Profile of biological aging in first primary cancers: a pan-cancer analysis of two large-scale cohorts from the UK and Hong Kong

Abstract

Background: Aging is a major risk factor for cancer, but the landscape of biological aging across different cancer types and its interplay with genetic risk remains unclear. This study aims to depict the biological aging profiles in specific cancers across diverse populations and investigate the bidirectional relationship between aging and cancer.

Methods: This study included 414,599 participants from the UK Biobank (UKB) and 83,788 participants from the electronic health record database of Hong Kong Hospital Authority (EHR-HK). Multivariable Cox and logistic regression models were used to evaluate associations between biological age acceleration (BioAgeAccel) and site-specific cancers in the UKB and EHR-HK, respectively. In the UKB cohort (n = 387,066), we further computed cancer-specific polygenic risk scores (PRSs) and calculated population attributable fractions (PAFs) to quantify the relative contributions of aging and genetics to cancer incidence and mortality. A nested two-sample bidirectional Mendelian randomization (MR) analysis within one-sample setting was employed to explore the reciprocal causality between aging and cancer.

Results: Compared to cancer-free individuals, the most pronounced BioAgeAccel disparities were observed in liver cancer (mean difference (MD): 5.9 years) within the UKB, and oesophageal cancer (MD = 18.4 years) within the EHR-HK. A 5-year increment in BioAgeAccel was associated with elevated overall cancer risk, with leukaemia demonstrating the highest hazard ratio in the UKB (HR = 1.13, 95% CI: 1.11-1.15) and oesophageal cancer exhibiting the highest odds ratio in the EHR-HK (OR = 1.55, 95% CI: 1.33-1.81). PAF analyses revealed that BioAgeAccel contributed to 47% of lung cancer incidence and 60% of lung cancer-specific mortality, exceeding contributions from genetic risk. Significant interactions between genetics and aging were identified for colorectal, lung and non-melanoma skin cancer. Bidirectional MR analyses demonstrated the reciprocal relationship between BioAgeAccel and lung cancer (aging-to-cancer nexus: OR = 1.30, 95% CI: 1.11-1.51; cancer-to-aging nexus: 1.05 (1.02-1.08)), female breast cancer (aging-to-cancer nexus: 1.09 (1.02-1.15); cancer-to-aging nexus: 1.05 (1.03-1.07)), and prostate cancer (aging-to-cancer nexus: 1.08 (1.01-1.16); cancer-to-aging nexus: 1.02 (1.00-1.03)).

Conclusions: This pan-cancer study reveals intricate interrelationships between biological aging and cancer, particularly in lung, prostate, and female breast cancer, with population-specific patterns and synergistic genetic interactions. Findings underscore the potential for aging-targeted strategies in cancer prevention and treatment.

Keywords: Bidirectional Mendelian randomization; Biological age; Cohort; Genetic; Pan-cancer; Reciprocal relationship.

Fig. 1

IRR and PRR of different cancers across ChronAge and BioAgeAccel groups in different population cohorts (A UKB; B EHR-HK; IRR, incidence rate ratio; PRR, prevalence rate ratio; ChronAge, chronological age; BioAgeAccel, biological age acceleration)

Fig. 2

Stratification analysis across genetic risk and biological aging on cancer incidence in the UKB (PRS, polygenic risk score; models were adjusted for ChronAge, sex, education level, household income, employment status, social connection, tobacco smoking, alcohol drinking, physical activity, sedentary behavior, vegetable intake, fruit intake, processed meat consumption, variation in diet, comorbidities, 10 principal components and genotyping chip batches)

Fig. 3

Population attributable fraction of BioAgeAccel and PRS on cancer incidence and mortality (BioAgeAccel, biological age acceleration; PRS, polygenic risk score; Models were adjusted for ChronAge, sex, education level, household income, employment status, social connection, tobacco smoking, alcohol drinking, physical activity, sedentary behavior, vegetable intake, fruit intake, processed meat consumption, variation in diet, comorbidities, 10 principal components and genotyping chip batches)