Statistical methods for clinical trials of novel PrEP agents

Investigator: David Glidden, PhD
Sponsor: NIH National Institute of Allergy and Infectious Disease

Location(s): United States


This project will develop statistical approaches to the analysis of clinical trials of novel PrEP agents. It will enrich intent to treat and as treated analysis of the novel agent by reconstructing the hypothetical placebo HIV rate. It will also build tools to understand who will prefer a novel PrEP agent over oral truvada®.

HIV pre-exposure prophylaxis (PrEP) with co-formulated emtricitabine tenofovir (TDF/FTC) in HIV negative persons is safe and effective for preventing HIV infection. Despite notable successes, PrEP has important limitations as a broad prevention tool. Some people who would benefit from PrEP refuse it or are poorly adherent. Surveys indicate that many of those at risk for HIV believe that daily oral TDF/FTC to be difficult to adhere to or unappealing and would be prefer a non-oral alternative. This has also been shown to be true in contraception. To maximize the benefits of biomedical prevention, we need alternatives to oral TDF/FTC. There is a robust pipeline of novel PrEP methods (NPM) in development. There are antiretrovirals formulated as injectables, infusions, implantables and microbicides. Some of these have entered phase III randomized trials and others could enter trials over the next 2-4 years. Since PrEP has shown effectiveness, the trials must make some provision for it ethically. For injections and implants, TDF/FTC will likely be a comparator arm and the degree of TDF/FTC adherence in these future studies is very uncertain – greatly complicating the comparison of the two arms. Standard methods lead to long expensive studies which will slow PrEP innovation. We believe can significantly increase power and/or lower the cost of future active-controlled randomized trials. By innovatively using post-baseline TDF/FTC pharmacology, which is well-studied in previous trials, to reconstruct a counterfactual placebo arm. This permits comparisons of the novel PrEP method to putative placebo in both intent to treat (ITT) and as-treated (AT) analyses – gaining substantial insight and statistical efficiency. We will also develop models and methods to understand who will sustain engagement with a NPM but not oral PrEP. By leveraging the well-studied TDF/FTC pharmacology and modern causal inference methods, we will to develop analysis methods and non-inferiority frameworks that allow next generation PrEP trials to be smaller, quicker and more informative.