Retrospective Evaluation of an Observational Cohort by the Central and Eastern Europe Network Group Shows a High Frequency of Potential Drug–Drug Interactions Among HIV-Positive Patients Receiving Treatment for Coronavirus Disease 2019 (COVID-19)

Botond Lakatos; Justyna Kowalska; Sergii Antoniak; Deniz Gokengin; Josip Begovac; Anna Vassilenko; Piotr Wasilewski; Lukas Fleischhans; David Jilich; Raimonda Matulionyte; Kerstin Kase; Antonios Papadopoulus; Nino Rukhadze; Arjan Harxhi; Sam Hofman; Gordana Dragovic; Marta Vasyliev; Antonija Verhaz; Nina Yancheva; Cristiana Oprea


HIV Medicine. 2022;23(6):693-700. 

In This Article


We found that potential DDIs were common in this cohort of HIV-positive patients on cART who had COVID-19-specific treatment.

To date, little is known about the frequency of potential DDIs for cART and COVID-19-specific treatments in routine practice. Diverse treatment strategies were implemented early in the COVID-19 pandemic, including many experimental drugs. To properly identify potential DDIs in experimental drugs, one needs time, strict pharmacovigilance, and careful clinical monitoring of patients. All these were hardly achievable during the pandemic, when restructured health care, increased patient inflow, and dynamically changing protocols were features of this period. However, the team at Liverpool University provided a data set for COVID-19-specific medication and their potential interactions,[11] which informed the above analyses.

In an era when INSTIs are the central component of cART, potential DDIs have decreased given the beneficial interaction profile of the nonboosted INSTIs.[12] In our study, two out of three patients took an INSTI as a core agent, but only a single potential interaction was identified with a nonboosted INSTI (bictegravir and hydroxychloroquine). Although only 17.7% of the patients received NNRTIs and 13.7% received boosted PIs as core agents, these accounted for the vast majority of DDIs.

We found that the most commonly interacting drugs were corticosteroids, hydroxychloroquine and favipiravir (Figure S1). Of note, hydroxychloroquine was frequently administered in the early stage of the pandemic in the light of in vitro results and uncontrolled observational studies; later guidelines, based on accumulated evidence, recommended against its use. Although favipiravir is not part of the standard of care in Europe, in some countries, such as Turkey, Romania and Hungary, it was widely introduced. The use of corticosteroids – either methylprednisolone or dexamethasone – became a standard in the treatment of moderate or severe COVID-19 requiring oxygen therapy.[6,7] COVID-19 indications for dosing and duration of corticosteroid use are ambiguous. The largest trials used 6 mg of dexamethasone; however, this is not universal.[13,14] Optimal dosing is still not established, and DDIs are dose-dependent, so the level of clinical consequences is not well defined. The optimal duration for corticosteroid administration also remains undefined, but, in general, it is usually given for at least 10 days. We found two 'red labelled' potential DDIs, between corticosteroid and rilpivirine, with significant serious risk. This coadministration should definitely be avoided as plasma levels of rilpivirine (as a result of the induction of cytochrome P450 3A4 enzyme (CYP3A4)) are expected to decrease.[9] Furthermore, in routine clinical practice, proton pump inhibitors (PPIs) are indicated along with corticosteroids to prevent peptic ulcers, which reduce the absorption of rilpivirine as a result of a gastric pH increase. Even if dexamethasone is given for 10 days and in low doses with PPIs, in the context of a severe or critically ill patient with COVID-19, the risk of cART therapeutic failure should not be ignored, and hence close monitoring is warranted.[15]

'Amber labelled' potential DDIs for corticosteroids included boosted PIs (n = 19) and elvitegravir (n = 3). In both these cases, the concentrations of corticosteroids and antiretrovirals may be compromised during chronic administration. Ritonavir- and cobicistat-boosted antiretrovirals can increase corticosteroid levels, and therefore monitoring for symptoms of hypercorticism is needed if steroid use is long term. Also, concentrations of antiretrovirals may decrease as a result of CYP3A4 induction.[16]

Further potential DDIs with corticosteroids were identified for efavirenz, and one case each for etravirine and nevirapine. Within this constellation, corticosteroid levels may be reduced as a consequence of the inducing effect of NNRTIs. This can result in a loss of steroid effect and the clinical progression of COVID-19, and therefore close monitoring and potential dose adjustment of the steroids may be needed to achieve the required clinical effect.

Hydroxychloroquine, an antiparasitic and anti-inflammatory drug initially used in the early COVID-19 era on the basis of in vitro efficacy results, has a relatively high potential DDI if coadministered with NNRTIs or boosted antiretrovirals.[9] In our study, one case each for boosted atazanavir, bictegravir and rilpivirine was identified. Atazanavir and rilpivirine may elevate hydroxychloroquine concentrations as a result of the inhibition of CYP3A4, which can potentially increase corrected QT interval prolongation.[9] Hydroxychloroquine inhibits P-glycoprotein, which may increase concentrations of bictegravir; however, the clinical relevance of this is unclear. In consequence, these drugs should be used with caution, notably in those cases where long-term electrocardiogram (ECG) monitoring is needed, even after discontinuing hydroxychloroquine as it has a particularly long half-life of 40–50 days. Further potential weak interactions found in this study were with boosted darunavir, elvitegravir and efavirenz. As hydroxychloroquine is metabolized by CYP 3A4/2D6/2C8, these antiretrovirals can potentially increase hydroxychloroquine levels by inhibiting these enzymes. Interestingly, efavirenz has the opposite effect (inhibition of 2C8 and induction of 3A4), but clinical monitoring is needed.[10]

Favipiravir is a selective RNA-dependent RNA polymerase inhibitor approved to treat influenza in Japan. Early reports showed in vitro effects against SARS-CoV-2, and limited and controversial in vivo clinical benefits in small and uncontrolled trials.[17] In certain ECEE Network Group countries such as Romania, Turkey and Hungary, favipiravir has been introduced as an experimental COVID-19 drug on the basis of the early results. Favipiravir is a moderate inhibitor of the organic anion transporter 1 renal transporter, which affects the renal elimination of TDF and increases the risk of renal toxicity.[10] Therefore, monitoring the estimated glomerular filtration rate (eGFR) closely during coadministration of these drugs is important.

We did not find any potential DDIs with remdesivir in our study. Remdesivir is a prodrug that transforms into a ribonucleotide analogue inhibitor of viral RNA polymerase. Its use against SARS-CoV-2 is widespread; however, the results of efficacy trials have not consistently supported its use.[17] Despite inducing CYP1A2, CYP2B6 and P-glycoprotein substrates, the potential for significant clinical interactions is low.[10]

There are currently few reports available assessing the potential DDIs in cART and COVID-19-specific treatments. A study group from Spain published a list of potential DDIs for lopinavir/ritonavir when used to treat COVID-19, and found high levels of potential interaction, mainly with concomitant use of corticosteroids.[18] Clinical data and guidelines do not support the use of lopinavir/ritonavir as a COVID-19-specific treatment any more, but data highlight, again, the high DDI profile of this boosted PI.[5,6,9] This problem was also brought up from a pharmaceutical point of view in a letter written by Turkish authors, but none of their own data were presented.[19]

There are some limitations to our study, mainly as a consequence of its retrospective nature. Although we assessed the DDI data for cART and COVID-19-specific treatments, there was no detailed analysis of each patient's comedication. The focus of this study was not to assess the full profile of the cohort's DDIs, but to identify the most common and clinically relevant potential DDIs in order to inform medical practice in the region. Also, no prospective clinical follow-up is available to assess the potential long-term effects of the DDIs. Finally, because the standard of care changed over the waves of the pandemic differently in the different countries of the Network Group and because of lack of data, we could not link specific potential DDIs to the waves of the pandemic.

We strongly believe that, in spite of the above-mentioned limitations, the findings of this real-life retrospective analysis from the Central and Eastern European region of 524 HIV-positive persons who were on cART while receiving COVID-19-specific treatments may encourage clinicians to focus more on the drug interaction challenge. We found that potential DDIs were common, and that corticosteroids coadministered with boosted PIs and INSTIs, or NNRTIs seem to have the most frequent potential DDIs.