Sex Dimorphism in DEN-Induced Hepatocellular Carcinoma
Background
Hepatocellular carcinoma is markedly more common in males, and the obvious way to look for why is to compare metabolites between the sexes. The obvious way is also wrong: male mice carry more tumour, so anything that merely tracks tumour burden will present itself as a sex difference. Telling the two apart is the whole problem.
Method
Paired serum untargeted metabolomics and liver transcriptomics from a DEN-induced mouse model. The raw acquisition gives 9,082 spectral features, most of them noise, adducts or fragments of the same compound; curation reduces that to a 468-metabolite matrix across 174 samples.
On top of it sits a sex × disease interaction model with tumour burden as a covariate, so a metabolite only scores if its behaviour differs by sex beyond what its tumour load explains. Significance comes from 1,000 permutations rather than a parametric assumption the data does not support.
Tools: R (DESeq2, WGCNA), Python, untargeted metabolomics, transcriptomics.
Results
The interaction test leaves roughly 190 candidates — too many to chase, and most of them will not survive contact with an orthogonal method. So the shortlist demands four independent lines of evidence from each metabolite:
- a significant sex × disease interaction,
- independence from tumour burden,
- reversal under intervention, and
- a corroborating enzyme in the transcriptome.
About 20 clear all four (permutation p = 0.001), and roughly 13 of those have not been reported before in the context of HCC sex dimorphism.
Together they outline an estrogen (ERα)-driven metabolic reprogramming axis, whose direction was confirmed rather than assumed: the intervention arm pushes the system both ways, and the markers move accordingly.
The panel is now queued for targeted MS/MS, which is the measurement that would turn these from candidates into findings.
The project is ongoing.