Estimated Glomerular Filtration Rate Slopes on Tenofovir Alafenamide

F Ibrahim; L Campbell; AC Bailey; S Stockwell; L Waters; C Orkin; M Johnson; M Gompels; A De Burgh-Thomas; R Jones; G Schembri; PW Mallon; FA Post


HIV Medicine. 2020;21(9):607-612. 

In This Article


We conducted an observational cohort study of individuals with HIV infection who had initiated TAF-containing ART prior to 1 October 2016 as part of routine clinical care and not through participation in randomized controlled clinical trials. This date was chosen to include those with early access to TAF (and thus more likely to have or be at risk of comorbid kidney or bone disease) while ensuring adequate follow-up time on TAF-containing regimens. Participants could be enrolled irrespective of ART and prior TDF exposure status (naïve or experienced: previous or current). Participating clinics retrospectively identified eligible subjects and provided anonymized demographic, clinical and laboratory data, including serial creatinine and proteinuria measurements, and a full ART history from 1 January 2000 onwards. The study was approved by the UK Health Research Authority; as per local guidance, Research Ethics Committee approval and informed consent were not required, with exception of the Irish site where data for these analyses were derived from a pre-existing, Institutional Review Board-approved, prospective cohort study.

The chronic kidney disease-epidemiology formulation was used to convert creatinine measurements to eGFR,[14] and serial eGFR measurements were used to generate eGFR slopes which capture the direction and rate of change of eGFR and as such provide an indicator of changing renal function over time.[15] Follow-up was stratified by time exposed to TDF (from first to last exposure irrespective of interruptions), time exposed to TAF (censored at TAF discontinuation or last visit on TAF), and time on no ART or non-tenofovir-containing ART (excluded from the analyses). To estimate and compare the rates of eGFR change per year on TDF and TAF, piecewise mixed effects models with random intercept and time terms were fitted with eGFR as the dependent variable. The models were fully adjusted for the following factors: age at switch, gender, ethnicity, eGFR at TDF/TAF initiation, and time-updated CD4 cell count, HIV RNA [undetectable (< 50 HIV-1 RNA copies/mL) or detectable] and cumulative TDF use (where relevant). Subgroup analyses were performed that restricted analyses to those with rapid eGFR decline (we considered two definitions: decline in eGFR of > 3 or 5 mL/min/1.73 m2/year[15] as estimated from unadjusted eGFR slopes in the mixed effects model) or CKD (defined as eGFR < 60 mL/min/1.73 m2) while receiving TDF. All analyses were performed using STATA 15 (StataCorp LLC, College Station, TX).