Citations

Predicting the times of milestone events, ie, interim and final analyses in clinical trials, helps resource planning. This manuscript presents and compares several easily implemented methods for predicting when a milestone event is achieved. We show that it is beneficial to combine the predictions from different models to craft a better predictor through prediction synthesis. Furthermore, a Bayesian approach provides a better measure of the uncertainty involved in prediction of milestone events. We compare the methods through two simulations where the model has been correctly specified and where the models are a mixture of three incorrectly specified model classes. We then apply the methods on two real clinical trial data, North Central Cancer Treatment Group (NCCTG) N0147 and N9841. In summary, the Bayesian prediction synthesis methods automatically perform well even when the data collection is far from homogeneous. AnRshiny app is under development to carry out the prediction in a user‐friendly fashion.

Background More than half of the 40,000 incident rectal cancer patients in the United States each year are diagnosed at clinical stage II and III (locally advanced stage). For this group, high rates of cure can be achieved with the combination of pelvic radiation and sensitizing 5-fluorouracil (chemoradiation), surgery and chemotherapy, but treatment is long, arduous and toxicities are substantial. The PROSPECT trial (N1048, NCT01515787) was designed to determine whether neoadjuvant chemotherapy with 5-fluorouracil and oxaliplatin (FOLFOX) could be used as an alternative to neoadjuvant chemoradiation without compromising treatment outcomes and to spare these patients excess toxicity. The statistical design balanced the twin co-primary goals of achieving low local and distant recurrence rates. Study design features contended with the need for stringent safeguards given limited phase II data, the need for straightforward criteria to facilitate both accrual and protocol fidelity and the importance of patients’ perspectives on symptom burden and treatment toxicity. Methods PROSPECT is an ongoing multi-site two-group seamless phase II/III randomized trial comparing standard neoadjuvant chemoradiation versus neoadjuvant chemotherapy with selective use of chemoradiation for patients with locally advanced rectal cancer. Challenges addressed in the design and conduct of PROSPECT have included the following: (1) setting safety thresholds given limited single-center phase II data, (2) establishing workable eligibility criteria, (3) balancing competing time to local and distant recurrence as co-primary endpoints and (4) obtaining reliable and complete data for patients’ symptom burden. The design and implementation challenges, choices, modifications and their implications for the design of future national cooperative group clinical trials are presented. Results PROSPECT incorporated stringent thresholds for both complete surgical resection (R0) and the time to local recurrence as early stopping rules. When predetermined stopping criteria were not met after evaluation of the first 366 participants in the randomized phase II, the study transitioned seamlessly to phase III with cumulative accrual of over 1000 participants. Eligibility criteria stipulating rectal tumor location based on distance from the anal verge were unworkable, and the protocol was amended to a more pragmatic approach that assigned surgeons with primary responsibility for determining eligibility. Central radiology review was feasible and in some cases prompted discontinuation of protocol treatment. Participation in toxicity reporting using the National Cancer Institute’s Patient-Reported Outcomes Version of the Common Terminology Criteria for Adverse Events was uniformly high and was well accepted by participants from over 200 sites in the United States, Canada and Switzerland. Conclusion The strategies used to overcome these obstacles may inform the design of other studies that involve multi-modality treatment interventions, particularly trials where implementation of consistent criteria for eligibility and outcomes across hundreds of practice settings is necessary.