Residual cardiovascular and kidney risk often persists after successful weight loss. The authors investigated whether molecular features that are genetically nominated for cardiovascular‑kidney‑metabolic (CKM) outcomes show larger short‑term responses to weight‑loss and cardiometabolic interventions than other adequately instrumented features. The question addresses whether genetic target support can serve as a signal for pharmacodynamic responsiveness, reversibility, or potential treatment‑response biomarkers.
Genetic nomination was performed across four molecular layers: the proteome, transcriptome, methylome, and metabolome. The authors applied layer‑specific cis‑Mendelian randomization together with colocalization or summary‑data shared‑signal filtering. These analyses were conducted against eight CKM genome‑wide association studies (GWAS) drawn from public repositories. The nomination step identified molecular features with genetic evidence linking them to CKM outcomes using publicly available cis‑QTL atlases and GWAS summary statistics.
Response to interventions was assessed across 13 omics‑by‑intervention analyses. Interventions included dietary interventions, bariatric surgery, empagliflozin exposure, and behavioral weight‑loss programs. Molecular response data were taken from public transcriptome, methylome, proteome, and metabolome resources, including GEO and ArrayExpress datasets and published supplementary materials of referenced intervention studies. For each analysis, nominated features were compared to adequately‑instrumented non‑nominated features to evaluate whether genetic nomination predicted larger observed molecular change.
The primary comparison metric was rank‑based Cliff’s δ, used to compare response magnitude distributions between nominated and non‑nominated features. For methylation analyses, baseline inter‑individual variance matching was performed because nominated CpG sites had different baseline variance characteristics; directional analyses of methylation were exploratory. Label‑permutation testing was used to assess significance of observed differences.
Across all 13 omics‑by‑intervention comparisons, effect size estimates were uniformly small: absolute values of Cliff’s δ were less than 0.08 for every analysis, with a median |δ| = 0.03. The pooled descriptive δ across comparisons was −0.003 with a 95% confidence interval of −0.016 to +0.010 and I2 = 0%, indicating minimal pooled effect and no detectable heterogeneity by that measure. None of the observed differences achieved significance by label‑permutation testing.
Despite the lack of a consistent larger response among nominated features, many nominated proteins and metabolites did exhibit measurable change after interventions: reported response rates ranged from 48% to 71% in those layers, similar to response rates among comparison features. Thus, genetic nomination was not associated with universal non‑responsiveness, nor did it reliably identify features with larger short‑term intervention responses in the datasets analyzed.
Nominated CpG sites were found to have lower baseline inter‑individual variance compared with other sites (Mann–Whitney P values ranged from 1.7 × 10−6 to 4 × 10−3). The authors observed apparent persistence of methylation signals, but this persistence attenuated when analyses were variance‑matched at baseline. The report emphasizes that baseline dynamic range can confound comparisons of molecular change and should therefore be assessed and controlled when evaluating responsiveness across selected features.
Under the definitions, nomination criteria, and datasets studied, genetic target support does not consistently predict pharmacodynamic responsiveness. The authors caution that genetic nomination should not be read as evidence that a molecular feature is a treatment‑response or reversibility biomarker. They also note that nominated features are not demonstrated to be unchanged by interventions, nor are they shown to explain residual clinical risk. The study highlights the importance of evaluating baseline dynamic range and recommends same‑subject longitudinal studies to determine prognostic or therapeutic value of nominated features.
All inputs to the study were publicly available summary statistics or de‑identified molecular data. Genetic nomination utilized CKM GWAS summary statistics from the EBI GWAS Catalog and cis‑QTL atlases including UKB‑PPP, deCODE, GTEx, and the eQTL Catalogue; cis‑mQTL data came from GoDMC; cis‑metaboQTL inputs were from published supplementary materials. Molecular response data were obtained from GEO and ArrayExpress accessioned intervention datasets and from supplementary materials of cited proteomic and metabolomic studies. Derived source data and analysis code are available at the linked GitHub repository and archived on Zenodo.
The authors declare no competing interests and state that only publicly available, de‑identified data were used. The manuscript is a preprint and has not been peer reviewed; the authors note it should not be used to guide clinical practice. The study was funded in part by the National Science and Technology Council and Chang Gung Memorial Hospital, as declared in the source.