SUMMARY We propose a Bayesian optimal phase II (BOP2) design for clinical trials with a time‐to‐event endpoint (eg, progression‐free survival [PFS]) or co‐primary endpoints consisted of a time‐to‐event endpoint and a categorical endpoint (eg, PFS and toxicity). We use an exponential‐inverse gamma model to model the time to event. At each interim, the go/no‐go decision is made by comparing the posterior probabilities of the event of interest with an adaptive probability cutoff. The BOP2 design is flexible in the number of interim looks and applicable to both single‐arm and two‐arm trials. The design maximizes the power for detecting effective treatments, with a well‐controlled type I error, thereby bridging the gap between Bayesian designs and frequentist designs. The BOP2 design is easy to implement. Its stopping boundary can be enumerated and included in study protocol before the onset of the trial for single‐arm studies. Simulation studies show that the BOP2 design has favorable operating characteristics, with higher power and lower risk of incorrectly terminating the trial than some Bayesian phase II designs. The software to implement the BOP2 design will be freely available at www.trialdesign.org .
This short communication concerns a biomarker adaptive Phase 2/3 design for new oncology drugs with an uncertain biomarker effect. Depending on the outcome of an interim analysis for adaptive decision, a Phase 2 study that starts in a biomarker enriched subpopulation may continue to the end without expansion to Phase 3, expand to Phase 3 in the same population or expand to Phase 3 in a broader population. Each path can enjoy full alpha for hypothesis testing without inflating the overall Type I error.
TPS4137 Background: Pembrolizumab (pembro) is a humanized monoclonal antibody against PD-1 that prevents binding of PD-1 to PD-L1 and PD-L2 and permits activation of an antitumor immune response. In KEYNOTE-012, pembro showed a manageable safety profile and a 22% ORR in pts with advanced gastric cancer. The randomized, open-label, phase 3 KEYNOTE-061 study (NCT02370498) is designed to compare the efficacy and safety of pembro with those of standard-of-care paclitaxel in the second-line treatment of advanced gastric cancer. Methods: KEYNOTE-061 is designed for patients with metastatic or unresectable gastric or GEJ adenocarcinoma that progressed after first-line treatment with platinum + fluoropyrimidine doublet chemotherapy. Patients with HER2/neu-positive tumors are eligible if they have documented progression on a regimen that also included trastuzumab. Other key eligibility criteria include measurable disease per RECIST v1.1, ECOG PS 0-1, no chemotherapy within 2 wk of first dose of study drug, and provision of a newly obtained or archival tumor sample for assessing PD-L1 status. Eligible pts are randomized 1:1 to receive pembro 200 mg Q3W or paclitaxel 80 mg/m2 IV on days 1, 8, and 15 of each 28-d cycle. Treatment will continue until disease progression, intolerable toxicity, refusal by pt or investigator, or completion of 24 mo of therapy (pembro arm only). Pts in the pembro arm who have a CR after ≥ 24 wk may discontinue after ≥ 2 doses following initial CR. Clinically stable pts who progress per RECIST v1.1 may continue pembro at the discretion of the investigator until a confirmatory CT scan performed ≥ 4 wk later. Response will be assessed every 6 wk per RECIST v1.1 by central imaging vendor review and per RECIST adapted for immunotherapy response patterns. AEs will be assessed throughout treatment and for 30 d thereafter (90 d for serious AEs). Primary efficacy end points are PFS per RECIST v1.1 and OS in pts with PD-L1+ tumors. Secondary end points include PFS and OS in all pts, time to progression, ORR, and duration of response. Enrollment in KEYNOTE-061 is ongoing and will continue until up to 720 pts are enrolled. Clinical trial information: NCT02370498.
Biomarker subpopulations have become increasingly important for drug development in targeted therapies. The use of biomarkers has the potential to facilitate more effective outcomes by guiding patient selection appropriately, thus enhancing the benefit-risk profile and improving trial power. Studying a broad population simultaneously with a more targeted one allows the trial to determine the population for which a treatment is effective and allows a goal of making approved regulatory labeling as inclusive as is appropriate. We examine new methods accounting for the complete correlation structure in group sequential designs with hypotheses in nested subgroups. The designs provide full control of family-wise Type I error rate. This extension of previous methods accounting for either group sequential design or correlation between subgroups improves efficiency (power or sample size) over a typical Bonferroni approach for testing nested populations.
2517 Background: Phase I studies in oncology patients seek to identify a dose of a new investigational agent that is tolerable and has the potential to be effective. Toxicity data are the key to defining dose levels of the new agent that are acceptable for further study. Several Phase I designs are available to researchers. Some are practical and easy to implement. Others are complex but yield better information on the toxicity of studied doses. We sought a Phase I design that had a high probability of selecting a dose with acceptable toxicity, allowed adaptation based on observed dose-limiting toxicity (DLT) data and did not create considerable enrollment delays. Methods: We compared the 3+3 design, an algorithmically-determined adaptive method proposed by Ji, et al. (2007), and variations on Ji's design including a 2-stage design combining a 3+3 with a modified Ji design. We simulated Phase I trials with 6 dose levels and 12 dose response relationships and studied the properties of the designs including how often the dose with the target toxicity level was selected, the expected percentage of patients with a DLT and the expected sample size. We also implemented operational conventions on top of the preferred design to enhance the practicality and clinical usefulness of the design. Results: In general, the adaptive, Ji-based designs outperformed the 3+3, selecting the target dose up to 18% more often. The 2- stage and other Ji-based designs were comparable in most scenarios and performed the best when the target dose was among the middle doses in the dose range tested. When the target dose was the highest dose or higher than all of the doses tested, the 3+3 and 2-stage designs outperformed some of the other Ji-based designs. Conclusions: For single agent studies, we prefer a 2-stage design in which a 3+3 is implemented until a preliminary maximum tolerated dose (MTD) is identified. Then, a modified Ji design (target toxicity rate = 20%) is used to confirm the MTD and allow further dose refinement. While this design may underperform slightly compared with the complex continuous reassessment methods, this design is easier to implement and more intuitive to investigators and ethical and regulatory review boards. In addition, it maximizes enrollment continuity while ensuring appropriate patient safety. [Table: see text]
Single-arm proof-of-concept (PoC) clinical studies are widely used to accelerate the signal-finding process in oncology drug development before or in lieu of randomized PoC studies. Traditionally the primary endpoint for single-arm PoC studies is objective response rate (ORR). However, in cases that ORR is not applicable or not clinically relevant, time-to-event (TTE) endpoint is used instead. One conventional approach is to dichotomize the TTE endpoint as a binary endpoint to assess the survival rate, which may compromise the testing efficiency due to the requirement of minimum follow-up without censoring. Alternatively, we can use the non-parametric one-sample log-rank test (OSLRT) to evaluate the survival curve difference compared with historical control. This approach can incorporate censoring and all time-point information on the survival curve, but the test statistic may be difficult to interpret and quantify the magnitude of treatment effect. Given that clinicians are more interested in the survival rate at a clinically relevant landmark timepoint, we can also use a landmark Kaplan-Meier method (LMKM) to estimate the survival rate at a landmark timepoint for design and analysis of single-arm proof-of-concept oncology trials with TTE endpoint. This non-parametric method is straightforward to clinicians and shows favorable operating characteristics from simulation studies. We also developed an R package for the implementation of these mainstream designs, which fills the gap of available software for design and analysis of single-arm studies with TTE endpoint.
A new two-stage design is proposed that is suitable for early detection of the anticancer activity of experimental therapies in Phase II oncology trials. The endpoints of interest are response rate and early progression rate. The anticancer activity is defined by a positive signal in one endpoint and a non-negative signal in the other endpoint. The two endpoints are modeled by the multinomial distribution. The design is optimal in that it minimizes the patient exposure when the experimental therapies are inactive. The design parameters are found by a grid searching algorithm under type I and type II error rate constraints. Examples of the design are also presented in this paper.
'%3e%3cpath fill='%23012169' d='M0 0v30h60V0z'/%3e%3cpath stroke='%23fff' stroke-width='6' d='m0 0 60 30m0-30L0 30'/%3e%3cpath stroke='%23C8102E' stroke-width='4' d='m0 0 60 30m0-30L0 30' clip-path='url(%23b)'/%3e%3cpath stroke='%23fff' stroke-width='10' d='M30 0v30M0 15h60'/%3e%3cpath stroke='%23C8102E' stroke-width='6' d='M30 0v30M0 15h60'/%3e%3c/g%3e%3c/svg%3e)