Presentation (CTI)

OBJECTIVES: We investigated drivers of improvement in median observed survival (OS) over three decades in malignant tumour sites.
METHODS: The median OS for 48 malignant tumour sites, diagnosed between 1975 and 2022, was derived using Surveillance, Epidemiology, and End Results Program (SEER) data. We carried out an analysis of the trends in median OS according to diagnostic year, including: the absolute and proportional change in OS (calculated using median OS from the first recorded diagnostic year and last observed median OS) and the annual change in OS (calculated by fitting linear regression models). Tumours with median OS available over a period of 25 diagnostic years were eligible for inclusion. The correlation between the risk of disease (incidence per 100,000 patients per year [SEER]) and the change in median OS was estimated using Spearman's rank and Kendall's tau. The correlation between the number of clinical trials and change in OS was estimated, utilising Cortellis Clinical Trial Intelligence to summarise the number of trials across the tumour sites. Sensitivity analyses were undertaken, restricting site inclusion based on the observation interval width to enhance the comparability of the timeframes used to model the trajectories in median OS.
RESULTS: Prostate, bones and joints, tonsil and breast cancer ranked in the top five malignant tumour sites showing improvement according to both absolute and annual change in median OS by diagnostic year. Vulva and ureter cancer ranked in the bottom five according to absolute and annual change. The correlation between the change in median OS and risk of disease was positive and statistically significant.
CONCLUSIONS: High-risk and high-profile cancers (e.g. prostate cancer and breast cancer) tended to experience greater improvements in OS than lower risk, rarer cancers. These higher risk cancers also tended to have higher clinical trial counts. Recent improvements in OS are likely driven by innovative targeted therapies.