999 - Using the F/TDF Adherence-Efficacy Relationship to Calculate Background HIV incidence: Results from the DISCOVER trial

Background: RRandomized trials of new PrEP agents compare to oral emtricitabine+tenofovir disoproxil fumarate (F/TDF) and do not have a placebo arm. We used the well-characterized adherence-efficacy relationship for F/TDF from iPrEX OLE, to back-calculate the (non-PrEP) background HIV incidence (bHIV) in the F/TDF arm of DISCOVER and estimate comparative efficacy (to bHIV).

Methods: TDISCOVER is an ongoing randomized active-controlled trial in 5,387 men who have sex with men and transgender women that demonstrated non-inferiority of F+tenofovir alafenamide (F/TAF) to F/TDF (IRR 0.47 (95% CI 0.19, 1.15).  TFV-DP levels in DBS were assessed for all diagnosed with HIV and in a randomized subset of 10%. We used a Bayesian model with a prior distribution, derived from iPrEx OLE, relating TFV-DP levels to HIV prevention efficacy: eg TFV-DP levels of < 350 (low), 350 to < 700 (medium) and ≥700 (high) fmol/punch were assumed to provide 0%, 86% and 98% HIV protection, respectively. This prior, combined with F/TDF seroconversion rate and TFV-DP levels, yields Bayesian inferences on the bHIV. In R, STAN was used to sample 10,000 realizations from the posterior distribution.

Results: There were 6 vs. 11 post-baseline HIV infections (0.14 v. 0.25 per 100 person-years [PY]) on F/TAF and F/TDF. Of the 11 on F/TDF, 10 had low, 0 had medium, and 1 had high TFV-DP levels; among HIV-negative controls, 5% of the person-time had low, 9% had medium, and 86% had high TFV-DP levels. A non-informative prior distribution for bHIV, combined with the prior for TFV-DP level-efficacy relationship, yielded a posterior bHIV incidence [0.80 Bayesian credible interval (CrI)] of 3.4/100 [1.9, 6.0/100] PY; which suggests a median F/TAF efficacy [0.95 CrI] of 96% [88%,99%] and 93% [87%,96%] for F/TDF compared to bHIV. If we chose a conservative prior distribution for bHIV of 1.0/100 PY, the model yields a median posterior bHIV [0.80 CrI] of 2.8/100 [1.7, 4.7/100] PY; which suggests a median efficacy [0.95 Cr] of 95% [86%, 99%] for F/TAF and 92% [86%, 67%] for F/TDF compared to bHIV with corresponding number of HIV infections averted of 117 and 114, respectively (Figure).Figure.


Conclusion: The F/TDF adherence-efficacy relationship can be used to back-calculate bHIV incidence in MSM/TW PrEP trials and assess the efficacy of new PrEP agents compared to bHIV.