OR WAIT null SECS
One of the primal questions clinical trialists engaged in clinical research and development always face is do we “keep or kill” a drug based on the data in hand.
One of the primal questions clinical trialists engaged in clinical research and development always face is do we “keep or kill” a drug based on the data in hand, specifically when we are left with the decision of whether to pursue an asset or not at the end of a Phase II trial.
Key approaches in reducing Phase III attrition include a better target selection (selection of more validated and druggable targets) and routine pursuit of early POC studies in the clinic, especially in Phase I, for which biomarkers and surrogate endpoints can often be employed¹. A robust early phase trial design will help lay the groundwork for a successful Phase III. A Phase III trial may not be even warranted after a successful Phase II² for certain indications provided all the appropriate questions are asked and answered through a clinically sound Phase II trial design.
The robustness of historical data and their application in a clinical trial design especially when go and no go decision is based on these trials are not only critical but extremely valuable. However, there appears to be a gap that exists. This gap can be addressed fairly easily. For example, in oncology trials we always look for “null” and “alternative” response rates of an outcome like the tumor response when studying an agent. As null responses require us to understand the historical data, it is critical that they become an integral part of a study design. Disappointingly, that always is not the case.
One of the very first reviews, exploring the utility of historical data in go and no go decisions was published in the Journal of Clinical Cancer Research³. Based on their conclusion, many of the Phase II trials require historical data to determine null response rates but very few did so appropriately and those that did not concluded that the experimental regimen was worth studying further. Of the 134 qualifying trials (out of the 251 that were chosen for this review), 70 (52%) of the trials required historical data for study design. However, 32 (46%) of these trials did not cite the source of the historical data used and only 9 (13%) of the trials had clearly provided a single estimate as a rationale for the null based on historical data. Trials that did not cite historical data appropriately were significantly more likely to declare an agent to be active (82% versus 33%). Therefore, careful consideration should be given to historical data and their relevance when designing clinical trials especially while testing compounds in oncology.
Clinically Meaningful Outcomes:
In 2014, ASCOâ´ undertook an exercise convening working groups for improving the design of future clinical trials yielding results that are clinically meaningful in patients (significantly improving survival, quality of life (QOL) or both). One of the key outcomes from this exercise was that patient symptoms from cancer progression and tolerance to treatment related toxicities were critical when considering whether a new treatment produces a clinically meaningful outcome. For example, a less toxic compound with negligible increase in efficacy is acceptable whereas conversely a highly toxic compound need to have established a substantial increase in efficacy to be clinically meaningful for patients. Overall, there seems to have been a consensus that overall survival will continue to serve as the primary measure of clinically meaningful outcome while the value of progression free survival and other surrogate end points as valid end points may need to be applied in certain clinical situations.
As such, appropriately utilizing historical data and employing clinically meaningful outcomes in a clinical trial design can help improve the quality of data and potentially address this attrition. Although these are not all encompassing strategies or solutions, these measures will result in generating trial data that is more robust and evidence based eventually helping us within clinical research and development in the go and no go decision making process. So I encourage, as clincial trialists engaged in clinical research and development, we begin not only to explore but also begin to adopt strategies and design considerations including the aforementioned in practice, while continually looking for ways and means of further optimizing “go and no go decisions”.