Proteomic aging clocks and the risk of mortality among long-term cancer survivors

BACKGROUND: To estimate biological age, we developed a proteomic aging clock in cancer-free participants (CaPAC) and examined its association with mortality in long-term cancer survivors (LTCS, >2 years between cancer diagnosis and blood collection) and cancer-free participants in the Atherosclerosis Risk in Communities (ARIC) and Multi-Ethnic Study of Atherosclerosis (MESA) studies. METHODS: ARIC measured 4712 proteins using SomaScan in plasma samples collected at 3 visits, including Visit 5 (2011-2013) from 806 LTCS and 3699 cancer-free participants, all aged 66-90. Among 2466 randomly selected cancer-free participants, we developed CaPAC using elastic net regression. Age acceleration was calculated as residuals of CaPAC regressed on chronological age (CaPACAccel). We used multivariable Cox proportional hazards regression to calculate hazard ratios (HRs) for the associations of CaPACAccel with all-cause and cancer mortality in LTCS and all-cause mortality in the remaining cancer-free participants. We replicated the analysis of all-cause mortality in MESA. RESULTS: In LCTS, CaPACAccel was associated with increased all-cause mortality in both ARIC (HR per 1 SD = 1.42, 95% confidence interval [CI] = 1.24 to 1.62; P < .001) and MESA (1.62, 1.12 to 2.33; P = .009). Also, in ARIC, CaPACAccel was associated with all-cause mortality in breast (1.54, 1.05 to 2.25; P = .028) and colorectal LTCS (1.96, 1.19 to 3.22; P = .008). Additionally, CaPACAccel was associated with cancer mortality in LTCS (1.34, 1.09 to 1.64; P = .005) in ARIC. In MESA, limited sample size precluded us from examining individual cancers and cause-specific mortality. In cancer-free participants, the associations of CaPACAccel with all-cause mortality were similar across studies. CONCLUSION: Proteomic aging clocks hold promise as a predictor of all-cause and cancer mortality in LTCS.