Predictors of toxicity with hydroxychloroquine and chloroquine use: A systematic review

ABSTRACT

Background: Hydroxychloroquine (HCQ) has attracted much attention especially during the COVID-19 pandemic. HCQ and closely related chloroquine (CQ) have known ocular and cardiac toxicity. However, although screening guidance now exists from the Royal College of Ophthalmologists for the former (RCOphth)1, little is known regarding predictors of both forms of toxicity. Objectives: To systematically explore the literature on predictors of toxicity limited to cardiac and ocular toxicity. Methods: A detailed search of the following databases was conducted PubMed, Medline, Embase, Web of Science, The Cochrane library, EMCare and Academic Search Premier. Studies addressing predictors of HCQ toxicity with relevant search terms used were included. Exclusion criteria were non-English articles, pre-clinical and paediatric studies. Three authors (SR, NC, PK) independently screened titles and abstracts for inclusion, ensuring each article was screened twice. Disagreement over inclusion was adjudicated by senior reviewers (EN, JG). Data extraction (SR, NC, PK) focused on predictors of toxicity. Results: The search strategy retrieved 3103 studies. 147 studies were included for data extraction, of which 92 were eventually excluded due to not identifying predictors (n=17), reviews (n=19), not ocular or cardiovascular toxicity (n=18), case reports (n=3), paediatrics (n=1), screening articles that focused on detecting retinopathy (n=30), article unobtainable at time of abstract submission (n=4), leaving 55 studies for full review. Studies addressing cardiac toxicity (n=16) included cohort, retrospective observational, a comparative pharmacovigilance, systematic monitoring protocol and a randomised control trial (RCT). The majority of these involved high-dose (>5mg/ kg/day) use for acute COVID-19 infection. The main significant predictors identified were Hydroxychloroquine (HCQ) use in combination with azithromycin (6/7 studies), cumulative dose (2 studies), pre-existing cardiovascular morbidity (2/3 studies) and prolonged baseline QTc (2 studies). Individual associations were also identified in the following longer treatment duration, daily dosage, increased age, male gender, severe COVID-19 infection, abnormal liver function tests (LFTs) and concurrent use of loop-diuretics. Regarding predictors of ocular toxicity (n=39), only one study was a RCT. The remainder were observational studies case-control, cohort, retrospective chart reviews and letters to the editor which included original patient data. Several predictors of retinopathy and maculopathy were examined in two or more studies and included duration of use (16/18 studies), daily dosage (7/13 studies), cumulative dose (11/14 studies), increased age (10/11 studies), body weight or BMI (3/5 studies), renal impairment (5 studies), HCQ blood levels (2 studies), keratopathy (2 studies) and tamoxifen use (2 studies). Sex (2 studies) and history of cataract surgery (2 studies) were not found to be predictors of toxicity. Only few studies performed regression analysis presenting odds and/or hazard ratios with confidence intervals. Conclusion: The most recognised predictor of cardiac toxicity was co-administration with azithromycin. In ocular toxicity, commonly cited predictors included those already recognised by the RCOphth1, as well as cumulative dose, increased age, weight considerations, HCQ blood levels and keratopathy. Further research is warranted on better characterising predictors of cardiac and ocular toxicity in patients on HCQ and CQ therapy.