Reproducibility and surrogacy of cancer trials, and disparities in cancer outcomes using registry data

Abstract

Cancer outcomes research is an important topic in health and medicine since cancer is the second leading cause of death globally. In addition to clinical trials for developing new treatments that are crucial to cancer outcomes research, secondary data analysis of observational data is also critical. Besides research gaps in trial designs, assessing surrogacy, and reporting practices in clinical trials, unidentified trends, and disparities in second primary cancers exist. In this thesis, four studies covering different aspects—significance threshold, strategies for assessing level of surrogacy, reporting quality, and secondary data analysis—in cancer outcomes research were conducted. Sample size is important in trial design. In recent years, the influence of the significance threshold on reproducibility has been considered. To investigate the impact of lowering the significance threshold on sample size and trial duration, phase II clinical trial articles published between 2015 and 2022 were collected. The findings indicated that lowering the threshold results in a median of 125.4% change increase in the sample size and a median of 2.47 additional required years. This would delay bringing new therapeutics to patients and affect the possibility of progressing to phase III trials. Applying progression-free survival as a surrogate endpoint for overall survival in immunotherapy clinical trials has been widely discussed. Moreover, since non-proportional hazard is common in cancer immunotherapy trials, the restricted mean survival time ratio has been proposed as an alternative to the hazard ratio for treatment evaluations. In this thesis, eligible phase II or III hepatocellular carcinoma immunotherapy studies published between January 2000 and August 2022 were analyzed using weighted least-square regression with the restricted mean survival time ratio and hazard ratio for assessing the surrogacy level. The results demonstrated the potential of the restricted mean survival time ratio in addition to the hazard ratio for evaluating the surrogacy level in immunotherapy trials. In addition to trial design, appropriate analyses and reporting are essential to the reproducibility and interpretation of clinical trials. However, the COVID-19 pandemic and other force majeure events, such as the war in Ukraine, have impacted the conduct of clinical trials. Here, randomized phase III trials from three top oncology journals were collected to develop and validate a checklist with ten new and four modified items, with the aim of improving the reporting of randomized clinical trials impacted by COVID-19 and force majeure events. With more patients surviving their primary cancers, a better understanding of the trends and disparities in this population of developing second primary cancers is needed. A retrospective analysis using competing risk analysis was conducted based on the Surveillance, Epidemiology, and End Results program. The results demonstrated that over the past three decades, the time to second primary cancer has noticeably shortened among lung primary cancer survivors. However, disparities in age, sex, and race still exist, indicating that further effort is needed to address the gap. In conclusion, this thesis contributes to closing some key research gaps in clinical trial designs, reporting practices, and primary cancer survivorships that are influential to cancer outcomes research.