Modeling early-onset cancer kinetics to study changes in underlying risk, detection, and impact of population screening.

Abstract

Recent studies have reported increases in early-onset cancer cases (diagnosed under age 50) and call into question whether the increase is related to earlier diagnosis from other medical tests and reflected by decreasing tumor-size-at-diagnosis (apparent effects) or actual increases in underlying cancer risk (true effects), or both. The classic Multi-Stage Clonal Expansion (MSCE) model assumes cancer detection at the emergence of the first malignant cell, although later modifications have included lag-times or stochasticity in detection to more realistically represent tumor detection requiring a certain size threshold. Here, we introduce an approach to explicitly incorporate tumor-size-at-diagnosis in the MSCE framework and account for improvements in cancer detection over time to distinguish between apparent and true increases in early-onset cancer incidence. We demonstrate that our model is structurally identifiable and provides better parameter estimation than the classic model. Applying this model to colorectal, female breast, and thyroid cancers, we examine changes in cancer risk while accounting for detection improvements over time in three representative birth cohorts (1950-1954, 1965-1969, and 1980-1984). Our analyses suggest accelerated carcinogenic events and shorter mean sojourn times in more recent cohorts. We further use this model to examine the screening impact on the incidence of breast and colorectal cancers, both having established screening protocols. Our results align with well-documented differences in screening effects between these two cancers. These findings underscore the importance of accounting for tumor-size-at-diagnosis in cancer modeling and support true increases in early-onset cancer risk in recent years for breast, colorectal, and thyroid cancer.

Significance: This study models recent increases in early-onset cancers, accounting for both true factors contributing to cancer risk and those caused by improved detection. We show that while advancement in detection has led to earlier detection, our model estimates shorter sojourn times and more aggressive carcinogenic events for recent cohorts, suggesting faster tumor progression. Further, a counterfactual analysis using this model reveals the known statistically significant reduction in colorectal cancer incidence (supporting a robust modeling approach), likely due to screening and timely removal of precancerous polyps. Overall, we introduce an enhanced model to detect subtle trends in cancer risk and demonstrate its ability to provide valuable insights into cancer progression and highlight areas for future refinement and application.