We recently reported in a previous blog post that the number of Phase I studies decreased in 2015 over previous years.
We proposed a number of possible reasons that may have contributed to the decrease in absolute numbers, including that the size of the Phase I studies was increasing. We have reviewed the data in Clinicaltrials.gov and found that as shown in Figure 2 the average number of subjects per trial has increased from an average of 50 subjects per trial in 2013 to 57 subjects per trial in 2015.
The number of subjects enrolled in total in 2015 has grown from 101,615 in 2013 to 123,403 subjects, an increase of 21.4% while the numbers of studies only increased 7.2%, further providing evidence that the Phase I study is increasing in size and complexity.
We provide this in light of the tragic events in Rennes, France with the Bial trial where a subject dies and several volunteers were hospitalized. The obvious question that has been raised with the current debate, are Phase I trials becoming too complex and almost becoming Phase 2 type studies to compress timelines?
The information that we have seen on clinicaltrials.gov certainly supports that the thesis that Phase I studies are becoming larger. However, since Phase I studies do not have to be registered on CT.g then this conclusion is speculative. It could be that in prior years Phase I studies were not registered, and probably the small phase I studies are probably not reported on CT.g. I would suggest that the main reason Phase I studies are reported are to obtain an NCT number so the results can be published in a peer review journal. Most of the small phase I studies conducted on healthy volunteers are rarely published in peer reviewed journals and therefore we may well be seeing a selection bias in Phase I registrations in CT.g: Only the large Phase I studies are registered.
While most phase I studies conducted in 2013 and 2014 have concluded, we can also see on Figure 2 that a large number of them still report enrollment as anticipated, indicated that the sponsor did not either update the data on CT.g once the study ended or it is truly still ongoing. This may be taken as an indicator reinforcing the above theory.
In reviewing the data, it seems that the anticipated enrollment is consistently significantly above actual enrollment for phase I studies submitted to CT.g as shown in Fig 3. However, in 2015 the forecast versus actual numbers of subjects enrolled are becoming much closer in alignment. It begs the question is the Biopharmaceutical industry improving its forecasting?
It should be noted that we would never expect total concordance in these numbers since there will always be a number of studies which are forecasted and are canceled and stopped before completion. In my view 2015 numbers are close to parity with the allowance for cancellation.
I would be interested in other opinions, so please comment on these concepts.