The article in this issue by Browning examining the impact of the revised American Academy of Ophthalmology (AAO) guidelines for hydroxychloroquine screening raises important points with respect to the goals, cost effectiveness, and present effectiveness of screening practice. However, I take issue with the conclusion that objective screening tests (such as spectral-domain optical coherence tomography [SD OCT], multifocal electroretinography [mfERG], or fundus autofluorescence) are neither needed nor cost effective.
The AAO document illustrates how subjective visual field changes can be misread or ignored in early hydroxychloroquine toxicity, and Browning’s examples actually confirm this concern. This is why objective testing was recommended. However, I agree that we may be able to screen more efficiently while still following the basic principles of the AAO recommendations. The most important AAO message may be to add additional tests for verification in the case of suspicious findings. Effective screening not only should detect early toxicity, it also should ensure that patients continue a very good drug with low systemic toxicity. Browning implies that fields alone are sufficient for the diagnosis of toxicity, but his patients really make the opposite point.
Patient 1 may not be toxic at all, because the 24–2 field defects illustrated are far outside the typical 2- to 6-degree eccentricity range for hydroxychloroquine damage. The mfERG image shown for verification is of very poor quality (unreadable in my laboratory). An SD OCT image in fact is desperately needed for verification, as is a 10–2 field.
Patient 2 showed sensitivity losses on white 10–2 pattern deviation plots that mostly are at the margins (usually artifact). The mfERG image is ambiguous, and Browning discusses changes that are well outside the region of hydroxychloroquine toxicity. The OCT findings are subtle, and we are not given cube thickness values that may confirm early parafoveal thinning. Red field testing showed a probable ring scotoma in 1 eye, but the initial data argue more for confirmatory tests than against them.
Browning points out that very thin patients may be overdosed on the basis of ideal rather than real weight, but he used a National Heart, Lung and Blood Institute table that gives upper limits for acceptable weight. Preferable for ideal weight are these simple and widely used calculations: (1) women = 100 lbs + 5 lbs for each inch over 5 feet and (2) men = 110 lbs + 5 lbs for each inch over 5 feet. These ideal values are significantly lower, and it is rare to find lighter patients. However, a significant number of patients indeed are prescribed an excessive dosage (more than 6.5 mg/kg ideal weight), and ophthalmologists should educate other physicians about proper levels. Overdosing is a risk factor, because lowering the current dose in a long-term patient does not remove cumulative risk from past exposure. The screening checklist Browning proposes is a nice suggestion and may help some practices to remember to check height and to calculate dosage by ideal weight.
The risk of missing toxicity by using only one screening test is unknown. Multiple tests may be overkill for young patients with short use on a proper dosage, but valuable for higher-risk patients (who presumably are only a portion of Browning’s population). Furthermore, different tests are more sensitive for different patients, and the value of a second test is as important for ruling out toxicity (eg, when fields are suggestive) as for finding it. The risk of toxicity in long-term users rises to several percent, and at that level, societal gain justifies use of the most effective screening tools.
Browning notes, perhaps accidentally, that the use of ring ratios in the mfERG (R1/R2) is “less sensitive” to hydroxychloroquine toxicity than response densities. The truth is exactly the opposite. Since the AAO revision, I have continued to study patients with hydroxychloroquine toxicity to compare test methods and to look at the relative value of red versus white fields. My personal guidelines are changing with this experience. The AAO document is an excellent approach to screening and errs on the side of completeness to minimize errors or overlooked toxic cases. These practices can be modulated for efficiency and cost effectiveness, but only if and to the extent that practitioners are truly familiar with the test procedures.
This caveat is important. The AAO document illustrated missed fields, and Browning’s patients showed field losses that do not fit typical hydroxychloroquine patterns. General ophthalmologists performing screening fields need to recognize the pattern of hydroxychloroquine toxicity just as they do an arcuate scotomas or hemianopia. Damage is seen best in 10–2 white pattern deviation plots or red 10–2 fields as a loss of sensitivity 2 to 6 degrees off center. Similarly, specialists who perform OCT, fundus autofluorescence, and mfERG tests need to be skilled at performing and reading them. The mfERG should follow International Society for Clinical Electrophysiology of Vision principles and should look for relative signal weakness in parafoveal rings. SD OCT readers should recognize the specific region of early thinning, which on OCT typically is just 2 to 3 degrees off center.
Here are some personal comments, which respond to many of Browning’s concerns:
Baseline testing is mainly to rule out macular disease that could contraindicate use or could complicate screening. A small old scar or scattered hard drusen should not prevent hydroxychloroquine use and screening. I often omit baseline fields and almost never order baseline mfERG testing.
No testing is needed for the first 5 years for a routine patient. Many patients begin annual testing anyway to become accustomed to the procedure, which is not cost effective but may prevent patient loss to follow-up later.
I am trying a new simplified procedure for routine patients to save cost and time: obtaining annual undilated, nonphysician visits with 10–2 fields and SD OCT. We read the tests and call with results. I find OCT sensitive and less variable for many patients—and it often saves money because I can defer repeating a questionable field until next year if the OCT results clearly are normal.
I add mfERG at some point between 5 and 10 years and usually order it only every few years or when something is suspicious on other tests.
Field testing requires awareness of the hydroxychloroquine pattern of damage. I almost never stop the drug for fields alone, but any defects in the critical 2- to 6-degree area are taken seriously. White 10–2 testing must include pattern deviation plots. Red testing can be more sensitive than white, but is less specific. User experience is critical.
Screening annually is sufficient, except for rare very high-risk cases. With good screening, one should never see bull’s-eye retinopathy.
My experience so far with early recognition of toxicity has been that progression is mild for a year or so after stopping hydroxychloroquine and is stable thereafter. (Older literature showing severe progression after stopping largely was based on cases with visible bull’s-eye damage.)
I normally require at least 2 methods (1 objective) to confirm toxicity before stopping hydroxychloroquine. I want to be sure field defects are not subjective variation.