The prevalence of hydroxychloroquine retinopathy (HCR) is important to know because it provides the justification for screening the estimated 350 000 Americans who take the drug. No methodologically sound epidemiologic study of the issue has been done, but many estimates of prevalence have been made, and in the last 20 years their size has increased dramatically. From 1983 to 2003, the estimates were 0%–0.5% for patients taking ≤6.5 mg/kg of actual body weight/day for less than 6 years, and very low for longer durations of use. However, in 2010 Wolfe and Marmor estimated that the prevalence of HCR among those taking the drug for over 5 years was 1% and higher with longer-duration use. In 2014, Melles and Marmor reported a prevalence of 7.5% after 5 years of use, based on 10-2 visual field (VF) and spectral-domain optical coherence tomography (SDOCT) testing. When the definition of retinopathy has been based on multifocal electroretinography (mfERG), reported prevalences have ranged as high as 33%.
These increasing estimated prevalences do not imply a growing public health problem. There is no evidence that patients are losing vision more frequently now than in the past. Instead, the increased estimates reflect more sensitive ancillary testing for retinopathy. The 10-2 VF testing, SDOCT, fundus autofluorescence, and mfERG permit the diagnosis of HCR earlier than do the older methods of color vision testing and Amsler grid testing. As ancillary testing has changed, the diagnostic criteria for HCR have changed, so that the estimated prevalences over time cannot be compared. The estimated prevalences arising from the different methods of defining retinopathy parallel the reported relative sensitivities of the tests. mfERG is the most sensitive test, followed by 10-2 VF testing, followed by SDOCT testing.
Beyond the changing definitions of HCR, the estimated prevalences arise from studies with methodological flaws. Most were retrospective chart reviews that used nonstandardized definitions of retinopathy, provided no evidence supporting the reproducibility of the definitions in their application, and had large proportions of missing data, leading to selection bias. The single prospective study was based on static automated perimetry, but it used a version that tested the central 25 degrees of field with 78 test points rather than the central 10 degrees of field with 68 test points, as in the more widely used 10-2 VF. To properly determine the prevalence of HCR would require a population-based study that could be replicated by others, would use standardized examinations and prospective follow-up, and would have a prespecified definition of HCR with reproducibility in application. It is unlikely that a study will ever be done.
Although it is impossible to accurately measure HCR prevalence from published reports, it is straightforward to calculate the prevalence of toxic daily dosing. Unfortunately, it has consistently been high over time. Despite the knowledge for 3 decades that daily dosing greater than 6.5 mg/kg of ideal body weight (IBW)/day is associated with increased risk of HCR, treating physicians continue to prescribe daily doses associated with increased risk; reported rates of toxic dosing range from 12.8% to 74.7%. Were this rate to decrease, the rate of HCR would fall. In patients receiving nontoxic daily doses of hydroxychloroquine, the prevalence of retinopathy is so low that the Royal College of Ophthalmologists does not recommend routine screening. The case for universal screening, as is recommended in the United States, rests on the fact that toxic dosing is so prevalent. Statements that HCR cannot be prevented put an unrealistically pessimistic spin on the facts. Most cases of HCR can be prevented by paying attention to proper dosing. HCR in patients taking nontoxic doses does exist, but the number is minute.
It follows that ophthalmologists who screen for HCR understand that their most important task is to evaluate the patient’s daily dose as a risk factor for eventual retinopathy. Of the known risk factors—daily dose, cumulative dose, renal disease, liver disease, and age—daily dose is the only modifiable one. The key fact to keep in mind is that hydroxychloroquine is not stored in fat, but mostly in lean tissue. Therefore, the daily dose must be normalized by lean body mass. If the patient is asthenic, one uses the actual body weight (ABW) as the denominator. If the patient is not asthenic, the ideal body weight (IBW) based on height is used.
Ophthalmologists have used at least 7 algorithms for IBW in the past, and there is controversy over which one is best. Without getting bogged down in the controversy, as a practical matter any of the algorithms is probably adequate. If the patient is taking greater than 6.5 mg/kg/day based on the lesser of IBW and ABW, the ophthalmologist has a responsibility to communicate with the prescribing physician to determine if daily dosing can be reduced to a safer range. In most cases it can, and prescribing physicians are happy to make the change. In some cases, such as in cases of treating graft-vs-host disease or cancer chemotherapy, in which expressly toxic doses are used for life-threatening reasons, a reduction is not prudent, and more frequent screening examinations will be needed, and the patient informed of the increased risk of retinopathy.
As supplied, hydroxychloroquine pill size is not friendly for optimizing safe dosing. It comes as 200 mg pills and most patients encountered in the clinic are either taking 200 mg or 400 mg daily. Therefore if a safe dosage is calculated to be 325 mg, rarely is the patient taking 300 mg, which would be safe. Ophthalmologists have lobbied drug companies to provide hydroxychloroquine in other pill sizes (eg, 50 mg, 100 mg), and the companies would perform a service if they complied, but they have not. Absent that reform, to reduce dosing, an easy method is to exclude dosing for 1–2 days of the week—typically the weekend. Because of the long half-life of the drug, no important fluctuations in plasma concentrations are effected by this method. The average daily dose is calculated by taking the cumulative weekly dose and dividing by 7.
Properly carried out, screening for HCR and toxic dosing will lead to an adjustment of daily dosing as the most common response to detecting unnecessarily high daily doses before retinopathy occurs. Recommending cessation of hydroxychloroquine is a far less common response that is reserved for cases with retinopathy, not simply toxic dosing.
In 2014, it was suggested that one could dispense with IBW and simply dose based on ABW, but use a value of 5 mg/kg/d as the multiplier. Adopting this suggestion could endanger obese patients. For example, consider a patient who is 5 feet 1 inch and weighs 200 lb. Her IBW using the National Heart Lung and Blood Institute table is 127 lb and her maximal safe dose is 375 mg/day. According to the 5 mg/kg of ABW rule, she could be safely dosed at 454 mg/d. The prescribing doctor might feel safe giving her the usual 400 mg/d, but she would be in a toxic dosing range based on the IBW rule that has stood the test of 30 years’ usage. On the other hand, using the lesser of ABW and IBW would keep this patient’s dosing in a low-risk range. The median weight of populations for a given height is increasing over time, reflecting the global epidemic of obesity. A conversion factor of 5 mg/kg, based on ABW taken from a data set obtained in 2004–2009, progressively becomes outdated, whereas the conversion factor based on IBW does not.
Previous American Academy of Ophthalmology (AAO) guidelines have indicated the lack of need for yearly examinations through 5 years if the patient is low risk. However, toxic dosing puts the patient in a high-risk group, and yearly follow-up is needed. When all the risk factors are tallied for an individual patient (eg, age >60, toxic daily dosing, and duration of therapy >5 years), a large fraction of patients fall into a high-risk group, for which yearly screening visits are indicated. As a practical matter, most American ophthalmologists have been reluctant to accept the hiatus in screening visits even for low-risk patients; multiple studies document their usual practice of annual visits. This avoids the potential problem of the patient who received her baseline screening examination and then is lost to follow-up. It also allows for multiple opportunities to reduce the dosing to a level associated with lower risk at a time before the cumulative dose has climbed to a riskier level after 5 years of use.
Since the revised guidelines on screening for HCR were published by the AAO in 2011, the emphasis has been on newer forms of ancillary testing useful for detecting HCR at an earlier stage. This is welcome, but even more important is the ophthalmologist’s attempts to prevent HCR in the first place. In most cases, the goal is possible, is inexpensive, and will not compromise the efficacy of the patient’s treatment of autoimmune disease.
Appendix
