Drug interactions are not always predictable

Caution is necessary when selecting anticonvulsants for people living with HIV because of potential interactions with antiretrovirals [1]. The anticonvulsant valproic acid is metabolized by uridine diphosphate glucuronosyltransferase (UGT) enzymes, with beta oxidation and cytochrome P450 (CYP450) oxidation as minor pathways [2]. Dolutegravir is metabolized via UGT1A1 and CYP3A4 and is a substrate of Pgp and BCRP and inhibitor of OCT2 [3]. Although a significant pharmacokinetic interaction is not anticipated, we report a case of subtherapeutic dolutegravir concentrations in a patient taking valproic acid. A 43-year-old woman from Ethiopia was diagnosed with HIV-1 in 2010 and was virally suppressed after starting efavirenz-based treatment in 2016. Her medical history was significant for a possible stroke and a seizure disorder for which she was taking valproic acid 250 mg daily. In July 2018, she experienced a seizure with a suboptimal valproic acid concentration of 114 umol/l. An interaction with efavirenz was suspected; valproic acid was increased to 1250 mg daily and she was switched to a regimen of dolutegravir, lamivudine and abacavir. In September 2018, valproic acid concentrations increased to 288 umol/l, and the patient had no further seizures. However, dolutegravir concentrations 7 h postdose were subtherapeutic at 0.13 mg/l (target trough: 0.3 mg/l). Efavirenz concentrations were undetectable in this sample, ruling out efavirenz-mediated induction as a cause. In October 2018, dolutegravir concentrations remained subtherapeutic, with a 16 h postdose level of 0.18 mg/l. The patient reported 100% adherence and was not taking any other prescription or over-the-counter medications. Valproic acid was discontinued and replaced with levetiracetam. One month later, dolutegravir concentrations 13 h postdose were therapeutic at 0.98 mg/l (Fig. 1). The patient remained virally suppressed throughout this period.Fig. 1: Plasma concentrations of dolutegravir during concomitant valproic acid therapy and in the absence of valproic acid.According to the Drug Interaction Probability Scale [4], our patient experienced a probable drug interaction between dolutegravir and valproic acid. Similarly, Palazzo et al.[5] observed lower dolutegravir trough concentrations in seven patients on valproic acid (median 0.068 μg/ml), compared with 142 patients not on valproic acid (median 0.557 μg/ml). Subsequent pharmacokinetic analysis of two patients revealed dolutegravir AUC 80–90.5% lower compared with mean reference values. The extent of reduction in dolutegravir concentrations in our case is similar with those observed by Palazzo et al.[5] Importantly, in our patient, dolutegravir concentrations increased to therapeutic levels following discontinuation of valproic acid, further supporting an interaction hypothesis. Palazzo et al.[5] suggested that dolutegravir absorption may have been reduced secondary to chelation with magnesium-containing long-acting valproic acid formulations. As their findings revealed lower dolutegravir levels at later time points, they also hypothesized that progressive enzymatic induction of dolutegravir-metabolizing enzymes or transporters may have occurred. The formulation of valproic acid that our patient was taking did not contain magnesium or any other cations as excipients. At low doses, valproic acid is a mild auto-inducer of beta-oxidation [6] and reduces exposures of some antipsychotics by 24–50% [7]. However, valproic acid auto-induction does not involve 4-hydroxylation or glucuronidation, which is the main pathway of dolutegravir metabolism [6], and the reductions in dolutegravir are significantly greater than the reductions in antipsychotics seen with valproic acid. More importantly, valproic acid is generally considered an inhibitor of UGT and CYP enzymes [2], and significantly increases concentrations of UGT and CYP substrates, such as lamotrigine, phenobarbarbital, and zidovudine by 80–100% [8]. Another potential mechanism could be displacement of dolutegravir protein binding by valproic acid, as both are highly bound drugs. Protein-binding interactions are generally not clinically significant, as increased fraction of free drug is available for elimination, and thus steady-state concentrations of unbound drug remain unchanged. However, total concentrations of the displaced drug would decrease, which would impact therapeutic drug monitoring interpretation if total rather than free concentrations are measured [9]. Valproic acid displaces the protein bound drugs diazepam and phenytoin, reducing total concentrations by 22–31% without similar decreases in free concentrations. This may explain the continued viral suppression in our patient despite significant reductions in dolutegravir. Nonetheless, this theory is not entirely satisfactory as the reductions in total diazepam and phenytoin are not as great as the reductions in dolutegravir observed by us and Palazzo et al.[5] Finally, if valproic acid did displace dolutegravir, why has this interaction not been described with other highly bound antiretrovirals? In a pharmacokinetic study, valproic acid did not significantly alter efavirenz pharmacokinetics and there was a trend to increased lopinavir AUC [10]. Similarly, our patient had therapeutic efavirenz concentrations while on valproic acid. In conclusion, the mechanism for the interaction between valproic acid and dolutegravir remains unclear. Measuring free dolutegravir concentrations would help to elucidate whether unbound dolutegravir exposures are affected. Close monitoring is recommended to ensure viral suppression. Acknowledgements Conflicts of interest A.T. has received speaker and consultant honoraria from Abbvie, Gilead, Merck, and ViiV. A.W. has received speaking and/or consultant honoraria from Merck and ViiV. C.M. declares no conflicts of interest. S.W. has spoken at CME events, and received consultant fees from Gilead, Merck, ViiV and Janssen. S.W. holds a chair in HIV clinical management and aging from the Ontario HIV Treatment Network, Canada.

Récupéré en direct depuis OpenAlex et désinversé. Les résumés ne sont pas conservés dans cette base de données : les index inversés représentent 8,6 Go des 9,3 Go de texte de la base, et le serveur dispose de 13 Go libres.