Cyclophosphamide for Ocular Inflammatory Diseases

Discussion 

This report confirms the beneficial effects of cyclophosphamide therapy for ocular inflammation. Cyclophosphamide was successful in achieving complete control of inflammation in 49.2% and 76.3% by 6 months and 12 months, respectively. Similarly, corticosteroid-sparing success (sustained control of inflammation while tapering prednisone to 10 mg or less) was gained by 30.1% and 61.2% by 6 and 12 months, respectively. At or before 2 years after initiation of treatment, 63% were able to discontinue therapy because of disease remission. Other studies have suggested varying success rates of cyclophosphamide in treating different forms of ocular inflammation with small sample sizes, using different outcome definitions, which makes comparison between studies difficult.9, 22, 36, 37, 38, 39, 40, 41

However, the success in terms of control of inflammation observed in this study seems lower than in some of these prior reports,36, 38, 39 likely because of the more stringent definition of success, the inclusion of only patients observed from the initiation of therapy who did not have the benefit of treatment before the initiation of observation time, and perhaps publication bias (particularly for the small case series reported). Although the conservative success criterion, requiring documentation of success at visits spanning at least 28 days, might have resulted in a lower success rate, it is arguably a more satisfactory definition of success. Sensitivity analysis at which success in control of inflammation or corticosteroid sparing omitted the requirement for sustained success indeed improved success to levels more similar comparable with some of the above studies (76%). Although all centers participating in the study were tertiary centers, which tend to see more severe disease than less specialized centers, most other reports derive from tertiary centers as well. Nevertheless, all the available reports suggest that cyclophosphamide is effective for the control of most, but not all, patients with ocular inflammation.

Sixty-one patients discontinued cyclophosphamide after achieving remission at the rate of 0.32 remissions/person-year, which is lower than the 0.50/ person-year (95% CI, 0.37–0.67/person-year) rate in a report of an overlapping group of ocular pemphigoid patients,38 although the former reflects the rate of remission among all patients treated and the latter is the rate of remission only among the 82.9% subset of patients whose disease initially was controlled by cyclophosphamide. Thus, although the estimate of the remission rate is lower, it is unlikely that the result is different to a statistically significant degree. Two possible reasons why this study observed a lower remission rate include the possibility that some patients scored as discontinuing cyclophosphamide for toxicity might have gone on to have disease remission but not be counted as such and the exclusion of patients who had the benefit of starting cyclophosphamide therapy before cohort entry—who may have reached remission sooner. In addition, the definition of remission was based on the reason for discontinuation of therapy, rather than using a definition based on follow-up after discontinuation,38 because of constraints of the data available. In any case, the results suggest that the substantial majority of patients able to continue therapy are likely to achieve medication-free remission in 2 years or less, a much higher rate of remission than that observed in studies of methotrexate,42 azathioprine,43 mycophenolate mofetil (to be published separately), and cyclosporine (to be published separately).

Full doses in the range of 100 to 150 mg were more effective in controlling inflammation compared with doses of <100 mg, but were more likely to lead to dose-limiting toxicity, confirming that clinicians should use full dosing (1–2 mg/kg) whenever it can be tolerated.5 Guidelines about how to implement such treatment are available.2 The observation that black persons tended to be less likely to achieve corticosteroid-sparing success compared with white persons, but had no significant difference with respect to control of inflammation, may be a random effect or a true difference and requires confirmation by supplemental studies. Neither the site of ocular inflammation nor prior use of immunosuppressive therapy seemed to affect the likelihood of success with cyclophosphamide.

A consensus panel on immunosuppression for ocular disease concluded, based on previous available studies,41, 44 that pulsed cyclophosphamide therapy for uveitis is less effective than oral cyclophosphamide.3 A randomized clinical trial45 in Wegener's granulomatosis patients concluded that pulse cyclophosphamide was as effective as oral cyclophosphamide in achieving initial remission and was associated with fewer side effects and lower mortality. However, in the long term, treatment with pulse cyclophosphamide neither maintained remission nor prevented relapses as well as oral cyclophosphamide had. In the authors' experience, the likelihood of treatment success tended to be higher with oral cyclophosphamide, but not to a statistically significantly degree. Bladder toxicity and bladder cancer risk, some of the major toxicities of cyclophosphamide, may be reduced when the drug is administered intermittently via the intravenous route, compared with oral daily dosing.46, 47 Thus, although the available information suggests that oral administration may be more effective than intravenous administration of cyclophosphamide, considerations regarding the potentially lower risks of side effects with intravenous cyclophosphamide leave open the question as to which should be the preferred approach for ocular inflammation.

Although—given enough time—cyclophosphamide usually was successful in controlling ocular inflammation, a clinically important degree of side effects occurred, requiring discontinuation of therapy in a large minority of patients and leading to 7 (3.3%) opportunistic infections with 1 death. The most common side effects leading to drug discontinuation were leukopenia (18.1%) and cystitis or hematuria (7.7%). Various other studies have reported a higher incidence of the side effects in the range of 18% to 46% for leukopenia and 8% to 33% for hemorrhagic cystitis.25, 48, 49, 50, 51 These differences in results probably derive from the fact that only problems resulting in discontinuation of therapy were recorded. Gonadal dysfunction has been observed in 60% of the patients after 6 months of treatment with cyclophosphamide.52 In this study, only 1 patient discontinued because of sterility, although patients likely anticipated this risk when starting the medication. Ocular side effects including dry eyes, blurred vision, and rise in intraocular pressure have been noted,53 which this study did not address. Based on the authors' experience with opportunistic infections, they have adopted the routine use of trimethoprim-sulfamethoxazole prophylaxis in patients treated with cyclophosphamide. This approach is frequent, but not universal, among rheumatologists using cyclophosphamide for systemic inflammatory diseases.54

There is considerable evidence suggesting that cyclophosphamide increases the risk of certain kinds of malignancy, and perhaps the overall risk of malignancy.47, 55, 56 A study of ocular inflammation patients also demonstrated that cyclophosphamide is not associated with a statistically significant increase in overall mortality (adjusted HR, 1.14; P = 0.45), but found that cancer mortality tended to be higher with respect to unexposed cohort (adjusted cancer mortality HR, 1.61; P = 0.17) and the United States general population (cancer-specific standard mortality ratio, 1.42; P = 0.056).57 Thus, the results may be consistent with a clinically important increase in overall cancer mortality, as suggested by a minority of reports based on the clinical experience in other fields.47, 58, 59 These toxicity considerations suggest that use of cyclophosphamide should be limited to the most vision-threatening cases of ocular inflammation and to cases where associated systemic disease provides an indication for the use of cyclophosphamide.

Cyclophosphamide also is teratogenic, causing skeletal and central nervous system abnormalities.60, 61 Therefore, use of effective contraception during cyclophosphamide therapy is required. Cyclophosphamide also can be excreted in breast milk, suggesting that mothers of infants should not breastfeed if cyclophosphamide therapy must be used.60, 62

Limitations of this retrospective, observational study include potential indications-for-treatment bias, missing data in chart notes, incomplete follow-up, and potential referral bias. Alkylating agents typically have been reserved for severe cases, and it is possible that results would have been better if the drugs were used in milder cases. However, the side effect concerns suggest that limiting use to more severe cases is appropriate. The centers involved were selected in part because of their habits of maintaining complete records to minimize missing data problems. Data were collected by trained expert ophthalmologist reviewers as per protocol in all the centers63 to minimize ascertainment bias. Ascertainment of treatment success and side effects likely was good because the patients typically are assessed every 4 to 6 weeks with monitoring blood work more often than that at all centers, although occasional successes and adverse effects might have been missed. The survival analysis approach assumes that patients lost to follow-up are similar to patients continuing in the study; that patients starting a second immunosuppressive drug were censored might have resulted in a slight overestimation of benefits in the analyses of successes. Referral bias is a concern in studies from tertiary care centers, but these results should be generalizable to tertiary ocular inflammation centers, where aggressive immunosuppression with cyclophosphamide typically is managed.

Strengths of the study include the large size of the cohort, assessment of the effects of cyclophosphamide as a single agent to avoid ascribing effects from second agents to cyclophosphamide, observation of patients from the time of initiation of therapy, and the ability to compare oral and intravenous therapies with respectable statistical power. Uniform data collection was promoted by quality control checks within the data system and by using data centers with extensive ophthalmologic clinical experience.33 More comprehensive analyses were carried out than have been used by most prior reports; also, a more realistic measure of treatment success than some prior reports was used. In addition, sensitivity analyses were conducted by changing the treatment success criteria to assess the robustness of the results.

In summary, these data suggest that—given enough time—cyclophosphamide is effective for most patients with uveitis, scleritis, ocular MMP, and other forms of ocular inflammation. Cyclophosphamide also has the advantage of a high rate of medication-free remission after treatment, which in the authors' experience usually is not seen with alternative immunosuppressive treatments. However, the risk of side effects is substantially greater than that associated with alternative agents and requires very careful monitoring and possibly preemptive antiopportunistic infection prophylaxis.64, 65 This concern, along with the apparent increase in the risk of cancer after therapy, suggests that cyclophosphamide is best reserved for patients at high risk of substantial vision loss for whom other forms of treatment have failed or are unlikely to succeed. However, clinicians should not be hesitant to use cyclophosphamide in instances where underlying systemic inflammatory diseases require such therapy, where cyclophosphamide may be life saving. If tolerable, doses in the 100 to 150 mg/day range seem to be more likely to succeed than lower doses.