Though overall survival has traditionally been considered the benchmark endpoint in determining efficacy of a treatment in large randomized controlled trials, new data analyzing outcomes of trials in soft tissue sarcomas where progression-free survival and response rate were the primary endpoints found those endpoints to be appropriate surrogates in trials. The findings were published online ahead of print in the Journal of Clinical Oncology (doi: 10.1200/JCO.2015.64.3437).
"Soft tissue sarcomas are a cluster of rare and heterogenic tumors, making trial design and interpretation particularly challenging," the study's lead author, Alona Zer, MD, of the Division of Medical Oncology at Princess Margaret Cancer Center in Toronto, explained in an email. "Many sarcoma randomized controlled trials used progression-free survival as a primary endpoint. We found that overall survival-the most definitive endpoint in oncology-was used as primary endpoint in only four percent of randomized controlled trials.
"It was reassuring to discover that progression-free survival is a good surrogate for overall survival," she said. The data also showed that other endpoints increasingly being used in clinical trials-three-month progression-free survival and six-month progression-free survival-have no correlation with overall survival.
Study Details
For the study, the researchers analyzed data from 52 randomized controlled trials in locally advanced/metastatic soft tissue sarcoma. The studies included 9,762 patients and 45 of the studies evaluated cytotoxic agents-and 11 studies included three arms (two experimental arms and one control arm), bringing the total number of comparisons in the analysis to 63.
Some of the key findings and trends about the soft tissue sarcoma randomized controlled trials that the researchers reported were the following:
* The number of trials in soft tissue sarcoma has increased substantially over time with 19 studies being published between 1974 and 1994 and 33 from 1994 to 2014.
* Overall survival was the primary endpoint in two studies, whereas 23 studies used other time-to-event primary endpoints.
* Over time, there was an increase in the frequency of progression-free survival as a primary endpoint, as well as a reduction in the use of response rate as an endpoint.
* Secondary endpoints included overall survival in 39 studies, toxicity/safety in 35 studies, other time-based events in 24 studies, and response rate in 23 studies.
* Three studies included quality of life as a secondary endpoint.
* Sponsorship was reported in 42 studies, with 10 studies being sponsored by industry, 23 being sponsored by a cooperative group or an institution, and nine receiving combined funding.
* Compared with studies published between 1974 and 1994, the studies published in the past two decades were more likely to be supported by industry.
* Toxicity was reported comprehensively in 20 studies and reported poorly in six studies-with no observed changes in toxicity reporting over time.
Surrogacy Analysis
The researchers analyzed the surrogacy of the progression-free survival and response rate endpoints for overall survival using weighted linear regression (including only those studies reporting the hazard ratio for overall survival and the hazard ratio for either progression-free survival or data allowing the calculation of overall response for response rate, three-month progression-free survival, or six-month progression-free survival). The data showed a highly significant correlation between overall survival and progression-free survival, a substantially significant correlation between response rate and overall survival, and no significant correlation between three-month progression-free survival and overall survival or between six-month progression-free survival and overall survival.
The eligible trials included in this study were found by a search of the MEDLINE, MEDLINE in-process, and EMBASE clinical trial databases, as well as meeting abstracts from the American Society of Clinical Oncology, the European Society for Medical Oncology, and the European Cancer Congress. The search included studies conducted between 1974 and December 2014. Because of their distinct biologic characteristics, trials involving bone sarcoma or gastrointestinal stromal tumors were excluded. The researchers also excluded review articles, meta-analyses, systematic reviews, phase I and II nonrandomized clinical trials, observational studies, and case reports.
Key Messages
"The main message from this study is that the new emerging endpoints-three- and six-month progression-free survival-should be used with caution, and probably not as the primary endpoint in randomized controlled trials for soft tissue sarcomas," Zer said. "And overall survival is still the ultimate primary endpoint, even though it requires larger, longer, and more expensive studies.
"And if a surrogate endpoint is chosen in place of overall survival, our data support the use of progression-free survival."
Additionally, Zer noted that a practical message from the study is that when reading a phase III randomized controlled trial abstract clinicians should pay attention to what primary endpoint was chosen for the study, she said. "Is it legitimate? Does it represent effectiveness of the treatment? Was it validated?"
In an accompanying editorial also published online ahead of print in JCO, Fengmin Zhao, MHS, PhD, of the Department of Biostatistics & Computational Biology at Dana-Farber Cancer Institute, discusses a few limitations of the study by Zer et al., including cautions about properly validating surrogate endpoints (doi: 10.1200/JCO.2016.66.4581). But, he notes in the editorial: "Continuous efforts to improve the validation of surrogate endpoints are important in several ways.
"In this era of personalized medicine and rapid development of oncology drugs, many drugs are in the pipeline, waiting for testing in phase III clinical trials. It is difficult, if not impossible, to conduct a phase III trial with overall survival as the primary endpoint for a rare disease. Surrogate endpoints usually allow for smaller trials and shorter completion times and, once validated, could help resolve these issues."
Sarah DiGiulio is a contributing writer.