4 Discussion

HCQ is approved and used worldwide for treatment of malaria, and RA (SLE and CLE). The approved doses are higher in magnitude and of shorter duration in malaria (1200mg on day 1 followed by 400mg daily over 10 days ) as compared to RA (loading dose of 400-800mg BID followed by maintenance dose of 200mg BID chronically with sometimes therapeutic drug monitoring targeting trough plasma concentrations of 0.6-1mg/L [4]). Several PK models are available in the literature for HCQ in these indications. [1, 2, 9] Our choice to use the PK model by Carmichael et al was motivated by the fact that it was developed to describe a relatively large cohort of blood concentrations (known to be less variable). The estimated apparent clearance values were quite consistent across previously published popPK models: 10-11L/h when whole blood concentrations were analysed [3,4] vs 51-68L/h when plasma concentrations were analysed [13-15]. The predictive performances of Caramichael model were confirmed on external data including data in CLE patients and in COVID-19 patients with overall satisfactory fitting of digitized concentrations. It is therefore judged adequate be used to explore the differential/relative impact of alternative dosing regimens in COVID-19 patients in the absence of a refined popPK model developed using COVID-19 patient data. It should however be noted that this model still carries an high unexplained variability component on the volume of distribution and clearance parameters: there is therefore a need for refinement of this model and better characterisation of PK in COVID-19 patients, including by adequate description of covariate effects. Potential covariates include bodyweight, CYP2D6 modulators and underlying renal impairment.
In the absence of a high loading dose, the results of the dosing simulations scenarios show that the drug progressively accumulates over the dosing periods of 5 or 10 days: safety monitoring can therefore be needed all the dosing time along and even after. This is confirmed by case reports of patients experiencing adverse drug reactions such as QT prolongations even after drug withdrawal. The appropriate characterization of the loading and maintenance doses needed is therefore important not only for drug efficacy but also for drug safety. The use of high loading doses need to be justified in view of the hazard for serious adverse events. There is still uncertainty on the target/relevant systemic concentrations for drug efficacy and safety. This is an important gap to be filled in the current situation because systemic concentrations are more accessible for monitoring than could be lung concentrations. There is an unmet need for adequately conducted clinical PK and exposure-response studies.
Yao and al [5] have shown in their recent publication that in vitro EC50 for prophylactic and treatment antiviral effects on SARS COV2 were 0.72 and 5.74 µM, respectively. Based on a PBPK modelling approach they have proposed dosing regimens that allow reaching empirically determined ratios between free lung concentrations and the in vitro EC50. However, in addition to the fact that this model was not validated using clinical data in COVID-19 patients, the recommended doses should still be cautiously considered because the relevant target ratios between lung or systemic concentrations and in vitro active levels are still to be established as well as the ranges for effective whole blood and plasma total concentrations. A more recent paper [12] was published in this sense using a model-based approach and PK/PD modelling of viral load and QT prolongation. However, it should be noted that this was a retrospective analysis of either aggregated or limited previously published data generated in different settings and for different purposes. Several unverified assumptions were therefore needed for the PK/viral load and the PK/QT modelling. Of note, the assumed/modelled QT prolonging effects were those of choloroquine and not hydroxychloroquine. Moreover, The overall unexplained variability was very high and covariate modelling was not implemented. Research is still needed to determine target HCQ level for in vivo (human) antiviral effect in COVID-19 and the link with clinically relevant outcome such as patient cure and survival for the different disease stages. Given the known multiphasic features of the COVID-19 disease and the importance of the inflammatory component of the disease, it is still unclear how relevant are viral load clearance by antiviral drugs for the patient clinical outcome in early vs later stages of the disease.
While it is not possible to identify the optimal dose in the absence of properly conducted dose-exposure-response analyses using relevant data in the target indication, the currently available clinical efficacy and safety data in different doses used in COVID-19 patients can already provide some useful information on the dose requirement when interpreted in link with the related PK information. High rates of positive clinical outcomes have been reported with doses of 600mg to 800mg daily on day 1 followed by 400 to 600mg daily for a total treatment duration of 5 to 10 days [6-10], also confirmed in the cohort of 172 patients treated at Saint Pierre hospital (see Table 2 and Figure 3). While these studies were all either single arm (no placebo arm), uncontrolled or of limited size, and therefore precluding the robust identification of the actual drug effect size, the important learning from these data is that higher doses might not needed for an important proportion of the patients. The determinants of positive patient outcomes are still to be identified, and HCQ dose optimization can certainly be one of them. Additionally, as extensively discussed in the recent literature, disease stage, patient age, and comorbidities might also play key roles[17-19]
As regards safety, the overall safety profile seems quite good when the drug is given at dose of 400mg - 800 mg on day 1 followed by 400 – 600 mg daily during 5 to 10 days, under close clinical monitoring. Available concerning cases reported in EV or in the literature are consistent with the known safety concerns with HCQ which are potentiated by either PK overexposure due to pharmacokinetic drug-drug interactions and/or renal impairment or PD drug-drug interactions due to additive toxicities with co-medications. Aggravation of the toxicity due to comorbidities or underlying renal or liver diseases related to COVID-19 pathophysiology cannot be excluded either [17-19] (also see Table 3). It is therefore essential that patient treated with HCQ are closely monitored for these risk factors, and that appropriate risk minimisation measures are implemented as needed.
It should however be noted that the clinical safety data from EV should be cautiously interpreted due to the potential bias related to spontaneous underreporting.