Methotrexate Decreases Tenofovir Exposure in Antiretroviral-suppressed Individuals Living With HIV

David Gingrich, BS; Amelia N. Deitchman, PharmD, PhD; Amy Kantor, MS; Liusheng Huang, PhD; James H. Stein, MD; Judith S. Currier, MD; Priscilla Y. Hsue, MD; Heather J. Ribaudo, PhD; Francesca T. Aweeka, PharmD


J Acquir Immune Defic Syndr. 2020;85(5):651-658. 

In This Article


A5314 Study Design

A5314 is a phase II double-blind, randomized placebo-controlled trial assessing the safety and efficacy of LDMTX on endothelial function and inflammation in PWH who have been virologically suppressed with continuous ART (NCT01949116).[1] The study included men and women at least 40 years old, with or at risk for atherosclerotic cardiovascular disease with ART-suppressed HIV (CD4+ T-cell count ≥ 400 cells/mm3 and HIV-1 RNA level < 40 copies/mL for at least 24 weeks prior to study entry). Participants who met the enrollment criteria were randomized 1:1 to LDMTX or placebo. From entry through week 1 (lead-in period), participants took 5 mg once weekly by mouth of either MTX or placebo. Participants were then titrated to 10 mg/wk over 12 weeks, at which point they received the maximum dose of 15 mg/wk until week 24. If a participant did not meet the criteria for dose escalation at the protocol-defined dose escalation time, then the participant remained on his/her current dose until the next study visit at which time the participant was re-evaluated for dose escalation. All participants received 1 mg folic acid daily from study entry until 4 weeks after completion of study treatment, regardless of randomization.

PK Study Design

Participants were required to be on steady state TFV (defined as continuous ART for ≥24 weeks before enrollment without any changes to their basic regimen in the prior 12 weeks) and self-reported adherence for the last 4 doses. Intensive PK sampling was performed after the second dose of MTX (10 mg). LDMTX (or placebo) and TFV were administered at the same time and serial sampling was conducted at 0 (predose), 0.5, 1, 2, 4, and 6 hours postdosing (protocol version 2.0) for analysis of both TFV and MTX (for those on active drug) levels; under version 1.0 of the protocol some participants were also sampled at 8, 12, and 24 hours postdosing. The reason for the 6-hour sampling was to enhance enrollment by limiting the time commitment for the study, but to still allow intensive PK sampling during the day while clinics remained open.

A sample size of approximately 21 evaluable participants per treatment group was chosen to provide 95% power to detect a clinically relevant 40% increase in TFV AUC (in LDMTX versus placebo arms) with a relaxed type-one error rate (1-sided 5%). Given the absence of appropriate control within the A5314, analysis plans for MTX exposure involved a simple characterization of the MTX AUC to be compared descriptively against historical controls.

MTX and TFV Assay Development and Sample Quantification

For study participants randomized to placebo, plasma samples were assayed only for TFV and not MTX, whereas those randomized to LDMTX were analyzed for both TFV and MTX using validated liquid chromatography coupled with tandem mass spectrometry methods. MTX was fortified with a deuterated internal standard, and extracted from 50 μL of plasma by protein precipitation with acetonitrile (ACN). The extracted samples were separated on an Agilent Zorbax XDB-C8 high-performance liquid chromatography column (2.1 × 50 mm, 5 μm), and detected on a Sciex API5000 mass spectrometer. The method had a lower limit of quantitation (LLOQ) of 5 ng/mL, with a calibration range of 5–500 ng/mL. During sample analysis, the coefficient of variation for quality control samples ranged from 3.15% to 4.39%. Similar to MTX, TFV was fortified with a deuterated internal standard, and then 50 μL of sample was extracted by protein precipitation with ACN. The sample extracts were separated on a Phenomenex Synergi Polar-RP high-performance liquid chromatography column (150 × 2.0 mm, 4 μm), then detected on a Sciex API 5000 mass spectrometer. The LLOQ for TFV was 5 ng/mL, with a calibration range of 5–1000 ng/mL. The coefficient of variation of the quality control during TFV sample analysis ranged from 5.18% to 8.36%.

PK and Statistical Analysis

PK parameter outcomes included the area under the plasma concentration vs. time curve (AUC), and peak concentration (Cmax) for MTX and TFV, and the trough concentration (Cmin) for TFV. Parameters were estimated using non-compartmental analysis in WinNonlin v.6.2.1 (Certara L.P., Princeton, NJ). For AUC calculations, samples below the LLOQ were treated as missing except for the pre-peak TFV or MTX concentrations, which was set to 0 if below the LLOQ. AUC was calculated using the linear up-log down trapezoidal rule from 0 to 6 hours (AUC6) and 0–24 hours (AUC24 and AUC24i, where AUC24 was available only in participants enrolled under Version 1.0 and AUC24i was estimated for all participants using the predose concentration as the imputed 24-hour TFV concentration for participants enrolled under Version 2.0). Imputed concentrations assume there is no difference between the concentrations between 0 and 24 hours for steady state dosing. For TFV, Cmin was calculated from the raw data at time 0.

AUC6 was the primary endpoint, whereas all other PK parameters (AUC24, AUC24i, Cmax, Cmin) were secondary endpoints for TFV exposure. The primary endpoint for MTX was AUC6 and the secondary endpoint was Cmin. TFV PK parameters were compared between placebo and LDMTX treatment groups, whereas MTX PK was characterized only in the context of TFV-containing ART. PK parameters were summarized using geometric means (GMs) with 90% confidence intervals (CIs); imputed zero concentrations were set to 0.1 ng/mL to facilitate log transformation. Distributions were compared between LDMTX and placebo groups using 2 sample t-tests with unequal variance and were interpreted at the 10% nominal level of significance (2-sided test) without adjustment for multiple comparisons. This change to the analysis plan specified in the protocol (use of a 10% two-sided test versus a 5% one-sided test) was made before review of the data. All tests were performed on natural log-transformed PK parameters. Statistical analyses were conducted with SAS Version 9.4 (SAS Institute Inc, Cary, NC).