Edward L. Korn

Patient withdrawal from a randomized clinical trial before starting the study treatment complicates the interpretation of the trial results, especially if a larger proportion of patients withdraws from the control arm than the experimental arm. Such withdrawals may occur in an open-label trial if the patients are disappointed with their treatment assignment. It has been hypothesized that 2:1 randomization may lead to more disappointment-related withdrawals. To investigate this phenomenon, we examined a cohort of 396 published oncology trials from 4 journals (2022-2025) to assess the differential pretreatment withdrawal rates. We found that in 35 open-label trials with 2:1 randomization, patients were, on average, 3.72 times more likely to withdraw from the control arm (95% CI = 2.66 to 5.22). This rate poses a larger relative risk than was seen in the 181 open-label trials with 1:1 randomization (relative risk = 1.99, 95% CI = 1.67 to 2.37). To avoid differential withdrawal, trials should be blinded when possible.

In the article that accompanies this editorial, Sicklick et al. provide the results of their work to match next-generation sequencing to novel therapeutic drugs/combinations in metastatic cancer patients. While one of the largest studies of its kind, this application of “precision medicine” is ultimately limited by the need for a randomized trial to show that using molecular data for a very heterogeneous group of patients identifies therapies that are superior to existing standard-of-care agents.

Improvements in cancer treatment are guided by randomized clinical trials (RCTs) that reliably assess the therapeutic impact of adding experimental treatments into clinical practice. The fundamental value of RCTs is that the difference in outcomes between the experimental and control arms provides an unbiased estimate of the risk-benefit ratio for the experimental treatment. However, the clinical relevance of trial results depends on the control arm accurately reflecting the appropriate standard of care (SOC) in the intended-use population. Due to the accelerating pace of cancer treatment development, it is becoming more common for the SOC to change during an ongoing randomized trial. This increasingly dynamic SOC landscape may have a nontrivial impact on the conduct and interpretation of future RCTs, and thus it is important to appreciate these issues and, when possible, prospectively address them when designing and conducting clinical trials. We review common scenarios where the SOC changes and provide recommendations on how to design a trial to minimize the impact of such changes when the trial is ongoing.

Background: In randomized trials where some standard-treatment arm patients cross to the experimental treatment, it is frequently of interest to estimate the between-arm survival difference as if no patients on the standard-treatment arm had crossed over to the experimental treatment. Rank-preserving structural failure time models, an extension of semiparametric accelerated-failure-time models, are a popular method for accomplishing this because they do not require modeling which patients will crossover. Methods: In trying to apply the rank-preserving structural failure time model in practice, we noted some unusual behavior of the estimated acceleration parameter (differential treatment effect). Simple examples and limited simulations are provided to examine and understand this behavior. Results: The simulations show that rank-preserving structural failure time model estimator of the acceleration parameter can take on extreme values, especially when the intent-to-treat analysis favors the standard-treatment arm. Furthermore, the addition of censoring is paradoxically shown to reduce the estimator’s variability compared to the uncensored data when the underlying observations are exponentially distributed. Use of a Weibull distribution with short tails for the survival times eliminates this unusual behavior. Conclusion: The rank-preserving structural failure time model estimators of the acceleration parameter are not based on the joint ranks of the original data, and it is suggested that this makes acceleration-parameter estimator unstable with long-tailed survival distributions.

Phase III trials that randomly assign patients to a control treatment (C), an experimental treatment (A), or a combination treatment (AB) should be designed with the goal to recommend the best treatment: AB (if it is better than A and C), A (if it is better than C, and AB is not better than A), or C (if neither AB nor A is better than C). However, this goal can be challenging to achieve with statistical confidence. We performed a survey of cancer trials published in five journals from January 2018 to May 2024 to assess the trial designs being used in this setting and found that three quarters of them did not have a provision for a formal comparison of the AB treatment arm with the A treatment arm, a possible shortcoming. A limited simulation evaluates two analysis strategies that incorporate an AB versus A comparison and is used to formulate some recommendations for designing these types of trials.

PURPOSE: A phase II/III trial is a type of phase III trial that has embedded in it an intermediate phase II go/no-go decision as to whether to continue the accrual to the phase III sample size. We examine the design characteristics and experience of a well-defined set of National Cancer Institute phase II/III trials, with special emphasis on designed accrual suspensions while awaiting the data to become mature enough for the phase II analysis. This experience is used to highlight the potential of using a calendar backstop to avoid an inordinately long accrual suspension. METHODS: We identified all phase II/III trials conducted by NRG Oncology or its precursor National Cancer Institute Cooperative Groups (Radiation Therapy Oncology Group, Gynecologic Oncology Group, and National Surgical Adjuvant Breast and Bowel Project). The design characteristics were recorded, and, for completed trials, the trial results in terms of sample sizes and timing of analyses were tabulated. RESULTS: Twenty-two trials were identified, 14 of which had a time-to-event end point for their phase II component. Thirteen of these 14 trials had designed accrual suspensions. Seven of the eight completed trials had designed accrual suspensions, all of which went on longer than their projected suspension times (3-20 months longer than planned). The trade-offs for using a backstop are discussed using one of these trials as an example. CONCLUSION: Phase II/III trials with an accrual suspension and a predefined backstop for the phase II analysis can be a useful tool for minimizing patient exposure to ineffective experimental treatments while still obtaining the trial results in a timely fashion.

In a randomized clinical trial, instead of allocating patients equally between the treatment arms, some trials in oncology assign a higher proportion of patients to receive the experimental treatment arm (eg, a two-to-one randomization). In this commentary, we first briefly review the common reasons given for the use of a two-to-one randomization and provide some examples of trials using these designs. We then explain why the risk-benefit ratio of this approach may not be favorable as is commonly assumed.

New oncology therapies that extend patients' lives beyond initial expectations and improving later-line treatments can lead to complications in clinical trial design and conduct. In particular, for trials with event-based analyses, the time to observe all the protocol-specified events can exceed the finite follow-up of a clinical trial or can lead to much delayed release of outcome data. With the advent of multiple classes of oncology therapies leading to much longer survival than in the past, this issue in clinical trial design and conduct has become increasingly important in recent years. We propose a straightforward prespecified backstop rule for trials with a time-to-event analysis and evaluate the impact of the rule with both simulated and real-world trial data. We then provide recommendations for implementing the rule across a range of oncology clinical trial settings.

When designing a randomized clinical trial to compare two treatments, the sample size required to have desired power with a specified type 1 error depends on the hypothesis testing procedure. With a binary endpoint (e.g., response), the trial results can be displayed in a 2 × 2 table. If one does the analysis conditional on the number of positive responses, then using Fisher's exact test has an actual type 1 error less than or equal to the specified nominal type 1 error. Alternatively, one can use one of many unconditional “exact” tests that also preserve the type 1 error and are less conservative than Fisher's exact test. In particular, the unconditional test of Boschloo is always at least as powerful as Fisher's exact test, leading to smaller required sample sizes for clinical trials. However, many statisticians have argued over the years that the conditional analysis with Fisher's exact test is the only appropriate procedure. Since having smaller clinical trials is an extremely important consideration, we review the general arguments given for the conditional analysis of a 2 × 2 table in the context of a randomized clinical trial. We find the arguments not relevant in this context, or, if relevant, not completely convincing, suggesting the sample‐size advantage of the unconditional tests should lead to their recommended use. We also briefly suggest that since designers of clinical trials practically always have target null and alternative response rates, there is the possibility of using this information to improve the power of the unconditional tests.