A new microsimulation model of HPV natural history for comparison of novel cervical screening strategies

The cost-effectiveness of cervical cancer screening strategies is estimated by health decision models. Fidelity to HPV natural history and multi-state carcinogenesis is critical to improve evaluations of new screening tests.

We developed a new health decision model of the HPV genotype-specific causal pathway to cervical cancer based on cohort, trial, and cross-sectional data. The novel model presents several new approaches in representing early HPV natural history: 1) transitions between health states are estimated by new statistical approaches adapted to handle intermittently-observed data; 2) transitions are directly estimated from observational data (i.e., statistically identifiable) rather than calibrated to fit; and 3) precancer is rigorously defined using histopathology and HPV genotyping as a surrogate for cancer risk.

We present a model for pending policy decisions in Brazil. Model outputs closely match age-specific prevalence data from São Paulo for thirteen carcinogenic HPV types 16/18/31/33/35/45/52/58 (modeled individually) and 39/51/56/59/68 (grouped). The model-projected median age of precancer is 30.0 years (95% CI: 27.9-32.3 years).

By restricting model health states and transitions to the essential, directly observable causal pathway, our approach can generate metrics with measurable standard errors for cost-effectiveness analyses. Comparative modeling analyses are underway to assess differences in results relative to existing state-of-the-art models. Our approach can be generalized to evaluate the cost-effectiveness of an HPV-based screen-triage-treat approach with extended genotyping that has been demonstrated efficacious in nine PAVE Consortium countries.