OBJECTIVES: Traditional cancer screening tests are designed for specific cancer types, whereas multi-cancer early detection (MCED) tests offer a broader approach. Blood-based MCED tests have generated significant interest, yet their long-term impact and economic viability remain underexplored. Given the lack of empirical data, modelling the effects of MCED screening on life expectancy and lifetime costs is crucial for evaluation. With project aimed to compare and appraise multi-cancer early detection modelling approaches.

METHODS: A targeted literature search was conducted in 2023 aimed to identify MCED models with a natural history disease component. We assessed various modelling options reported in the literature, focusing on assumptions regarding cancer's natural progression, screening impact, and approaches to handle heterogeneity and uncertainty.

RESULTS: Out of 11 identified models, one was excluded due to its limitations. Most (n=6) models adopted a cohort design, modelled only incident population (n=8) and delineated cancer progression into stages 1 to 4 (n=7). Five approaches for modelling natural history disease in MCED models were identified. The models addressed heterogeneity between cancer types by independently modelling parameters such as incidence, progression, and mortality. Progression times across stages were assumed to follow exponential distributions (in all except one model), with limited consideration for within-cancer type heterogeneity. The models focused on predicting the stage-shift impacts of screening based on test accuracy, with lead time or length time bias included. Explicit consideration of over diagnosis was present in only two models. Furthermore, uncertainty in the natural history of disease was infrequently evaluated, particularly in models employing prediction-based parameterization.

CONCLUSIONS: The heterogeneity of assumptions across MCED models underscores the need for comprehensive simulation exercises to establish their potential impact.