By Anna Rose Welch, Director, Cell & Gene Collaborative
Subscribe to my blog ARW on CGT here!
There are few instances in which an FDA advisory committee meeting can be compared to a bingeable Netflix series. But having recently re-watched the Sarepta adcom meeting for its AAV gene therapy in Duchenne Muscular Dystrophy, I’m confident in declaring this adcom one of the finest and most riveting regulatory dramas I’ve ever streamed.
We already know how the advisory committee voted: 8 voted in support of an accelerated approval of Sarepta’s gene therapy, while 6 voted to hold off until the conclusion of Part 1 of the company’s ongoing Phase 3 study. (FYI, part 1 of the study will conclude at the end of September, and data is expected by the end of this year.) We also know that the company faced an additional bump in the road in the form of a one-month-long delay on the FDA’s action date. But in true Deus Ex Machina fashion, Peter Marks swooped in and declared that the therapy’s quality and safety and — despite some doubts on the review team — its clinical efficacy data were supportive of an accelerated approval.
I daresay we could’ve perhaps foreseen this conclusion based on Marks’ beautiful closing remarks about the work we and the FDA are doing to weigh the risks vs. benefits of our advanced therapies.
“What I hear coming out of all of this is that we are still in an area where there is a lot of uncertainty, but that the small majority [in the advisory committee] believe there is enough compelling information to make them feel comfortable moving forward with an accelerated approval [provided that the confirmatory trial is completed]. Now we will take this back and do what we do every day at the FDA: We will manage through the uncertainty.”
There was a lot that caught my attention throughout this epic 9-hour meeting, including the fact that the product underwent a significant manufacturing change late in clinical development which altered the product’s purity. Not to worry, I discussed how comparability was approached in this adcom much more closely in part two of this article. But I felt I’d be remiss if I didn’t also provide you with two additional “plot points” that nicely reflect how the often-theoretical regulatory and CMC discussions/debates around protein expression and potency are being approached in the “real” CGT world. Though we cling to — and simultaneously despair over — the fact that regulatory experiences are product-specific, the hurdles/debates Sarepta faced remind us of the scientific and clinical complexities our future CGT products will no doubt be faced with during their own BLA reviews.
- Endogenous Vs. “Engineered”: The CGT’s Own “Nature vs. Nurture” Debate
It’s a well-known fact that AAV can only package DNA sequences that are smaller than 4,700 DNA base pairs (i.e., 4.7 kb). Thanks to its two million-plus DNA base pairs, dystrophin — a critical protein for healthy muscle function that is altered/missing in DMD patients — wins the trophy for being the largest human gene. As such, it’s impossible to “pack” the dystrophin-encoding gene sequence within the too-small “suitcase” of an AAV vector.
In response to these packaging constraints, Sarepta did exactly what brilliant scientists do: They find creative ways to overcome the limits of biology to safely treat patients. Instead of encoding the full dystrophin protein, Sarepta’s gene therapy encodes a shortened form of dystrophin — a.k.a. micro-dystrophin — which contains what the company identified as the key functional domains within the full-sized dystrophin protein itself. As Sarepta’s (fascinating) analysis of the dystrophin protein revealed, not all portions/sections of the long protein are critical to have a beneficial clinical effect.
Now, it’s important to note that there are some instances in which shortened forms of dystrophin occur in nature — particularly in patients with Becker Muscular Dystrophy ([BMD]; a less debilitating form of DMD). Much of Sarepta’s natural history research delved into the function of micro-dystrophin proteins in BMD patients. This natural history was integral in informing and justifying Sarepta’s own micro-dystrophin design decisions.
However, what I found most noteworthy throughout the adcom were the FDA reviewers’ concerns over whether Sarepta’s micro-dystrophin protein was highly similar in terms of expression profile and function to naturally occurring micro-dystrophin (and full-length dystrophin) proteins. Overall, the FDA was laser focused on the fact that Sarepta’s micro-dystrophin were “engineered” and “don’t occur naturally.” This discomfort was made particularly clear in the FDA’s ongoing reference to the micro-dystrophin in question as “Sarepta’s micro-dystrophin.”
In particular, the FDA homed in on some of the unique structural differences between Sarepta’s micro-dystrophin and the truncated proteins produced by BMD patients. The FDA’s assessment of the nonclinical data also indicated that Sarepta’s micro-dystrophin has a “very different” protein expression level in mice compared to the full-length dystrophin protein expressed by the endogenous DMD gene. And, in the FDA’s eyes, said nonclinical data — accompanied by the clinical data — did not suggest that the product’s protein expression alone could predict a beneficial functional/clinical response in patients.
Now, if you watched the adcom, you’ll know that this hesitance raised some committee members’ eyebrows, considering that the literature and the experience of BMD patients support the notion that “a little bit of truncated protein can go a long way.” As one particularly salient adcom member keenly pointed out, the FDA and sponsor had intriguingly different interpretations of Sarepta’s nonclinical data. But the FDA remained relatively steadfast in its hesitance, reiterating that, “Shortened dystrophin are very different from one another. Since this is a novel engineered protein that isn’t in nature, we don’t have the kind of information outside of clinical trials to be able to assess what the effect of various levels of Sarepta’s micro-dystrophin may be.”
Admittedly, there is so much more biological and clinical nuance to be offered here. But I felt the aforementioned examples clearly demonstrate the “nature vs. nurture” debate facing us as an industry. We may work with elements that appear in nature — whether they be patients’ cells (or cells in general), DNA, mRNA, and viral vectors. Thanks to our “nurturing” or engineering of these materials into therapeutic products, they may go on to encode a naturally occurring protein in vivo. But the FDA’s hesitance over Sarepta’s micro-dystrophin suggests we may face an even more complicated justification beyond how our design decisions and processes impact our product’s structure/function. We may also have to better articulate that our “nurtured” products’ biological cascades are comparable to those that occur in nature — particularly in indications in which protein function remains mysterious.
This brings me to my next important point: This adcom clearly demonstrated we’re still struggling to wrap our brains around…
- The (Tenuous) Relationship Between “Expression” & “Function”
Late last year, I shared a few articles featuring my takeaways from an ARM/ASGCT workshop on potency assays (here and here). This workshop comprised several hours of discussion between FDA and industry about the importance of and difficulties we face in analytically measuring the individual steps of a product’s biological cascade (i.e., infectivity, expression, & function/biological effect). As was made clear throughout that workshop, we have many questions about how to best show analytically that our therapies work. These questions include but are hardly limited to whether we always must demonstrate our product’s functional activity (as FDA seems partial to today), or if it’s possible to link a product’s MOA/potency solely to the expression of a protein.
Of course, what is or is not acceptable will be product-specific and will depend on the totality of the evidence (both analytical and clinical). But the Sarepta adcom gave us a fantastic look into how the agency and industry are continuing to grapple with (and butting heads over) how we demonstrate, measure, and predict the clinical effect of our products — particularly when looking at protein expression as a mechanism of action.
Those of us on the CMC side of the industry might have expected that such a protein expression-heavy discussion would henceforth go deep into the weeds of potency assays. But this was not exactly the case. Though there were brief references to the assays used to quantitatively measure Sarepta’s micro-dystrophin expression (i.e., Western blot and immunofluorescence staining), the true star of the discussion was the company’s decision to use expression as a surrogate endpoint in their trials.
For the non-clinical gurus among us, I’d refer you to the FDA’s helpful resources on what a surrogate endpoint is, how it functions, and why it exists. But what you ultimately need to know right now is this: In choosing expression as a surrogate endpoint, Sarepta was making the argument that expression of the micro-dystrophin protein could be used to predict the product’s clinical effect/benefit (i.e., an improvement in a patient’s North Star Ambulatory Assessment score). And, to some adcom members and FDA reviewers, this correlation was tenuous at best.
As it turns out, the CMC space is not the only one suffering from “quantitative” woes; as this article nicely summarizes, one of the biggest sticking points for the FDA was the lack of “quantitative correlation” between Sarepta’s surrogate endpoint and clinical outcome. The ability to clearly correlate expression with a change/increase in the patient’s NSAA score was — in the FDA’s eyes — complicated by quite a few factors. These factors included: the limited amount of clinical data in certain age groups; high variability in micro-dystrophin expression between patients of different ages and in different arms of the study; the potential for bias due to Sarepta’s trial design; and an overall lack of statistical rigor.
I encourage you to check out the FDA’s materials and/or watch snippets of the adcom to get all the juicy clinical/statistical details; I (sadly) cannot succinctly summarize them here. But I wanted to single-out this expression-as-function debate because it unites the (often-siloed) goals of CMC and clinical development under the same umbrella.
In the ATMP space, we regularly proclaim the importance of CMC and clinical development working together. After all, we can’t supply patients with a high-quality product — or a drug product at all — without CMC. Likewise, we can’t arrive at a well characterized process and product without the insights provided from our patients’ clinical data. But despite our best efforts to remain a united front, we can easily get lost in our own CMC-centric discussions of potency, whether they be hypothetical arguments about why “expression (should) = clinical benefit” or why a potency assay matrix should/should not include certain assays.
It's not often we get the chance to see how our CMC challenges translate into (or are shared) in the bigger picture of our products’ clinical development. But this adcom, as mind-bendingly complex as it was, did just this for us — and that alone is noteworthy.
Check out part 2 in which I share a few manufacturing-related takeaways from the Sarepta adcom meeting!
Like what you’re reading? Sign up to continue following ARW’s writings on the CGT + mRNA manufacturing paradigm.