Fetal sex confounds placental transcriptomic signatures of gestational diabetes: a six-cohort audit of public data

Background The placenta is sexually dimorphic: fetal sex shapes placental gene expression, growth and immunity. Public placental transcriptomes are increasingly reused to define gestational diabetes mellitus (GDM) signatures; whether fetal sex can be separated from maternal hyperglycemia and fetal overgrowth in these datasets has not been tested. Methods Six public cohorts (bulk and single-cell RNA-seq, spatial transcriptomics, expression array) were reanalyzed under an a priori-defined Plan–Do–Check–Act audit (C1–C17). Fetal sex was inferred per sample from seven Y-linked markers and XIST; sex-linked, copy-number and segmental-duplication genes were excluded; X enrichment was tested rather than assumed; signatures were tested for cross-cohort transfer; compartment mapping used within-sample correlations in four control villi; cis-Mendelian randomization was exploratory. Results In the only cohort with a growth contrast, sex separated the groups completely: both cases were female (Y-marker counts 24 and 12; XIST 14,129 and 18,191), whereas the control and the insulin-treated sample were male (Y 29,124 and 26,300). Twelve of the 20 largest uncorrected effects were Y-linked or copy-number artifacts. Under the corrected specification the overgrowth-associated set fell from 2,930 to 1,872 genes, and the growth-discordant set (1,014 mapped genes) was unstable across leave-one-out (1,376 vs 665; Jaccard 0.55–0.70). The overgrowth-associated set showed no X enrichment (1.02×, P = 0.89). Transferability was limited: subtype signatures did not transfer to an independent hyperglycemia cohort (rho − 0.03 to − 0.06) and replicated only weakly on an array (rho 0.105–0.132); the shared-direction set reversed in both external cohorts; mitochondrial-RNA down-regulation tracked preeclampsia (13/13) rather than GDM (8/13). In control villi the overgrowth-associated set localized to cytotrophoblast (r = 0.42) and the growth-discordant set to fibroblast (r = 0.40). Conclusions In public placental transcriptomes fetal sex is not a covariate that can be adjusted away: it is confounded with, and in the critical arm identical to, the exposure contrast. GDM signature and sex-difference claims derived from single-sex arms are non-identifiable. Cohorts should report fetal sex and growth per sample, balance sex within arm, and pre-specify sex-aware filters.

Authors

Institutions

Publication Details

Journal
Research Square
Published
2026-11-18
DOI
https://doi.org/10.21203/rs.3.rs-11583612/v1
Primary Topic
Gestational Diabetes Research and Management
Type
preprint
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
preprint

Fetal sex confounds placental transcriptomic signatures of gestational diabetes: a six-cohort audit of public data

Suting LI, Songming He, Zhi Wang, Hongming Li et al.
3 links
Research Square
Gestational Diabetes Research and Management
preprint

Fetal sex confounds placental transcriptomic signatures of gestational diabetes: a six-cohort audit of public data

Suting LI, Songming He, Zhi Wang, Hongming Li, Li Hong, Jinhua Tian
preprint en

Abstract

Background The placenta is sexually dimorphic: fetal sex shapes placental gene expression, growth and immunity. Public placental transcriptomes are increasingly reused to define gestational diabetes mellitus (GDM) signatures; whether fetal sex can be separated from maternal hyperglycemia and fetal overgrowth in these datasets has not been tested. Methods Six public cohorts (bulk and single-cell RNA-seq, spatial transcriptomics, expression array) were reanalyzed under an a priori-defined Plan–Do–Check–Act audit (C1–C17). Fetal sex was inferred per sample from seven Y-linked markers and XIST; sex-linked, copy-number and segmental-duplication genes were excluded; X enrichment was tested rather than assumed; signatures were tested for cross-cohort transfer; compartment mapping used within-sample correlations in four control villi; cis-Mendelian randomization was exploratory. Results In the only cohort with a growth contrast, sex separated the groups completely: both cases were female (Y-marker counts 24 and 12; XIST 14,129 and 18,191), whereas the control and the insulin-treated sample were male (Y 29,124 and 26,300). Twelve of the 20 largest uncorrected effects were Y-linked or copy-number artifacts. Under the corrected specification the overgrowth-associated set fell from 2,930 to 1,872 genes, and the growth-discordant set (1,014 mapped genes) was unstable across leave-one-out (1,376 vs 665; Jaccard 0.55–0.70). The overgrowth-associated set showed no X enrichment (1.02×, P = 0.89). Transferability was limited: subtype signatures did not transfer to an independent hyperglycemia cohort (rho − 0.03 to − 0.06) and replicated only weakly on an array (rho 0.105–0.132); the shared-direction set reversed in both external cohorts; mitochondrial-RNA down-regulation tracked preeclampsia (13/13) rather than GDM (8/13). In control villi the overgrowth-associated set localized to cytotrophoblast (r = 0.42) and the growth-discordant set to fibroblast (r = 0.40). Conclusions In public placental transcriptomes fetal sex is not a covariate that can be adjusted away: it is confounded with, and in the critical arm identical to, the exposure contrast. GDM signature and sex-difference claims derived from single-sex arms are non-identifiable. Cohorts should report fetal sex and growth per sample, balance sex within arm, and pre-specify sex-aware filters.

Research Square
Renmin Hospital of Wuhan University (CN)
Good health and well-being
Gestational Diabetes Research and Management
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon