Rho experts recently attended the FDA-AACR-ASA Workshop on the use of overall survival as an endpoint in oncology trials.  In a series of five blogposts, they will discuss key topics to consider when designing such a study as presented at the workshop. These topics will include study design, endpoint construction, subgroup considerations, and analysis interpretation. This is the third blog in the series.

Overall survival (OS) is an endpoint of significant importance when assessing safety and efficacy in oncology studies. Multiple trial design decisions are made when planning the analysis of OS and analysis of OS is even more complicated when it is analyzed as a non-primary endpoint. Below are some common trial design elements to consider that may have potential pitfalls for your next study.

1) Crossover design

When the protocol-specified study treatment has concluded, physicians and patients can be faced with a dilemma. If the patient did not respond to treatment, the natural urge is to try another intervention (i.e., a crossover). For oncology studies, crossover designs are attractive for patients who are guaranteed to receive the experimental treatment at some point during the study. While accessing additional therapies has potential benefit for patients, crossing over has the potential to confound the assessment of overall survival. Analysis plans will need to address how crossover will be handled in the analysis of overall survival.

2) Follow-up duration

Many trials use a short-term endpoint (e.g., progression-free survival) for the primary analysis. It is usually desirable to continue to follow subjects for overall survival beyond the time required to assess the short-term endpoint. Determining an appropriate amount of follow-up time depends on many factors, including the trial setting and the anticipated pattern/shape of the survival curve. For instance, the survival timeline is significantly longer for indolent cancers (e.g., prostate) as compared to more aggressive cancers such as acute myeloid leukemia.

3) Independent Data Monitoring Committee (IDMC) data review

The IDMC should be provided with clear instructions as to how they are meant to use overall survival data. If the IDMC is to be tasked with deciding about early stopping (for efficacy or safety), formal interim analysis boundaries should be calculated. Providing stopping boundaries helps to prevent kneejerk reactions and emotion from derailing an otherwise promising trial.

4) 2:1 randomization

Unequal randomization schemes (i.e., 2:1 randomization) reflect “preference” of the experimental arm and could be useful when there is prior evidence of a favorable benefit-risk profile (e.g., promising results found in earlier trials or in other disease settings). Unequal randomization schemes are often favorable for boosting patient enrollment, by giving patients an increased likelihood of randomization to the experimental treatment. While the use of 2:1 randomization can sometimes be justified, this approach comes with potential pitfalls. When compared to 1:1 randomization allocation, 2:1 allocation requires larger sample sizes to achieve the same power and will likely lead to greater imbalance in smaller subgroups. When the experimental treatment yields an increased toxicity, unequal randomization may magnify safety concerns.

As described, when designing a clinical trial in which OS will be analyzed, sponsors need to carefully consider the pros and cons of various study design decisions. Collaborating with a CRO that has oncology expertise and regulatory body experience can minimize these potential pitfalls and facilitate an efficient pathway to market approval.

Jessica Gladstone, Ph.D., Senior Biostatistician, has over 10 years of statistical research experience, with 3 years of experience working in the pharmaceutical industry. She has experience working in all phases of clinical trials, as well as in a variety of therapeutic areas. At Rho, Dr. Gladstone serves as the Biostatistics team lead on numerous studies serving multiple clients. Ms. Gladstone received her Ph.D. in Quantitative Methods at the University of Pennsylvania in 2020.

Shane Rosanbalm, MS, Principal Biostatistician, holds advanced degrees in Mathematics and Biostatistics. He has over 20 years of experience as a statistician, providing support for Phase I-IV clinical trials. Mr. Rosanbalm’s experience includes writing detailed statistical analysis plans, preparing CDISC-compliant specifications for analysis (ADaM) databases, specifying and performing statistical analyses, preparing displays and reports, communicating with sponsors, and leading teams of Rho statisticians and programmers. He has collaborated with researchers in several areas including Parkinson’s, hypertension, neonatal sepsis, HIV, hepatitis C, rheumatoid arthritis, diabetes, HAE, cardiovascular disease, and oncology.