Background: In DISCOVER, F/TAF was statistically noninferior to F/TDF. However, there were numerically 53% fewer HIV infections in the F/TAF arm vs. F/TDF . We explored HIV risk, STIs, adherence, and pharmacokinetic (PK) data to evaluate this imbalance.
Methods: In all participants (N=5,387), we compared HIV risk from computer-aided self-interview, lab-assessed STIs, and adherence by pill count and self-report. In a randomized subset, we measured adherence by intracellular TFV-DP concentrations in peripheral blood mononuclear cells (PBMCs) (Week 4; n=324) and in dried blood spots (DBS) (every 12 weeks; n=309). We assessed the relationship between adherence (TFV-DP in DBS) and efficacy using exact conditional logistic regression in a nested case-control-study; every incident HIV case paired with 5 controls matched by treatment arm, diagnosis date, rectal STI, and geography (cases=22, controls=109). We estimated duration of protection using PK data from historic Phase 1 studies to simulate TFV-DP concentrations for TAF and TDF.
Results: There were no differences in HIV risk, STIs, or adherence by pill count or self-report (N=5,387). Week 4 PBMC TFV-DP levels were 6.3 fold higher with F/TAF vs. F/TDF; 98% of F/TAF vs 65% of F/TDF participants had TFV-DP above the protective threshold (p< 0.001). The median duration of protection after last dose at steady state was 60% longer for F/TAF vs. F/TDF. Low DBS TFV-DP levels (adherence < 2 doses/week) were associated with increased risk of HIV for F/TAF: odds ratio [OR] 29.4 and F/TDF: OR 13.2 (p< 0.001 for both) with similar results from sensitivity analyses excluding suspected baseline infections.
Conclusions: Low TFV-DP concentrations were associated with an increased risk of HIV acquisition.There was no difference in HIV risk or adherence between arms, but there were significant differences in TFV-DP levels. TAF has advantageous PK parameters for HIV prevention compared to TDF including a more rapid, higher, and longer sustained duration of TFV-DP levels above the protective threshold in PBMCs, which may explain the lower number of HIV infections in the F/TAF vs. F/TDF in DISCOVER.

Drug (Active Moiety)TAF 25 mg (TFV-DP)TDF 300 mg (TFV-DP)
Median (IQR) PBMC Ctau fmol/106 cells (Ctau [trough] is the concentration at 24 hrs post dose)404 (226, 711)61 (34, 105)
Protective Threshold in PBMCs (EC90) fmol/106cells40 (Anderson, 2012)40 (Anderson, 2012)
Time to Protective Threshold1-2 hrs post single dose (historic Phase 1 data from GS-US-380-4017, Schwartz, 2018)3-4 days (Anderson, 2012)
Median Duration of Protection after Last Daily Dose (after steady state achieved)16 days10 days
[Table 1: Key HIV Prevention Clinical Pharmacology Parameters: TFV-DP with F/TAF and F/TDF]