Paper
- Alzforum Recommends
Steward A, Biel D, Dewenter A, Roemer S, Wagner F, Dehsarvi A, Rathore S, Otero Svaldi D, Higgins I, Brendel M, Dichgans M, Shcherbinin S, Ewers M, Franzmeier N. ApoE4 and Connectivity-Mediated Spreading of Tau Pathology at Lower Amyloid Levels. JAMA Neurol. 2023 Dec 1;80(12):1295-1306. PubMed.
Recommends
Please login to recommend the paper.
Comments
Stanford University
While it is encouraging to see alternative hypotheses being generated from post hoc data analyses, I'm not certain the field has fully come to grips with the lack of correlation between amount of amyloid removal and clinical efficacy. This is certainly what one would expect of a surrogate biomarker deemed "likely to predict a clinical benefit" by the FDA. We almost never see, either in papers or presentations, scatterplots showing within trial correlations between amyloid plaque removal and clinical outcome across individuals. What is frequently shown, instead, is a correlation plot across trials suggesting that those trials that achieved a greater amount of amyloid removal tend to have better clinical outcomes. The only within trial data we could obtain (from the two Phase 3 aducanumab studies) show no significant correlation between amyloid removal and change in CDR-SB (Figure 1B in the paper cited below).
Furthermore, just as there is a correlation between amyloid removal and clinical response across trials, we have shown a correlation between percentage of subjects with ARIA-E and clinical response across trials (Figure 2 in the paper cited below). This raises the concerning possibility that functional unblinding due to ARIA is driving some or all of the clinical effect in these studies. Given the arguable clinical relevance of the outcome data reported, even some contribution from a functional unblinding/placebo effect would be worrisome. The utter lack of movement in the tau PET measures in the donanemab study (which had the largest effect on CDR-SB and had the largest reported tau PET sub-study) would be consistent with such a non-biologically-mediated effect on clinical outcome measures.
It is certainly possible that the correlation between ARIA and clinical outcomes across trials reflects some co-linearity, across trials, between amyloid removal and ARIA. The pharmaceutical companies could, easily, shed light on this topic in several different ways. Providing spaghetti plots of each subject's longitudinal course with time stamps indicating which subjects had ARIA and when would be incredibly informative. Is there a presumably non-biological, post-ARIA "bump" in CDR-SB that would support a functional unblinding/placebo effect? It would also be helpful, in each trial, to compare participants in the active treatment arm who did have ARIA with participants in the active treatment arm who did not have ARIA, matched for relevant variables like amount of amyloid removed (or speed of removal), age, sex, and APOE4 dose. If ARIA is not inflating clinical outcomes these two groups should have similar clinical responses. As discussed in our commentary, the sensitivity analyses designed to address the potential impact of ARIA on functional unblinding in the lecanemab and donanemab studies are hobbled by the greater risk of ARIA in APOE4 carriers.
The "speed of amyloid removal" hypothesis strikes me as entirely plausible but it has been generated, appropriately, from post hoc analyses. The field would advance more quickly if we had a better understanding of within-trial associations between amyloid removal, ARIA and outcomes.
References:
Digma LA, Winer JR, Greicius MD. Substantial Doubt Remains about the Efficacy of Anti-Amyloid Antibodies. arXiv:2310.15456 [q-bio.TO], November 19, 2023 Cornell University, Quantitative Biology
View all comments by Michael GreiciusMake a Comment
To make a comment you must login or register.