Association of autosomal mosaic chromosomal alterations with risk of bladder cancer in Chinese adults: a prospective cohort study

Abstract

Little is known about the prospective association between autosomal mosaic chromosomal alterations (mCAs), a group of large-scale somatic mutations on autosomes, and bladder cancer. Here we utilized data from 99,877 participants who were free of physician-diagnosed cancer at baseline (2004-2008) of the China Kadoorie Biobank to estimate the associations between autosomal mCAs and bladder cancer (ICD-10: C67). A total of 2874 autosomal mCAs events among 2612 carriers (2.6%) were detected. After a median follow-up of 12.4 years, we discovered that participants with all autosomal mCAs exhibited higher risks of bladder cancer, with a multivariable-adjusted hazard ratio (HR) (95% confidence interval [CI]) of 2.60 (1.44, 4.70). The estimate of such association was even stronger for mosaic loss events (HR [95% CI]: 6.68 [2.92, 15.30]), while it was not significant for CN-LOH events. Both expanded (cell fraction ≥10%) and non-expanded autosomal mCAs, as well as mosaic loss, were associated with increased risks of bladder cancer. Of interest, physical activity (PA) significantly modified the associations of autosomal mCAs and mosaic loss (P_interaction = 0.038 and 0.012, respectively) with bladder cancer. The increased risks of bladder cancer were only observed with mCAs and mosaic loss among participants with a lower level of PA (HR [95% CI]: 5.11 [2.36, 11.09] and 16.30 [6.06, 43.81]), but not among participants with a higher level of PA. Our findings suggest that peripheral leukocyte autosomal mCAs may represent a novel risk factor for bladder cancer, and PA may serve as a potential intervention target for mCAs carriers.

Fig. 1. Cumulative incidence curves of bladder cancer by mCAs status.

Cumulative incidence of bladder cancer for participants with or without autosomal mCAs (A) and mosaic loss (B). Cause-specific hazard models were used to account for the competing risk of death. Model for bladder cancer and model for death from other causes were separately developed, with the same set of variables (as same as model 3 in Table 2) included in both models. Cumulative incidences were estimated and plotted by the status of mCAs and mosaic loss, with all covariates being set to the average level of the study population. mCAs, mosaic chromosomal alterations.

Fig. 2. Associations of all autosomal mCAs and mCAs subtypes with bladder cancer by baseline subgroups.

Adjusted covariates in the models were consistent with model 3 in Table 2, as appropriate. ^aThe P values for interaction were computed with likelihood-ratio tests comparing models with and without multiplicative interaction terms of the baseline stratifying variable and types of mCAs. ^bCurrent-smoker included former smokers who had stopped smoking because of illness. ^cThere were no bladder cancer cases among current smokers with CN-LOH. HR hazard ratios, CI confidence interval, mCAs mosaic chromosomal alterations, CN-LOH copy-neutral loss of heterozygosity, PA physical activity, MET metabolic equivalent of task.

Fig. 3. Cumulative incidence curves of bladder cancer by mCAs status and physical activity levels.

Cumulative incidence of bladder cancer according to joint categories of level of physical activities and autosomal mCAs (A) or mosaic loss (B). Cause-specific hazard models were used to account for the competing risk of death. Model for bladder cancer and model for death from other causes were separately developed, with the same set of variables (as same as model 3 in Table 2) included in both models. Cumulative incidences were estimated and plotted by joint categories of level of PA (classified into low and high groups by the 25^th percentile of sex-specific MET-h per day) and autosomal mCAs or mosaic loss, with all covariates being set to the average level of the study population. PA physical activity, mCAs mosaic chromosomal alterations, MET metabolic equivalent of task.