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

Discussion

Common renal elimination transporter pathways of TFV and MTX raise concern for potential drug–drug interactions. In this study, we investigated the impact of LDMTX on TFV PK via intensive venous sampling in a subset of participants chronically suppressed with TFV-containing ART who were randomized to either active LDMTX or placebo as part of the A5314 study.^[1] Our results demonstrate decreased TFV exposure in the presence of LDMTX, including a 22% reduction in the GM AUC₆. Similar results were seen for AUC₂₄, AUC_24i, and C_max (with 36%, 24%, and 27% reductions, respectively). Although the differences in GMR for some PK parameters did not reach statistical significance, these results demonstrate an overall trend toward decreased TFV exposure in the presence of MTX. However, this magnitude of decrease in TFV exposure is likely not clinically significant toward providing adequate viral suppression. As reported for the parent trial,^[1] very few participants in the trial as a whole experienced a HIV-1 RNA level above the assay limit of quantification (40 copies/mL), and all were evenly distributed by treatment group. Among participants with measures above the assay limit of quantification, none had confirmed viral load failures (HIV-1 RNA >200 copies/mL). Visually, the terminal slopes appear identical and rate of absorption similar. In support of this observation, reductions in TFV exposure are not a result of inhibition of renal transporters during excretion, and are likely attributable to decreased absorption, as supported by a significant decrease in TFV C_max with LDMTX. Within the placebo arm, TFV C_max averaged 315 ng/mL, consistent with published TFV exposure.^[18]

Within PI subgroups, comparisons between study arms showed that TFV AUC₆, AUC_24i, and C_max trended toward higher values in participants on PIs, consistent with a previous report.^[19,20] This potential PI-based TFV interaction is likely driven by increased absorption by PI-related inhibition of P-glycoprotein (P-gp) and intestinal esterase,^[20–22] resulting in higher TFV exposure estimates in the context of PIs which partially compensates for the overall lower TFV exposure measured during MTX co-administration. This trend toward lower TFV exposure is likely driven by inclusion of 5 participants not receiving PIs in the LDMTX group who exhibited particularly low TFV exposure. These participants were demographically similar to the PK substudy population with consistent body mass indices, age, creatinine clearance and mixed regarding race. Overall, when the TFV PK parameters are stratified by use of concomitant PIs, a modest shift is seen in the distributions in the absence of PIs; however, this shift is not apparent in the co-administration of concomitant PIs and LDMTX.

MTX levels were also quantified using a precise analytical method but no statistical comparisons were possible because of lack of a control group not receiving TFV. However, comparisons were made to parameters published previously for low-dose MTX (Table 3). One would anticipate lower peak concentrations for intramuscular versus oral (PO) dosing and higher peak concentrations for 10 mg versus 7.5 mg PO doses. The MTX C_max observed in this study was about 40%–60% and 34% lower than published studies of weekly 10 mg intramuscular dosing^[23–25] and weekly 7.5 mg PO dosing,^[26] respectively. Of note, studies among individuals with rheumatoid arthritis report a wide range of bioavailability of oral methotrexate ranging from 20% to 100%.^[27,28] It is also possible that TFV and/or HIV disease may decrease MTX exposure. The earlier studies used fluorescence polarization immunoassays (FPIA) or radioimmunoassay; a recent comparison of FPIA to liquid chromatography coupled with tandem mass spectrometry indicates FPIA overestimated concentrations in this range,^[29] which may explain the discrepancy with prior PK estimates. One study, using a homogenous enzymatic immunoassay, reported no effect of TFV on MTX elimination.^[17] However, that study examined high doses of intravenous MTX, supporting that a potential interaction is likely driven by an absorption process, rather than elimination. Further studies with a non-TFV control are needed to confirm the effect of TFV on MTX exposure.

One of the limitations of study A5314 was that the targeted LDMTX was on the low end of the dose range that had been effective in reducing cardiovascular events in larger cohort studies of MTX.^[30–34] Although exposure–response analysis was not the focus of the primary study, it is possible that if the 57/86 participants assigned to LDMTX who were taking TDF had lower than anticipated exposure to MTX, its effects on the primary efficacy and safety endpoints may have been reduced, contributing to decreased power to detect differences between arms.

Understanding the underlying mechanisms of TFV- and MTX-mediated drug–drug interactions and toxicities, particularly the role of renal and intestinal transporters, continues to evolve and is a rich area of research and scientific debate. TFV is a substrate of renal transporters [OAT1, OAT3, multidrug resistance protein (MRP) 4] and TDF is subject to intestinal transport via P-gp, and likely MRP4. MRP2's role in TFV PK has been debated in the literature.^[21,35] More recently, TFV was identified as a substrate of MRP8 for which MTX is also a known substrate.^[36] Other transporters, such as MRP7, have been associated with renal injury but mechanistic studies are lacking.^[37] MTX is a substrate for numerous transporters;^[38] those shared with known or potential TFV or TDF pathways include OAT1 and OAT3,^[2,3] MRP2,^[39] MRP4,^[39] MRP8,^[36] and P-gp.^[40]

Of these pathways MRP4, mechanistically, has the most potential for a TDF–MTX PK interaction. MRP4 is an efflux transporter; although since it is located on the basolateral side of enterocytes, it ultimately facilitates drug entry to the blood and serves to mediate intestinal basolateral influx of TDF and MTX. The observed results of this study would be supported by this mechanism given lower C_max for TFV in the presence of MTX and for MTX as compared with prior studies.

Although the role of intestinal MRP4 in TDF absorption has not been well described, it was found to be a major contributor to adefovir dipivoxil uptake,^[41] a compound similar structurally and pharmacokinetically to TDF. Furthermore, Vitamin D₃, a known inducer of MRP4 expression, enhanced adefovir exposure in rats.^[42] Mouse models have also demonstrated the importance of MRP4 in cefadroxil absorption^[43] and dasatinib absorption and efficacy.^[44] In vitro and in situ interaction studies of TDF on atazanavir absorption supported clinical findings that TDF decreases atazanavir bioavailability; the authors hypothesized that OATP or MRP-mediated absorption pathways were likely involved.^[45]

Although MRP4 present in proximal renal cells is considered important in TFV-mediated kidney injury,^[46] intestinal MRP4 may be more so subject to saturation by MTX and TDF, given its lower expression frequency.^[47] Although preclinical experiments in MRP4 knockout mice failed to show importance of MRP4 in MTX absorption,^[48] in vitro interaction assays demonstrated MRP4-mediated interactions between MTX and nonsteroidal anti-inflammatory drugs.^[39] There are known inconsistencies in drug transporter expression in in vitro and preclinical models versus in human tissues^[49] limiting clinical translation of preclinical findings. MRP4 is expressed in the human jejunum,^[49] the likely site of MTX absorption.^[5,50]

Furthermore, genetic polymorphisms in transporter expression may have contributed to observed between-subject PK variability. Both MTX and TFV have documented polymorphisms that contribute to differences in PK, efficacy, or toxicity in varying patient populations and indications.^[51] For instance, genotype differences in MRP4 have been linked to altered TFV clearance in PWH and differential MTX plasma exposure in pediatric patients being treated for acute lymphoblastic leukemia.^[19,52]

Aside from transporters, an emerging field, the gastrointestinal microbiome, which may also play a key role in HIV disease pathogenesis,^[53] has been shown to metabolize and/or be altered by exposure to both TDF^[54] and MTX.^[55,56] Future studies may investigate the interplay between the microbiome and pharmacokinetics of multiple co-administered drugs.

The clinical implications of these results on the use of tenofovir alafenamide (TAF) are unclear. Unlike TDF, TAF is relatively stable in plasma and exhibits lower TFV plasma levels and higher intracellular tenofovir diphosphate levels than TDF. Although sharing some transport pathways with TDF (ie, P-gp and breast cancer resistance protein), TAF is additionally a substrate of OATP1B1 and OATP1B3, but not subject to OAT1 or OAT3 transport.^[57,58] Current literature does not indicate whether TAF is a substrate of MRP4. Given these differences, separate preclinical and/or clinical PK investigations of TAF and its potential in transporter-mediated interactions with MTX are warranted.

In summary, LDMTX resulted in modest reductions in TFV exposure that are not expected to be clinically significant for maintenance of HIV-1 viral suppression. However, these decreases seem driven by a subset of participants who were not on PIs, which indicates a change in TFV dosing is not warranted for those taking PIs. Further confirmatory and mechanistic studies of TFV PK in the setting of LDMTX without PI co-administration are warranted. Although MTX exposure was only characterized in the context of TFV co-administration, exposure was lower than anticipated from prior reports, which may have impacted effects of LDMTX on cardiovascular outcomes. This study will help inform dosing during MTX-TFV co-administration particularly for PWH who have rheumatoid arthritis or psoriasis.

J Acquir Immune Defic Syndr. 2020;85(5):651-658. © 2020 Lippincott Williams & Wilkins