Published today, three-plus years in the making, is an article decribing our clinical strategy for preventing genetic prion disease and our engagement with regulators to chart a path towards this goal [Vallabh 2020]. This article lays out many of the same arguments we made in our Scientific American article earlier this month, but with all the scientific details to back it up. (See also: source code and an accompanying commentary [Aguzzi & Frontzek 2020]). In this post I’ll explain why we wrote this article and what it means to us.
When we joined the Broad Institute in 2014 as PhD students, the only goal on our minds was to identify the therapeutic molecule that would be the first drug for prion disease. Then, Eric Lander sat us down and asked a tough question:
How will you show that it works in humans?
We were wrongfooted. We had never thought once about how clinical trials were designed or conducted or how it should be done in our disease. But the more we reflected on our own experience with Sonia’s mom’s disease, and the more that Eric talked us through the way drugs get made and approved, the more we realized that designing the right clinical trial might be just as challenging as finding the drug molecule in the first place.
Business as usual, we came to learn, is that clinical trials are done in sick patients. People with symptoms are randomized to drug or placebo and followed to see how long they survive or how they perform on some other clinical endpoint such as cognitive function. That’s one model for prion disease trials, it has been done before and will likely be done again, but we came to see a few reasons why it shouldn’t be our only option.
One is the rapid course of prion disease. Our experience of Sonia’s mom’s disease was that it was so mysterious, and so rapid, that she was never correctly diagnosed in her lifetime. Indeed, by the time she even saw a neurologist, she already had no quality of life left. Sonia has said many times: “There was no moment in time where both of the following were true: first, that it was clear something was really wrong, and second, that if it was me, I would have chosen to take a drug to stabilize me in her condition.” To be sure, some prion disease patients do get an accurate diagnosis earlier in their disease course, but our situation is not uncommon. Every year at the CJD Foundation conference, we hear many stories of the endless diagnostic odysseys that patients and their families have been on, with the patient’s condition becoming so advanced that by the time a diagnosis does come, they move immediately to hospice. Thus, when considering clinical trials in sick patients, we must contend with the questions of whether even the most rational, targeted therapeutic could still move the needle on their disease course, and, if it did extend survival but without bringing back lost cognitive function, whether that would be a benefit to the patients and their families.
Another reason to explore other options is the evidence from animal models that drugs that work early may not work if given too late. Several therapeutic small molecules that have been tested in prion-infected mice were able to double, triple, or even quadruple survival time if given early enough, but all of them were less effective the later they were given, and none showed any convincing efficacy at a clearly symptomatic stage. Of course, mice aren’t humans and we don’t know how their disease stages map onto ours, but prion disease is a disease that naturally afflicts many mammals, and we know that at least the underlying molecular mechanisms are the same. This is reason enough to envisage the following scenario: that we could hold in our hands the long-sought therapeutic molecule, one that could extend Sonia’s life by decades, but then put it into a clinical trial only in sick patients, and thereby convince ourselves that it does nothing.
But while the unique situation of prion disease presents these enormous challenges, we also have an enormous advantage: the biology of prion disease is really clear. We know, from decades of research and from all available lines of evidence, that the prion protein (PrP) is at the heart of this disease. This points us both toward what a drug must do — lower PrP — and what we must do to determine whether our drug did its job: measure PrP.
Beginning in 2016, Eric Lander mentored us through every step of formulating our thought process and figuring out what clincial trial design it necessitated. He guided us through putting the arguments into a persuasive writing piece — an Ames brief, as he called it, or a plaidoyer, as a peer reviewer would later call it. We eventually submitted the document as a white paper to FDA in a request for a Critical Path Innovation Meeting (CPIM). CPIM is a mechanism FDA established to allow groups, including researchers or patient groups, to interface with regulatory scientists at very early preclinical drug development stages and get informal, non-binding advice on how best to work towards a drug.
FDA Critical Path Innovation Meeting on genetic prion disease in Silver Spring, MD, November 14, 2017. From left to right: me, Eric Lander, Margaret Orseth, Sonia Vallabh, and Steven Arnold.
FDA generously granted us a CPIM on genetic prion disease, and when it convened in November 2017, we went in nervous as could be. We knew this might well be the most important meeting of our lives. But we needn’t have worried: the 25 FDA scientists we met with that day were wonderful. They were incredibly well-prepared, having read every word of material we’d sent, and they asked all the right questions. CPIMs are non-regulatory and non-binding, so there were of course no promises or commitments to a particular drug development path. But at a high level, what we heard that day was enthusiasm for our goal of preventing prion disease, and agreement in principle that our strategy of lowering PrP in people at high genetic risk, and measuring PrP in spinal fluid to show that we did it, made sense. FDA offered lots of informal advice on what we would need to do next to further validate our therapeutic hypothesis, to further credential our biomarker, and to prepare for trials. They offered to partner with us going forward, and they have continued to generously make time to advise us and work with us to make prevention a reality in prion disease.
Now, at long last, the arguments we presented to FDA that day are published in the form of a scientific article that you can read [Vallabh 2020]. I won’t repeat every detail of the logical arc here, but here is a quick summary of the five key points we argue:
- Prion disease is well-understood. We know, beyond a shadow of doubt, that PrP is responsible.
- Lowering PrP is very likely to be effective at preventing, delaying, or treating prion disease.
- In pre-symptomatic people at high genetic risk, there is an opportunity to extend healthy life, but there exists a risk that a therapeutic agent that can do so might not work after symptom onset.
- Lowering PrP may now be realistically achievable in the human brain using antisense oligonucleotides.
- Measurement of CSF PrP as a biomarker in pre-symptomatic at-risk people should be considered as a surrogate endpoint for FDA’s Accelerated Approval program.
A bit of background here for those who don’t know: Accelerated Approval is an FDA program that allows new medicines to be approved quickly for serious diseases based on a surrogate endpoint, meaning, something other than the outcome you truly care about. The program was originally established during the HIV/AIDS crisis, and an early precedent was the approval of AIDS drugs based, not on increasing survival in AIDS patients, but lowering HIV viral load. Patients cannot directly sense what their HIV viral load is, but we know that HIV causes AIDS, and so it was reasoned — and later clinically validated — that lowering viral load predicts increased survival. Likewise, in prion disease, what we truly care about is disease onset — we want to keep the disease from developing. But, following pre-symptomatic people all the way to disease onset in a trial before drug approval is numerically infeasible [Minikel 2019]. So instead, we could measure PrP in spinal fluid. A patient can’t directly feel what their PrP level is, but from a molecular standpoint, if we can show we lowered PrP, that will be very likely to predict a delayed disease onset.
To be sure, Accelerated Approval will take more than just a demonstration of lowered PrP in spinal fluid. Any drug will still have to prove safe in human trials, a wealth of animal data will be required to help establish the therapeutic hypothesis, and people will have to be followed long term, after approval of the drug, to ultimately confirm that it does indeed delay disease onset. But Accelerated Approval, if achievable, would be a watershed for prion disease: an opportunity to intervene early, before anything goes wrong, and preserve full quality of life.
There are certainly many other therapeutic approaches and strategies in prion disease, and people can and should pursue many options in parallel. But the strategy laid out in our article today is the one that we currently believe is most likely to succeed, to yield an effective drug in our lifetimes, and it has shaped all aspects of our research agenda for the past few years. To us, this preventive strategy represents an opportunity to make good on the years of advance notice — the letter from the future — that Sonia’s genetic test report gave us, way back in 2011. Devastating as that piece of news was for us, we now recognize that embedded in it were two shining pieces of good news: we can identify people at risk for this disease, and we know the molecular cause. Putting those two pieces together, we are optimistic that we have a path forward to prevent her disease.