE score and therefore show proof-of-concept as surrogate measures of cognitive performance. These panels may Declarations be relevant for future clinical trials targeting TDP-43 or Ethics approval and consent to participate tau pathology, where there will be a need for a surro- All participants gave written informed consent to participate. The local ethics committee at each center approved the. gate measure of cognitive performance, independent of study. the drug target. In addition, the postsynaptic modulator, calsyntenin-1 both alone and in combination with other Consent for publication Not applicable. synaptic proteins, has potential as an objective measure of TDP-43-mediated degeneration in FTLD. Validation Competing interests of these findings in independent cohorts is needed to OB, AL, JF and DA declare a filed patent application (pending) to the European Patent Office for Markers of synaptopathy neurodegenerative disease. fully determine their clinical potential. Author details 1 Hospital de La Santa Creu I Sant Pau, Universitat Autonoma de Barcelona, Abbreviations Barcelona, Spain. 2 Centre of Biomedical Investigation Network for Neurode‑ AD: Alzheimer’s disease; ANCOVA: Analysis of Covariance; AUC​: Area- generative Diseases (CIBERNED), Madrid, Spain. 3 Memory Unit and Biomedi‑ under-the-curve; bvFTD: Behavioural variant; CSF: Cerebrospinal fluid; FTD: cal Research Institute, IIB Sant Pau, c/Sant Quintí 77, 08041 Barcelona, Spain. Frontotemporal dementia; FTLD: Frontotemporal lobar degeneration; ROC: 4 Penn FTD Center, Department of Neurology, Perelman School of Medicine Receiver operating characteristic; SRM: Selected Reaction Monitoring; TDP-43: at the University of Pennsylvania, Philadelphia, PA, USA. 5 Penn Alzheimer’s Dis‑ Transactive response DNA-binding protein-43. ease Research Center, Department of Neurology, Perelman School of Medicine Cervantes González et al. Molecular Neurodegeneration (2022) 17:29 Page 12 of 12 at the University of Pennsylvania, Philadelphia, PA, USA. 6 Center for Neuro‑ 17. Benagiano V, Lorusso L, Flace P, Girolamo F, Rizzi A, Bosco L, Cagiano degenerative Disease Research, Department of Pathology and Laboratory R, Nico B, Ribatti D, Ambrosi G. VAMP-2, SNAP-25A/B and syntaxin-1 in Medicine, Perelman School of Medicine at the University of Pennsylvania, glutamatergic and GABAergic synapses of the rat cerebellar cortex. BMC Philadelphia, PA, USA. Neurosci. 2011;12:118. 18. Lin RC, Scheller RH. Structural organization of the synaptic exocytosis Received: 13 September 2021 Accepted: 30 March 2022 core complex. Neuron. 1997;19:1087–94. 19. Gu Y, Chiu SL, Liu B, Wu PH, Delannoy M, Lin DT, Wirtz D, Huganir RL. Dif‑ ferential vesicular sorting of AMPA and GABAA receptors. Proc Natl Acad Sci U S A. 2016;113(7):E922-31. 20. Ullrich B, Ushkaryov YA, Sudhof TC. Cartography of neurexins: more than References 1000 isoforms generated by alternative splicing and expressed in distinct 1. Henstridge CM, Sideris DI, Carroll E, Rotariu S, Salomonsson S, Tzioras M, subsets of neurons. Neuron. 1995;14:497–507. McKenzie CA, Smith C, von Arnim CAF, Ludolph AC, et al. Synapse loss in 21. Graf ER, Zhang X, Jin SX, Linhoff MW, Craig AM. Neurexins induce dif‑ the prefrontal cortex is associated with cognitive decline in amyotrophic ferentiation of GABA and glutamate postsynaptic specializations via lateral sclerosis. Acta Neuropathol. 2018;135:213–26. neuroligins. Cell. 2004;119:1013–26. 2. Mackenzie IR, Neumann M. Molecular neuropathology of frontotemporal 22. Chih B, Gollan L, Scheiffele P. Alternative splicing controls selective dementia: insights into disease mechanisms from postmortem studies. J trans-synaptic interactions of the neuroligin-neurexin complex. Neuron. Neurochem. 2016;138(Suppl 1):54–70. 2006;51:171–8. 3. Lashley T, Rohrer JD, Mead S, Revesz T. Review: an update on clinical, 23. Benussi A, Alberici A, Buratti E, Ghidoni R, Gardoni F, Di Luca M, Padovani genetic and pathological aspects of frontotemporal lobar degenerations. A, Borroni B. Toward a glutamate hypothesis of frontotemporal dementia. Neuropathol Appl Neurobiol. 2015;41:858–81. Front Neurosci. 2019;13:304. 4. Rademakers R, Neumann M, Mackenzie IR. Advances in understand‑ 24. Remnestal J, Oijerstedt L, Ullgren A, Olofsson J, Bergstrom S, Kultima K, ing the molecular basis of frontotemporal dementia. Nat Rev Neurol. Ingelsson M, Kilander L, Uhlen M, Manberg A, et al. Altered levels of CSF 2012;8:423–34. proteins in patients with FTD, presymptomatic mutation carriers and 5. Lleo A, Nunez-Llaves R, Alcolea D, Chiva C, Balateu-Panos D, Colom- non-carriers. Transl Neurodegener. 2020;9:27. Cadena M, Gomez-Giro G, Munoz L, Querol-Vilaseca M, Pegueroles J, et al. 25. Missler M, Zhang W, Rohlmann A, Kattenstroth G, Hammer RE, Gottmann Changes in synaptic proteins precede neurodegeneration markers in K, Sudhof TC. Alpha-neurexins couple Ca2+ channels to synaptic vesicle preclinical Alzheimer’s disease cerebrospinal fluid. Mol Cell Proteomics. exocytosis. Nature. 2003;423:939–48. 2019;18:546–60. 26. Hyman BT, Phelps CH, Beach TG, Bigio EH, Cairns NJ, Carrillo MC, Dickson 6. Scheiffele P, Fan J, Choih J, Fetter R, Serafini T. Neuroligin expressed in DW, Duyckaerts C, Frosch MP, Masliah E, et al: National Institute on Aging- nonneuronal cells triggers presynaptic development in contacting axons. Alzheimer’s Association guidelines for the neuropathologic assessment Cell. 2000;101:657–69. of Alzheimer’s disease. Alzheimers Dement. 2012;8:1-13. 7. Vogt L, Schrimpf SP, Meskenaite V, Frischknecht R, Kinter J, Leone DP, 27. Brockhaus J, Schreitmuller M, Repetto D, Klatt O, Reissner C, Elmslie K, Ziegler U, Sonderegger P. Calsyntenin-1, a proteolytically processed post‑ Heine M, Missler M. alpha-Neurexins together with alpha2delta-1 auxil‑ synaptic membrane protein with a cytoplasmic calcium-binding domain. iary subunits regulate Ca(2+) Influx through Cav2.1 channels. J Neurosci. Mol Cell Neurosci. 2001;17:151–66. 2018;38:8277–94. 8. Irwin DJ, McMillan CT, Toledo JB, Arnold SE, Shaw LM, Wang LS, Van 28. Tong XJ, Lopez-Soto EJ, Li L, Liu H, Nedelcu D, Lipscombe D, Hu Z, Kaplan Deerlin V, Lee VM, Trojanowski JQ, Grossman M. Comparison of cer‑ JM. Retrograde Synaptic Inhibition Is Mediated by alpha-Neurexin Bind‑ ebrospinal fluid levels of tau and Abeta 1–42 in Alzheimer disease and ing to the alpha2delta Subunits of N-Type Calcium Channels. Neuron. frontotemporal degeneration using 2 analytical platforms. Arch Neurol. 2017;95:326–40 e325. 2012;69:1018–25. 29. McAleese KE, Colloby SJ, Thomas AJ, Al-Sarraj S, Ansorge O, Neal J, 9. Irwin DJ, Lleo A, Xie SX, McMillan CT, Wolk DA, Lee EB, Van Deerlin VM, Roncaroli F, Love S, Francis PT, Attems J. Concomitant neurodegenerative Shaw LM, Trojanowski JQ, Grossman M. Ante mortem cerebrospinal fluid pathologies contribute to the transition from mild cognitive impairment tau levels correlate with postmortem tau pathology in frontotemporal to dementia. Alzheimers Dement. 2021;17:1121–33. lobar degeneration. Ann Neurol. 2017;82:247–58. 10. Toledo JB, Van Deerlin VM, Lee EB, Suh E, Baek Y, Robinson JL, Xie SX, McBride J, Wood EM, Schuck T, et al. A platform for discovery: the uni‑ Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in pub‑ versity of Pennsylvania integrated neurodegenerative disease biobank. lished maps and institutional affiliations. Alzheimers Dement. 2014;10:477–84 e471. 11. Brettschneider J, Del Tredici K, Irwin DJ, Grossman M, Robinson JL, Toledo JB, Fang L, Van Deerlin VM, Ludolph AC, Lee VM, et al. Sequential distribu‑ tion of pTDP-43 pathology in behavioral variant frontotemporal dementia (bvFTD). Acta Neuropathol. 2014;127:423–39. 12. Irwin DJ, Byrne MD, McMillan CT, Cooper F, Arnold SE, Lee EB, Van Deerlin VM, Xie SX, Lee VM, Grossman M, Trojanowski JQ. Semi-automated digital image analysis of pick’s disease and TDP-43 proteinopathy. J Histochem Cytochem. 2016;64:54–66. Ready to submit your research ? Choose BMC and benefit from: 13. Montine TJ, Phelps CH, Beach TG, Bigio EH, Cairns NJ, Dickson DW, Duyckaerts C, Frosch MP, Masliah E, Mirra SS, et al. National Institute on • fast, convenient online submission aging-Alzheimer’s association guidelines for the neuropathologic assess‑ ment of Alzheimer’s disease: a practical approach. Acta Neuropathol. • thorough peer review by experienced researchers in your field 2012;123:1–11. • rapid publication on acceptance 14. Choi M, Chang CY, Clough T, Broudy D, Killeen T, MacLean B, Vitek O. • support for research data, including large and complex data types MSstats: an R package for statistical analysis of quantitative mass spec‑ trometry-based proteomic experiments. Bioinformatics. 2014;30:2524–6. • gold Open Access which fosters wider collaboration and increased citations 15. R-Core-Team. R: A language and environment for statistical computing. • maximum visibility for your research: over 100M website views per year Vienna: R Foundation for Statistical Computing; 2018. 16. Robin X, Turck N, Hainard A, Tiberti N, Lisacek F, Sanchez JC, Muller M. At BMC, research is always in progress. pROC: an open-source package for R and S+ to analyze and compare ROC curves. BMC Bioinformatics. 2011;12:77. Learn more biomedcentral.com/submissions