So what’s the answer?

I’ve often seen that if you want to learn about the problems that people face, ask them for their recommendations. The converse also works, if you ask for potential solutions, people start describing their problems. We’ve employed both questions from a myriad of stakeholders as described in all our weeknotes. We’ve been hearing a number of issues again and again, but the one I am thinking about most this week is enrollment.

“We need more clinical trials.” “There need to be bigger and better clinical trials.” “Trials are going abroad because it is difficult to enroll patients in the United States.” “We need an ethical way to engage more research subjects and patients.” “We need to support new clinical trials such as n = 1 and organ-on-a-chip where enrollment is not a big factor to be able to minimize costs.” Those are the types of answers we have heard from almost everyone we have talked to. It’s not just the experts, each of our team members has been circling the issue of enrollment for quite a while. From the international team who see it as a trend of trials shifting abroad to the risk team who weigh the benefits and downfalls of the requirement of high enrollment numbers, our team has been tackling this issue from a number of angles. To get an idea of how important an issue of enrollment is, the following graphic shows the distribution for enrollment numbers in clinical trials.

Figure 1: Enrollment Numbers for Clinical Trials from the CT.GOV Database

It’s funny how we see bumps at whole number enrollments (100, 200, 300, 400… 1500), which shows that enrollment number for a clinical trial is an issue that is determined by more than just math. And, in fact, it influences multiple factors and gives rise to several inefficiencies of the system. Trial cost, and time to completion are directly related to enrollment. But other issues our research group has found are not far behind. The success of CRRI that we visited over the summer was directly related to them being able to recruit relevant research subjects quickly. The issues with international trials revolve around the ethics associated with international research subjects, but also the ease with which enrollment happens abroad. Here’s a graphic showing how international trials tend to be completed quicker than US trials, even though enrollment numbers for international trials are slightly higher.

Figure 2: Time to Completion for Clinical Trials in the US and Abroad

Our data team used a machine learning algorithm to try and see what the most important factors are that influence a trial being terminated. With ten factors, we can predict with 75% accuracy if a trial will be terminated or not given our model. The most important factor that determines if a trial will be terminated? You guessed it: enrollment. Enrollment is directly relevant to our team members interested in personalized medicine, n = 1 trials, innovations that reduce requirements for research subjects, and patient-centered trial design.

So what’s the answer? “We need to tell everyone that clinical trials are the best way to get care, and they need to enroll in as many as possible to help society.” “Find a way to get research subjects in trials and keep them.” “Society needs to realize the importance of clinical research.” Again, we’ve heard these ideas several times, but they don’t really answer the question about enrollment. They say the answer to bad enrollment is more enrollment. Of course, we can come up with ideas on how to increase enrollment, but solutions like this for society rarely work. So what would I do if I could control the clinical trial system?

*The following reflect only the opinions of the author, and not necessarily the MIT Collaborative clinical trials team

Let’s divide the histogram above into the phases of clinical trials. Here are Phase 1,2, and 3 next to each other:

Figure 3: Enrollment Distribution for Phase 1 Trials

Figure 4: Enrollment Distribution for Phase 2 Trials

Figure 5: Enrollment Distribution for Phase 3 Trials

Phase 1 tests the safety of the drug. Phase 2 tests efficacy and safety. By the time a trial passes Phase 2, it is supposed to be safe and have some efficacy. Phase 3 involves safety, efficacy, and effectiveness and is by far the biggest driver of high enrollment numbers.

But our team knows that Phase 1,2,3 don’t happen one after the other. There are many reasons for this, but again, enrollment is a big issue. It’s smart for companies to begin enrollment for Phase 3 early, as those trials are the most expensive and take the longest time to complete. Not only that, but many Phase 3 trials continue well after the drug has been approved. So really, Phase 3 trials are where the largest issue with enrollment takes place, mostly because they seem to be the most important part of the process.

The first part of my proposal is to isolate Phase 3 trials so that they cannot be done before Phase 1 and 2 are complete and approved by FDA. This will ensure 1) the drugs that have been through Phase 1 and 2 are at least safe and effective by FDA standards so can be consumed by anyone, 2) Expensive Phase 3 trials do not get started and time is not spent designing them if not needed, and 3) it allows a better filter in a system clogged with clinical trials and long regulatory time.

Note that this will result in larger enrollment numbers for Phase 1 and 2, as I am sure the FDA will need to be more stringent for approving Phase 1 and 2 drugs for being safe and effective without additional data from Phase 3. This is fine, this is the entire purpose of Phase 1 and 2. They have been designed to test safety and efficacy before the entire population can consume the drug, and this is what I believe will make the system more efficient.

The second part of my proposal could be more controversial. My last post about medicine being an art and a science hinted at how each patient is different and the drug that might work “on average” wouldn’t work for an individual. Similarly, a drug that might not work “on average” could save someone’s life. I believe to improve the efficiency of the system, Phase 1 and 2 approved drugs need to be made available to patients immediately so that their physician can prescribe them. Thus, Phase 3 of a trial will turn into something like a Phase 4 trial, which tests post-market drugs. What does the histogram for Phase 4 look like?

Figure 6: Enrollment Distribution for Phase 4 Trials

This already looks more like Phase 3 trials than Phase 1 or 2 trials. Phase 4 trials are conducted primarily by physicians for their patients. Enrollment in these trials is not a big issue, because they don’t need to be held primarily at a research site. Any drugs in Phase 4 trials have already been approved for use by the general public. They further test safety, efficacy, and effectiveness like Phase 3, but for more long term. Of course, like people, all drugs are different. So medical experts will need to be careful about which drugs in Phase 3 can be made freely available vs which ones can’t. But I will repeat this here: if Phase 1 and Phase 2 do what they are designed to do, there should be no reason why these drugs cannot be consumed by the general public.

The other advantage is that we can then set a price cap for drugs in Phase 3, while still allowing companies to sell the drug for a profit. In economic terms, a drug in Phase 3 should be priced at or below standard of care, even if an official cap is not set. This allows companies to make profits earlier, patients to get access to drugs cheaply, and also encourages patients to sign up for a Phase 3 trial thus increasing enrollment.

The missing data from my analysis is the number of adverse events for a drug in Phase 3 that has been through Phase 1 and 2 without an adverse event.  This data can further enlighten the ways in which the proposal can take force. If, in fact, there are not that many adverse events in Phase 3, then Phase 1 and 2 are working, and the proposal is implementable. If not, then it shows we need to improve Phase 1 and 2 before proceeding with any such idea.

Another issue will be that doctors normally have a lot of evidence from Phase 3 trials when they decide to prescribe a particular drug. They will now have to do this without that evidence, thus making clinical practice more difficult. But I see this as a positive, as it allows doctors to be more flexible in having different therapies for different patients. In this age of big data, the results accumulated from these new Phase 3 trials can be made quickly available for doctors to get the evidence as the trial is ongoing. It also brings the doctors close to research, and allows easier participation of academics within the clinical trial system. It can also lead to more translational research as opposed to just about the drugs, as now more individuals have access to new therapies and data.

If this became reality without adverse events to patients, then we would have reduced the time for drugs to get approved, reduced the overall cost and waste in the system, satisfied every stakeholder (quicker approval to market, less regulatory time, more patient agency), and also not have a big bill on our hands.

I have to admit, this isn’t my idea. The idea of revisiting the regulation around efficacy and effectiveness came from Debashish Roychowdhury at our first round table last year in Cambridge. The idea lingered in my head for a while, and I think there is enough evidence out there to support it.

Of course, my recommendation may not be an idea that’s easily implementable. That’s the sad part, very little research is done on implementation out there. And as Justin likes to remind me, policy recommendations are not great recommendations. But we’ll keep working on from here, and any ideas on how to tackle these issues will be greatly appreciated!