Comparison of outcomes of phase II trials (P2Ts) and subsequent randomized control trials (RCTs) using identical therapeutic regimens

Reviewer: Maria Luisa Veronese, MD
Abramson Cancer Center of the University of Pennsylvania
Last Modified: June 6, 2004

Share article

Presenter: M.I. Zia
Presenter's Affiliation: Princess Margaret Hospital, Toronto, ON, Canada
Type of Session: Scientific


Phase I, Phase II and Phase III clinical trials represent the paradigm of clinical drug development.  Phase I studies are usually conducted in patients with refractory diseases and the endpoints are safety, tolerability and an indication of efficacy.  Phase II studies are conducted in patients with a specific tumor type to evaluate efficacy and toxicity and the endpoint is usually response rate (RR).  Phase III trials are large, usually comparative randomized studies and have as main endpoint overall survival (OS).  Randomized clinical trials (RCTs) represent the basis for the clinical management of the patients.  Phase III studies are expensive and, unfortunately, most of them are negative.  They are often based on promising results from phase II studies (P2Ts).  There are no proven strategies that ensure success of RCTs.  This analysis was conducted to determine 1) whether promising results of P2Ts can be reproduced in randomized phase 3 trials (P3Ts) ; 2) which P2Ts characteristics can predict success of P3Ts.

Materials and Methods

  • All RCTs of chemotherapy or combined chemotherapy and radiation therapy published between 7/1998 and 6/2003 were analyzed
  • Only trials conducted in patients with solid malignancies were included in this analysis
  • Only RCTs published in the English language literature
  • Eligible RCTs were reviewed to identify preceding P2Ts 
  • Only P2Ts and RCTs with identical therapeutic regimens and conducted in the same patient populations were included in this analysis
  • RCTs were considered to be positive if the experimental regimen was significantly better than the control in terms of primary endpoints
  • Response rates from both P2Ts and RCTs were analyzed.  If the endpoint of a RCT was not specified, overall survival was considered the endpoint
  • Variables from P2Ts included in this study were : number of patients, whether randomization was applied, whether it was a multicenter study,  and the impact factor of the journal in which a P2T was published
  • Comparison between P2Ts and P3Ts and analysis of the influences of these variables from P2Ts on outcomes of subsequent RCTs were performed using the paired t-test and logistic regression


  • 181 RCTs were identified
  • 52 preceding P2Ts
  • 43 RCTs used the same therapeutic regimen as those in 52 preceding P2Ts
  • 10/43 (23%) RCTs were considered positive RCTs 
  • 8/43 (18.6%) RCTs had RR > P2Ts
  • Mean difference in response rates between P2Ts and RCTs was 12.5% (p=0.0001)
  • 1.9% of P2Ts were randomized
  • P2Ts RR was superior to RR of the corresponding RCTs across all tumor types
  • The number of patients entered in preceding P2Ts was the only statistically significant predictor of a successful RCTs 

Author's Conclusions

  • Positive P2Ts are not predictive of positive RCTs
  • RR in most RCTs are lower than those in preceding P2Ts
  • Number of patients in preceding P2Ts is the only significant predictor of a positive RCT
  • P2Ts should include an adequate number of patients
  • Small P2Ts should be considered with caution when designing RCTs

Clinical/Scientific Implications

RCTs are large and expensive studies that represent the basis of clinical management of patients.  Unfortunately, very few RCTs are positive and promising results from P2Ts are not predictive of a successful RCT.  This analysis was conducted to evaluate whether promising results of P2Ts can be reproduced in randomized phase 3 trials (P3Ts) and to determine which P2Ts characteristics can predict success of P3Ts.  Less than half of RCTs were preceeded by a P2Ts with the same regimen and patient population.  Only 20% of RCTs were positive and the RR of RCTs is lower that the RR in the preceding P2T.  The number of patients in P2Ts was the only statistically significant predictor of a positive RCTs.  This study did not include trials conducted in patients with hematologic malignancies, only studies published in the English language were included and unpublished RCTs were not included.  A careful analysis of P2Ts is crucial prior to designing RCTs and future P2Ts should include adequate number of patients.  Additionally, randomized P2Ts may provide a more efficient trial design.

Oncolink's ASCO Coverage made possible by an unrestricted Educational Grant from Bristol-Myers Squibb Oncology.

I Wish You Knew

How cancer patients have changed my life

View More

Blogs and Web Chats

OncoLink Blogs give our readers a chance to react to and comment on key cancer news topics and provides a forum for OncoLink Experts and readers to share opinions and learn from each other.

OncoLink OncoPilot

Facing a new cancer diagnosis or changing the course of your current treatment? Let our cancer nurses help you through!

Learn More