Renal Safety of Tenofovir Alafenamide vs. Tenofovir Disoproxil Fumarate

A Pooled Analysis of 26 Clinical Trials

Samir K. Gupta; Frank A. Post; José R. Arribas; Joseph J. Eron Jr; David A. Wohl; Amanda E. Clarke; Paul E. Sax; Hans-Jürgen Stellbrink; Stefan Esser; Anton L. Pozniak; Daniel Podzamczer; Laura Waters; Chloe Orkin; Jürgen K. Rockstroh; Tatiana Mudrikova; Eugenia Negredo; Richard A. Elion; Susan Guo; Lijie Zhong; Christoph Carter; Hal Martin; Diana Brainard; Devi SenGupta; Moupali Das


AIDS. 2019;33(9):1455-1465. 

In This Article


Study Design and Participants

We included 26 phases 2 and 3 multicenter, multinational, clinical studies of TAF-containing regimens in PLH including adults, adolescents, and children (aged ≥6 years) who were either ART-naive or virologically suppressed on a stable ART regimens containing TDF. These studies were conducted between 28 December 2011 and 4 December 2017. Study design and inclusion criteria, including minimum renal function, of each trial are described in Appendix Table 1, Of the 26 studies, 14 were double blinded and randomized, six were open label and randomized, and six were single arm. All trials were undertaken in accordance with the Declaration of Helsinki and approved by central or site-specific review boards or ethics committees. All participants or their legal guardians (if minors) provided written, informed consent.


Postbaseline study visits were conducted at weeks 4, 8, 12, 24, 36, and 48 and every 12 weeks thereafter until week 96. Renal laboratory tests included serum creatinine (SCr), CrCl by Cockcroft–Gault, treatment-emergent proteinuria by dipstick, urine albumin-to-creatinine ratio (UACR), and tubular proteinuria [urine retinol binding protein-to-creatinine ratio (RBP : Cr) and β2-microglobulin-to-creatinine ratio (β2M : Cr)] (Covance Laboratories, Indianapolis, Indiana, USA).

Renal safety was assessed by recording of adverse events, which were coded using the Medical Dictionary for Regulatory Activities (MedDRA, version 18.1–19.1) (Appendix Table 2,

Analysis of Primary Renal Safety Outcomes

The primary renal safety outcomes were incidence of PRT events, and study drug renal discontinuation events. For primary outcomes analysis, we pooled all participants from the 26 available trials who received at least one dose of study drug (safety analysis set). We derived safety measures data using all data collected on or after study drug was first given up to either the data cut date for participants still on study drug or up to 30 days after the last dose of study drug for participants who permanently discontinued treatment early. We summarized baseline demographics and characteristics of the included participants with descriptive statistics.

We defined 'renal discontinuation events' as investigator-reported discontinuation events for which the attributable MedDRA code exists in selected renal preferred terms from the 'renal and urinary disorders' System Organ Class (Appendix Table 2, Similarly, PRT cases were defined as investigator-reported adverse events indicative of tubular disorders, including reported terms of PRT and Fanconi syndrome (preferred terms are provided in Appendix Table 3,, regardless of study drug relatedness. The cumulative incidence rates of investigator-reported cases of PRT and renal adverse events leading to study drug discontinuation were calculated as the number of events divided by the total numbers of participants pooled from the 26 trials treated with TAF-containing or TDF-containing regimens, respectively. The differences in the cumulative incidence rates between treatment groups were compared using Fisher's exact test. To minimize type I error resulting from multiple hypothesis testing, we performed primary endpoint analysis in a predetermined sequence, only proceeding to the second endpoint (renal discontinuation events) if the first endpoint (PRT events) analysis demonstrated statistical significance with α = 0.05.

Analysis of Secondary Renal Outcomes

We assessed secondary renal outcomes including treatment-emergent renal adverse events, SCr, CrCl, treatment-emergent gross proteinuria (by dipstick), UACR, and tubular proteinuria (urine RBP : Cr and β2M : Cr). Treatment-emergent proteinuria was defined as 1+ or greater proteinuria by dipstick on any occasion during trial follow-up, regardless of persistence. Urine protein-to-creatinine ratio was monitored during the trials, but a change in assay methodology occurring partway through several trials resulted in data unsuitable for integrated analysis. For the analysis of these secondary renal outcomes, we selected a subset of trials that satisfied the following predetermined criteria: randomized design; TAF and TDF arms; and at least 48 weeks of follow-up. Based on these criteria, a total of seven trials were selected, including two treatment-naive studies and five virologically suppressed studies (referred to as switch studies) (Figure 1). To facilitate accurate assessment of CrCl changes in study participants, we excluded participants who switched from an ART regimen lacking a known creatinine transport inhibitor to a regimen containing a known creatinine transport inhibitor (rilpivirine, dolutegravir, bictegravir, COBI, or RTV).[36–41] This approach allowed us to reduce confounding caused by SCr increases attributable to initiation of a creatinine transport inhibitor.

Figure 1.

Characteristics of studies included in the integrated analysis. Treatment-naive studies included in the secondary analysis are highlighted in blue, virologically suppressed people living with HIV studies are highlighted in green. 3TC, lamivudine; ATV, atazanavir; AE, adverse event; B, BIC, bictegravir; C, COBI, cobicistat; DRV, darunavir; DTG, dolutegravir; DB, double blind; E, elvitegravir; FTC, emtricitabine; OL, open label; PI, protease inhibitor; R, randomized; R, RPV, rilpivirine; RTV, ritonavir; STR, single tablet regimen; TE, treatment-experienced; TN, treatment-naive; VS, virologically suppressed.

Using these data, we evaluated the incidence rates of treatment-emergent renal adverse events (Appendix Table 2, and of proteinuria by dipstick. We also summarized change from baseline in SCr and CrCl and percentage change from baseline in UACR, RBP : Cr, and β2M : Cr. We used logistic regression models to compare the differences in incidence rates between treatment groups and linear regression and rank analysis of covariance (adjusted for baseline demographics and disease characteristics selected from step-wise procedure) for change and percentage change from baseline in renal parameters, respectively.

To control for type I error in the testing of multiple secondary renal outcomes hypotheses, we employed the following testing strategies. First, the primary comparisons of PRT and renal discontinuation events in all 26 studies were analyzed using a predefined sequence as described above. Subsequently, hypothesis testing for secondary outcomes was performed using the Holm–Bonferroni method; P values reported in the text are Holm–Bonferroni adjusted.[42,43] We used SAS Software Version 9.4 (SAS Institute Inc., Cary, North Carolina, USA) for all analyses. All studies were conducted according to protocol without substantial deviations.