Abstract
Importance:
TMEM106B is a frontotemporal lobar degeneration (FTLD) genetic susceptibility factor, and TMEM106B protein aggregates are a feature of aging and neurodegeneration. Whether TMEM106B protein levels are associated with clinical features is unknown.
Objective:
To investigate the clinical associations of cerebrospinal fluid (CSF) TMEM106B in FTLD.
Design:
Cross-sectional study in two independent frontotemporal dementia (FTD) cohorts (recruitment April 2009 - July 2023, analyses Jan 2025 – April 2026), with a 2-year follow up.
Setting:
Multicenter clinical study with clinical, genetic, biomarker, and neuroimaging data.
Participants:
Individuals recruited through the University of California, San Francisco (n=3733) or ALLFTD (n=2343). Participants with available CSF were included. A discovery cohort (n=271) included participants with sporadic neuropathology-confirmed FTLD, presymptomatic or symptomatic carriers of pathogenic variants in C9orf72, GRN, or MAPT or controls. An independent validation cohort (n=383) included participants with clinically-diagnosed sporadic FTD, Alzheimer’s disease (AD) and controls.
Exposures:
CSF samples for TMEM106B quantification with aptamer proteomics [SomaScan v3.0 (discovery) and SomaScan v4.1 (validation)].
Main Outcomes and Measures:
Parametric tests compared CSF TMEM106B by disease severity, TMEM106B rs1990622 genotype, sex, clinical syndrome, pathological diagnosis, and pathogenic variant and determined associations with brain volume.
Results:
In discovery [n=271 (51% women, median age 59±21 years)] and validation [n=383 (48% women, 64±14 years)] cohorts, lower CSF TMEM106B was associated with more severe disease (β=−0.15, 95% CI −0.24 to −0.04, P<0.01), lower frontotemporal brain volumes (β=0.42; 95% CI, 0.24 to 0.61; P<.001) and faster clinical progression (β=−2.21, 95% CI −3.70 to −0.72, P = 0.001). Associations of TMEM106B with clinical disease severity were independent of those with neurofilament light chain. TMEM106B levels were influenced by TMEM106B rs1990622 genotype, where individuals with the protective G/G genotype had lower levels than the risk A/A genotype. CSF TMEM106B does not differentiate between FTLD subtypes or between FTLD and AD.
Conclusions and Relevance:
TMEM106B is detectable in CSF and levels reflect disease severity, in sporadic and genetic FTLD and AD, but are also influenced by the TMEM106B rs1990622 genotype. CSF TMEM106B could support further studies to understand the mechanisms of disease and develop clinical tools in FTLD and other neurodegenerative diseases.
Introduction
Frontotemporal lobar degeneration (FTLD) causes aggressive forms of dementia featuring behavioral, language, and motor deficits. FTLD is associated with greater disability, caregiver burden and familiar prevalence than Alzheimer’s disease (AD)1, with about 30% of cases being caused by autosomal dominant pathogenic variants in the C9orf72, GRN, or MAPT genes.2 FTLD has no cure, and a barrier to therapeutic development is the paucity of biomarkers for early diagnosis and disease monitoring. Early GWAS identified the lysosomal type II transmembrane protein 106B (TMEM106B) gene as a key genetic susceptibility factor for FTLD.3 In GRN and C9orf72 pathogenic variant carriers, TMEM106B rs1990622 (a lead variant of ~100 strongly linked SNPs in the TMEM106B locus) is a strong modulator of FTLD clinical expression, similar to the protective effect of the APOE3 Christchurch mutation on PSEN1 genetic AD.4–9 The rs1990622 major allele (A) is associated with an increased risk for developing with TDP-43 pathology (e.g. FTLD-TDP), whereas the minor allele (G) confers protection.5,6
In FTLD, TMEM106B expression is dysregulated and exacerbates TDP-43 aggregation.10–12 The TMEM106B protein is as a major component of amyloid fibrils in FTLD-TDP,11,13,14 although TMEM106B-rich fibrils are also present in FTLD-tau, AD, alpha synucleinopathies, and normal aging.15–17 We hypothesized that TMEM106B protein concentrations in biofluids may track with the clinical severity of FTLD. The value of TMEM106B as a fluid biomarker has remained unexplored because methodological limitations prevented its reliable quantification. In this study, we measured cerebrospinal fluid (CSF) TMEM106B with aptamer-based proteomics in FTLD, AD and controls. We compared CSF TMEM106B levels across FTLD clinical phenotypes, genotypes, neuropathologies, and TMEM106B rs1990622 genotypes. We also examined the association between CSF TMEM106B and clinical measures of disease severity and brain volumes, in comparison to that of neurofilament light chain (NfL), a non-specific marker of neurodegeneration.
Methods
Participants
This cross-sectional study recruited research participants with available CSF from two independent cohorts between April 2009 and July 2023. The discovery cohort (n=271) included sporadic and familial FTLD participants recruited through the UCSF Memory and Aging Center and ALLFTD, respectively (Table 1).18 These consisted of sporadic neuropathology-confirmed FTLD-tau (n=34), FTLD-TDP (n=13), clinically-diagnosed progressive supranuclear palsy-Richardson’s syndrome (PSP-RS, n=19), a proxy of FTLD-tau PSP,19 healthy controls (n=27), presymptomatic (n=54) or symptomatic (n=65) carriers of FTLD-causing pathogenic variants [C9orf72 (n=49), GRN (n=33), MAPT (n=37)], and related confirmed non-carrier controls (n=59). The validation cohort (n=383) included participants with clinically-diagnosed sporadic FTD syndromes (n=124) or familial FTLD (n=22), biomarker-confirmed AD (n=160) and cognitively healthy controls (n=77) recruited through UCSF (eTables 1–2 in Supplement). All participants or their surrogates provided written informed consent for study participation. This study followed STROBE reporting guidelines, was consistent with the Declaration of Helsinki and approved by the local Investigational Review Boards of participating research centers.
Table 1.
Cohort Characteristics
| Variable | Discovery Cohort | Validation Cohort | |||||||
|---|---|---|---|---|---|---|---|---|---|
| Control (n = 86) | Sporadic FTD | Familial FTD | Control (n = 77) | FTD (n = 146) | AD (n = 160) | ||||
| PSP-RS (n = 19) | FTLD-tau (n = 34) | FTLD-TDP (n = 13) | PreSx Carrier (n = 54) | Sx Carrier (n = 65) | |||||
| Age, y | 59 (28) | 71 (8) | 69 (13) | 65 (8) | 43 (20) | 57 (14) | 68 (22) | 64 (12) | 63 (12) |
| Sex, M/F, n | 41/45 | 11/8 | 18/16 | 8/5 | 27/26 | 29/36 | 43/34 | 75/71 | 82/78 |
| Education, y | 16 (5) | 16 (4) | 16 (5) | 14 (6) | 16 (4) | 15 (3) | 18 (3) | 16 (4) | 16 (2) |
| Race, n (%) | |||||||||
| White | 36 (42) | 15 (79) | 27 (79) | 12 (92) | NA | NA | 55 (71) | 121 (83) | 132 (81) |
| Chinese | 1 (1) | 0 (0) | 1 (3) | 0 (0) | NA | NA | 4 (5) | 5 (3) | 4 (3) |
| Black | 0 (0) | 0 (0) | 0 (0) | 0 (0) | NA | NA | 0 (0) | 2 (1) | 3 (2) |
| Other | 0 (0) | 1 (5) | 3 (9) | 0 (0) | NA | NA | 9 (12) | 7 (5) | 5 (4) |
| Unknown | 49 (57) | 3 (16) | 3 (9) | 1 (8) | NA | NA | 9 (12) | 11 (8) | 16 (10) |
| CSF NfL [platform], log2 pg/mL | 9.02 (1.76) [Simoa] n = 57 | 11.3 (.90) [Simoa] n = 14 | 12.2 (1.21) [Simoa] n = 32 | 12.7 (1.45) [Simoa] n = 9 | 8.37 (1.36) [Simoa] n = 45 | 10.6 (2.43) [Simoa] n = 52 | 9.35 (.81) [Elecsys] n = 35 | 11.4 (1.26) [Elecsys] n = 90 | 10.1 (0.66) [Elecsys] n = 113 |
| CSF NfL, log2 RFU | 8.89 (0.72) | 9.34 (0.42) | 9.59 (0.58) | 9.79 (1.06) | 8.77 (0.78) | 9.26 (1.14) | 9.98 (0.45) | 11.4 (1.01) | 10.5 (0.49) |
| CSF TMEM106B, log2 RFU | 7.48 (0.28) | 7.18 (0.34) | 7.26 (0.27) | 7.22 (0.43) | 7.52 (0.33) | 7.34 (0.31) | 7.23 (0.13) | 7.19 (0.12) | 7.16 (0.11) |
| FTD Pathogenic Variant, n (%) | |||||||||
| C9orf72 | 0 (0) | 0 (0) | 0 (0) | 0 (0) | 25 (46) | 24 (37) | 0 (0) | 9 (6) | 0 (0) |
| GRN | 0 (0) | 0 (0) | 0 (0) | 0 (0) | 11 (21) | 22 (34) | 0 (0) | 9 (6) | 0 (0) |
| MAPT | 0 (0) | 0 (0) | 0 (0) | 0 (0) | 18 (33) | 19 (29) | 0 (0) | 4 (3) | 0 (0) |
| Pathology confirmed, n (%) | NA | 4 (20) | 34 (100) | 13 (100) | NA | NA | NA | 58 (40) | 44 (27) |
| TMEM106B rs1990622 Genotype, n (%) | |||||||||
| A/A | 20 (23) | 7 (37) | 15 (44) | 7 (54) | 22 (41) | 22 (34) | 24 (31) | 39 (27) | 50 (31) |
| A/G | 45 (52) | 7 (37) | 16 (47) | 3 (23) | 27 (50) | 33 (51) | 27 (35) | 80 (55) | 70 (44) |
| G/G | 12 (14) | 4 (21) | 3 (9) | 3 (23) | 5 (9) | 10 (15) | 5 (7) | 24 (16) | 33 (21) |
| Unknown | 9 (11) | 1 (5) | 0 (0) | 0 (0) | 0 (0) | 0 (0) | 21 (27) | 3 (2) | 7 (4) |
| CDR+NACC-FTLDsb | 0 (0) | 5.5 (6) | 5.3 (5) | 9.5 (8) | 0 (0) | 5 (8) | 0 (0) | 5.5 (6) | 4 (2.5) |
| APOE ε4 carrier, n (%) | 25 (29) | 3 (16) | 6 (18) | 5 (38) | 19 (35) | 22 (34) | 24 (31) | 41 (28) | 84 (52) |
| T1-weighted MRI, n (%) | 70 (81) | 15 (79) | 25 (74) | 8 (62) | 41 (76) | 46 (71) | 73 (95) | 136 (93) | 145 (89) |
PreSx = Presymptomatic. Sx = Symptomatic. AD = Alzheimer’s Disease. RFU = relative fluorescent units. CDR®+NACC-FTLDsb = Clinical Dementia Rating scale (CDR®) plus Behavioral and Language Domains from the National Alzheimer’s Coordinating Center (NACC) FTLD module sum of boxes score. NC = non-carriers. PSP-RS = Progressive supranuclear palsy – Richardson’s Syndrome. Unless noted otherwise, all measures are expressed as median (interquartile range).
Clinical procedures
Participants underwent standardized neurological exams, functional interviews, neuropsychological testing, lumbar puncture for CSF collection, and brain MRI. Disease severity was determined with the Clinical Dementia Rating scale (CDR®) plus Behavioral and Language Domains from the National Alzheimer’s Coordinating Center (NACC) FTLD module (CDR®+NACC-FTLD).20 Cases were staged with the CDR®+NACC-FTLD global score as presymptomatic (global CDR®+NACC-FTLD score of 0), mild cognitive impairment (MCI, global CDR®+NACC-FTLD score of 0.5) or dementia (global CDR®+NACC-FTLD score ≥1). Participants with AD followed the same staging scores but used global CDR® scores. Additional clinical scales used in this study are described in the supplement.
CSF TMEM106B, NfL and p-tau181 measurements
CSF TMEM106B and NfL levels were quantified with aptamer-based proteomics (SomaScan, SomaLogic, Boulder, CO), with target values expressed as log2 relative fluorescent units (RFU, Supplement).21 The discovery cohort used SomaScan v3.0 (~5000 targets) and the validation cohort used SomaScan v4.1 (~7000 targets). The SomaScan assay used an aptamer targeting amino acids 118-274 by the luminal C-terminus of the human TMEM106B protein. These residues are part of the TMEM106B amyloid fibril core and most likely detect the full-length protein, as opposed to C-terminal cleaved fragments.15 The specificity of the TMEM106B aptamer (SomaScan sequence ID 8687-26) was verified via an electrochemiluminescence-based immunoassay in HEK293T cell lysates transfected with full-length TMEM106B (eFigure 1 in Supplement). An in-house polyclonal antibody was used to detect the TMEM106B filament core sequence.11 CSF NfL was measured with SomaScan and in subsets of participants, also with Simoa (discovery cohort), as published before,22 or Elecsys (validation cohort, Roche Diagnostics, Rotkreuz, Switzerland). The overall relationship between NfL measured by SomaScan v3.0 and Simoa was r =0.72. The overall relationship between NfL measured by SomaScan v4.1 and Elecsys was r =0.89. CSF total tau and p-tau181 measured by Elecsys were available in a subset of AD cases in the validation cohort (n=109).
Genotyping
Genetic screening for FTLD pathogenic variants was conducted in all subjects at the University of California, Los Angeles.23 TMEM106B rs1990622 and rs3173615 genotyping was performed by real-time PCR on a LightCycler 480 System using TaqMan SNP Genotyping Assays (#C__11171598_20). Assays were run in duplicate.
Neuroimaging
Participants from discovery (n=205) and validation (n=354) cohorts with available T1-weighted imaging data were included in the neuroimaging analysis (Supplement).
Statistical Analyses
Prior to comparative analyses, CSF TMEM106B and NfL levels were adjusted for principal component 1 (PC1) values across the full SomaScan proteome (eFigure 2 and eTable 3 in Supplement), consistent with recent work.24 TMEM106B levels were compared by disease severity, TMEM106B rs1990622 genotype, clinical phenotype, pathological diagnosis, and FTLD genotype with one-way ANOVA with post hoc Tukey’s multiple comparisons test. TMEM106B levels in models by disease severity and TMEM106B rs1990622 genotype were corrected for age and sex. Linear regression models analyzed the independent effects of age, sex, CSF TMEM106B and CSF NfL on disease severity outcomes. Linear regression models also analyzed the independent effects of age, sex, disease severity and TMEM106B rs1990622 genotype on CSF TMEM106B protein levels. In the discovery cohort, the relationship between baseline CSF biomarker levels and the longitudinal change of clinical severity was tested with mixed linear models corrected for baseline age, sex, and disease duration (Supplement).
Results
TMEM106B and demographic features
The median disease severity of individuals with dementia was similar in both cohorts. There were no differences in CSF TMEM106B levels by sex [discovery: male 7.37 ± 0.28 log2 RFU vs. female 7.41 ± 0.27 log2 RFU; P =.320; validation: male 7.18 ± 0.10 log2 RFU vs. female 7.19 ± 0.09 log2 RFU; P =.263]. TMEM106B levels associated negatively with age (discovery: β=−0.36; 95% CI, −0.48 to −0.25; P<.001; validation: β=−0.23; 95% CI, −0.33 to −0.13; P<.001) (eFigures 3–4 in Supplement). CSF TMEM106B associated negatively with CSF NfL (discovery: β=−0.18; 95% CI, −0.30 to −0.06; P=.003; validation: β=−0.22; 95% CI, −0.32 to −0.13; P<.001) and showed no associations with total tau [(validation): β=−0.09; 95% CI, −0.21 to 0.04; P=.168] or p-tau181 [(validation): β=−0.06; 95% CI, −0.19 to 0.07; P=.343, eFigure 5 in Supplement].
In both cohorts, CSF TMEM106B showed higher levels in controls and presymptomatic participants than those with symptomatic disease across numerous phenotypes, but discrimination between controls from symptomatic participants with ROC analyses was poor with AUC ranging between 0.70 and 0.78 (eFigure 6 in Supplement). There were no differences in TMEM106B levels between genetic pathogenic variants or between pathological groups (e.g. FTLD-tau, FTLD-TDP, or AD) (eFigures 7–9 in Supplement).
TMEM106B and disease severity
There were effects of disease severity on CSF TMEM106B protein levels in both cohorts, including in AD. In the discovery cohort, participants with dementia had lower TMEM106B levels than participants with MCI (7.59 ± 0.25 log2 RFU vs. 7.75 ± 0.26 log2 RFU; P=.003), presymptomatic disease (7.82 ± 0.20 log2 RFU; P<.001) and controls (7.81 ± 0.22 log2 RFU; P<.001; Figure 1A). Disease severity effects were also observed separately in sporadic and familial FTLD groups (eFigure 10 in Supplement). Lower TMEM106B levels with more severe disease were also seen in both validation cohort groups (Figure 1B). In FTD, participants with dementia had lower TMEM106B than controls (7.28 ± 0.09 log2 RFU vs. 7.35 ± 0.08 log2 RFU; P<.001). In AD, controls had higher TMEM106B than participants with MCI (7.35 ± 0.08 log2 RFU vs. 7.28 ± 0.08 log2 RFU; P<.001) and AD dementia (7.29 ± 0.10 log2 RFU; P<.001).
Figure 1. Box plots showing cerebrospinal fluid (CSF) levels of TMEM106B across clinical disease severity groups in an (A) discovery cohort of sporadic and familial FTD and a (B) validation cohort of FTD and AD groups.
CSF TMEM106B levels (corrected for age, sex and PC1; log2 transformed [RFU]) are shown in relation to global CDR®+NACC-FTLD scores (Control [0], Presymptomatic [0], 0.5, 1+] for FTD groups and CDR® global scores (Control [0], 0.5, 1+] for the AD group. Higher disease severity, indicated by higher CDR scores, is associated with lower CSF TMEM106B levels across both FTD cohorts (discovery: η2 = 0.15; validation: η2 = 0.12) and the AD group (η2 = 0.10). Significant differences are marked by asterisks: * = P <.05, ** = P <.01, *** = P <.001 after Tukey’s multiple comparisons test. Global CDR® = Clinical Dementia Rating scale global score. Global CDR®+NACC-FTLD = Clinical Dementia Rating scale (CDR®) plus Behavioral and Language Domains from the National Alzheimer’s Coordinating Center (NACC) FTLD module global score.
CSF TMEM106B was strongly associated with CDR®+NACC-FTLDsb scores (eTable 4 and eFigure 11 in Supplement). Subgroup linear regression models tested whether CSF TMEM106B explained additional variance in disease severity beyond age, sex and CSF NfL (Table 2). In symptomatic subgroups of both cohorts CSF TMEM106B explained additional variance in CDR®+NACC-FTLDsb scores independent of CSF NfL (discovery: β=−0.41; 95% CI, −0.57 to −0.25; P<.001; accounts for 15% of variation; validation: β=−0.23; 95% CI, −0.40 to −0.07; P=.005; accounts for 5.1% of variation).
Table 2.
Associations of fluid biomarkers with disease severity in participants with symptomatic disease.
| Cohort | Discovery Cohort | Validation Cohort | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Symptomatic FTD Group | Symptomatic FTD Group | Symptomatic AD Group | ||||||||||
| Variables | β | 95% CI | R2 | β | 95% CI | R2 | β | 95% CI | R2 | |||
| Linear Regression Model: CDR®+NACC-FTLDsb ~ CSF TMEM106B + Age + Sex | ||||||||||||
| CSF TMEM106B | −0.37 | −0.53 to −0.20 | 0.20 | −0.27 | −0.42 to −0.11 | 0.15 | 0.05 | −0.11 to 0.21 | 0.03 | |||
| Age | 0.13 | −0.03 to 0.29 | −0.18 | −0.33 to −0.02 | −0.16 | −0.32 to −0.01 | ||||||
| Sex | −0.10 | −0.26 to 0.06 | −0.25 | −0.40 to −0.09 | −0.10 | −0.26 to 0.05 | ||||||
| Linear Regression Model: CDR®+NACC-FTLDsb ~ CSF NfL + Age + Sex | ||||||||||||
| CSF NfL | 0.26 | 0.10 to 0.43 | 0.14 | 0.19 | 0.03 to 0.35 | 0.12 | 0.24 | 0.09 to 0.39 | 0.09 | |||
| Age | 0.16 | −0.01 to 0.33 | −0.11 | −0.27 to 0.05 | −0.19 | −0.34 to −0.03 | ||||||
| Sex | −0.17 | −0.33 to −0.01 | −0.25 | −0.41 to −0.09 | −0.10 | −0.25 to 0.06 | ||||||
| Linear Regression Model: CDR®+NACC-FTLDsb ~ CSF TMEM106B + CSF NfL + Age + Sex | ||||||||||||
| CSF TMEM106B | −0.41 | −0.57 to −0.25 | 0.29 | −0.23 | −0.40 to −0.07 | 0.17 | 0.04 | −0.11 to 0.20 | 0.09 | |||
| CSF NfL | 0.31 | 0.16 to 0.47 | 0.13 | −0.04 to 0.29 | 0.24 | 0.09 to 0.39 | ||||||
| Age | 0.05 | −0.10 to 0.21 | −0.15 | −0.31 to 0.01 | −0.18 | −0.34 to −0.03 | ||||||
| Sex | −0.08 | −0.24 to 0.07 | −0.25 | −0.40 to −0.09 | −0.10 | −0.25 to 0.06 | ||||||
Abbreviations: AD = Alzheimer’s Disease; FTD = Frontotemporal Dementia; CDR®+NACC-FTLDsb = Clinical Dementia Rating scale (CDR®) plus Behavioral and Language Domains from the National Alzheimer’s Coordinating Center (NACC) FTLD module sum of boxes score. β = standardized beta coefficient from the linear regression model. CDR®+NACC-FTLDsb scores were used in FTD groups and CDR®sb scores were used in the AD group
In the discovery cohort, lower baseline TMEM106B levels used as a continuous variable was associated with faster longitudinal disease progression. For every lower log unit in baseline TMEM106B, there was a 2.25-point increase in the CDR®+NACC-FTLDsb score after two years (β=−2.25; 95% CI, −3.71 to −0.76; P=.001, eTable 5 in Supplement). Lower baseline TMEM106B was also associated with rates of change in measures of instrumental and basic function, global cognition, executive function, and depression. When used as a categorical variable, the effects of baseline CSF TMEM106B on disease progression were marginal (eFigure 12 and eTable 6 in Supplement). Higher baseline NfL was associated with more severe CDR®+NACC-FTLDsb worsening, instrumental and basic function, global cognition, and memory, executive, language, and motor function, regardless of whether it was used as a continuous or categorical variable (eTables 4,5,7 in Supplement).
CSF TMEM106B and TMEM106B genotype
In both cohorts, TMEM106B rs1990622 genotype was associated with CSF TMEM106B levels after correction for age, sex and disease severity. Age, disease severity, and TMEM106B rs1990622 were independently associated with CSF TMEM106B (eTable 8 in Supplement). In the discovery cohort, participants with the protective G/G genotype had lower TMEM106B than those with the A/G or A/A genotypes (G/G: 7.58 ± 0.27 log2 RFU vs. A/A: 7.78 ± 0.24 log2 RFU; P<.001 and A/G: 7.74 ± 0.24 log2 RFU; P =.002; Figure 2A). Within the discovery cohort, this effect was seen in the familial FTLD group (G/G: 7.60 ± 0.28 log2 RFU vs. A/A: 7.81 ± 0.20 log2 RFU; P=.002), with a trend in the sporadic FTLD group (G/G: 7.10 ± 0.26 log2 RFU vs. A/A: 7.28 ± 0.29 log2 RFU; P=.095).
Figure 2. CSF TMEM106B levels by TMEM106B rs1990622 Genotype in (A) Discovery and (B) Validation Cohorts.
CSF TMEM106B levels (corrected for age, sex and PC1; log2 transformed [RFU]) in the discovery cohort and stratified by TMEM106B rs1990622 genotype (A/A, A/G, G/G) and color coded by global CDR®+NACC-FTLD score (Control [0], Presymptomatic FTLD pathogenic variant carrier [0], 0.5, 1+] for FTD groups and by global CDR® score (Control [0], 0.5, 1+) for the AD group. Lower CSF TMEM106B levels correspond with the G/G TMEM106B genotype compared to A/A or A/G (discovery: η2 = 0.06; validation FTD: η2 = 0.06; validation AD: η2 = 0.06). Significant differences are marked by asterisks: * = P <.05, ** = P <.01, *** = P <.001 after Tukey’s multiple comparisons test. Global CDR® = Clinical Dementia Rating scale global score. Global CDR®+NACC-FTLD = Clinical Dementia Rating scale (CDR®) plus Behavioral and Language Domains from the National Alzheimer’s Coordinating Center (NACC) FTLD module global score.
The TMEM106B rs1990622 genotype effect replicated in the validation cohort (Figure 2B). Participants with FTD and the risk A/A TMEM106B rs1990622 genotype had higher TMEM106B than those with the A/G or G/G genotypes (A/A: 7.21 ± 0.09 log2 RFU vs. A/G: 7.17 ± 0.09 log2 RFU; P=.030 and G/G: 7.15 ± 0.10 log2 RFU; P=.014). Participants with AD and the protective G/G TMEM106B rs1990622 genotype also had lower TMEM106B than the A/A genotype (G/G: 7.12 ± 0.09 log2 RFU vs. A/A: 7.18 ± 0.10 log2 RFU; P=.005). TMEM106B genotype effects on CSF TMEM106B protein levels were also observed with the rs3173615 SNP, which is in high linkage disequilibrium with rs1990622 (eFigure 13).
These effects were observed in all three FTLD-causing pathogenic variant carrier groups in the discovery cohort (eFigure 14 in Supplement). There were no TMEM106B rs1990622 G/G GRN pathogenic variant carriers who had symptomatic disease. Sample sizes for the FTLD pathogenic variants in the validation cohort were small and no effect was observed. The effects of TMEM106B rs1990622 genotype were also seen regardless of pathological diagnosis (i.e., FTLD-tau or FTLD-TDP) in both cohorts (eFigure 15 in Supplement).
TMEM106B and brain volume
In symptomatic FTD subgroups of both cohorts, TMEM106B levels were associated with frontotemporal (discovery: β=0.42; 95% CI, 0.24 to 0.61; P<.001; validation: β=0.27; 95% CI, 0.10 to 0.43; P =.002; Figure 3) and hippocampal volumes (eFigure 16). In the AD group, however, TMEM106B showed no relationship with frontotemporal brain volume (β=0.05; 95% CI, −0.12 to 0.21; P =.583; eFigure 17G in Supplement), but showed a trend for association with hippocampal volume (β=0.15; 95% CI, −0.01 to 0.32; P =.063). CSF NfL was associated with brain volumes in both symptomatic FTD groups (discovery: β=−0.26; 95% CI, −0.47 to −0.05; P=.016; validation: β=−0.35; 95% CI, −0.52 to −0.17; P<.001; eFigure 16B and E in Supplement), but not in AD (β=−0.35; 95% CI, −0.50 to −0.19; P<.001; eFigure 16H in Supplement). In symptomatic FTD, TMEM106B and NfL were independently associated with brain volume, together with sex and age (eFigure 16C and F in Supplement), but not in symptomatic AD (eFigure 16I in Supplement). The strongest relationship between TMEM106B and brain volume was seen in symptomatic FTD (discovery, β=0.42; 95% CI, 0.24 to 0.61; P<.001), but associations, although weaker, were also observed in controls (discovery: β=0.21; 95% CI, −0.02 to 0.44; P=.067; validation: β=0.51; 95% CI, 0.31 to 0.72; P<.001; eFigure 18 in Supplement). Similar relationships were observed between TMEM106B and mesial temporal lobe volumes (eFigures 19–20 in Supplement). Associations with brain volumes were stronger in TMEM106B rs1990622 G/G carriers compared to A/A and A/G (eTable 9 in Supplement). TMEM106B and NfL showed strong topographically overlapping, but inversely related associations with frontotemporal volumes (eFigure 22 in Supplement).
Figure 3. Linear regression models with regional brain volume outcomes within participants with symptomatic disease in Discovery and Validation Cohorts.
Scatterplot showing the relationship between CSF TMEM106B levels (PC1 corrected and log2 transformed [RFU]) and standardized frontotemporal brain volumes within the (A) symptomatic FTD group of the discovery cohort and (C) symptomatic FTD group of the validation cohort. Scatterplot showing the relationship between CSF NfL levels (PC1 corrected and log2 transformed [RFU]) and standardized frontotemporal brain volumes within the (B) symptomatic FTD group of the discovery cohort and (D) symptomatic FTD group of the validation cohort The gray line represents the linear regression fit with 95% confidence intervals. Standardized beta coefficients and p-values of the regression are shown.
Discussion
We measured CSF TMEM106B protein levels with aptamer-based proteomics in FTLD and AD. TMEM106B levels were independently influenced by age, disease severity, and TMEM106B rs1990622 genotype, regardless of FTD phenotype, pathogenic variant or neuropathological diagnosis. Lower TMEM106B levels were observed in individuals with worse clinical disease severity. TMEM106B levels were positively associated with regional brain volumes. Carriers of the protective G/G TMEM106B rs1990622 variant also showed low TMEM106B compared to A allele carriers. These effects were strong in FTLD, although they were also seen in AD. TMEM106B was associated with age and NfL, but its associations with disease severity and brain volume were independent of age, sex, and NfL (eFigure 22 in Supplement). The findings suggest that CSF TMEM106B levels reflect clinical and pathophysiological aspects of FTLD different from those of NfL.
Our findings align with a previous study that detected lower CSF TMEM106B measured with aptamer-based proteomics in AD compared to healthy controls.25 Low CSF TMEM106B levels were observed in healthy individuals with the TMEM106B rs1990622 G/G variant, compared to carriers of the high-risk A allele.25 We observed that TMEM106B levels correlate with age, which mirrors previous findings of high TMEM106B brain fibril aggregation with aging.17 It seems paradoxical that TMEM106B levels are low in individuals with more severe disease and, at the same time, in those with the lowest risk of disease expression due to carriership of the protective G/G TMEM106B rs1990622 variant. Disease severity had the strongest effect on CSF TMEM106B levels, compared to TMEM106B rs1990622 genotype, and age. In contrast, the strongest associations of CSF TMEM106B in AD were with age and TMEM106B rs1990622 genotype. We suspect that these associations are mediated by independent mechanisms. Low TMEM106B levels likely reflect a decrease in soluble full-length TMEM106B, which results from increases in brain TMEM106B amyloid fibrils composed of cleaved C-terminal fragments. In turn, carriers of the protective TMEM106B genotype variant may have constitutionally lower protein expression levels.28 Carriers of the protective variant have previously shown no changes in the abundance of brain monomeric full-length TMEM106B, but an increase in dimeric full-length TMEM106B and minimal TMEM106B aggregation.11 The high-risk A/A TMEM106B genetic variant has been predicted to confer high genetic susceptibility to symptomatic disease by increasing TMEM106B expression.3,29–31 Neuropathology studies support that high TMEM106B expression leads to endosomal-lysosomal dysfunction in FTLD.32 Preclinical studies, however, support that the high-risk variant increases TMEM106B protein aggregation and subsequent TDP-43 dysfunction and aggregation without affecting TMEM106B RNA or monomeric protein expression.11 Other studies have documented a post-transcriptional mechanism that induces resistance to protein degradation.33 It is unlikely that low CSF TMEM106B levels are just a direct result of decreased protein expression with neurodegeneration. Although studies should further explore the mechanisms by which CSF TMEM106B levels are altered in FTLD, we provide evidence that TMEM106B protein levels are detectable in CSF, influenced by both disease severity and the TMEM106B genotype, and have meaningful FTLD and AD clinical correlates.
We explored the clinical correlates of CSF TMEM106B levels in FTLD and AD in contrast to those of NfL, an established, but non-specific, marker of axonal injury and neurodegeneration.22,34,35 Distinctive aspects of the clinical associations of CSF TMEM106B and NfL included 1) their independence, 2) their opposite directions, 3) associations with clinical scales of disease severity that were weaker for TMEM106B than NfL, and 4) associations of CSF TMEM106B with brain volumes that were topographically more extensive but weaker and opposite direction than those of NfL. The NfL associations were also influenced by the TMEM106B rs1990622 genotype and were in line with previous studies demonstrating the effects of the protective variant on lower CSF NfL and greater brain volumes in FTLD.36,37 CSF TMEM106B could have value in monitoring the presence of FTLD and other neurodegenerative diseases or be of interest for prospective FTLD disease-modifying interventions that modulate lysosomal function and TMEM106B biology.38,39
This study has limitations. The sample sizes were small and not ethnically diverse. The discovery cohort primarily comprised highly-educated White individuals and participants without a high prevalence of cardiovascular comorbidity. The effects of other comorbidities and concomitant medications were not analyzed. Not every case had neuropathological confirmation of the diagnosis. Participants were recruited from memory and aging cohorts, with potential for selection bias. This could affect both the internal validity and generalizability of the observed associations. Two different versions of the SomaScan platform were used, which together with differences in cohort composition may affect results. This was a cross-sectional study and further work investigating the longitudinal trajectories of TMEM106B is needed to clarify its role in the natural history of neurodegenerative disease. Aptamer proteomics are not readily accessible, and single-analyte tests with better availability are preferable. Despite these limitations, this study offers evidence that TMEM106B is detectable in CSF and its levels reflect independent effects of disease severity and the TMEM106B rs1990622 genotype. The data implicates TMEM106B in the pathophysiology of sporadic and familial FLTD and less strongly in AD. CSF TMEM106B is not currently used in clinical practice and further studies should investigate its role in FTLD pathophysiology and its potential to support advancement of FTLD diagnostics and therapeutics.
Supplementary Material
Key Points.
Question:
What are the clinical correlates of cerebrospinal fluid (CSF) TMEM106B protein levels in sporadic and familial frontotemporal lobar degeneration (FTLD)?
Findings:
CSF TMEM106B levels do not differentiate FTLD from Alzheimer’s disease or between FTLD subtypes, but are lower in patients with more severe disease and those with the protective TMEM106B rs1990622 genotype, regardless of underlying neuropathology or FTLD-causing pathogenic variant. Low CSF TMEM106B is associated with smaller regional brain volumes, and faster clinical decline.
Meaning:
TMEM106B is detectable in CSF and its clinical associations may shed light on the pathophysiology of FTLD and other neurodegenerative diseases.
Acknowledgments
Data collection and dissemination of the data presented in this manuscript was supported by the ALLFTD Consortium (U19: AG063911, funded by the National Institute on Aging and the National Institute of Neurological Diseases and Stroke) and the former ARTFL & LEFFTDS Consortia (ARTFL: U54 NS092089, funded by the National Institute of Neurological Diseases and Stroke and National Center for Advancing Translational Sciences; LEFFTDS: U01 AG045390, funded by the National Institute on Aging and the National Institute of Neurological Diseases and Stroke). Additional funding was provided by the Rainwater Charitable Foundation and the Bluefield Project to Cure FTD. Samples from the NCRAD, which receives government support under a cooperative agreement grant (U24AG021886) awarded by the NIA, were used in this study. The manuscript has been reviewed by the ALLFTD Executive Committee for scientific content. The authors acknowledge the invaluable contributions of the study participants and families, as well as the assistance of the support staff at each of the participating sites. The John Douglas French Alzheimer Foundation and AlzOut (JCR).
Disclosures
Molly Olzinski – Nothing to disclose
Joshua Downer – Nothing to disclose
Yann Cobigo – Nothing to disclose
Binita Rajbanshi – Nothing to disclose
Amy Wise – Nothing to disclose
Julia Webb – Nothing to disclose
Jingyao Li – Nothing to disclose
Joseph Loureiro – Nothing to disclose
Kathleen A. Worringer – Nothing to disclose
Hilary Heuer – Nothing to disclose
Peter Ljubenkov – Nothing to disclose
Lawren Vandevrede – Nothing to disclose
Adam Staffaroni – Served as a paid consultant to Alector, Aviado Bio, CervoMed, Eli Lilly/Prevail Therapeutics, Passage Bio, Takeda, and Vesper Bio. Research support from Association for Frontotemporal Degeneration, NIH, Bluefield Project to Cure FTD.
Argentina Lario-Lago – Nothing to disclose
Mark Sanderson-Cimino – Nothing to disclose
Eden Barragan – Nothing to disclose
Dana Leichter – Nothing to disclose
Amy Wolf – Nothing to disclose
Salvatore Spina – Nothing to disclose
Lea T. Grinberg – Receives research support from K24AG350405
William W. Seeley – Nothing to disclose
Kaitlin B. Casaletto – Nothing to disclose
Joel Kramer – Nothing to disclose
Eliana Marisa Ramos – Receives research support from NIH.
Daniel Geschwind – Nothing to disclose
Casey Cook – Nothing to disclose
Leonard Petrucelli – Nothing to disclose
Rosa Rademakers – Nothing to disclose
Leah K. Forsberg – Nothing to disclose
Tania Gendron – Nothing to disclose
Bradley F. Boeve – Nothing to disclose
Howard J. Rosen – Nothing to disclose
Rowan Saloner – Nothing to disclose
Adam L. Boxer – Served as a paid consultant to Alector, Alexion, Arrowhead, Arvinas, Eli Lilly, Merck, Neurocrine, Ono, Oscotec, Switch and Transposon. Research support from Biogen, Eisai and Regeneron.
Julio C. Rojas – Site PI for clinical trials sponsored by Eli Lilly, Eisai and Amylyx. Consulting fees from Ferrer International, Roon Health, Clarivate and Adept Field Solutions.
References
- 1.Liu S, Jin Y, Shi Z, et al. The effects of behavioral and psychological symptoms on caregiver burden in frontotemporal dementia, Lewy body dementia, and Alzheimer’s disease: clinical experience in China. Aging Ment Health. 2017;21(6):651–657. [DOI] [PubMed] [Google Scholar]
- 2.Olszewska DA, Lonergan R, Fallon EM, Lynch T. Genetics of Frontotemporal Dementia. Curr Neurol Neurosci Rep. 2016;16(12):107. [DOI] [PubMed] [Google Scholar]
- 3.Van Deerlin VM, Sleiman PM, Martinez-Lage M, et al. Common variants at 7p21 are associated with frontotemporal lobar degeneration with TDP-43 inclusions. Nat Genet. 2010;42(3):234–239. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Arboleda-Velasquez JF, Lopera F, O’Hare M, et al. Resistance to autosomal dominant Alzheimer’s disease in an APOE3 Christchurch homozygote: a case report. Nat Med. 2019;25(11):1680–1683. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Gallagher MD, Suh E, Grossman M, et al. TMEM106B is a genetic modifier of frontotemporal lobar degeneration with C9orf72 hexanucleotide repeat expansions. Acta Neuropathol. 2014;127(3):407–418. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Pottier C, Zhou X, Perkerson RB 3rd, et al. Potential genetic modifiers of disease risk and age at onset in patients with frontotemporal lobar degeneration and GRN mutations: a genome-wide association study. Lancet Neurol. 2018;17(6):548–558. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.van Blitterswijk M, Mullen B, Nicholson AM, et al. TMEM106B protects C9ORF72 expansion carriers against frontotemporal dementia. Acta Neuropathol. 2014;127(3):397–406. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Finch N, Carrasquillo MM, Baker M, et al. TMEM106B regulates progranulin levels and the penetrance of FTLD in GRN mutation carriers. Neurology. 2011;76(5):467–474. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Cruchaga C, Graff C, Chiang HH, et al. Association of TMEM106B gene polymorphism with age at onset in granulin mutation carriers and plasma granulin protein levels. Arch Neurol. 2011;68(5):581–586. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Busch JI, Martinez-Lage M, Ashbridge E, et al. Expression of TMEM106B, the frontotemporal lobar degeneration-associated protein, in normal and diseased human brain. Acta Neuropathol Commun. 2013;1:36. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Marks JD, Ayuso VE, Carlomagno Y, et al. TMEM106B core deposition associates with TDP-43 pathology and is increased in risk SNP carriers for frontotemporal dementia. Sci Transl Med. 2024;16(730):eadf9735. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Werner G, Damme M, Schludi M, et al. Loss of TMEM106B potentiates lysosomal and FTLD-like pathology in progranulin-deficient mice. EMBO Rep. 2020;21(10):e50241. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Jiang YX, Cao Q, Sawaya MR, et al. Amyloid fibrils in FTLD-TDP are composed of TMEM106B and not TDP-43. Nature. 2022;605(7909):304–309. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Vicente CT, Perneel J, Wynants S, et al. C-terminal TMEM106B fragments in human brain correlate with disease-associated TMEM106B haplotypes. Brain. 2023;146(10):4055–4064. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Chang A, Xiang X, Wang J, et al. Homotypic fibrillization of TMEM106B across diverse neurodegenerative diseases. Cell. 2022;185(8):1346–1355 e1315. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Fan Y, Zhao Q, Xia W, et al. Generic amyloid fibrillation of TMEM106B in patient with Parkinson’s disease dementia and normal elders. Cell Res. 2022;32(6):585–588. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Schweighauser M, Arseni D, Bacioglu M, et al. Age-dependent formation of TMEM106B amyloid filaments in human brains. Nature. 2022;605(7909):310–314. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Boeve B, Bove J, Brannelly P, et al. The longitudinal evaluation of familial frontotemporal dementia subjects protocol: Framework and methodology. Alzheimers Dement 2020;16(1):22–36. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Ali F, Martin PR, Botha H, et al. Sensitivity and Specificity of Diagnostic Criteria for Progressive Supranuclear Palsy. Mov Disord. 2019;34(8):1144–1153. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Miyagawa T, Brushaber D, Syrjanen J, et al. Utility of the global CDR((R)) plus NACC FTLD rating and development of scoring rules: Data from the ARTFL/LEFFTDS Consortium. Alzheimers Dement. 2020;16(1):106–117. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Gold L, Ayers D, Bertino J, et al. Aptamer-based multiplexed proteomic technology for biomarker discovery. PLoS One. 2010;5(12):e15004. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Rojas JC, Wang P, Staffaroni AM, et al. Plasma Neurofilament Light for Prediction of Disease Progression in Familial Frontotemporal Lobar Degeneration. Neurology. 2021;96(18):e2296–e2312. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Ramos EM, Dokuru DR, Van Berlo V, et al. Genetic screening of a large series of North American sporadic and familial frontotemporal dementia cases. Alzheimers Dement. 2020;16(1):118–130. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Oh HS, Urey DY, Karlsson L, et al. A cerebrospinal fluid synaptic protein biomarker for prediction of cognitive resilience versus decline in Alzheimer’s disease. Nat Med. 2025;31(5):1592–1603. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Chemparathy A, Guen YL, Zeng Y, et al. A 3’UTR Insertion Is a Candidate Causal Variant at the TMEM106B Locus Associated with Increased Risk for FTLD-TDP. Neurol Genet 2024;10:e200124. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Grimmer T, Riemenschneider M, Forstl H, et al. Beta amyloid in Alzheimer’s disease: increased deposition in brain is reflected in reduced concentration in cerebrospinal fluid. Biol Psychiatry. 2009;65(11):927–934. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Horie K, Barthelemy NR, Spina S, et al. CSF tau microtubule-binding region identifies pathological changes in primary tauopathies. Nat Med. 2022;28(12):2547–2554. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Lee JY, Harney DJ, Teo JD, et al. The major TMEM106B dementia risk allele affects TMEM106B protein levels, fibril formation, and myelin lipid homeostasis in the ageing human hippocampus. Mol Neurodegener. 2023;18(1):63. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Fujita M, Gao Z, Zeng L, et al. Cell subtype-specific effects of genetic variation in the Alzheimer’s disease brain. Nat Genet. 2024;56(4):605–614. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Gallagher MD, Posavi M, Huang P, et al. A Dementia-Associated Risk Variant near TMEM106B Alters Chromatin Architecture and Gene Expression. Am J Hum Genet. 2017;101(5):643–663. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Yu L, De Jager PL, Yang J, Trojanowski JQ, Bennett DA, Schneider JA. The TMEM106B locus and TDP-43 pathology in older persons without FTLD. Neurology. 2015;84(9):927–934. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Chen-Plotkin AS, Unger TL, Gallagher MD, et al. TMEM106B, the risk gene for frontotemporal dementia, is regulated by the microRNA-132/212 cluster and affects progranulin pathways. J Neurosci. 2012;32(33):11213–11227. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Nicholson AM, Finch NA, Wojtas A, et al. TMEM106B p.T185S regulates TMEM106B protein levels: implications for frontotemporal dementia. J Neurochem. 2013;126(6):781–791. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Gendron TF, Heckman MG, White LJ, et al. Comprehensive cross-sectional and longitudinal analyses of plasma neurofilament light across FTD spectrum disorders. Cell Rep Med. 2022;3(4):100607. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Staffaroni AM, Quintana M, Wendelberger B, et al. Temporal order of clinical and biomarker changes in familial frontotemporal dementia. Nat Med. 2022;28(10):2194–2206. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Harding SR, Bocchetta M, Gordon E, et al. The TMEM106B risk allele is associated with lower cortical volumes in a clinically diagnosed frontotemporal dementia cohort. J Neurol Neurosurg Psychiatry. 2017;88(11):997–998. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Vandebergh M, Ramos EM, Corriveau-Lecavalier N, et al. Gene-Specific Effects on Brain Volume and Cognition of TMEM106B in Frontotemporal Lobar Degeneration. Neurology. 2024;103(8):e209832. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Kurnellas M, Mitra A, Schwabe T, et al. Latozinemab, a novel progranulin-elevating therapy for frontotemporal dementia. J Transl Med. 2023;21(1):387. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Liu Y, Qin K, Jiang C, Gao J, Hou B, Xie A. TMEM106B Knockdown Exhibits a Neuroprotective Effect in Parkinson’s Disease via Decreasing Inflammation and Iron Deposition. Mol Neurobiol. 2025;62(2):1813–1825. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.