CANCER CLINICAL TRIAL DESIGN IN THE ERA OF TARGETED AND IMMUNOLOGICAL THERAPIES

Abstract

The aim of this thesis was to understand and improve the design of clinical trials testing targeted and immunological anticancer therapies. My systematic review of contemporary phase 3 trials of targeted and immunological therapies identified that almost a quarter of trials (23%) with a statistically significant result for the primary endpoint of overall survival reported an observed effect size that was probably not clinically meaningful. The findings of preceding trials (n=542) that were cited as the justification for a subsequent phase 3 trial (n=205) were reported ‘positively’ by their authors in 87%, but a priori criteria for judging their findings ‘positive’ were specified in only 74%, and met in only 53%. Characteristics of the preceding trials that were associated with a positive primary outcome in the subsequent phase 3 trial were industry-sponsorship (odds ratio 2.0, p=0.01) and conduct in the same tumour type (odds ratio 2.4, p=0.05). I analysed the views of investigators and patients participating in a randomised, phase 3 trial of adjuvant targeted therapy for renal cell carcinoma. Investigators judged larger survival benefits than their patients necessary to make adjuvant sorafenib worthwhile (2% of investigators vs 21% of patients judged sufficient an extra 1% beyond a five year survival rate of 65%, p<0.0001). Phase 3 trials should be designed and powered such that observed effect sizes which are statistically significant are also clinically meaningful. Researchers should pre-specify and report the target level of activity in a phase 2 trial that would warrant further research. Researchers testing targeted or immunological therapies should obtain preliminary evidence of activity in the same tumour type prior to embarking on a large scale, phase 3 trial. Researchers and clinicians should appreciate that their judgements may differ from those of their patients and colleagues, and be wary of overestimating, or overstating, their potential benefits.