DNA methylation-predicted GDF-15 and mortality in cancer survivors: a cohort study.

Abstract

Developing non-invasive prognostic biomarkers remains critical to improving personalized cancer care. Growth differentiation factor-15 (GDF-15), a TGF-β family cytokine, plays a key role in tumorigenesis and immune evasion. Circulating GDF-15 serves as a biomarker for cancer prognosis, and DNA methylation (DNAm)-predicted GDF-15 has been linked to mortality risk in the general population. However, the association between DNAm-predicted GDF-15 and mortality risk in cancer survivors remains unexplored. We analyzed the association between DNAm-predicted GDF-15 and all-cause, long-term all-cause, and cancer mortality risks using a cohort of 343 cancer survivors from the National Health and Nutrition Examination Survey (NHANES) 1999-2002 with a median follow-up of 138 months. Multivariable Cox regression reporting hazard ratios (HRs) and 95% confidence intervals (CIs) demonstrated that each 1-standard deviation (SD) increment in DNAm-predicted GDF-15 was associated with a 60% higher all-cause mortality risk adjusted with model 1 of age and sex, and a 54% greater all-cause mortality risk in model 2 adjusted additionally for ethnicity, education, smoking, and coronary heart disease. Participants in the high GDF-15 tertile showed a 201% and 166% higher mortality risk in model 1 and model 2, respectively (both p for trend <0.0001) compared to the low tertile. Its association with long-term mortality risk remains unchanged. Stratified analyses indicated consistent relationships across multiple subgroups. Kaplan-Meier and competing risk analyses revealed a graded increase in cancer mortality risk across ascending GDF-15 tertiles; Cox models confirmed a significant positive association per 1-SD increment in the unadjusted model and model 1, which remained consistent in direction and magnitude in model 2, with a marginally significant (p=0.052). The current study provided evidence that DNAm-predicted GDF-15, an alternative and precise estimate of GDF-15 based on DNA methylation, is positively associated with all-cause and long-term all-cause mortality risks and showed a trend of positive association with cancer mortality among cancer survivors. Future larger longitudinal studies with serial DNAm-predicted GDF-15 assessments are needed to verify potential causal links.