“Medicine is a science of uncertainty and an art of probability”, stated the Regius Professor of Medicine at Oxford William Osler (1849-1919) more than a century ago. After producing hundreds of thousands of papers on COVID-19, scientists have no clear answer to its origin and unpredictable individual responses following the infection. It is the task of physicians and scientists to reduce the uncertainty and apply the best knowledge in caring for their patients, guided by the Hippocratic aphorism primum non nocere.
The article on the critical appraisal of the therapeutic use of chloroquine diphosphate (CQ) and hydroxychloroquine (HCQ) in COVID-19 is a rich account of the state-of-the-art. It illustrates the importance of in-depth and comprehensive literature search and the need for preparatory work before conducting a randomized clinical trial. As informed in the article, it was already known before COVID-19 that: ”1) The lethality of chloroquine in overdose is over six times higher than with other drugs [ref 116]. Mortality is proportional to peak blood concentrations; Outcome is dependent on the dose retained, the blood concentrations that result, and the delay in reaching supportive intensive care; 2) Hydroxychloroquine was slightly safer than chloroquine in preclinical testing and is considered better tolerated over the long term. Both drugs are dangerous when overdosed; 3) The main concern with high doses is cardiovascular toxicity. In overdose, a variety of arrhythmias have been observed, including sino-atrial and atrioventricular block, bundle branch block, and different ventricular arrhythmias (including TdP). The ECG commonly shows QRS widening and QT prolongation. 4) There is a significant contribution of azithromycin to arrhythmia risk [refs 52, 78, 108]. Individually, azithromycin use may carry a greater risk of TdP than chloroquine or hydroxychloroquine; 5) The main drug-drug interactions causing concern are pharmacodynamic interactions with other hERG channel blocking (QT prolonging) drugs—notably, azithromycin. If such drugs are added during or shortly after the 4-aminoquinolines have been given, then plasma potassium concentrations should be over 4 mmol/L, calcium and magnesium plasma concentrations should be in the normal range, and ECGs should be monitored for QT prolongation; 6) Hypokalaemia, resulting from intracellular accumulation, is a consistent feature, important complication and an indicator of prognosis of chloroquine poisoning, contributing to arrhythmias [ref 93]; 7) The plasma potassium concentration on admission correlated inversely with QRS widening and QT prolongation. Good intensive care with prompt management of hypokalaemia were important contributors to survival [refs 113–115]; 8) In patients with an admission whole-blood chloroquine concentration less than 15 μM, no deaths occurred. Mortality rose sharply at higher concentrations, with greater than 5% mortality for concentrations above 20 μM; 9) Sudden, unexplained death has not been associated with previously approved dosage regimens of chloroquine, despite administration of billions of prophylaxis and treatment courses and wide variation in dosing [refs 72, 73]; 10) Confusion and concern have arisen by extrapolating from the undoubted cumulative long-term risks of myocardial damage with chronic dosing to short-term exposures, overestimating the risk of ventricular arrhythmias resulting from moderate QT prolongation; 11) Chloroquine and hydroxychloroquine both have unusual pharmacokinetic properties with enormous apparent volumes of distribution (chloroquine > hydroxychloroquine) and very slow elimination from the body (terminal elimination half-lives > 1 month); 12) Monitoring for cardiovascular adverse events (QRS widening, QT prolongation, arrhythmias) and modifiable risk factors (i.e., plasma concentrations of potassium/calcium/phosphate/magnesium, severely impaired renal function and coadministration of drugs which prolong the QT interval) is advisable in patients receiving high doses of chloroquine or hydroxychloroquine and severely ill (hospitalized); 13) Although chloroquine and hydroxychloroquine do prolong the ECG J to T-peak interval and are potentially ‘torsadogenic’, the extent to which the risk of TdP is increased is unclear. Furthermore, reports of QT prolongation commonly omit measurement of QRS prolongation and thus overestimate JT prolongation; 14) The most common adverse reactions reported are dyspepsia, nausea, occasionally vomiting, visual disturbances (particularly transient accommodation difficulties), and headache [refs 58–62]. Pruritus is particularly troublesome in dark-skinned patients and may be dose limiting [ref 96]. 15) Orally administered chloroquine is well absorbed, even in unconscious patients. Accumulation occurs slowly with repeated dosing (as in rheumatological conditions or continuous antimalarial) because of the very slow terminal elimination rate. Higher levels are reached in patients with renal failure [refs 37, 40]”.
A critical and vital issue in the study of repurposed drugs in highly uncertain situations, as COVID-19 soon after its occurrence, is the selection of the drug concentration to be tested in randomized clinical trials. For hydroxychloroquine, the critical analysis quotes the study by Yao X, Ye F, Zhang M, Cui C, Huang B, Niu P, et al.. They found HCQ more potent than chloroquine by developing a physiologically-based pharmacokinetic (PBPK) model (published on March 9, 2020, before the publication of results of randomized clinical trials in respected journals). They suggested a loading dose of 400 mg twice daily of hydroxychloroquine sulfate given orally, followed by a maintenance dose of 200 mg given twice daily for 4 days to treat SARS-CoV-2 infection [total dosage 2.4g). This dosage regimen, according to them, reached 3 times the potency of chloroquine phosphate when given 500 mg twice daily 5 days in advance and was adopted in the RECOVERY study but for its duration (5 vs 10 days) and 3 times higher loading dose in the first day (800mg vs 2,400mg). Horby P, White NJ and Landray MJ, in a reply on the NEJM, stated that HCQ was evaluated primarily as an antiviral drug in the UK RECOVERY study. “The dosing schedule of hydroxychloroquine in the RECOVERY trial was therefore designed to provide the highest tissue concentrations that were safe in order to provide the maximum antiviral activity and thus the best chance of therapeutic benefit. These doses were based on pharmacokinetic modeling in malaria and rheumatologic conditions and were informed by safety thresholds determined from toxicokinetic data on intentional overdoses with the closely related chloroquine”.
In another paper (accepted March 25, 2020) by Cui C, Zhang M, Yao X, Tu S, Hou, En VSJ et al., the authors applied the same PBPK model above referred and studied the dose selection of chloroquine phosphate (CQ) for the treatment of COVID-19. According to the authors, “the novel PBPK model allows in-depth description of the pharmacokinetics of CQ in several key organs (lung, heart, liver and kidney), and was applied to design dosing strategies in patients with acute COVID-19”. After combining the PBPK model with known clinical exposure-response relationships, they proposed CQ dosages for acute COVID-19 (Day 1: 750 mg BID, Days 2–5: 500 mg BID, total dosage 5.5 g), moderate COVID-19 (Day 1: 750 mg and 500 mg, Days 2–3: 500 mg BID, Days 4–5: 250 mg BID, total dosage 4.25 g), and other vulnerable populations (e.g., renal and hepatic impairment and elderly patients, Days 1–5: 250 mg BID, total dosage 2.5 g ). It is noticeable that as the disease progress, or in vulnerable populations, the CQ total dose decreases in their fixed 5-day treatment course. This dosage regimen appears to fit well with known pathophysiological knowledge on viral diseases. The antiviral therapeutical action is expected to take place mainly in the initial replication phase. The early recognition of the progression of COVID-19 (viral replication/prodromal stage, pneumonia/inflammatory stage and cytokine storm/critical illness) would recommend testing primarily CQ and HCQ in the first 5 days after symptoms onset and not in later phases of hospitalized patients with a hyper-inflammatory immune response secondary to the virus aggression. The PLoS Medicine paper reasserts that “the earlier in the evolving disease process that pathogen multiplication is inhibited, the better is the outcome”. The effectiveness of any antiviral agent can vary if it is prescribed early, at the prodromal stage, or later, with inflammatory and immunopathologic phenomena.
Both WHO SOLIDARITY / UK RECOVERY (total dosage 9.6g of HCQ, equivalent to 7.44g of chloroquine base) and the Brazilian CloroCOVID19 trials (20g total dosage of CQ in the high dose group, equivalent to 12g of chloroquine base) adopted a flat 10-day prescription for hospitalized and severe patients with clinically suspected or laboratory-confirmed SARS-CoV-2 infection. It is well known that patients prescribed higher chloroquine dosages are at higher risk of overdose death, as shown in the CloroCOVID19 trial and the self-poisoning subjects studied by Clemessy and collaborators in France. The HCQ total dose used in RECOVERY/SOLIDARITY trials were 2.83 times superior to the estimated dosage by Yao X et al. study published almost simultaneously with the two protocols development. In the CloroCOVID19 high dose group (severe or critically ill patients), the total dose was double the suggested by Guangdong expert consensus, 2.85 superior to the guidelines proposed by the Health Commission of China, 4.44 higher than the dosage regimen permitted by the Brazilian Ministry of Health for hospitalized patients and 3.63 times superior to the highest dose recommended in Cui C et al. PBPK model for acute severe disease, 4.7 times superior to moderate COVID-19 and 8 times superior to their dosage for vulnerable patients.
It is unclear if the dosage regimens in the WHO SOLIDARITY / UK RECOVERY trials were defined after detailed pharmacokinetic-pharmacodynamic modelling. On the other side, it is clear that there is no report of a dosage simulation model in the CloroCOVID19 trial, only a reference to the dosage suggested by the Guangdong expert consensus. The consensus based their recommendations solely on in vitro studies and unpublished case series of patients seen in China. The total CloroCOVID19 dosage was 68% stronger than the one used in the WHO SOLIDARITY / UK RECOVERY trials, in case hydroxychloroquine and chloroquine have molar equivalent toxicities. About the high dosage in the CloroCOVID19 trial, it is stated in the article that “serious cardiotoxicity was encountered in a prematurely terminated COVID-19 trial in which the chloroquine malaria treatment loading dose of 10 mg base/kg was given twice daily for a week”. It needs a correction: the high-dose group in the CloroCOVID19 trial ingested 8 CQ tablets daily for 10 days (the loading dose for uncomplicated P. vivax malaria in Brazil is 4 tablets) without adjustment per weight. The patients in their high dose group would have to weigh 120kg to properly receive the 10 mg base/kg malaria treatment loading dose. One CQ treated patient in the CloroCOVID19 trial was less than 50kg. This high dosage could likely contribute to the aggravation of the disease instead of being therapeutical.
Extra caution should be used to determine dosage for converting chloroquine base to chloroquine diphosphate to avoid unintentional overdose caused by adverse effects of CQ that could lead to hypokalaemia, heart rhythm changes and deaths, intensified in patients with renal problems. On the same day (April 24, 2020) of the publication of the CloroCOVID19 trial in the JAMA Network Open (preceded by the April 12 interview to the NYT), the FDA announced that “we are aware of reports of serious heart rhythm problems in patients with COVID-19 treated with hydroxychloroquine or chloroquine, often in combination with azithromycin and other QT prolonging medicines”. Commenting on the paper, the second most cited in 2020 in JAMA Network Open, its editor stated that the discontinuation of the Brazilian trial was decided because the high-dose regimen’s risk of toxic effects outweighed its benefits and that “Pretty much after this, all randomized clinical trials of this were stopped.” It is weird that a small non-controlled trial, too full of bias, comparing two highly different dosage regimens of CQ as an adjuvant to other drugs that prolong QT interval in vulnerable Amazonian patients with suspected severe COVID-19, has gained so much acknowledgement from the scientific community. Are medical scientists doing blind clinical trials and also becoming blinded to its critical analysis and interpretation?
Besides the total dosage of CQ, it is also crucial to evaluate the clinical status of the patients and the level of care delivered in the hospitals. For the CloroCOVID19 trial, there were no exclusion criteria. For the UK RECOVERY trial, only patients with a known prolonged corrected QT interval on electrocardiography were ineligible to receive hydroxychloroquine and if, in the opinion of the attending clinician after taking the medical history, the patients were not exposed to substantial risk (not defined) if they were to participate in the trial. A critical inclusion criterion, informed on the registration at clinicaltrials.gov, was not disclosed in the CloroCOVID19 report on JAMA Network Open. This criterion only allowed the participation of patients aged <51 years if they had a known risk factor for COVID-19 or severe comorbidity (cancer, decompensated diabetes, systemic arterial hypertension, cardiovascular disease, decompensated kidney disease, pre-existing pulmonary disease, use of immunobiological drugs, use of immunosuppressive drugs and HIV/AIDS patients with CD4+ lymphocytes below 250).
According to the Social Vulnerability Index (SVI), based on urban infrastructure, human capital and income and work, Manaus was the metropolitan region with the worst results among the 16 studied by the Institute of Applied Economic Research in Brazil. Among other big cities in Brazil, its local health system has no good reputation. At the time of the CloroCOVID19 trial, the Hospital Delphina Aziz was the only hospital to admit COVID-19 cases in Manaus. The quality of health attention in 176 hospitals in the United Kingdom, where the RECOVERY trial was conducted, is not comparable to the hospital in Manaus, an overloaded hospital in a low-income setting. Nevertheless, the hospital protocol in Manaus for the treatment of suspected cases of COVID-19 included the use of azithromycin, ceftriaxone and oseltamivir, with the adjuvant use of chloroquine as an experimental drug. The CloroCOVID19 protocol did not systematically preview serum electrolyte analysis (especially potassium) at entry and as routine care in the following days after admission. There was no report of dropouts due to intense nausea, vomiting or diarrhoea as one would expect in trials using such a CQ high dosage regimen, particularly after the obligatory administration of another CQ dose, by protocol, after 30 minutes, in case of rejection or vomiting.
Except for the CloroCOVID19 trial in Brazil, Watson, Tarning, Hoglund, Baud, Megarbane, Clemessy and White concluded that “pharmacokinetic modeling predicts that most high dose regimens trialled in COVID-19 are unlikely to cause serious cardiovascular toxicity”. For the dosage prescribed in the Brazilian trial, the authors stated that it could be directly lethal, as suggested by a pharmacometrics analysis of chloroquine poisoning cohorts. However, it must be remembered that the simulation did not consider the damages of the SARS-CoV-2 in the cardiovascular system and associated comorbidities in fragile/elderly patients treated with a cocktail of other drugs. In the Brazilian trial, chloroquine diphosphate was administrated as adjuvant to azithromycin, ceftriaxone and oseltamivir in patients possibly using other drugs (e.g. digitalis, antidiabetics and diuretics) with significant and potentially lethal interactions. It could then result in hypokalaemia, acid-base imbalance, heart disorders, cardiac arrest and death. Therefore it is improper to include the findings of the Brazilian CloroCOVID19 trial as done in the meta-analysis of RCT published on mortality outcomes with HCQ and CQ in COVID-19 or to mix evidence from patients in different stages of COVID-19, as proposed by the WHO COVID-19 therapeutic guidelines, as if there was conclusive evidence on the effectiveness and safety of CQ/HCQ for the treatment of COVID-19 for all stages.
Dosis sola facit venenum”, stated Paracelsus in 1538. For drugs presenting a narrow therapeutic index, such as chloroquine, careful attention must be taken to find the limits of therapeutic and toxic doses to run a randomized clinical trial. However, in clinical judgment, the ethical rule of prudence imposes prescribing a not so strong dose of some powerful medicine to vulnerable and fragile patients facing a lethal disease in its final stage. The publication of the findings of the Brazilian CloroCOVID19 trial in the JAMA Network Open (April 24, 2020), which suggested a cardiotoxic effect of chloroquine diphosphate in 2 sudden death cases, and the retracted multinational registry analysis published in The Lancet (May 22, 2020), “impede conduct of the very randomized controlled trials needed to provide the robust evidence on risks and benefits” of chloroquine and hydroxychloroquine, as the impartial and balanced critical review published in PLoS Medicine rightly pointed out. Increased politicization and unsound science combined with unspeakable interests are accountable for making the fear triumph over the hope of doctors and patients to find effective, safe and cheap treatments for the prodromal phase of COVID-19.