Back in 2013, NIR used a picture of an ill-fated South Pole expedition in a presentation at CNS Summit. The caption read: “Lilly announced it will run another major Phase III program for solanezumab. Patients must show scan or CSF evidence of amyloid pathology, but have only mild cognitive symptoms. Early intervention? Not really.”
Three years later, the top-line results are in, and indeed the EXPEDITION3 trial failed. Solanezumab is being terminated by Lilly, after a total expenditure that surely must have easily exceeded a billion dollars, with room to spare.
Over those three years, NIR had concluded that Lilly’s task was to power the trial large enough to squeeze a mediocre signal past the threshold of statistical significance. As we said in the NP September/October review of Alzheimer’s: “A key change was the inclusion requirement that amyloid pathology be shown via imaging or CSF assay, thus enhancing the homogeneity of the population, but also meaning that the disease have reached the stage of beta-amyloid plaque. This is a critical conundrum: Patients must be advanced enough to have significant plaque deposits that show on PET-scan, but not so advanced that they have anything more than mild cognitive impairment. Given the very modest treatment effect Lilly gleaned from data-mining the substantially-scaled first two trials, we suspect that even this ‘enriched’ population might be sufficiently advanced in disease progression that nothing more than a minimal treatment effect can be hoped for…..Since EXPEDITION3 powered up and enrolled 2100 patients, it could end up with statistically significant results, but it is very questionable whether a case for clinical significance will be convincingly made based on a signal of very modest impact–less than is seen with cholinesterase inhibitors.”
That is essentially how it turned out: There was a trend towards superiority for the drug group, with a p value of .095. Given that p=.05 is the arbitrary, binary benchmark for statistical ‘truth’, this does save us from a painful debate over whether a clinically meaningless intervention should be approved on the basis of achieving statistical significance.
Lilly will present more complete results on December 8. There are many things that we do not yet know, these are just are a few:
- Was the failure due to the aforementioned risk that the treatment population, already showing accumulated plaque, was too advanced in their disease to benefit?
- Did solanezumab achieve sufficient brain penetrance to have a shot at neutralizing enough beta-amyloid to change the course of the disease? And is it reasonable to deliver an antibody against a large pool of beta-amyloid monomers, rather than aggregated forms?
- Is beta-amyloid central enough to the core pathology of Alzheimer’s to be a viable target for intervention? This treatment failure does not demolish the ‘amyloid hypothesis’, but it does not lend any support to its historical role as the gold-standard for AD targets.
- Will reducing the production of beta-amyloid via BACE inhibitors or the inhibition of pyroglutamic beta-amyloid be more efficient and effective than trying to clean up the amyloid mess after the fact? Merck‘s BACEi data next year will be the first major test of this hypothesis.
- There has been growing support for the hypothesis that beta-amyloid sets the stage for tau to serve as the toxic blow to the brain. The tau hypothesis has not had a real test in the clinic as of yet, with all due respect to the failed TauRx Phase III trials.
Since Alzheimer’s is unfairly and inaccurately depicted as if it is a proxy for the ability of clinical neuroscience to deliver tangible benefit to patients, we had hoped that our trepidation regarding solanezumab would prove to have been overblown. Now, we will wait to see Lilly’s more detailed data presentation, and whether this further shakes their already wavering dedication to neuroscience. They have other AB antibody programs in the clinic, a tau antibody, and a BACEi being developed in partnership with AstraZeneca–we hope that solanezumab’s failure does not wreak collateral damage within Lilly’s remaining neuroscience R&D portfolio and resourcing.