Loss of Y chromosome and its implications in male amyotrophic lateral sclerosis: insights from the UK Biobank

Abstract

Background

Amyotrophic lateral sclerosis (ALS) shows a male predominance, yet the underlying mechanism remains unclear. Although the loss of Y chromosome (LOY) in peripheral blood — a male-specific genetic alteration — has been implicated in certain neurodegenerative disorders (NDDs), its association with ALS in men remains unexplored and has not been explored.

Methods

We focused on men in the UK Biobank to investigate whether LOY influences the risk and prognosis of ALS. Initially, the LOY level for each male participant was determined using sequencing data. Subsequently, Cox proportional hazards (Cox PH) model analysis was used to assess LOY-associated risk of ALS; thirdly, piecewise linear regression, Kaplan–Meier, and Cox PH analysis assessed LOY’s associations with ALS age at onset (AAO) and survival. Fourthly, multiple analytical methods were implemented to explore the relationship between LOY and ALS indicators, including plasma GFAP (glial fibrillary acidic protein) and NfL (neurofilament light chain). Finally, sensitivity analysis was carried out.

Results

Our final cohort consisted of 158,953 male participants, with a mean follow-up of 11.7 years. Among them, 297 individuals developed ALS. After adjusted multiple confounding factors, including C9orf72 hexanucleotide repeat expansion (HRE), male participants with LOY exhibited an elevated risk of developing ALS (HR [95% CI]: 1.619 [1.059–2.475], p = 0.026). LOY carrier may be more likely to be associated with a later AAO and shorter survival; however, this association did not reach statistical significance in multivariate models. Additionally, our findings revealed that LOY was significantly associated with elevated plasma NfL levels (p = 0.004). Moreover, the median Log2 R ratios of Y chromosome (mLRRY value) exhibited a modest inverse correlation with plasma GFAP levels (Pearson’s r = − 0.059). Nevertheless, LOY did not exert an influence on the longitudinal trends of NfL and GFAP and was not clearly associated with C9orf72 HRE status.

Conclusions

Our results indicate that LOY makes a potential contribution to the risk of ALS and the elevation of plasma NfL levels. While LOY’s impact on ALS AAO and survival requires further validation, these findings identify it as a promising sex‑specific therapeutic target and support its potential for stratifying male ALS patients toward personalized treatments.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12916-025-04594-x.

Background

Amyotrophic lateral sclerosis (ALS) is a fatal neurodegenerative disease characterized by the gradual loss of motor neurons. This leads to debilitating muscle atrophy, weakness, and the loss of voluntary movement [1]. At present, the cause and mechanism of ALS remain unclear [2–4], and there are no effective treatment measures [5]. The median survival time after symptom onset ranges from 2 to 5 years [3]. ALS shows a male predominance, with reported gender ratios varying from 1:1 to 3:1 (male–female), indicating that sex-specific factors might play a role in disease susceptibility [6]. Males with ALS generally experience earlier symptom onset [7] and have shorter survival times compared to females [8, 9]. The reasons for these sex differences remain unclear. However, understanding them could expedite the discovery of treatments, guide patient care, and improve clinical trial design.

The Y chromosome, being the male-specific sex chromosome, contains a specialized set of genes mainly devoted to testicular development and male reproductive function [10]. Loss of Y chromosome (LOY) is the most prevalent postzygotic genetic alteration in aging males. Its prevalence, which depends on age, exceeds 40% in men over 70 years old [11]. This widespread chromosomal aberration has become a biologically significant event with substantial multisystem clinical implications. Over the past decade, accumulating evidence has shown that LOY has a profound impact on immune system regulation and is mechanistically associated with various pathological processes [12], including cancers [13], neurodegenerative disorders [14, 15], cardiovascular disease [16], and altered responses to infectious pathogens [17]. Although LOY shows significant associations with various disease states, the etiological mechanisms remain elusive. Besides age, smoking may also contribute to the development of LOY in men [18]. Moreover, genome-wide association studies (GWASs) in UK and Japanese populations respectively identified 156 and 46 LOY-associated loci, both of which link LOY to blood cell differentiation [11, 19]. Despite the increasing recognition of LOY’s clinical significance in male aging and disease, its potential role as a risk modifier or prognostic factor in ALS remains unexplored. To date, no large-scale epidemiological or mechanistic studies have investigated whether LOY contributes to the susceptibility, onset time, or disease progression of ALS in male patients.

Thus, we put forward the hypothesis that somatic LOY might play a role in male-specific ALS susceptibility. To verify this hypothesis, we employed Y-chromosome probe intensity data (log-R ratios, LRRs) from UK Biobank genotyping arrays to quantify the levels of blood LOY. We further carried out a comprehensive evaluation of the role of LOY in ALS by assessing its associations with both disease susceptibility and clinical progression. Simultaneously, we explored its potential biological mechanisms through the analysis of established fluid biomarkers, such as glial fibrillary acidic protein (GFAP) and neurofilament light chain protein (NfL).

Methods

Study design and participants

The present analysis leverages the UK Biobank resource (under application 194,434), a population-based longitudinal study that has established a comprehensive biomedical database of over 500,000 participants who were recruited during 2006–2010. All participants provided informed consent and underwent detailed baseline assessments according to standardized protocols. From an initial cohort of 502,163 individuals, we identified 158,953 male participants after applying the following exclusion criteria: (1) lack of follow-up records, (2) missing genetic data, (3) incomplete covariate information, (4) the presence of sex chromosome abnormalities of 47XYY, and (5) a prior diagnosis of ALS at the time of blood sample collection (Fig. 1). Participants were prospectively followed until the occurrence of the first ALS diagnosis, death, loss to follow-up, or the study censoring date (September 2023), whichever came first.

Fig. 1.

The overall flow chart of this study. Abbreviations: ALS, amyotrophic lateral sclerosis; LOY, loss of Y chromosome; Cox PH, Cox proportional hazards analysis; CIC, cumulative incidence curve; TDI, Townsend deprivation index; HRE, hexanucleotide repeat expansion; AAO, age at onset; ANCOVA, analysis of covariance; PLR, piecewise linear regression; KM, Kaplan–Meier analysis; LMM, linear mixed models

Our study systematically investigated the association between LOY and ALS through four key analyses. First, we examined the impact of LOY on the risk of ALS. Then, we evaluated whether LOY modifies the age at onset (AAO) and survival duration in patients diagnosed with ALS during the follow-up. Finally, we assessed the relationship between LOY and the levels of NfL and GFAP (Fig. 1).

Estimation of LOY carrier

For all the participants included, we had access to both LRR and B allele frequency (BAF) measurements. The detection of Y-chromosome mosaicism was performed using LRR values. These values represent normalized total signal intensity measurements, which reflect relative copy number variations. The LOY was scored from the median LRR (mLRR) of approximately 691 SNP probes (Additional file 1: Table S1) in the male-specific region of Y chromosome (MSY) (ChrY: 2,658,271–28,767,492, GRCh37/hg19) obtained from genome-wide arrays in UK Biobank, as before [20, 21]. To detect potential MSY deletions, we quantified genomic abnormalities by analyzing the degree of divergence between an individual’s LRR measurements and the expected mLRR value. Typically, mLRR indicates a loss when it is negative and a gain when it is positive. Using the analytical framework interval for experimental noise in genotyping data established by Forsberg et al. [20], we determined the lower limit of the 99% confidence interval (CI) threshold for LOY detection. Samples below this cutoff were categorized as LOY carriers, whereas those above were considered non-carriers. Samples with mLRR > 0.15 were excluded because of potential chromosomal abnormalities (e.g., XYY) [22]. Based on the above method, a mLRR of MSY (mLRRY) threshold of − 0.096 was used to detect LOY carriers, reflecting an estimated 12.161% of affected cells (LOY (%) = 100 × (1–2^{2 × mLRRY})) [23].

Estimation of C9orf72 HRE status

The UK Biobank employed Illumina NovaSeq technology to conduct whole-genome sequencing (WGS) on blood samples from over 490,000 participants. ExpansionHunter [24] was utilized to identify C9orf72 hexanucleotide repeat expansion (HRE) carriers (UKB Field ID: 24,062), as previously reported [25]. This approach has been demonstrated to exhibit excellent performance in detecting C9orf72 HRE. Specifically, it shows 100% sensitivity and 99.8–99.9% specificity when compared to repeat-primed PCR [24]. As in our previous studies [26, 27], all individuals with > 30 repeats, which is the commonly adopted pathogenicity threshold, were designated as “C9orf72 HRE carriers.”

Measurement of NfL and GFAP

The plasma level of NfL and GFAP was quantified through the UK Biobank Pharma Proteomics Project (UKB-PPP) using the Olink Explore platform. Proteomic measurements were expressed as log2-transformed normalized protein expression (NPX) units, providing standardized quantification for cross-sample analytical comparisons. We extracted NPX values for NfL and GFAP obtained at baseline recruitment and during two subsequent follow-up visits for analysis. The details of Olink data processing can be found on the UK Biobank website (https://biobank.ndph.ox.ac.uk/showcase/refer.cgi?id=4656).

Ascertainment of ALS incidence, age at onset, and survival duration

To guarantee the accuracy of the diagnosis, ALS cases were identified only in hospital inpatient records using the International Classification of Diseases, Tenth Revision (ICD-10) coding system. These records incorporated admission and diagnostic data retrieved from the Hospital Episode Statistics database. The AAO of ALS cases was calculated as the time interval between the reported date of ALS and the participant's birth date. Survival duration was determined as the period from the reported ALS date to the date of death.

Covariates

Age at blood sampling, Townsend deprivation index (TDI, a composite measure of socioeconomic status incorporating social class, employment, car ownership, and housing conditions), smoking status, hypertension, and diabetes status were extracted from the UK Biobank database. The age at blood sampling for genotyping was calculated as the time interval between the first blood collection date and the participant’s birth date. The C9orf72 HRE status was obtain via the method described above. To account for potential confounders in the NfL and GFAP analyses, covariates were incorporated for baseline neurodegenerative conditions and age at sampling. The neurodegenerative conditions encompassed ALS, Alzheimer’s disease (AD), Parkinson’s disease (PD), vascular dementia, all-cause dementia, unspecified dementia, multiple sclerosis, multiple system atrophy, progressive supranuclear palsy, dystonia, and all-cause parkinsonism.

Statistical analysis

We employed both univariable and multivariable Cox proportional hazards (Cox PH) models to estimate hazard ratios (HRs) with 95% CI for ALS risk associated with LOY and ALS survival outcomes in relation to LOY status. The cumulative incidence curves and Kaplan–Meier survival curves were generated to visualize the outcomes. The Ramsey Regression Equation Specification Error Test (RESET) was to detect violations of linearity assumptions in regression models. Piecewise linear regression (PLR) models (unadjusted and covariate adjusted) were implemented to examine the relationships between LOY status and AAO of ALS, and longitudinal change of circulating level of neurodegeneration biomarkers (NfL and GFAP) was analyzed by LMM (linear mixed models). All continuous variables were assessed for normality using Shapiro–Wilk tests. Non-normally distributed variables were log-transformed when appropriate. For normally distributed data, between-group differences were analyzed using Student’s t-tests, while Wilcoxon rank-sum test was employed for non-normally distributed data. All statistical analyses were conducted using R software (version 4.2.2), with two-tailed p-values < 0.05 considered statistically significant.

Sensitivity analysis

1. To comprehensively evaluate potential effect modification, we conducted stratified analyses by age at blood sampling (< 60 vs ≥ 60 years), TDI (quartiles: Q1–Q4), and smoking history (never vs ever smoked), diabetes (with vs without), hypertension (with vs without), and C9orf72 HRE (> 30 vs ≤ 30 repeats) with formal interaction testing in multivariable models.

2. To guarantee a more precise assessment and minimize the risk of overestimating its corrective effect, we adopted a three-level categorization of the C9orf72 HRE, namely ≤ 30 repeats, 30–100 repeats, and > 100 repeats.

3. To address potential reverse causality, we performed sensitivity analyses using extended lag periods (1-, 3- and 5-year follow-up from blood sampling) in comparison to our primary no lag analysis.

4. We examined dose-dependent relationships by categorizing LOY into three tiers (< 12.161%, 12.161%–40%, and ≥ 40% of cells affected) [20] to assess potential threshold effects on the studied outcomes.

Results

Baseline characteristics

After strictly applying the inclusion and exclusion criteria, our final cohort comprised 158,953 male participants. The mean follow-up duration was 11.7 ± 3.5 years (Table 1). During the study period, a total of 297 incident cases of ALS were identified and confirmed through hospital-based diagnoses. Among these cases, 29 were carriers of C9orf72 HRE. For these ALS cases, the mean AAO was 68.6 ± 7.8 years (Additional file 2: Table S1). Survival analysis indicated the mean follow-up duration of 1.6 ± 2.2 years for ALS patients, during which 257 deaths (86.5% of cases) were recorded. The median survival time from ALS reported was 8.0 months (Additional file 2: Table S1). For biomarker analyses, baseline measurements of GFAP were available for 16,503 participants, 122 of whom subsequently developed ALS. Baseline measurements of NfL were available for 16,511 participants, 125 of whom developed ALS. Longitudinal data were accessible for a subset of participants (n≈500) with repeated measurements of GFAP and/or NfL across two follow-up visits (Additional file 2: Table S2).

Table 1.

Baseline characteristics of study participants stratified by ALS status during follow-up

Characteristics	Overall	Non-ALS	ALS	p-value
Number of participants, n	158,953	158,656	297
Age at baseline, years (mean ± SD)	56.952 ± 7.998	56.945 ± 7.997	60.892 ± 6.869	< 0.001
TDI (mean ± SD)	− 1.527 ± 3.004	− 1.528 ± 3.004	− 1.297 ± 3.091	0.199
Hypertension (%)	49,049 (30.9%)	48,954 (30.9%)	95 (32.0%)	0.720
Diabetes (%)	6365 (4.0%)	6360 (4.0%)	5 (1.7%)	0.058
mLRRY (median)	− 0.002	− 0.002	− 0.016	0.013
Ever smoke, n (%)	104,725 (65.9%)	104,526 (65.9%)	199 (67.0%)	0.729
C9orf72 HRE (G4C2 > 30) (%)	349 (0.2%)	320 (0.2%)	29 (9.8%)	< 0.001
The percent of blood cells with LOY, % (median)	1.2	0.9	1.6	0.013
LOY status (%)				< 0.001
Loss	5888 (3.7%)	5864 (3.7%)	24 (8.1%)
Normal	153,065 (96.3%)	152,792 (96.3%)	273 (91.9%)
Follow-up duration, years (mean ± SD)	11.716 ± 3.486	11.724 ± 3.481	7.614 ± 3.670	< 0.001

Abbreviations: ALS amyotrophic lateral sclerosis, SD standard deviation, TDI Townsend deprivation index, mLRRY median Log2 R ratios of Y chromosome, HRE hexanucleotide repeat expansion, G4C2 GGGGCC, LOY loss of Y chromosome; between-group comparisons were performed using the Wilcoxon rank-sum test for LOY-positive blood cell percentages and mLRRY, and either Student’s t-test or χ² tests for other variables, with statistical significance defined as p < 0.05

Association between LOY and risk for ALS

To evaluate the association between blood LOY levels and the risk of ALS, we performed a prospective cohort analysis using data from the UK Biobank. Additional file 3: Fig. S1A depicts the histogram of the frequency distribution of mLRRY in the UK Biobank. Based on this, a participant was considered a LOY carrier when the mLRRY was less than − 0.096. Among 158,953 male participants who were free of ALS diagnosis at the time of baseline blood collection, 5888 (3.7%) participants were identified as LOY carriers (Fig. 2A, Table 1). We also observed a significant positive correlation between LOY carrier status and increasing age (Additional file 3: Fig. S1B). Moreover, compared with non-ALS controls, the mLRRY value in participants who subsequently developed ALS was significantly reduced (p < 0.001, Fig. 2B). But we did not find the association between mLRRY value and C9orf72 HRE status (Additional file 3: Fig. S1C).

Fig. 2.

Identification of LOY carriers in the male participants of the UK Biobank using the mLRRY threshold. A Scatter plot of mLRRY distribution. The red line shows the 99% confidence interval, and the LOY cutoff is shown by the red line on the left (mLRRY = − 0.096). The red dots represent LOY carriers, and blue dots represent normal participants. B Box plot of distribution of baseline mLRRY in male ALS and non-ALS group. The difference between groups was not statistically significant as assessed by the Wilcoxon rank-sum test (p < 0.001). The central black line represents the median, box boundaries show the interquartile ranges (IQR, 25th–75th percentiles), and outliers are shown as individual points. Abbreviations: ALS, amyotrophic lateral sclerosis; LOY, loss of Y chromosome; mLRRY, median Log2 R ratios of Y chromosome

Cumulative incidence analysis indicated that the risk of ALS was significantly higher in individuals with LOY than in non-carriers (Gray’s test, p < 0.001, Fig. 3A). Additionally, a univariate Cox PH regression analysis was performed, revealing that LOY carriers had a significantly elevated risk of developing ALS compared to non-carriers (HR [95% CI]: 2.507 [1.651–3.805], p < 0.001) (Additional file 2: Table S3). More significantly, after comprehensive adjustment for potential confounders including age at blood sampling, TDI, diabetes, hypertension, smoke status, and C9orf72 HRE status, the multivariable Cox PH model also demonstrated a significant association between LOY carrier status and the risk of incident ALS (HR [95% CI]: 1.619 [1.059–2.475], p = 0.026) (Fig. 3B).

Fig. 3.

The association between LOY and risk for ALS. A Cumulative incidence curve for ALS incident by LOY status. Gray’s test (p < 0.001). B The forest plot between LOY and risk for ALS via multivariable Cox PH. Abbreviations: LOY, loss of Y chromosome; TDI, Townsend deprivation index; HR, hazard ratio; 95% CI, 95% confidence interval; HRE, hexanucleotide repeat expansion

Association between LOY and AAO of ALS

Among the 297 participants diagnosed with ALS, LOY carriers had a significantly higher average AAO compared to non-carriers (71.61 ± 6.53 years vs. 68.38 ± 7.91 years, respectively; t-test, p = 0.032, Additional file 2: Table S1) (Fig. 4A). Nevertheless, after adjusting for potential confounders, including age at blood sampling, TDI, diabetes, hypertension, smoking status, and C9orf72 HRE status, no significant association was detected between LOY status and ALS AAO (Fig. 4A, ANCOVA (analysis of covariance), p = 0.662). We further investigated the association between mLRRY (analyzed as a continuous variable) and the AAO of ALS. Given the non-normal distribution of mLRRY with potential breakpoints (RESET, p = 0.047, Additional file 3: Fig. S2), we performed threshold-based piecewise linear regression at mLRRY = − 0.096 (the established LOY carrier cutoff). Although there was a certain positive linear correlation trend between the onset age in LOY carriers and mLRRY, no significant association between mLRRY and AAO was observed in either regression segment in LOY carriers and non-LOY carriers (Fig. 4B, Additional file 2: Table S4).

Fig. 4.

The association between LOY and AAO survival for ALS. A Box plot of AAO of AAO in LOY and non-LOY group. Student’s t-test revealed a statistically significant difference between groups (p < 0.001), whereas ANCOVA adjusting for age at blood sampling, TDI, diabetes, hypertension, smoking status, and C9orf72 HRE showed no significant effect (p = 0.662). The central black line represents the median, box boundaries show the interquartile range s (IQR, 25th–75th percentiles), and outliers are shown as individual points. B The result of PLR by mLRRY = − 0.096. For samples with mLRRY < − 0.096, neither the univariate linear regression (unadj.p1 = 0.586) nor the multivariate linear regression (adjusted for age at blood sampling and C9orf72 HRE, adj.p1 = 0.156) reached statistical significance. For samples with mLRRY ≥ − 0.096, the univariate linear regression showed a statistically significant association (unadj.p1 = 0.013), whereas this association was attenuated to non-significance in the multivariate model after adjusting for age at blood sampling and C9orf72 HRE (adj.p1 = 0.817). C Kaplan–Meier survival curve for ALS by LOY Status.log-rank test, p = 0.082. D The forest plot between LOY and survival for ALS via multivariable Cox PH. Abbreviations: ALS, amyotrophic lateral sclerosis; LOY, loss of Y chromosome; ANCOVA, analysis of covariance; PLR, piecewise linear regression; AAO, age at onset; HR, hazard ratio; 95% CI, 95% confidence interval; HRE, hexanucleotide repeat expansion

Association between LOY and survival of ALS

Among the 297 ALS patients included in survival analysis, 257 mortality events (86.5%) were recorded during the follow-up period. Kaplan–Meier analysis demonstrated no statistically significant association between LOY status and ALS survival (log-rank test, p = 0.082, Fig. 4C). The median survival of ALS patients with LOY was 6.912 months, approximately 1.5 months shorter than that of patients without LOY (8.340 months) (Additional file 2: Table S1). This finding was consistent in multivariate Cox PH regression adjusting for AAO, age at blood sampling, and C9orf72 HRE status, which showed no significant survival difference between LOY carriers and non-carriers (HR [95% CI]:1.413 [0.912–2.187], p = 0.122) (Fig. 4D).

Association between LOY and GFAP

Among all the participants, 16,503 individuals, including 696 (4.3%) LOY carriers with baseline GFAP measurements, were involved. Eventually, 122 of them subsequently developed ALS. At baseline, in unadjusted analysis, LOY carriers demonstrated significantly higher plasma GFAP levels compared to non-carriers (mean ± SD: − 0.051 ± 0.540 vs. 0.085 ± 0.552, p < 0.001). However, after adjusting for age at blood sampling and neurodegenerative condition, this difference was no longer statistically significant (ANCOVA: p = 0.723) (Fig. 5A). When using mLRRY as a continuous variable, a significant negative correlation with GFAP levels was observed (Pearson’s r = − 0.059, p < 0.001) (Fig. 5B). Longitudinal analysis indicated a significant age-associated increase in plasma GFAP levels (p < 0.001). Nevertheless, after adjusting age at blood sampling and neurodegenerative condition, the rate of GFAP accumulation did not differ significantly between LOY carriers and non-carriers (interaction β = − 0.022, 95% CI [− 0.0625, 0.0172], p = 0.282), as presented in Fig. 5C. Additional file 3: Fig. S3A indicates no difference of GFAP at baseline between ALS and non-ALS groups (p = 0.051). Additionally, when comparing ALS patients with LOY (n = 11) to those without LOY (n = 111), no statistically significant difference in baseline GFAP concentrations was detected (p = 0.720, Additional file 3: Fig. S3B and Additional file 2: Table S5).

Fig. 5.

The association between LOY and GFAP/NfL. A Box plot of distribution of baseline GFAP in LOY carriers and normal group. Student’s t-test revealed a statistically significant difference between groups (p < 0.001), whereas ANCOVA adjusting for age at blood sampling (> 60 vs ≤ 60) and neurodegenerative conditions showed no significant effect (p = 0.723). The central black line represents the median, box boundaries show the interquartile ranges (IQR, 25th–75th percentiles), and outliers are shown as individual points. B Correlation between mLRRY and plasma GFAP level. Unadjusted: r = − 0.059, p < 0.001. Age (per year) and neurodegenerative conditions adjusted: adj.p = 0.017. C Temporal changes in mean (± SE) plasma GFAP level in LOY carriers and normal group. The rate of GFAP accumulation did not differ significantly between LOY carriers and non-carriers via LMM (unadj.p1 < 0.001, adjusting age at blood sampling (per year), and neurodegenerative conditions: adj.p1 = 0.264). D Box plot of distribution of baseline NfL in LOY carriers and normal group. Student’s t-test revealed a statistically significant difference between groups (p < 0.001), and ANCOVA adjusting for age at blood sampling (> 60 vs ≤ 60) and neurodegenerative conditions also showed significant effect (p = 0.004). The central black line represents the median, box boundaries show the interquartile ranges (IQR, 25th–75th percentiles), and outliers are shown as individual points. E Correlation between mLRRY and plasma NfL level. Unadjusted: r = − 0.11, p < 0.001; age (per year) and neurodegenerative conditions adjusted: adj.p = 0.336. F Temporal changes in mean (± SE) plasma NfL level in LOY carriers and normal group. The rate of NfL accumulation did not differ significantly between LOY carriers and non-carriers via LMM (unadj.p1 < 0.001, adjusting age (per year) and neurodegenerative conditions: adj.p1 = 0.458). Abbreviations: LOY, loss of Y chromosome; GFAP, glial fibrillary acidic protein; NPX, normalized protein eXpression; NfL, neurofilament light chain protein; SE, standard error; ANCOVA, analysis of covariance; LMM, linear mixed models

Association between LOY and NfL

Among all the participants, 16,511 individuals, including 702 (4.3%) LOY carriers with baseline plasma NfL measurements, were involved. Eventually, 125 of them developed ALS. LOY carriers had significantly higher NfL levels compared to non-carriers (mean ± SD: 0.292 ± 0.555 vs. 0.044 ± 0.591, p < 0.001). This difference persisted even after adjusting for the age at blood sampling and neurodegenerative condition adjustment (ANCOVA, p = 0.004) (Fig. 5D). When analyzed as a continuous variable, mLRRY was found to have a significant negative correlation with NfL levels (Pearson’s r = − 0.11, p < 0.001). However, this association became non-significant after adjusting for age (adjusted p = 0.336) (Fig. 5E). Additionally, after adjusting for age at sampling, LOY status did not show a significant association with the rate of longitudinal NfL change (interaction β = 0.015, 95% CI [− 0.025, 0.056], p = 0.458) (Fig. 5F). Additional file 3: Fig. S3C displays the baseline NfL distributions between the ALS and non-ALS groups during the follow-up period (p = 0.070). Similarly, no significant difference in NfL levels was observed between ALS patients with (n = 10) versus without LOY (n = 115) (p = 0.870; Additional file 3: Fig. S3D, Additional file 2: Table S5).

Sensitivity analysis

The primary association remained robust across subgroups categorized by age at blood sampling, socioeconomic status (TDI), comorbidities (diabetes, hypertension), smoking status, and C9orf72 HRE status (Additional file 2: Table S6). When C9orf72 repeat expansion was considered as a three-level covariate (≤ 30, 30–100, and > 100 repeats), the results remain consistent (Additional file 2: Table S7). Sensitivity analysis employing different lag periods (1-, 3-, and 5-year intervals) produced relatively consistent results (Additional file 2: Table S8). Stratified analyses further characterized the dose–response relationships between LOY burden (low/high levels) and the risk of ALS (Additional file 2: Table S9).

Discussion

The reasons why men are more significantly affected by ALS than women, as well as why the survival time of men with ALS is shorter, remain poorly understood. The results of our study based on the UK Biobank provide the first evidence that LOY is associated with an increased risk of ALS. For the first time, this study also explores the relationships between LOY and disease onset, survival time, and neurodegeneration markers.

Previous studies have reported the linked between LOY and AD [15] or PD [14]. In this large prospective cohort study, we propose that higher LOY levels are associated with a slightly increased risk of developing ALS. Our findings expand its role to ALS, suggesting that LOY may be a pan-neurodegenerative risk factor. LOY has been found to be age related in the hematopoietic cells [20], and age is an unavoidable risk factor for ALS [3, 28, 29]. Beyond the well-documented association with aging, emerging evidence from cytogenetic studies implicates several intrinsic vulnerabilities in Y-chromosome loss, including genetic factors [11, 19], chromosome mis-segregation trapped in micronuclei [30], telomere shortening [31], structural instability conferred by the absence of CENP-B (centromere protein B)-binding elements [32], and reduced alpha-satellite repeats at the centromeric region [12]. Notably, these molecular mechanisms exhibit remarkable similarities to established pathways in ALS pathogenesis [33, 34]. This suggests that LOY might serve as one of the mediating factors of age. Nevertheless, further data analysis is needed. Furthermore, there seems to be some overlap between LOY-related genes and certain ALS-modifying genes, particularly genes associated with DNA damage, such as PARP1 (poly(ADP-ribose) polymerase 1) [35] and TP53 (tumor protein P53) [11, 36]. This indicates that there may be specific gene co-regulation between LOY and ALS. However, due to the limited number of ALS cases in our cohort, our study was unable to detect a statistically significant increase in ALS risk across different lag periods (1-, 3-, and 5-year intervals) or dose–response relationships between LOY burden (low/high levels) and ALS risk after adjusting for multiple factors. However, we did observe a consistent trend toward elevated risk. This potential association merits further investigation in larger ALS cohort studies. Furthermore, our analysis also took into account C9orf72 HRE status, which has been verified as a crucial factor promoting the risk of ALS. Nevertheless, no significant correlation was observed between C9orf72 HRE status and LOY.

ALS patients typically have an average AAO of over 60, with men showing an earlier disease onset compared to women [29]. The mean age at ALS onset in our cohort (68.6 years) is consistent with a previous report from the UK Biobank [37]. In unadjusted analyses, LOY carriers demonstrated a modest yet statistically significant delay in disease onset compared to non-carriers. However, this association did not hold in adjusted models, probably due to the limited statistical power resulting from the small LOY carrier subgroup (n = 24). Notably, in patients with LOY, lower mLRRY values were associated with premature ALS onset, implying that LOY might accelerate disease manifestation when it is below specific molecular thresholds. To our knowledge, no previous studies have systematically examined the relationship between pre-diagnostic LOY status and AAO across major neurodegenerative disorders. Nevertheless, the relationship between LOY and ALS onset age remains unclear and requires investigation in larger, adequately powered cohorts. Consistent with the previous study [25], we also did not identify an association between C9orf72 HRE status and an earlier age at onset among males.

As a fatal disease, ALS is characterized by a very short survival period [3]. Although not statistically significant, ALS patients with LOY showed a tendency towards shorter survival. Previous studies have reported that LOY was associated with more rapid motor impairment and cognitive decline in PD [14]. In addition, LOY has been widely linked to immune dysfunction and chronic inflammation [12], both of which are involved in ALS risk and disease progression [38–40]. It warrants further mechanistic exploration whether LOY contributes to ALS pathogenesis and progression through immune-mediated mechanisms. Owing to the population-based design of the UK Biobank, data regarding the ALS onset site (spinal or bulbar)—a known factor influencing survival [3]—were not accessible, which may potentially confound the study outcomes. Consistent with our previous work [2, 3], the associations between C9orf72 status, AAO, and ALS survival were also replicated in male subjects. Moreover, large-scale longitudinal analysis of the association between LOY and the decline rates of the ALS Functional Rating Scale-Revised (ALSFRS-R) may provide more critical insights into the dynamics of ALS progression.

We further explore the relationship between LOY and potential markers of ALS. LOY exhibits a strong correlation with elevated plasma NfL levels, a key biomarker of ALS risk assessment and prognosis. Despite the fact that NfL serves as a general biomarker that is elevated in numerous neurodegenerative diseases [41–43], our analysis reveals that even after preliminary adjustment for several such conditions, NfL levels are significantly higher in LOY carriers. Thus, our finding reveals a novel link between LOY and neurodegenerative processes. When analyzed as a continuous variable, lower mLRRY measurements may contribute to increased plasma NfL levels as well. We also noted an upward trend in pre-symptomatic NfL levels trend among male ALS patients, which is consistent with previous findings by Smith et al. [44]. Typically, NfL levels increase prior to the onset of the disease and subsequently remain stable throughout its course [45]. It is worthwhile to utilize the potential NfL cutoff points to distinguish ALS progression status (i.e., fast, intermediate, slow) and to determine whether the LOY status affects NfL levels associated with each subgroup or is more prevalent in any give subgroup. Our current study, however, lacked the statistical power to test this hypothesis owing to the limited sample size of ALS cases. This underscores the necessity for larger-scale studies to explore these nuanced relationships. Moreover, it has been reported that regulatory T cells had significantly more LOY than any other studied T lymphocytes subtype [13, 46, 47], and T-cell abnormalities are also prevalent in ALS [48, 49]. The interaction between LOY T cell and neuron might contribute to the elevation of NfL, similar to the situation in multiple sclerosis [50]. However, the spatial distribution and cell-type specificity of LOY in ALS-affected CNS (central nervous system) tissues, particularly in the motor cortex and anterior horn regions, still need to be systematically characterized at single-cell resolution level. Our primary analysis indicated no significant association between LOY/ALS status and plasma GFAP levels. However, continuous analysis of mLRRY demonstrated a statistically significant but weak negative correlation with GFAP levels. This modest association may reflect the relatively lower disease specificity of GFAP for ALS compared to other neuroinflammatory conditions [51–53]. This is further supported by its pronounced age-dependent dynamics. Additionally, NfL levels exhibited significant age-associated changes only in the LOY group. Overall, considering the limited sample size in the ALS subgroup, our analysis did not reveal significant associations between LOY status and either NfL or GFAP levels.

Significantly, our study has some advantages. Firstly, our research is the first to comprehensively examine the role of LOY in mediating sex-specific differences in ALS risk, clinical progression, and biomarker profiles. Secondly, the UK Biobank encompasses approximately over 240,000 male participants (baseline age 40–69 years), providing sufficient statistical power for the assessment of LOY. Thirdly, the extended follow-up period in the UK Biobank, combined with serial measurements of NfL/GFAP data, enables a robust analysis of the temporal effects of LOY on their longitudinal changes. Nevertheless, our study has several limitations. Firstly, as LOY is an age-dependent somatic mutation that accumulates gradually, our cross-sectional measurement at a single time point restricts the assessment of an individual’s LOY trajectory and its temporal relationship with ALS risk. Secondly, the limited number of incident male ALS cases (n = 297) in the UK Biobank cohort may decrease statistical power to detect modest associations, especially for AAO, survival, and subgroup analyses requiring stratification by LOY status. Thirdly, as a population-based cohort, the UK Biobank does not routinely collect granular clinical data for specific conditions like ALS, including the site of onset or ALSFRS-R scores. Last but not least, although our longitudinal analysis was conducted exclusively in individuals of European ancestry, validation in multiethnic populations is necessary to establish broader applicability of these findings.

Conclusions

In conclusion, the significant sex difference in ALS is related to a male-specific genetic risk factor, i.e., LOY. This data-driven investigation underscores the possible functions of LOY in male ALS patients. Our analysis findings suggest that LOY serves not only as a marker of aging in males but also as a contributing factor to the risk of ALS and elevate plasma NfL levels. Nevertheless, the influence of LOY on the AAO and survival in ALS requires additional validation. Understanding how LOY level affects ALS may aid in identifying biomarkers and developing effective treatments for male ALS patients.

Abbreviations

AAO

Age at onset

AD

Alzheimer’s disease

ALS

Amyotrophic lateral sclerosis

ALSFRS-R

The ALS Functional Rating Scale-Revised

ANCOVA

Analysis of covariance

CENP-B

Centromere protein B

CI

Confidence interval

Cox PH

Cox proportional hazards

GFAP

Glial fibrillary acidic protein

GWASs

Genome-wide association studies

HR

Hazard ratio

HRE

Hexanucleotide repeat expansion

ICD-10

The International Classification of Diseases, Tenth Revision

LMM

Linear mixed models

LOY

Loss of Y chromosome

mLRR

Median Log2 R ratios

mLRRY

Median Log2 R ratios of Y chromosome

MSY

Male-specific region of Y chromosome

NDDs

Neurodegenerative disorders

NfL

Neurofilament light chain

NPX

Normalized protein eXpression

PARP1

Poly(ADP-ribose) polymerase 1

PD

Parkinson’s disease

PLR

Piecewise linear regression

RESET

Ramsey Regression Equation Specification Error test

TDI

Townsend deprivation index

TP53

Tumor protein P53

UKB-PPP

UK Biobank Pharma Proteomics Project

WGS

Whole-genome sequencing

Authors' contributions

WMS: conception, methodology, analysis, and writing of the manuscript. QQD: methodology, analysis and validation. SYH, RYL, XJW, NZ and TC: validation. BC and YPC: wrote, revised and discussed the final edition. All authors read and approved the final version of the manuscript.

Funding

This study was supported by the National Natural Science Fund of China (Grant nos. 82371422 and 81971188 to Y. P. C.), the National Key Research and Development Program of China (Grant no. 2022YFC2703101 to Y. P. C.), and the 1·3·5 project for disciplines of excellence Clinical Research Fund, West China Hospital, Sichuan University (Grant no. 2023HXFH032 to Y. P. C.).

Data availability

This research was conducted using the UK Biobank data under application number 194434. Data access requests should be made to the UK Biobank.

Declarations

Ethics approval and consent to participate

UK Biobank’s ethical approval was granted by the North West Multi-Centre Research Ethics Committee in 2011 and is renewed every 5 years. The current approval is valid through the renewal application in 2021 (reference ID: 21/NW/0157). This broad authorization enables researchers to conduct studies without needing additional ethical review, as they fall within the scope of the existing approval. All participants provided informed consent for their anonymized data to be utilized in health-related research serving the public interest.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.