Biogen and Ionis announced Monday that they are ceasing development of the antisense oligonucleotide (ASO) BIIB078 for C9orf72 ALS. The press release does not contain a lot of detail, so this blog post is just a quick stub to unpack what little we know about this decision.

The most common genetic form of ALS is caused by a repeat expansion in the intron of the gene C9orf72 which leads to translation of a toxic dipeptide. BIIB078 was a drug candidate designed by Ionis and developed clinically by Biogen, aiming to lower the amount of C9orf72 RNA. It had two clinical trials posted to clinicaltrials.gov. One is a Phase 1/2 that began in August 2018 and completed in November 2021. It enrolled 106 participants who were symptomatic with ALS and confirmed C9orf72 mutations, who, depending on study arm, received a total of 5 to 8 doses of drug over 8.5 months. The primary endpoint was safety, with a battery of secondary endpoints concerning the drug’s pharmacokinetics as well as several different measures of ALS severity. The other trial is an open-label extension (OLE) follow-on open to those same participants, which began in Feb 2020 and was originally slated to run through July 2023.

After a trial finishes, it takes months to perform all the biofluid analyses, then analyze all the data and decide on a course of action. We can infer that while the official end of the Phase 1/2 was in November 2021, it must not have been until just now that Biogen had all the data in hand to make its decision. And that decision was to stop development of the drug. The press release states that while there were no safety issues per se — adverse events observed were not severe and were not enriched in drug arms as opposed to placebo arms of the trial — the participants in the highest dose group (90 mg) “trended toward a greater decline than those in the placebo group across secondary endpoints”. The two key words here are “trended” and “secondary”. The trial was designed around safety as its primary endpoint. It wasn’t designed around these ALS severity measures, and presumably was not well-powered to detect changes in them. Accordingly, the use of the term “trended” suggests that the mean values of these measures were worse in the 90 mg group than the placebo group but that the difference was not statistically significant, or at least not for all such measures. Thus, the trial results don’t prove that BIIB078 made people worse, but they look bad.

When I blogged about tofersen last year, I projected that we wouldn’t get any real new information about ASO trials for brain diseases until 2024, based on the completion dates for posted Phase 3 trials. It hadn’t occurred to me that an ASO candidate might be dropped after just Phase 1. To my knowledge this hadn’t happened for any previous ASO candidates in brain diseases. The decision to end development so early on, without a Phase 3 trial really powered to assess ALS endpoints, might mean one of two things. One possibility is that the difference between 90 mg and placebo looked really bad, bad enough to talk you out of wanting to touch the drug ever again. The use of the word “trended” argues against this, but you never know. Hopefully the full data package will be presented publicly soon. The other possibility is that Biogen’s threshold for abandoning ASO programs has been lowered as a result of the negative outcomes that Roche observed with tominersen and the middling outcome Biogen observed for tofersen. In other words, maybe they have adjusted their prior expectation that an ASO might prove harmful or at least non-beneficial to patients, such that now it doesn’t take as strong of evidence to talk them out of developing an ASO further.

The termination of the BIIB078 is certainly a devastating setback for the C9orf72 community. There are two other C9orf72 ASOs in development. One was produced by UMass Medical School and so far tested in rodents, non-human primates, and just one human patient [Tran & Moazami 2022]. That ASO potently lowered the amount of toxic dipeptide in the patient’s spinal fluid, but with just one patient, it is too early to make any inferences about its effect on disease severity. The other, developed by Wave Life Sciences, had some preclinical work in mice published [Liu 2021] and entered Phase 1 last summer (NCT04931862), with no human results reported yet.

How does the failure of BIIB078 adjust our own expectations in the prion disease community? It is certainly a reminder that drugs fail in clinical development all the time, and we should view a future clinical trial of a prion disease ASO as an experiment, and not as a therapy guaranteed to work. As for whether this affects our estimate of probability of success, it’s still hard to say. Let’s consider gapmer ASOs (ASOs designed to lower a target RNA) in the CNS. By my accounting it appears that 4 such drug candidates have gotten to Phase 1 trial readouts — tominersen (HTT), tofersen (SOD1), BIIB080 (MAPT), and now BIIB078 (C9orf72). (A number of others are still in Phase 1, as Holly Kordasiewicz outlined at last year’s Lancet Summit.) Of the drug candidates that completed Phase 1, 3/4 looked good enough to continue onward. Of those, only 2 have had Phase 3 readouts, of which tominersen made patients worse (though Roche thinks a low dose in young, mildly symptomatic patients could still work), while tofersen didn’t make patients obviously better or worse, though there were a few encouraging signs of “maybe better”. So if your fear is that all gapmer ASOs are poorly tolerated in the CNS and just make patients worse, I’d say the data so far don’t appear to support such an extreme conclusion. If you believe that each ASO is different and the question is whether any given ASO is likely to have a favorable risk-benefit balance, then the new failure shifts the calculus a bit, but it’s still small numbers on which to try to estimate a probability.

Meanwhile, it’s not as though the different ASOs are the only variables across these trials: each also tests a different therapeutic hypothesis in a different disease. For tominersen, the negative outcome led some observers to wonder whether Huntington’s disease is not purely a gain-of-function, and whether non-allele-specific lowering of HTT has a therapeutic window at all. For BIIB078, Frank Bennett at Ionis seems to be entertaining a similar possibility, that the trial’s failure could actually be related to therapeutic hypothesis and not to the specific compound tested in the clinic: the press release quotes him as saying “We designed BIIB078 to test the prevailing hypothesis that the mechanisms of disease for C9orf72-associated ALS were caused by toxicity associated with the repeat containing RNA and corresponding dipeptides. Unfortunately, this Phase 1 study did not support the hypothesis, suggesting that the disease mechanism is much more complex.”

Finally, for every ASO that’s been in development, one can ask the question of whether it was tested in the right population, at the right disease stage. Both for tominersen and tofersen there was speculation that maybe the window for efficacy was actually earlier in disease or perhaps before symptom onset. That could be true for BIIB078 as well. In a quick search I didn’t find any animal studies published on BIIB078 or any related tool compounds, so I don’t know how strong the toxicology data were in animals leading to first-in-human dose, nor do I know how good the evidence is that BIIB078 should have been effective, or at what disease stage it was beneficial in animals. In prion disease, we have evidence from animal models that earlier intervention is much better, and so we’ve advocated for a preventive trial in pre-symptomatic people from day one. But we don’t have any human data on earlier intervention with the ASOs that have been tested clinically yet. The ATLAS trial (NCT04856982) will test whether tofersen is effective in prodromal SOD1 ALS (people with biomarkers of pathology but no symptoms), but we won’t learn the results until 2026.

For us working on prion disease, I think the failure of BIIB078 is not a reason to lose hope in our program. The ASO we’re looking forward to for prion disease is a different compound, for a different indication, and we hope it will be tested at a different disease stage. But, this week’s news is a reason to remember that any drug development program is an experiment. We do clinical trials to make sure that we only give people drugs that benefit them and don’t harm them, and that requires incredible patience. As I argued when the disappointing news about tominersen first dropped, developing new medicines requires a “try, try again” approach, and our efforts to develop a PRNP ASO have already moved us forward in profound ways, such that, even if one drug candidate fails, we won’t be back to square one.