“Involvement of statisticians in clinical trials is very much essential, and they should be a part of the committee right from the beginning,” emphasised Dr. Cyrus R. Mehta, President of the Boston-based Cytel Incorporation. Cytel is involved in trial design, programming and analysing clinical data.
Fifty per cent of new molecules that graduate to Phase III fail outright as they do not meet the desired end point of preventing/treating diseases. Of this, 35 per cent fail on safety grounds. “And the remaining 15 per cent is due to poor trial design,” he said, highlighting the importance of statisticians’ involvement in clinical trials.
Statisticians play an important role at several stages of a trial — designing the trial, making midway corrections to the conduct of the trial, and finally in analysing the data. For instance, “even if the drug is effective, it will not be known if the trial design is not good,” Dr. Mehta said. Similarly, how many different drug doses should be tested, what dosage levels should be tested, the number of people to be studied, to name a few, are determined by these people along with clinicians. “The highest dosage [tested in earlier phases of the trial] may not be always be good as they may cause serious side effects,” he explained.
The FDA released last month the draft guidelines on a new trial methodology — Enrichment designs for clinical trials. Enrichment refers to the use of “any patient characteristic to select a study population so that detection of the drug’s effect is more likely to be seen than in an unselected population. It is possible to test a population with and without the enrichment characteristics and finally zero in on the subgroup that has the particular characteristics.
This is particularly becoming popular in the case of cancer where people with a certain gene are more likely to benefit from the drug. This is because the “drug has been designed to bind to a particular molecular target and only those who have the gene stand to benefit.” Colorectral cancer and non-small cell lung cancer are two such examples.
If patients have particular genetic biomarkers, then certain drugs bind to them, and as a result they will survive longer.
Since drug companies do not have enough data to target only a specific population, they tend to study a wider population of patients. “Then based on the interim data analysis, the trial may continue with a certain subgroup that has the gene,” Dr. Mehta said. “This is called precision medicine.” Statisticians play a role in deciding how to choose the subgroup.
Statisticians have been playing a big role for the last 10 years in designing adaptive trials. In this, it is possible to change the future conduct of a trial midway by altering the number of subjects and even dropping certain doses being tested.
In effect, the data monitoring committee can alter the design and conduct based on interim analysis of the unblinded data. “This is done without undermining the statistical properties of the trial design,” he said.
