Letter on ‘Downregulation of EPAS1 and EGLN1 mRNA Expression Associated With High‐Altitude Adaptive Genetic Variants in Sherpa Highlanders’

1.

To the Editor,

The study by Droma et al. (2026) reports that Sherpa highlanders carry significantly lower frequencies of the wild‐type alleles of two EPAS1 (rs13419896:G>A and rs4953354:A>G) and two EGLN1 (rs1435166:G>A and rs2153364:A>G) intronic variants, and that whole‐blood mRNA expression of both genes is markedly reduced in Sherpas relative to non‐Sherpa lowlanders (26.7% and 39.2% of control, respectively). The authors interpret this as transcriptional modulation of the HIF pathway that blunts the erythropoietin (EPO) response and protects against chronic mountain sickness. The hypothesis is biologically coherent and aligns with prior Tibetan data. The strength of the proposed causal chain (genotype → reduced mRNA → blunted EPO → protection), however, depends on several methodological points that the present design does not fully resolve.

First, mRNA was quantified from EDTA‐anti‐coagulated whole‐blood RNA without correction for differential leukocyte composition between groups. Sherpa highlanders chronically exposed to hypoxia at 3440 m may differ from Kathmandu lowlanders in granulocyte‐to‐lymphocyte proportions, and bulk whole‐blood expression is well documented to be confounded by underlying cell‐type admixture (Shannon et al. 2018). A 2.5‐fold expression gap between groups could reflect, at least partly, cellular composition rather than allele‐specific transcription. Complete blood counts with differential or an in silico deconvolution would have allowed adjustment.

Second, the comparison is altitude‐confounded with ethnicity. Sherpas were sampled at 3440 m and non‐Sherpas at 1300 m, so the observed expression difference combines genetic ancestry, current hypoxic state and any environmental exposures correlated with altitude (diet, infection burden, sleep architecture). Without a same‐altitude Sherpa or same‐altitude non‐Sherpa stratum, the genetic and environmental contributions cannot be partitioned. The authors' own observation that mRNA levels did not correlate with SpO₂ within Sherpas (r = 0.16 and 0.18) is informative but does not eliminate the between‐group altitude effect.

Third, the central inferential claim that reduced wild‐type allele frequency drives reduced mRNA rests on stratified expression means in subgroups of n = 2 to n = 14 (figs. 3 and 4 of Droma et al.). Several of the highlighted percentage reductions (e.g., 26.4% and 62.9%) are not accompanied by within‐stratum statistical tests, and the corresponding standard‐error bars overlap considerably across genotypes within each population. The post hoc power calculation was performed for between‐population allele‐frequency comparisons (OR ≥ 2.0), not for the genotype‐stratified mRNA comparisons that carry the mechanistic argument.

Fourth, population stratification was not formally addressed. Two ethnically distinct samples were compared without ancestry‐informative marker correction or principal‐components adjustment, the now‐standard safeguard against spurious genotype–phenotype associations in admixed or differentiated populations (Price et al. 2006). Sherpa and Kathmandu Valley populations differ across the genome, not only at the four candidate loci, and any locus reaching the genome‐wide Sherpa lowlander F _st observed here for EPAS1 (0.31–0.33) would generate similar between‐group expression contrasts by chance.

Fifth, the unaltered serum EPO across genotypes (figs. 5 and 6 of Droma et al.) is presented as evidence that the variants suppress EPO production, yet circulating EPO is regulated by additional layers, including soluble EPO receptor concentration, which is itself altered at high altitude (Villafuerte et al. 2014). A null EPO‐genotype association is equally compatible with post‐transcriptional or receptor‐level regulation that is independent of EPAS1/EGLN1 transcription.

The reduced wild‐type allele frequencies and downregulated EPAS1/EGLN1 transcripts described by Droma et al. are interesting and consistent with the broader literature on Tibetan‐lineage high‐altitude adaptation. Establishing the proposed transcriptional‐modulation model as causal will require cell‐type‐corrected expression measurement, a same‐altitude comparison cohort, and ancestry‐adjusted association testing in a larger sample.

Author Contributions

Maajid Mohi Ud Din Malik: conceptualisation, methodology, formal analysis, investigation, writing – original draft, writing – review and editing, supervision, validation, visualisation, final approval.

2. Funding

The author has nothing to report.

Conflicts of Interest

The author declares no conflicts of interest.

Data Availability Statement

The author has nothing to report.