
We asked Neil Parkin from Data First Consulting, Inc and Veronica Miller from UC Berkeley and the Forum for Collaborative Research a few questions on the study and research behind the abstract “HIV recent infection test-based incidence as a counter-factual for new PrEP trials” for IAS2021.
“The good news is that we have very good PrEP products; the bad news is that this makes it difficult to demonstrate efficacy of new products. Efficacy/safety requires a comparison to something – ideally that would be a placebo. If placebo [is] not possible (in this case not ethical) we rely on other comparators, such as the standard of care in a superiority trial; or a standard of care in a non-inferiority trial. The very high (>90%) efficacy of approved PrEP makes demonstration of superiority difficult because we would be dealing [with] small differences requiring large/larger sample sizes. Non-inferiority has [its] own limitations, including that we lack the appropriate “original” placebo data for women; furthermore, [non-inferiority] trial designs rely on a whole series of assumptions (including the constancy assumption) which cannot always be met.”
“The most obvious solutions are superiority and/or non-inferiority clinical trial designs as mentioned above, but these are difficult to implement and will become more and more difficult as new products reach the market. A key issue is that if both arms ([standard of care] and the new product) show high efficacy, we may not be able to interpret the finding convincingly. For example, it could be argued that the people enrolled in the trial were not sufficiently at risk for HIV acquisition. Thus, we need to find a way to demonstrate that HIV transmission was going on in the communities in which PrEP trials are being conducted, but without including a placebo as an internal control (randomization arm within the trial). Thus, the counterfactual placebo estimate is the closest we can come to a reasonable comparator arm. We must remember that a traditional placebo arm is but a counterfactual estimate (of what happens when the intervention is not given/experienced) – it happens to be a good one because through randomization we can eliminate some of the biases that we are concerned about. But given the impossibility of doing an internally randomized placebo, the counterfactual estimate using data from cross-sectional screenings such as an HIV recency test is the best available approach.”
“We have the most experience [with] the LAg [limiting antigen avidity assay] recency test and assay calibration in settings with subtypes A, B and C [of HIV]. We need to gain experience for regions with other viral subtypes. As we proceed to learn more about how a recency assay performs in the setting of PrEP trials over the next years, we need to monitor the experience with respect to regional epidemiologic and viral context; most importantly, we need to open to knowledge exchange and collaboration amongst all partners to ensure the most efficient, ethical, and responsible use of data.”
The Forum for Collaborative Research will also be sponsoring a satellite IAS 2021 session “Innovative Clinical Trial Designs to Accelerate Increase in PrEP Choices” on the issues and innovations discussed above. The session is chaired by Veronica Miller and takes place 18th July 21:00-22:30 (CEST Berlin time). Don’t miss it!