Comparison of statistical methodologies for comparing metastatic renal cell carcinoma treatments

Renal cell carcinoma (RCC) is a type of kidney cancer that significantly impacts patients' lives. Metastatic RCC occurs when the cancer spreads beyond the kidney, making it more challenging to treat effectively. Worldwide, thousands of people are affected by RCC annually. Various treatments have been assessed for improving survival rates, but comparing their effectiveness is complicated due to differences in how clinical trials are conducted and variations in patient characteristics.

Deciding the best treatment for metastatic RCC often relies on comparing results from multiple clinical trials. However, traditional statistical methods like network meta-analysis (NMA) can lead to uncertain conclusions because they don’t always account for differences in patient characteristics between trials. Our research focuses on newer methods, multilevel network meta-regression (ML-NMR) and matching-adjusted indirect comparison (MAIC). ML-NMR and MAIC are different population adjustment methods, but they essentially enable the comparison of different treatments across different studies by taking data and adjusting for variations between these studies (e.g. in populations and/or patient characteristics). By implementing these methods, we aim to produce more reliable results for comparing treatments. This research could improve how health agencies evaluate cancer treatments, ultimately benefiting patients.

We will use patient data from clinical trials involving metastatic RCC treatments. Our work will replicate and build on analyses conducted for the National Institute for Health and Care Excellence (NICE) appraisal of nivolumab, a drug which works by activating the immune system to attack cancer cells. Using both individual participant data (IPD) and summary data, we will apply ML-NMR and MAIC. These methods allow us to adjust for differences in patient characteristics, such as previous treatments and prognostic risk scores (score predicting patients’ likely outcomes), to reduce bias when comparing studies.