Finding the falsification threshold of the toxic proteinopathy hypothesis in neurodegeneration

A biomedical hypothesis is a theoretical assumption amenable to being tested in a randomized clinical trial. The main hypotheses in neurodegenerative disorders are based on the concept that proteins accumulate in an aggregated fashion and trigger toxicity. The toxic proteinopathy hypothesis posits that neurodegeneration is caused by toxicity of aggregated amyloid in Alzheimer's disease (toxic amyloid hypothesis), aggregated α-synuclein in Parkinson's disease (toxic synuclein hypothesis), and aggregated tau in progressive supranuclear palsy (toxic tau hypothesis). To date, we have accumulated 40 negative anti-amyloid randomized clinical, 2 anti-synuclein trials, and 4 anti-tau trials. These results have not prompted a major reconsideration of the toxic proteinopathy hypothesis of causality. Imperfections in trial design and execution (incorrect dosage, insensitive endpoints, too-advanced population) but not in the underlying hypotheses have prevailed as explaining the failures. We review here the evidence suggesting that the threshold of hypothesis falsifiability may be too high and advocate in favor of a minimal set of rules that facilitate the interpretation of negative clinical trials as falsifying the driving hypotheses, in particular if the desirable change in surrogate endpoints has been achieved. We propose four steps to refute a hypothesis in future-negative surrogate-backed trials and argue that for the actual rejection to take place, refutation must be accompanied by the proposal of an alternative hypothesis. The absence of alternative hypotheses may be the single greatest reason why there remains hesitancy in rejecting the toxic proteinopathy hypothesis: in the absence of alternatives, we have no clear guidance as to where to redirect or focus.