Pooling 16 public RNA-seq datasets (364 healthy men aged 18-92), a new R pipeline called MetAR found androgen receptor (AR) expression in skeletal muscle declining by 0.006 log2 CPM per year, roughly 4.4% per decade (p < 0.001). Genes governing AR stability, transcriptional activity and nuclear transport also fell with age, as did downstream AR-signalling regulators, suggesting coordinated remodelling of the whole pathway rather than loss of one component.
The finding matters because testosterone therapy in older men yields modest, inconsistent muscle benefits, and diminished receptor availability is one plausible reason: more hormone may not help if the signalling machinery is thinned out. Rodent data have long pointed this way; human evidence was muddy, and a pooled analysis across cohorts is a sensible way to test it. Important caveats apply. This is a bioRxiv preprint that has not been peer reviewed, so results may change. Heterogeneity was moderate (I-squared 53.8%), and the 95% prediction interval crosses zero (-0.0143 to +0.0014), meaning a future study could plausibly show no decline. The data are cross-sectional, mRNA rather than protein or receptor activity, men only, and cannot separate ageing from inactivity or comorbidity. Editorially, this is confirmatory and mechanistic. It offers no actionable intervention yet, but supports the case for combining resistance training, which can upregulate AR, with any hormonal strategy.