Trends in Young-Onset Cancer Incidence: A Modeling Perspective

Recent increases in the diagnosis of certain cancers among younger individuals are generating intense concern. Many studies attribute the increase in so-called "young-onset" cancer to an etiologic cause but questions have also arisen about the role of earlier diagnosis. We simulate incidence trends from a natural history model that includes healthy, preclinical, and clinical disease states, where transitions from a healthy to a preclinical state represent disease onset and transitions from the preclinical to the clinical state represent diagnosis. We superimpose birth cohort effects on the rate of disease onset and period effects on the rate of disease diagnosis to identify those that match patterns of relative incidence by age group and five-year calendar interval from 2000-2019 for six "young-onset" cancers (colon, rectum, female breast, gastric, pancreas, kidney). Two types of effects are broadly consistent with the observed increasing incidence trends in younger individuals: (1) A birth-cohort effect on disease onset that begins around 1970 and becomes more pronounced in later birth years, or (2) A period effect consistent with progressive reduction over time in the duration of preclinical disease. An earlier, protective birth cohort effect is consistent with recent declining trends in incidence in older individuals for colon, rectal, and gastric cancers. A disease model provides clues about the possible drivers of cancer incidence trends, suggests constraints on the patterns of exposures that might be implicated etiologically, and indicates that the role of diagnostic changes warrants consideration alongside potential etiologic explanations.