Charles Liu, PhD, Cytel Statistician and Product Manager
In the US, cancer is the most common cause of death after heart disease, accounting for nearly 1 of every 4 deaths [1]. Tackling the immense burden of this disease, researchers are currently investigating an estimated 900 novel cancer agents in over 6,000 clinical trials [2]. Unfortunately, the clinical success rate remains low, and failed trials amount to billions of dollars invested, while providing little direct benefit to patients. Such failures also discourage patients from participating in the testing of novel treatments: Currently, less than 2% of cancer patients enroll in clinical trials [3].
One contributing factor to high failures rates is the uncertainty about the treatment effect or expected benefit. In a paper published in the Journal of the National Cancer Institute, Gan and colleagues (2012) [4] analyzed over 250 published oncology trials. They found little correspondence between the expected and observed benefits, and more important, the expected benefits consistently overestimated the observed benefits often by a large margin. They concluded that, “the use of adaptive trials designs, although still relatively infrequent, offers potential advantages”, including early stopping rules for efficacy or futility, and adaptive increases in sample size. By capitalizing on information observed during the trial, the uncertainty inherent before the trial (and hence the probability and cost of failures) can be dramatically reduced.
Adaptive designs for Phase 3 trials have been a core component of East for many years. In addition, similar principles can be applied to early phase designs. For instance, Simon’s two-stage design is a common adaptive design for Phase 2 oncology trials [5], which features an early stopping rule for futility, while attempting to minimize the expected or maximum number of patients required. For Phase 1 dose escalation trials, a number of modern adaptive methods such as the - modified toxicity probability interval (mTPI) method [6], and the Bayesian logistic regression model (BLRM) [7] - have been shown to be more accurate than traditional methods at picking the maximum tolerated dose (MTD). These modern methods provide greater flexibility for adjusting specific decision rules, to maximize the eventual probability of success, and to minimize risk to patients.
--
1. Cancer Facts & Figures 2013
3. Participation in cancer clinical trials
4. Assumptions of expected benefits in randomized phase III trials evaluating systemic treatments for cancer
5. Optimal two-stage designs for phase II clinical trials
6. A modified toxicity probability interval method for dose-finding trials
7. Critical aspects of the Bayesian approach to phase I cancer trials
Related Items of Interest
Register now for our upcoming webinar:
A Complimentary, Instructional Webinar with Cytel Adaptive Trial Expert, Pantelis Vlachos PhD
East® ESCALATE is the new East software module to design, simulate, and operationally support the Phase 1 dose escalation trials in determining maximum-tolerated dose (MTD).
In addition to the popular 3+3 design, East® ESCALATE includes the modified Toxicity Probability Interval (mTPI) method, the Continual Reassessment Method (CRM), and the Bayesian Logistic Regression Model (BLRM).
Cytel adaptive trial expert Pantelis Vlachos will begin by reviewing the underlying theory, then demonstrate early phase design applications using East® ESCALATE.
This webinar will teach you how to:
- Evaluate and compare the operating characteristics of your designs under different dose-toxicity profile assumptions.
- Based on modeling the accumulating data, recommend the optimal dose for the next cohort of patients
- Clearly communicate critical information to clinicians to better guide dosing decisions.
- Improve screening and selection of active agents to increase clinical success rates.
