henhaobio
Sign in

medRxivGenetic & Genomic MedicinePreprint

Investigating the Y chromosome in complex disease: Phenome-wide scan across 104,334 Finnish men

Annina PreussneriD, Jaakko T. LeinoneniD, FinnGeniD, Aoxing LiuiD, Matti PirineniD, Taru TukiaineniD

PDFDOIData

In-depth summary

Claude Opus 5.5 · xhigh effort · 8 Nov 2026 · from the full text of version 2 (17 Jun 2026) · CC BY 4.0

Phenome-wide scan of Y-chromosomal variants against 1,426 traits in 104,334 Finnish men finds no significant hits; kin-based validation supports haplogroup I1 increasing coronary heart disease risk.

The Y chromosome is routinely excluded from genome-wide association studies (GWAS) because absent recombination, haploidy and strong stratification complicate analysis, and reported germline Y associations, such as haplogroup I1 with atherosclerosis in UK data, remain inconsistent. This work tests common Y variation phenome-wide in Finnish men, including the regionally restricted N1a1 lineage, with kin-based and geographic checks against confounding.

Key results

  1. ShownTesting nine haplogroup-tagging Y SNPs against 1,426 traits gave no phenome-wide significant association but 121 suggestive signals across many disease categories; among the major haplogroups, I1 and R1b each had 18, while N1a1 and R1a had the fewest.

    Bonferroni p<3.9x10^-6: none; suggestive p<5.6x10^-3: 121 (I1 18, R1b 18, I1a2-Z73 16, R1a1-Z284 10, N1a1 nine, R1a four). 80% power to detect OR from 1.13 for traits with ~10,000 cases (e.g., asthma). Traits with strong socioeconomic or regional patterning, such as number of children, showed no signals. · Significance threshold and power calculations; Y-chromosomal variation in FinnGen shows…

  2. SuggestedHaplogroup I1 had suggestive cardiometabolic associations. The coronary heart disease (CHD) signal was echoed in first-degree male relatives but not female relatives, as expected for a Y-linked effect, whereas the type 2 diabetes (T2D) signal was matched in neither.

    I1-CHD PheWAS OR=1.06 (95%CI 1.02-1.10) vs male relatives OR=1.05 (0.95-1.15, PP=0.57) and female relatives OR=0.97 (0.84-1.13, PP=0.36). I1-T2D PheWAS OR=1.06 (1.02-1.10) vs male OR=1.02 (PP=0.39) and female OR=1.03 (PP=0.48). Discussion gives I1-CHD p=3.5x10^-3. · Kin-based validation supports several suggestive…; DISCUSSION

  3. ShownThe I1-CHD estimate stayed stable across geographic strata, after adjustment for an autosomal CHD polygenic risk score (PRS), smoking or loss of Y (LOY), and under alternative reference haplogroups. This argues against stratification as the explanation.

    Northeast OR=1.08 (1.01-1.15) vs southwest OR=1.05 (1.01-1.10), z-test p=0.48; 20 sub-regions Q-test p=0.75; PRS-adjusted OR=1.06 (p=7.1x10^-3); smoking-adjusted OR=1.06 (p=0.053); LOY-adjusted OR=1.05 (p=0.071); adding haplogroup improved fit (LRT p=4.0x10^-3); OR=1.02-1.07 across reference haplogroups. · Adjusting for possible CHD confounders; Association between cardiovascular disease and…

  4. SuggestedKin-cohort validation classed 32% of followed-up suggestive associations as Y-consistent (male kin concordant, female kin null). The rest tracked female relatives, both sexes or neither, pointing to shared environment or chance. R1b links to rectal cancer held up in male kin.

    Concordant: 38/120 (32%); I1 38%, R1b 44%. Of the other 82: concordant with female kin only N=28, with both N=10, with neither N=44. R1b with rectal adenocarcinoma OR=1.62 (p=6.6x10^-4); male relatives OR=1.52, female relatives OR=0.63. Relative cohorts: N=24,712 male, N=19,275 female. · Kin-based validation - Effect size comparison…; Kin-based validation supports several suggestive…; DISCUSSION

  5. ShownLOY prevalence did not differ by haplogroup, but rare protein-altering Y variants were unevenly spread. N1a1 carried far fewer, and I1 carriers showed missense enrichment in KDM5D, driven by one Finnish-enriched variant.

    Age-adjusted LOY 15.9% (any) and 6.9% (large), haplogroup chi-square p=0.05 and p=0.06. Whole-exome data (9,736 samples): 300 Y variants in 45 genes, 31 missense and seven LoF; protein-altering variants per sample 0.004 in N1a1 vs 0.019 to 0.022 in others (p=1.43x10^-8); 18 of 20 KDM5D carriers had chrY:19706254 A>G. · Exome sequencing data QC and annotation; Loss of Y chromosome (LOY); LOY and protein-coding variation on Y…

Methods, data and reproducibility

REGENIE step-2 regressions, rescaled to haploid coding, of nine haplogroup-tagging Y SNPs on 1,426 FinnGen endpoints in unrelated men, adjusted for age, batch, PCs 1-20 and birth region. Suggestive hits were followed up by kin-cohort validation with Bayesian linemodels, plus regional, PRS, smoking, LOY and exome analyses.

  • FinnGen release 12 unrelated Finnish-ancestry men genotyped on custom array v2 (from 140,234 males; 554 Y markers before QC, 60 SNPs at >=1% frequency after QC) · human · n = 104,334
  • FinnGen endpoints: binary diagnoses with at least 100 cases plus 13 quantitative traits (height, weight, offspring number, 10 laboratory measures) · human · n = 1,426 traits (1,413 binary)
  • First-degree male relatives used as positive-control kin validation cohort · human · n = 24,712
  • First-degree female relatives used as negative-control kin validation cohort · human · n = 19,275
  • FinnGen whole-exome sequencing subset used to test rare Y coding-variant enrichment by haplogroup · human · n = 9,736 samples; 300 variants
  • LOY calls from FinnGen R9 generated with MoChA · human · n = 56,715 (LOY in >=10% of cells); 64,150 (any LOY)
  • Subsets with smoking or LOY status for CHD confounder-adjusted models · human · n = 64,244 (smoking); 64,249 (LOY)
  • Regional strata for the I1-CHD analysis: northeast vs southwest Finland · human · n = 40,328 (NE); 62,834 (SW)
Methods & tools
REGENIE (step 2 only; effects rescaled to haploid coding), linemodels (R package, Bayesian concordance models), R 4.5.0, Yhaplo (haplogroup calling), MoChA (loss-of-Y calls), KING kinship coefficient, YFull and ISOGG haplogroup annotation, gnomAD, CHD polygenic risk score, Estimated marginal means for age-adjusted LOY prevalence, FinnGen custom genotyping array v2, Whole-exome sequencing
Reagents & environment
R version 4.5.0 for LOY, CHD confounder and linemodels analyses; REGENIE used (version not stated); Y markers aligned to GRCh38.
Reproducibility gaps
Individual-level FinnGen genotype and registry data require an access application; no analysis code or repository is provided; detailed per-association results are in supplementary Tables S1-S9.
Limitations (6)
  • The array tagged only nine phylogenetically distinct Y lineages and missed 98/100 common Finnish sublineages (e.g., N1a1 lineages N-VL29 and N-Z1934). This restricts inference to main haplogroups and may hide N1a1 effects. · stated by the authors
  • The male-specific Y is a single non-recombining haplotype, so fine-mapping and colocalization are essentially unusable. The KDM5D lead conflicts with prior UTY findings, and I1 carriers in the exome data are young (mean age 44), with no CHD among variant carriers, so causality is unresolved. · stated by the authors
  • Male-only samples, lower Y allelic dosage and a few highly correlated variants limit power, and Bonferroni correction may be too strict. Lack of phenome-wide significance therefore does not argue against Y effects. · stated by the authors
  • Only SNPs were analysed. Structural and copy-number variation in Y repeats, amplicons and palindromes, which could alter gene dosage or regulation, was not assessed. · stated by the authors
  • No external replication cohort was analysed. Support for I1-CHD rests on modest kin evidence (male-relative PP=0.57, OR CI 0.95-1.15), and the smoking- and LOY-adjusted CIs reach 1.00, so the signal remains suggestive. · noted in this summary
  • Some reported values are not reconciled: 60 vs 58 common Y SNPs after QC, a suggestive threshold given as both 5.5x10^-3 and 5.6x10^-3, and an HbA1c 95% CI (-0.11 to 0.08) that spans zero despite p=1.6x10^-3. · noted in this summary

Read this if you study Y-chromosome or male-specific genetics in biobanks and want a practical approach (haploid rescaling in REGENIE, kin-cohort validation) plus confounder-checked evidence on haplogroup I1 and CHD.

Citations
0

Abstract

Although the Y chromosome represents roughly 2% of the male genome, it is often ignored in genome-wide association studies (GWAS). Subsequently, the potential health impacts of Y-chromosomal genetic variation remain incompletely understood. To fill this gap, we performed a phenome-wide association study (PheWAS) in FinnGen across 1,426 binary and quantitative traits using Y-chromosomal variation (frequency [≥] 1%) in 104,334 genotyped men. As Y chromosome variation is prone to population stratification, we performed carefully adjusted association analyses and further examined these through kin-based validation in 19,275 female and 24,712 male 1st degree relatives. We found 121 suggestive (p View larger version (26K): org.highwire.dtl.DTLVardef@1ca3f30org.highwire.dtl.DTLVardef@fa9557org.highwire.dtl.DTLVardef@16292a4org.highwire.dtl.DTLVardef@8978c7_HPS_FORMAT_FIGEXP M_FIG C_FIG Graphical Abstract Overview of the study. Using a dataset of 104,334 unrelated genotyped Finnish men we performed a PheWAS across 1,426 binary and quantitative traits and common Y-chromosomal variation (MAF [≥] 1%), adjusting for geographical confounding factors known to vary within the population. To confirm suggestive associations identified in the PheWAS, we performed kin-based validation to infer whether the association is likely linked to the Y chromosome, using the samples 1st degree male (positive control) and 1st degree female (negative control) relatives phenotype data.

Reproducibility ledger

Nobody has reported trying to reproduce this paper yet. Did you replicate an experiment or run the code? Your report saves the next lab a week.

Sign in to file a reproducibility report.

Related papers

Similarity Map
SelectedSame fieldCross-field
THISErzurumluog '16Davis '25Grenn '22Gao '15Sun '23Aquino '24
62% semantic matchGenetics
A. Mesut Erzurumluoglu, Denis Baird, Tom G Richardson+2 more·bioRxiv·Apr 19, 2016