In a recent issue of the Journal of the American Medical Association ( JAMA ), Etminan and associates report in a case-control study that the use of oral fluoroquinolones is associated with a significantly higher risk of rhegmatogenous retinal detachment (RRD), with an adjusted rate ratio (ARR) of 4.5. Oral fluoroquinolones are one of the most commonly prescribed classes of antibiotics in the United States. The study has important potential ramifications, including: 1) increased patient referrals to retina specialists for evaluation of retinal complications of fluoroquinolone use; 2) decreased use of fluoroquinolones in patients otherwise at risk for retinal detachment; and 3) medicolegal risks for physicians and pharmaceutical companies. Given these dramatic potential implications, a critical appraisal of the clinical findings of this article is warranted.
Fluoroquinolones provide broad-spectrum coverage for a number of infectious diseases, including sinus/respiratory tract infections and genitourinary tract infections. The mechanism of action of fluoroquinolones involves inhibition of deoxyribonucleic acid (DNA) gyrase and topoisomerase IV, which are essential enzymes in bacterial DNA replication. The fluoroquinolones are bactericidal agents since drug binding to the enzyme-DNA complex blocks progression of DNA polymerase, ultimately leading to bacterial DNA damage and cell death. Common nonocular adverse effects associated with fluoroquinolone use include gastrointestinal toxicity, typically manifesting as mild anorexia, nausea, vomiting, abdominal discomfort, or diarrhea; and central nervous system toxicity, presenting predominantly as mild headache, dizziness, insomnia, or mood alterations. Other adverse effects include tendinitis and tendon rupture, prolongation of the QT interval, hypoglycemia and hyperglycemia, elevated transaminases and liver failure, rashes and other allergic reactions, and hematologic toxicity. Reported ocular adverse events include corneal perforation from topical use and toxic optic neuropathy.
Retinal toxicity and degeneration, but not RRD, have been associated with the use of oral fluoroquinolones in animals. Albino mice that were administered oral fluoroquinolones followed by 4 hours of ultraviolet radiation demonstrated histologic vacuolation of photoreceptor segments and swelling of retinal pigment epithelium (RPE) cells. These changes eventually resulted in atrophy and disorganization of retinal architecture. Intravitreal trovafloxacin resulted in dose-dependent injury to the RPE and photoreceptors in rabbits.
Retinal hemorrhages were a reported adverse event associated with the use of gemifloxacin. However, there have been no documented reports in the literature of RRD associated with the use of oral fluoroquinolones. In their JAMA article, Etminan and associates cite 1 case of unspecified retinal detachment associated with oral ciprofloxacin that was reported in the 1999 Health Canada Adverse Reaction Report . Sirbat and associates reported 3 cases of bilateral serous retinal detachments (not RRD) related to an older fluoroquinolone known as flumequine. All 3 patients had chronic renal failure and were being treated for urinary tract infections, which may have increased the amount of active drug in their circulation. Within 2 days of drug cessation, the patients’ vision returned to baseline, and their serous retinal detachments completely resolved within 5 days of drug cessation. The authors noted that a quinolone treatment (nalidixic acid) has been shown to result in transient bullae on young animals’ articular cartilage. The direct cytotoxic effect of quinolones on the cartilage leads to interstitial edema and subsequent bullae. It was postulated that a similar cytotoxic effect occurred within the retina, resulting in serous retinal detachments. It is important to consider that these serous retinal detachments are not rhegmatogenous (ie, not caused by a break in the retina).
In their case-control study, Etminan and associates utilized the British Columbia Linked Health Database to review a cohort of 989 591 patients who had visited an ophthalmologist between January 2000 and December 2007. Exclusion criteria included previous RRD, previous endophthalmitis, and previous vitreous biopsy with intravitreal injection of antibiotics. Ten age-matched controls were selected for each patient found to have a diagnosis of RRD. A total of 4384 RRD cases were matched to 43 840 control cases. The most common indications for oral fluoroquinolone use among the RRD patients were respiratory tract infections (77%) and genitourinary infections (9%).
The authors classified fluoroquinolone exposure as follows: current use (date of termination of fluoroquinolone overlaps with date of RRD diagnosis); recent use (date of termination of fluoroquinolone was 1 to 7 days before date of RRD diagnosis); and past use (date of termination of fluoroquinolone was 8 to 365 days before date of RRD diagnosis). The current use of fluoroquinolones was associated with a significantly greater risk of RRD (ARR 4.50), whereas recent (ARR 0.92) and past (ARR 1.03) users did not have a significant RRD risk, suggesting that RRD was an acute adverse effect. The absolute increase in rate of RRD with fluoroquinolones was 4 per 10 000 person-years; consequently, 1 additional RRD would occur for every 2500 fluoroquinolone users. The population-attributable risk of RRD in the United States with fluoroquinolones was 4%. Thus, if all fluoroquinolones were removed from the United States, RRD cases would decrease by 4%. The authors estimated that 1440 cases of RRD diagnosed annually in the United States could be attributed to fluoroquinolone use. As a quality measure, the authors tested the RRD risk in their cohort with oral β-lactam antibiotics and short-acting β-agonists. They found no significant association with RRD for either class of medication.
The most obvious strength of this study is its power, with close to 1 million patients in its cohort. Infections that require oral fluoroquinolones are very unlikely to be related to higher risk of RRD. Consequently, oral fluoroquinolones are unlikely to be prescribed for a significant number of patients who are already at high risk for RRD. The study seemingly points out an unappreciated risk to the millions of patients commonly using fluoroquinolones.
However, it is paramount to know whether the association of fluoroquinolone use with RRD was the only hypothesis tested in this study. If multiple hypotheses were tested, special consideration may be necessary to determine proper analysis of statistical significance. Thus, it would not be surprising to observe several positive associations with P value < .05 by chance alone when hundreds of different hypotheses are simultaneously tested in a large data set such as the one used in this study.
Assuming that the findings presented are accurate, should these data limit the use of fluoroquinolones in patients at risk for RRD? Given the results of the study, 1 in 2500 patients using fluoroquinolones will have an RRD that they may not have otherwise had. By way of comparison, 1 in 1000 patients using fluoroquinolones will suffer tendinitis or tendon rupture. Given the known benefits of treatment, this level of absolute fluoroquinolone-associated risk is likely worth taking.
Assuming that the findings presented are accurate, should asymptomatic patients at risk for RRD and on fluoroquinolones be screened by a retinal specialist? The incidence of RRD in very highly myopic eyes (between −15 and −30 diopters) was observed to be 1.2% over 36 months or roughly 0.4% over 12 months. Assuming very highly myopic subjects have an ARR of 4.5, the risk of RRD in these patients on fluoroquinolones would be 1.8%. Even for very highly myopic subjects, who are at increased risk for RRD by several orders of magnitude relative to the general population, using fluoroquinolones would have a low risk of RRD. Patients progressing to RRD would almost certainly have symptoms, since asymptomatic retinal lesions only rarely progress to RRD. Moreover, RRDs that are asymptomatic or subclinical are best left observed. Therefore, only symptomatic patients are likely to benefit from retinal evaluation, even in very high-risk cases.
In conclusion, these data are of great interest and generate a plausible hypothesis that needs to be further tested and confirmed in future studies. Fluoroquinolones have saved the lives of thousands of patients with a diverse array of infections, including sepsis, community-acquired pneumonia, ventilator-associated pneumonia, and intra-abdominal infections in whom the benefit/risk ratio is immeasurable. The use of fluoroquinolones is best restricted to those patients in whom there is a clear indication of moderate or severe infection requiring such treatment. However, the current clinical implications of the data discussed here are limited because of the low absolute risk of RRD observed, and the lack of independent observations confirming this association. We believe this study does not support altering the current standard of care for patients using fluoroquinolone antibiotics, whether or not they are otherwise at increased risk for RRD.