As everyone knows, there have been a lot of attempts to repurpose existing therapies for the coronavirus pandemic. I’ve covered several of these along the way, but it’s time for some updates. The work that’s been going on not only adds to our knowledge about treatment for infected patients, but it should – ideally – also show what clinical research is like and why we have to do all these trials. Getting solid answers is a lot harder than it appears to be.
Antibodies Against IL-6 Activity
Case in point: the IL-6 story. Interleukin-6 is well-characterized as a pro-inflammatory cytokine signal, especially in the acute phase of the response, and that made this pathway a natural target for study after people realized that the “cytokine storm” immune reponse was getting coronavirus patients into trouble. Back in April, we had some numbers on the Roche/Genentech monoclonal antibody against the IL-6 receptor (Actemra, tocilizumab) that suggested that it might be beneficial – in fact, the French team conducting the study said that deaths were “substantially reduced” in the treatment group. But at the same time, another antibody against IL-6R, Sanofi and Regeneron’s Kevzara (sarilumab) looked like it wasn’t working. That was puzzling at the time, and investigations on both of these continued.
Well, earlier this month sarilumab failed to reach its endpoints in a trial adding it to standard-of-care for hospitalized coronavirus patients, and that pretty well took care of it idea of using Kevzara as a therapy. Meanwhile, tocilizumab had failed to help patients a bit earlier in the disease progression in a study in Italy, which was disappointing after the French results, but there was still some hope that it could help the most severely affected patients. But just today, Roche announced that no, Actemra had no effect on clinical status or mortality. So much for that.
This is instructive on several levels, especially for folks who have been looking in on drug development from outside during the pandemic. The IL-6 hypothesis was a perfectly reasonable one, a solid idea from what we know about inflammation and about what was going wrong with patients who were being put on ventilation. But it’s wrong. A lot of perfectly reasonable medical hypotheses are wrong; this happens all the time and it’s why we have to run controlled trials rather than just running with what looks like it should work. And to go further, that first trial of tocilizumab looked like it worked, didn’t it? Deaths were “substantially reduced”, and how can you argue with that, right? But it was a small trial and it was only one trial, at that. (Update: thanks to AndyBiotech on Twitter, I find that it’s even worse than that. Turns out that the trial’s safety data monitoring committee resigned in May over strong disagreements about how the readout of the trial was characterized. So there’s that!) Real treatments work again when you test them again, and they continue to work when you test them under more controlled and more statistically powerful conditions. Other results evaporate when you do that, and that’s what happened here. Those hopeful results early on were almost certainly an illusion, and this happens all the time. It’s why we run more than one trial, and why we make the later ones larger and more powerful.
You may be able to think of other high-profile therapeutic ideas that have had a similar course during the pandemic: promising early results in small trials followed by an inability to replicate them on a more robust scale. When such things don’t follow up, it’s not a conspiracy and it’s not malevolence and it’s not incompetence. Not usually. Because, and I cannot emphasize this enough, such things happen all the time. If you follow drug discovery and development, you’re used to it, because you’ve seen it happen over and over in all different disease areas. If you’re just tuning in, though, it can be hard to come to terms with.
There are plenty of other candidates out there. One that’s been getting attention is apilimod, a small molecule that’s been kicking around for some years now. It was originally investigated for its ability to lower IL-12 and IL-23 levels, a cytokine activity profile that looked like it could be useful in arthritis, psoriasis, and autoimmune diseases in general. If from that numbering you take away the idea that there are an awful lot of interleukins and that these cytokine signaling pathways must be rather a tangle, you are most extremely correct. The number of interactions in such systems (and with finer and finer distinctions of individual immune-responsive cell populations, too) is absolutely eye-watering.
But like many another small molecule, apilimod’s activity profile is a longer story than it first appears. It was identified in 2007 from a cell-based screen looking for modulators of interleukin activity. For some years, the only thing known about its mechanism was that it seemed to inhibit the activity of a transcription factor called c-Rel, preventing it from getting into the nucleus, and c-Rel was known to be essential for production of IL-12 and IL-23. But in 2017, a cell-based antiproliferative screen identified it further as an inhibitor of the enzyme phosphatidylinositol-3-phosphate 5-kinase (known as PIKfyve), and had just been found out (via other small-molecule PIKfyve inhibitors) that PIKfyve is essential for c-Rel activity, clearing up the mechanism a bit and giving apilimod a solid target and pathway to explain its actions. Apilimod has been suggested as an anticancer compound on the basis of such results.
So as for the pandemic, we’re back to inhibiting cytokine storms, right, this time by lowering IL-12 and IL-23 with this PIKfyve inhibitor compound, right? Maybe not! A recent paper in Nature (from a very large multicenter team) details another large-scale drug repurposing screen, this one done by looking for compounds that actually inhibit wild-type SARS-Cov2 virus infection of Vero-6 cells, a commonly used cell model from monkeys. Now, this is an in vitro screen, of course, but it’s a hard-core in vitro screen, because the team didn’t use a model for the virus itself (a pseudovirus or something of that sort), but rather went right to the real virus, which takes some serious screening facilities because of the serious containment needed. They followed up their best hits in human-derived cells and even in human lung tissue, so the results are quite solid.
And apilimod showed up as the best of the bunch, the only one to make it all the way through to good activity in the lung tissue assay. Not only did it inhibit viral replication strongly across all the whole screening cascade, but it also showed strong activity against the Ebola, Lassa, and Marburg viruses. (Told you that this was a hellacious containment facility, right? What a lineup.) That’s very interesting indeed, suggesting that this almost certainly has to be a host target involved in general viral infection mechanisms and not some specific protease or nonstructural protein of the pathogens themselves. Those viruses are all over the map phylogenetically; they really shouldn’t have much in common past their common use of an RNA genetic payload.
Indeed, PIKfyve has been shown to be an important regulator of endosome activity, particularly early endosome formation. That means that it could be right up there at the beginning stage of viral infection, because endosomes are how many viruses actually deliver their genetic material into a host cell. The data in this new Nature paper strongly suggest the apilimod be tested as a preventative of coronavirus infection and in patients who already have the disease.
But there’s more to the story – I haven’t seen this highlighted, but PIKfyve has also been shown (by the van den Bogaart and Bertozzi groups) to be a key part of the process whereby antigens are presented to the surfaces of dendritic cells (for T cells to then come along and recognize them). That is an absolutely crucial part of the immune response to new antigens, and that paper (which came out in January of last year) showed that apilimod and other PIKfive inhibitors do indeed impair immune function and T-cell activation. That’s just what you don’t want in a viral infection!
So clinical trials of apilimod are going to be rather interesting. Will it protect from viral entry, but make other (nonviral) infections more likely? The balancing act between the viral entry inhibition and immune system impairment is something that can only be evaluated for sure in human patients. It’s one of those little biochemistry jokes – there are several of these – that the relentless evolutionary repurposing of enzymes and mechanisms should come along and complicate the attempt to repurpose a drug when it’s needed the most.