Age-related somatic mutations in the cancer genome

Abstract

Aging is associated with an increased risk of cancer, possibly in part because of an age-related increase in mutations in normal tissues. Due to their extremely low abundance, somatic mutations in normal tissues frequently escape detection. Tumors, as clonal expansions of single cells, can provide information about the somatic mutations present in these cells prior to tumorigenesis. Here, we used data from The Cancer Genome Atlas (TCGA), to systematically study the frequency and spectrum of somatic mutations in a total of 6,969 patients and 34 different tumor types as a function of the age of the patient. After using linear modeling to control for the age structure of different tumor types, we found that the number of identified somatic mutations increases exponentially with age. Using additional data from the literature, we found that accumulation of somatic mutations is associated with cell division rate, cancer risk and cigarette smoking, with the latter also associated with a distinct spectrum of mutations. Our results confirm that aging is associated with the accumulation of somatic mutations, and strongly suggest that the level of genome instability of normal cells, modified by both endogenous and environmental factors, is the main risk factor for cancer.

Keywords: aging; bioinformatics; genomics; sequencing; somatic mutation.

Figure 1

(A) Mutation frequency versus age in tumors of 6,969 individuals. The relationship between the two variables can be expressed as an exponential increase (P < 2.2*10⁻¹⁶, r = 0.36). (B) Frequency of somatic mutations in different age groups. Subjects over 80 had a mutation frequency more than 5 times higher than that of subjects under 20; the differences between all age groups are significant as measured by the Wilcoxon rank sum test.

Figure 2

(A) Lifetime risk of cancer of a tissue type [10], as a function of the estimated lifetime mutation accumulation, i.e., the increase in mutation frequency calculated for the tissue type by the linear model between birth and age 80 (P = .079, r = 0.53). (B) Lifetime mutation accumulation, i.e. the increase in mutation frequency calculated for the tissue type by the linear model between birth and age 80, for different tissue types as a function of the estimated lifetime number of stem cell divisions (P = .019, r = 0.66). (Abbreviations: LAML=acute myeloid leukemia, COAD=colorectal adenocarcinoma, ESCA=esophageal squamous cell carcinoma, GBM=glioblastoma, HSNC=head and neck squamous cell carcinoma, LIHC=liver hepatocellular carcinoma, MD=medulloblastoma, SKCM=skin cutaneous melanoma, OV=ovarian, PAAD=pancreatic ductal adenocarcinoma, TGCT=testicular germ cell cancer, THCA=thyroid papillary/follicular carcinoma).

Figure 3. Mutation frequency as a function of age in 14 bladder tumors for which both whole exome and whole genome were available [11]

Whole exome data: r = 0.50; whole genome data: r = 0.29.

Figure 4. Correlation coefficient of mutation frequency increase with age as a function of median mutation frequency in the different tumors

For each tumor type, the exponential correlation coefficient for the association between mutation frequency and age was plotted against its median mutation frequency. The correlation coefficients were inversely correlated with median mutation frequency (P = 0.0059, r = −0.46).

Figure 5. Distinct mutation spectrum in lung adenocarcinoma

(A) Principal component analysis of the proportions of mutations reveals that lung adenocarcinoma tumors tend to have a spectrum of mutations not shared by other tumor types, including lung squamous cell carcinoma tumors. (B) Lung adenocarcinoma tumors have a larger proportion of C->A mutations than all tumor types combined (P < 2.2*10⁻¹⁶, Wilcoxon test).