Estimating the carcinogenesis timelines in early-onset versus late-onset cancers and changes across birth cohorts

Understanding the timing of key mutational events in cancer development is critical for informing cancer prevention and detection strategies, particularly for early-onset cases that have increased in recent years. Yet intermediate mutational events are challenging to observe in humans. Here, we extend a tumor kinetic model we recently developed and long-term cancer registries data to estimate the expected timing of intermediate mutational events for breast, colorectal, and thyroid cancers. We formulate three distinct systems of ordinary differential equations, each describing cell evolution sequentially through three stages of carcinogenesis, up to the first occurrence of malignant transition. We further apply a convolution-based method to derive probability distributions and compute the expected age for each transition, based on parameters fit to incidence and tumor size data for each cancer. The models estimate the initial mutation occurs early in life for all three cancer types. For breast and colorectal cancers, estimated malignant transformation occurs more than a decade faster and hence earlier in early-onset than late-onset cases (in late 30s vs. late 40s to early 50s). In contrast, early- and late-onset thyroid cancers show similar early timelines (malignant transformation in late 20s), consistent with known early-life thyroid clonal activity. We also quantify early-onset carcinogenesis timelines for three key birth cohorts (born during 1950-1954, 1965-1969, and 1980-1984) and identify a shift toward earlier malignant transformation in more recent cohorts, largely due to faster progressions of later stage transitions, for all three cancer types. These findings can inform early-onset cancer etiologic studies and intervention strategies.