Pharmacokinetics/Pharmacodynamics of Antiviral Agents Used to Treat SARS-CoV-2 and Their Potential Interaction with Drugs and Other Supportive Measures: A Comprehensive Review by the PK/PD of Anti-Infectives Study Group of the European Society of Antimicrobial Agents

Table 1 contains general information about the key drugs currently suggested for the treatment of COVID-19. Most agents were originally approved for the treatment of other viral infections. The exception is (hydroxy)chloroquine, an immunomodulatory drug with known antiviral activity that has been used for over 60 years, primarily to treat malaria and, more recently, autoimmune diseases, such as systemic lupus erythematosus and inflammatory arthritis. Some of the agents listed have been studied with related viruses, such as Middle East Respiratory Syndrome (MERS) coronavirus. Most agents discussed in this article purportedly have direct mechanisms of action against SARS-CoV-2. As COVID-19 disease has emerged only recently, drugs currently being used are in various phases of clinical trials and none are approved for use in COVID-19, except for the very recent approval of remdesivir. The dosing regimens are based on current knowledge, derived from other indications, and may change in the future when new data become available. Dosing regimens in children are unavailable for most agents. In general, the heterogeneous total daily doses for drugs against COVID-19 disease are similar to, or greater than, that used for other indications.

Table 2 summarizes the data on the PK/PD properties of the agents recommended for SARS-CoV-2 and other viruses. The available data are limited and are based primarily on in vitro studies in various cell lines. Drug potency is usually presented as the half maximal effective concentration (EC₅₀), which varies between viruses. The EC₅₀ values for other viruses are compared against SARS-CoV-2, with a lower EC₅₀ indicating increased potency. While EC₉₀ is usually preferred as a therapeutic target for antivirals, it can only be inferred from EC₅₀ when the Hill coefficient is 1 (in which case EC₉₀ is ninefold higher than EC₅₀). Since the Hill coefficient is not routinely reported, we used EC₅₀ to compare the relative potencies of the antivirals reviewed.

Data on other coronaviruses have been summarized; however, where data are unavailable for coronaviruses, other pathogens are reported. For chloroquine and PegIFN-α2β, measured intracellular concentrations are correlated with the in vitro EC₅₀ and, as such, serve as the PK/PD index [6, 7]; however, there are no studies comparing various PK/PD indices for these agents.

The EC₅₀ values for remdesivir, chloroquine, and ribavirin against SARS-CoV-2 were compared with those of MERS. For both remdesivir and ribavirin, the EC₅₀ values were higher than for MERS, indicating that a larger dose may be needed to treat COVID-19. The EC₅₀ value of chloroquine was within the same range for SARS-CoV-2 and MERS. In an in vitro study, Yao et al. compared chloroquine with hydroxychloroquine and reported that hydroxychloroquine was more potent than chloroquine [7], although caution interpreting these results is warranted since different EC₅₀ values were reported depending on whether experiments were conducted for 24 or 48 h. Since EC₅₀ is not a time-dependent parameter, this calls into question how reliable the estimate is and how well it may translate to an in vivo target. In addition, this study has recently undergone a critical review by authors from the US FDA [8]. Furthermore, other studies have conversely found chloroquine to be more potent than hydroxychloroquine [9], and emerging data from randomized controlled trials (RCTs) [10] and large observational studies [11] suggest that both chloroquine and hydroxychloroquine result in increased mortality when used in COVID-19. While we include chloroquine and hydroxychloroquine in the summary of evidence, there is great uncertainty as to the clinical role of these drugs in hospitalized COVID-19 patients. Indeed, meanwhile, several negative studies have led to discontinuation of the use of these two drugs in many clinical studies in many countries [12, 13], yet other countries still continue to use these widely available drugs due to a lack of alternatives.

Several interferons (IFNs), including IFN-α, PegIFN-α2β, IFN-α1β and IFN-β1β, have been examined for the treatment of COVID-19; however, they are administered as adjuvant therapy with other anti-COVID-19 drugs.

The currently available data on drug efficacy and PK/PD targets for COVID-19 are inadequate to support therapeutic drug monitoring; however, some data on plasma concentrations are available in the literature (Table 2). When drug concentrations are available in the literature, it may be prudent to evaluate individual concentrations in patients in which high variability in PK combined with an increased likelihood of DDIs and adverse effects can be expected, i.e. typically critically ill patients.

A shortcoming of the data presented in Table 2 is the fact that the total concentrations of the drug were reported by the majority of papers, or information as to whether total or free concentrations were reported is not available at all. This must be taken into account when PK/PD calculations are performed, since, unusually, only the free fraction will be active.

As there are no approved COVID-19 therapies, combination therapy against SARS-CoV-2 with agents exhibiting different modes of action may have a role in helping to optimize therapy until clinical trial data become available. This combination approach is consistent with the management of many viral, fungal, and bacterial infections where there are suboptimal single-agent treatment options. We aimed to assess the evidence of repurposed antiviral combinations that were not specifically designed to treat SARS-CoV-2.

The strongest RCT evidence exists for remdesivir, which has been shown to reduce the recovery time for moderate–severe COVID-19 in comparison with standard care (11 vs. 15 days [14]; p < 0.001). These data have now led some to declare remdesivir to be the standard of care for COVID-19 disease, even though there was no significant difference in mortality between the remdesivir and standard care groups.

Other, albeit less-compelling, data exist for LPV/r plus ribavirin therapy (retrospective, case–control study) resulting in a reduction in mortality, acute respiratory distress syndrome (ARDS), and viral shedding in the treatment of SARS (Table 3). However, extrapolating these data to SARS-CoV-2 should be undertaken with caution as LPV/r and another HIV protease inhibitor, nelfinavir, exhibit good activity against SARS [15, 16] but are less effective against MERS [17]. Another potential combination includes LPV/r, ribavirin and IFN (prospective, non-randomized, comparative controlled study), resulting in shorter duration of viral shedding and hospital stay when compared with LPV/r alone. Randomized trials involving these drugs, based on their promising in vitro activity, will provide important guidance. Of note, we warn against the use of hydroxychloroquine in combination with other drugs that may prolong the QT interval due to potentially life-threatening adverse effects [11]. In a large cohort study, all patients who received hydroxychloroquine for the treatment of pneumonia associated with COVID-19 were at high risk of QTc prolongation, but concurrent treatment with azithromycin was associated with greater changes in QTc [18]. However, since combinations of QTc-prolonging drugs do not necessarily result in additive QTc prolongation, a case-by-case evaluation seems warranted [19].

Supportive care with other pharmacological agents, including antibiotics, sedatives, analgesics, and anticoagulants, need to be considered when treating with antiviral and other repurposed agents, particularly in the critical care setting. Electronic supplementary Table 1 lists the major drugs by class and highlights potential interactions. In particular, caution should be exercised for CYP3A4 DDIs with narrow therapeutic index drugs, such as anticoagulants (warfarin, acenokumarol, dabigatran, rivaroxaban, and apixaban) and immunosuppressant agents in transplant recipients in whom additional monitoring may be required. For patients sedated with midazolam, dose reduction should be considered when patients are treated with CYP3A4 inhibitors, such as LPV/r, as there is a significant risk of oversedation, resulting in unnecessary prolongation of intensive care unit (ICU) stay [20]. There are multiple potential interactions resulting in cardiac complications, including prolonged QT interval with (hydroxy)chloroquine combinations and LPV/r, that necessitate monitoring. As these DDI are substantial, the balance of risk versus benefit should be carefully considered prior to drug administration to prevent significant morbidities.

According to a preliminary report in patients hospitalized with COVID-19, dexamethasone significantly reduced 28-day mortality among those patients receiving invasive mechanical ventilation or oxygen at randomization, but not among patients not receiving respiratory support [21]. If these data hold true, at the moment corticosteroids belong to the most active intervention in patients with severe COVID-19 disease. In this aspect, it should also be noted that while dexamethasone is a potent inductor of CYP3A4 induction, this effect seems to be less relevant for the common alternative, betamethasone [22].

The potential interactions for most drugs used to treat COVID-19, and comedications, are provided in detail in electronic supplementary Table 2. It is emphasized that not all described interactions are necessarily clinically relevant, and the ultimate decision on the need for avoiding a certain combination or dose adjustment must be taken by the treating physician.

DDIs are an important consideration for all therapies used to treat COVID-19. This is especially the case with repurposed antiretroviral drugs and (hydroxy)chloroquine, which have many potential DDIs. Clinicians treating patients infected with SARS-CoV-2 need to carefully consider the potential for DDIs before commencing therapy. Many DDIs may be mitigated by simple measures such as continuous electrocardiogram (ECG) monitoring or by having maximum allowable QT intervals (e.g. 450 ms) for the interacting combinations to be used. The ritonavir component of LPV/r is deliberately used to inhibit CYP3A4 and thereby increase antiretroviral drug concentrations; however, this leads to a significant potential to increase concentrations of other coadministered therapies that are CYP3A4 substrates. Notably, the interaction between (hydroxy)chloroquine and other agents that may inhibit drug CL can result in cardiac toxicity and patients should be monitored closely. Based on the long half-life of (hydroxy)chloroquine (in the magnitude of several weeks), interactions might persist for several days after treatment has ceased. This may be especially problematic in critically ill patients with pre-existing cardiovascular morbidity, and extreme caution should be observed. The use of triazole antifungals should be avoided or carefully monitored if administered concurrently with LPV/r due to DDIs. Potential DDIs are reported between either LPV/r and important drugs commonly used in the critical care setting, including ketamine, rocuronium, and many of the opioid agents. Utmost care is required when considering the coadministration of these agents with antiretroviral drugs in critically ill patients. The newer investigational antiviral agents, remdesivir and favipiravir, appear to have a lower potential for DDIs; however, the main concern with their use is decreased concentrations if coadministered with enzyme inducers. (Hydroxy)chloroquine may prolong the QT interval, therefore ECG monitoring is required when they are coadministered with other agents known to cause QT interval prolongation. Coadministration of (hydroxy)chloroquine with drugs that are known to prolong the QT interval, such as amiodarone and flecanide, is not recommended. However, clinical experience with these drugs is much less than the repurposed antiretrovirals. A comprehensive and evolving DDI database has been created by the University of Liverpool and this should be consulted for potential DDIs not covered in our review [3].

Systemic inflammatory response syndrome (SIRS) may occur with the use of extracorporeal treatments, and may cause alterations in CYP‐mediated metabolism. SIRS increases activity of all CYPs, except CYP3A4, which decreases. In children, CYP enzymes are commonly immature in neonates and take time to reach similar activity levels as adults [23]. In this section, as well as in Table 4, we summarize the principles of PK alterations that occur with extracorporeal membrane oxygenation (ECMO) and renal replacement therapy (RRT) to impact drug exposure and thereby dose. Although limited studies have been conducted on COVID-19 therapies, optimized dosing should consider these potential impacts on individual patients.

The most common mechanisms by which ECMO is likely to affect the PK of drugs are sequestration by the oxygenator and tubing in the ECMO circuit, leading to reduced circulating drug concentrations [23]. While lipophilic drugs and highly protein-bound drugs are more likely to be sequestered in the circuit, hydrophilic drugs can be significantly affected by hemodilution and changes in albumin concentration, potentially leading to altered protein binding and an increased Vd. Indeed, an increased free fraction means more distribution from the central compartment into the peripheral compartments (i.e. tissues), leading to an increased apparent Vd [24]

Sepsis-related acute kidney injury often develops in the context of multiple organ dysfunction syndrome and leads to relevant modifications of several PK parameters. Moreover, high volumes of fluid resuscitation, commonly required in critically ill patients [25], may significantly affect the Vd of several drugs [26]. When Vd of a drug is typically small since the drug is mostly retained in the intravascular compartment (where protein binding is high), clinically significant removal of small drugs by RRT is unlikely. Where Vd is < 1.0 L/kg and protein binding is not high (> 80%), the commencement of RRT adds further complexities for dosing, with possible extracorporeal CL. Renally excreted drugs are usually affected by RRT to a much greater extent than hepatically excreted drugs [27]. Only the free (i.e. unbound) drug is cleared across the RRT filter. With the exception of a few drugs, the molecular weight (MW) of the most commonly used antimicrobial agents is lower than 1000 Da and plays a key role, especially in diffusive RRT modalities, as the sieving coefficient (SC) of a molecule is inversely proportional to MW. The SC is generally similar for drugs with a MW around 1000–1500 Da in convective modalities. However, in diffusive techniques, the ratio of dialysate to plasma solute concentration (saturation coefficient [SA]) is more strictly dependent on MW and tends to decrease progressively as MW increases [28]. Whereas intermittent hemodialysis (IHD) or continuous venovenous hemodialysis (CVVHD) are essentially diffusive techniques, continuous venovenous hemofiltration (CVVH) is a convective technique, and continuous venovenous hemodiafiltration (CVVHDF) is a combination of both. As a general rule, the efficiency of drug removal by the different techniques is expected to be CVVHDF > CVVH > CVVHD/IHD, but this can still vary greatly depending on the physicochemical properties of each drug and the CRRT settings [29]. Dialyzer membrane characteristics (cut-off) may also play a key role.

Table 4 extrapolates basic drug physicochemistry and known PK data to predict the likely effects of ECMO on PK. Sparse data on the IHD CL of (hydroxy)chloroquine are available and suggest that the high Vd of (hydroxy)chloroquine limits significant alteration in drug concentrations [30]. As such, ECMO is predicted to have minimal impact on the drug concentrations of (hydroxy)chloroquine, although sequestration onto the oxygenator and circuit tubing cannot be excluded. There is also the hypothesis that (hydroxy)chloroquine rapidly partitions intracellularly, potentially resulting in minor effects overall and in the vascular compartment [25]. The PK of LPV/r in hemodialysis suggests that dosing adjustments are unnecessary in treatment-naive patients, in part due to the high protein binding of these drugs [31]. CVVH has no clinically relevant contribution to total CL of favipiravir [32]. For ribavirin, CL was reduced by 50% via IHD [33], which is not considered significant enough to justify increasing the dose based on increases in dialysate or RRT filtration flow rate [34]. No significant effect of RRT on IFN concentrations is predicted due to the MW of the IFN compounds which usually exceeds 15 kDa [35, 36].