Long-term remission after cessation of interferon-α treatment in patients with severe uveitis due to Behçet's disease

Authors


Abstract

Objective

To retrospectively assess the development of visual acuity and the frequency and duration of relapse-free periods in patients who were treated with interferon-α (IFNα) for severe uveitis due to Behçet's disease (BD) and who completed a followup period of ≥2 years.

Methods

IFN alfa-2a was administered at an initial dosage of 6 million IU per day, then tapered to a maintenance dosage of 3 million IU twice per week, and finally discontinued, if possible. In case of a relapse, IFN treatment was repeated. Visual acuity at the end of followup was compared with visual acuity when ocular disease was in remission.

Results

Of 53 patients (96 eyes), 52 (98.1%) responded to IFN. In 47 patients (88.7%), IFN could be discontinued when the disease was in remission. Twenty of these 47 (42.6%) needed a second treatment course during a median followup of 6.0 years (range 2.0–12.6 years). Visual acuity improved or remained unchanged in 91 eyes (94.8%). Ocular disease was still in remission in 50% of the patients 45.9 months after cessation of the first IFN course. The relapse rate tended to be lower in women than in men. The BD activity score decreased significantly during followup, but long-term remission of nonocular BD manifestations was not achieved. However, since local treatments were sufficient, no systemic treatment was administered.

Conclusion

Our findings indicate that IFNα induces long-lasting remission in patients with severe ocular BD, resulting in a notable improvement in visual prognosis.

Behçet's disease (BD) is a systemic vasculitis of unknown origin manifesting mainly with oral or genital aphthous ulcers and skin lesions. Any organ system can be affected. Ocular involvement is frequent and occurs in ≥50% of patients. The most common ocular findings are a relapsing, mostly bilateral, posterior or panuveitis, often accompanied by a cystoid macular edema, and an occlusive retinal vasculitis, which may affect both arteries and veins. Without treatment, patients will become blind an average of 3.36 years after onset of the first symptoms (1).

The use of modern immunosuppressive drugs has undoubtedly provided an improvement in the management of BD. In a retrospective study, Hamuryudan et al (2) re-evaluated patients who previously took part in a double-blind, placebo-controlled trial of azathioprine for a mean of 94 months. They showed that early treatment with azathioprine tends to favorably affect long-term visual prognosis of the disease. The frequency of the emergence of blindness after the trial was 40% among the patients who had initially been randomized to receive placebo, compared with 13% among those patients who had received azathioprine (2).

However, it has been reported that despite immunosuppressive treatment, a loss of central vision frequently occurs after a certain period of time. In 1985, Pivetti Pezzi et al (3) studied 51 patients who were treated with steroids and/or chlorambucil. Ten years after the onset of symptoms, 50% of the eyes had a visual acuity of 20/200 or less (3). Between 1975 and 1985, BenEzra and Cohen (4) treated 70 patients with systemic steroids and chlorambucil, colchicine, or azathioprine. Irrespective of the treatment regimen, 74% of the eyes showed a visual acuity of 20/200 or less after 6–10 years (4). More favorable results have been achieved with cyclosporine. However, daily dosages >3 mg/kg have usually been used, which, in our opinion, are not tolerable for long-term use due to nephrotoxic side effects (5). In 1994, Whitcup et al (6) described 19 patients (37 eyes) who were treated with cyclosporine either alone or in combination with steroids. During a followup period of 51 months, 24.3% of the eyes had a decrease in visual acuity of more than 2 lines (6). This situation requires a change in treatment paradigm, from managing flares of uveitis to aiming at a long-lasting preservation of good visual function in BD patients by preventing relapses (7).

In open studies and case series, interferon-α (IFNα) has been shown to be very effective for treating severe ocular BD. Response and remission rates >90%, the reperfusion of occluded retinal vessels, and the regression of retinal neovascularizations have been reported (8–16). As a consequence, IFNα has been included in the European League Against Rheumatism recommendations for the treatment of BD as equivalent to tumor necrosis factor (TNF) antagonists (17).

In this retrospective study, we aimed to confirm and to better define the favorable effects of IFNα. We assessed the frequency of long-term remission, particularly after cessation of treatment and with special attention to prognostic factors such as sex, ethnicity, and HLA–B51, in patients who received IFNα as therapy for severe inflammation of the posterior eye segment. We also wanted to gain more detailed information about the visual prognosis in these patients.

PATIENTS AND METHODS

Patients.

Patients were treated in a tertiary interdisciplinary uveitis center that treats ∼1,000 new uveitis patients per year. On a regular basis, 110 patients with Behçet's uveitis are seen and treated per year; 20 of these are new patients. Choice of treatment differs according to the underlying disease or uveitis entity. For Behçet's uveitis (posterior or panuveitis or retinal vasculitis), a standardized algorithm is used, starting with corticosteroids plus azathioprine or cyclosporine. When this is not efficacious, patients are switched to IFNα, and if this is ineffective or induces intolerable side effects, TNF antagonists are introduced. In patients presenting with an acute retinal vessel occlusion or with retinal neovascularizations, we switch directly from corticosteroids to IFNα, since conventional immunosuppressants are considered to be less effective in such cases.

Among all patients with BD in the uveitis clinic, 40% are treated with azathioprine, 10% with cyclosporin A, 1% with anti-TNF drugs, and 49% with IFNα. Of the latter patients, those who were followed up for ≥2 years after the first dose of IFNα, irrespective of how long they received IFNα and irrespective of whether this treatment had to be prematurely discontinued due to lack of efficacy or to side effects, were included in this analysis. Patients also had to fulfill the criteria of the International Study Group for BD (18). Active posterior uveitis or panuveitis and/or retinal vasculitis in at least 1 eye had to be present. The patients were required to have undergone a previous treatment consisting of ≥1 conventional immunosuppressive drug for a minimum of 3 months and/or systemic corticosteroids in a dose of ≥1 mg/kg of prednisolone equivalent.

Administration of IFN alfa-2a.

IFN alfa-2a (Roferon-A; Hoffmann-La Roche) was administered at an initial dosage of 6 million IU per day subcutaneously for ≥2 weeks. (Patient 1 was first treated with IFN alfa-2b at an initial dosage of 18 million IU daily due to Kaposi sarcoma, which had developed under triple immunosuppression [19]). Afterward, depending on the course of the disease, IFN alfa-2a was tapered stepwise over several months (4.5 million IU per day, 3 million IU per day, 3 million IU every other day, 3 million IU 3 times per week) to a maintenance dosage of 3 million IU 2 times per week, and finally discontinued, if possible. To avoid an antagonization of the effect of IFN alfa-2a, which is immunomodulatory and not immunosuppressive and works via NFκB, which is blocked by, for example, corticosteroids, previous immunosuppressive agents were stopped the day before initiation of IFN alfa-2a. Systemic corticosteroids were reduced to a maximum dosage of 10 mg of prednisolone equivalent per day (7). In cases of extraocular symptoms, treatment with low-dose corticosteroids (5–7.5 mg of prednisolone per day) was maintained during followup. To minimize the typical flu-like symptoms that occur at the beginning of IFN therapy, patients also received acetaminophen (paracetamol) at a dosage of 500 mg 3 times per day. In case of a relapse of uveitis, treatment with IFN alfa-2a was reinitiated as described above.

Assessment of response to treatment.

Response and remission of intraocular inflammation were assessed using the uveitis scoring system described by BenEzra et al (20). Response was defined as a reduction of scores by 50% from their initial values, and remission was defined as scores becoming 0. For evaluation of the frequency and duration of relapse-free intervals after IFNα therapy, the time from discontinuation of IFNα until the occurrence of a relapse of uveitis, which was defined as the point at which it became necessary to reinstitute treatment (time range from the end of the previous course of IFNα treatment until the beginning of the next treatment course) was assessed.

Visual acuity was expressed in European decimals and then converted to the logarithm of the minimum angle of resolution (logMAR) for computing. To evaluate long-term visual development, visual acuity at the end of followup (assessed by the mean of visual acuity during the last 3 visits) was compared with the visual acuity during remission of ocular disease (defined as baseline), calculated as the mean of the visual acuity measurements obtained at the visits during the time period from 0.25–0.75 years after initiation of IFN treatment. According to the recommendations of the Standardization of Uveitis Nomenclature (SUN) Working Group, worsening of visual acuity during followup was defined as an increase of ≥0.3 logMAR (worsening of ≥3 lines) and improvement of visual acuity as a decrease of ≥0.3 logMAR (improvement of ≥3 lines) (21). Visual acuity within these borders was defined as stable.

For the assessment of disease activity in general, the BD current activity form (BDCAF) (online at http://www.behcet.ws/pdf/BehcetsDiseaseActivityForm.pdf) was used, but the scores for ocular disease were omitted, since we used the uveitis scoring system. The maximum score for BDCAF was 24 since, in contrast to the original description, we also used a visual analog scale (scores of 0–6) for the patient's and physician's general estimation of disease activity.

Statistical analysis.

The Kaplan-Meier method was used to estimate the duration of the first and second IFNα treatment courses as well as the frequency and duration of relapse-free intervals after discontinuation of IFNα. In addition, the effects of the risk factors ethnicity, HLA–B51 status, and sex on duration of the first IFN treatment course and time to relapse after cessation of first IFN course were explored by Kaplan-Meier curves, and hazard ratios (HRs) with 95% confidence intervals (95% CIs) were calculated using Cox regression. Proportional hazards models were fitted with 1 risk factor at a time (bivariate) and with all 3 risk factors together in a Cox multiple regression for the duration of the first IFN treatment. Observations on other durations were too few for such explorations. Some combinations of risk factors were so rare that interaction effects could not be considered in multiple regression.

The BD activity score is an ordinal trait that should be summarized by a median. Because the number of different values was not too small, the quantile–quantile plot of the log (normal) distributions showed a reasonable fit, the measurements were not feasible in a strict time schedule, and the relative change was to be reported, geometric means and ratios of geometric means were computed. The 95% CIs assumed normality of logarithms, as did the coefficients of variation (CVs; SD as a percentage of the mean).

RESULTS

Patient characteristics.

As of December 2009, a total of 68 patients followed up for a mean of 58.5 months (range 1–153 months) have been treated with IFNα for severe ocular BD. Of these, 53 consecutive patients with 96 affected eyes (47 right and 49 left eyes) were eligible for analysis. Forty-one of the patients (77.4%) were men, and 12 (22.6%) were women. Twenty-four patients (45.3%) were Turkish, 18 patients (34.0%) were German, and 11 patients (20.8%) were of other origin. Forty-one patients (77.4%) were positive for HLA–B51. Eye involvement was bilateral in 43 patients (81.1%). Extraocular manifestations of BD included oral aphthous ulcers in all patients (100%), skin lesions in 43 (81.1%), genital ulcerations in 30 (56.6%), arthritis in 19 (35.8%), epididymitis in 5 (9.4%), thrombosis or thrombophlebitis in 4 (7.5%), and central nervous system involvement in 3 (5.7%). Pathergy test results were positive in 5 patients (9.4%).

In general, pretreatment (over the entire course of the patient's medical history) included corticosteroids alone in 18 patients (34.0%), corticosteroids plus 1 immunosuppressive drug in 27 patients (50.9%), and corticosteroids plus 2 immunosuppressive drugs in 8 patients (15.1%). Immediate pretreatment during at least 3 months prior to initiation of IFNα is shown in Table 1. If IFNα was started without immediate pretreatment or directly after treatment with corticosteroids, this was due to the presence of either an acute vessel occlusion or neovascularizations of the retina. Conventional immunosuppressive agents are not considered sufficient to control such lesions. Median age at initiation of IFNα therapy was 31.1 years (interquartile range [IQR] 27.0–38.1 years; range 21.0–58.8 years). Median followup after initiation of IFN therapy was 6.0 years (IQR 3.8–7.2 years; range 2.0–12.6 years). The study period ranged from April 1994 (entrance of the first patient) until December 2007 (last analyzed visit).

Table 1. Immediate pretreatment during at least 3 months prior to initiation of IFNα treatment in the 53 patients with BD*
TreatmentNo. (%) of patients
  • *

    IFNα = interferon-α; BD = Behçet's disease; CSA = cyclosporin A; AZA = azathioprine; MTX = methotrexate.

CSA + AZA + steroids1 (1.9)
CSA + steroids16 (30.2)
AZA + steroids3 (5.7)
MTX + steroids2 (3.8)
CSA alone2 (3.8)
Colchicine alone1 (1.9)
Steroids alone20 (37.7)
None8 (15.1)

Treatment with IFNα.

In 52 patients (98.1%), ocular disease responded to IFNα therapy, and remission was achieved. Only 1 patient (patient 42) had to be switched to azathioprine after 3 months due to a lack of efficacy of IFNα. Another patient (patient 35) was switched to azathioprine after 50 months of IFNα treatment because of side effects (aggressiveness and depression). In 4 patients (patients 3, 31, 38, and 53), IFNα could not be discontinued due to relapses, and thus, at the end of the followup period, these patients were still receiving treatment (minimum duration 26.2 months). However, in 47 patients (88.7%), IFNα could be discontinued during remission of ocular disease. Twenty of these patients (42.6%) needed a second treatment course, which consisted of IFNα in 17 patients and a different immunosuppressive drug in 3 patients (cyclosporine in 2 patients and azathioprine in 1 patient), and was still ongoing at the end of the followup period in 8 patients. In 11 of the 17 patients (64.7%) who received IFNα during the second treatment course, IFNα could again be discontinued during remission. Only 4 patients received a third treatment course. The sequence of treatment courses and relapse-free intervals is shown in Figure 1.

Figure 1.

Sequence of treatment courses and relapse-free intervals in the 53 patients with Behçet's disease (BD) treated with interferon (IFN). Patients are shown in the order that they started treatment.

When patients who experienced a relapse (n = 20) and patients who did not experience a relapse (n = 27) after cessation of their first IFN course were compared, only slight differences were detectable regarding mean age at the start of treatment (32.9 years [range 21–58 years] in patients who experienced a relapse versus 31.9 years [range 21–50 years] in those who did not) and sex distribution (75% men and 25% women in the group that experienced a relapse versus 74% men and 26% women in the group that did not). The only difference was in the mean followup time, which was shorter in patients who experienced a relapse (23.4 months [range 2.4–107.2 months]) compared with patients who did not (41.6 months [range 2.8–107.5 months]).

The median duration of IFNα treatment in the responders was estimated to be 22.4 months (IQR 15.5–36.4 months; range 9.2–79.9 months) for the initial course (Figure 2A) and 17.0 months (IQR 12.8–20.9 months; range 3.6–34.1 months) for a second course if applicable (Figure 2B). Female patients discontinued IFN treatment at ∼1.5 times the rate that male patients did (Table 2). HLA–B51 status was not helpful in predicting the duration of IFN treatment (Table 2). Median duration of the first IFN course was estimated to be 23.1 months (IQR 19.0–45.3 months) in men, 17.1 months (IQR 13.6–31.8 months) in women (Figure 3A), 23.1 months (IQR 17.1–36.4 months) in HLA–B51–positive patients, and 19.0 months (IQR 14.4–58.2 months) in HLA–B51–negative patients (Figure 3B). The duration of the first IFN course seemed to be longest in Turkish patients, followed by patients of German and other origin (Table 2 and Figure 3C). This trend was confounded by the small number of women of non-German origin.

Figure 2.

Kaplan-Meier curves for nonstratified data. A, Probability that the patients with Behçet's disease (BD) were still receiving the first course of interferon alfa-2a (IFN alfa-2a) treatment at the indicated time points (n = 52 patients). B, Probability that the patients with BD were still receiving the second course of IFN alfa-2a treatment at the indicated time points (n = 17 patients). C, Probability that the patients with BD were free of a uveitis relapse at the indicated time points after discontinuation of the first course of IFN alfa-2a (n = 47 patients). D, Probability that the patients with BD were free of a uveitis relapse at the indicated time points after discontinuation of the second course of IFN alfa-2a (n = 11 patients). The pointwise 95% confidence intervals (broken lines) and the fitted log-normal survival curve are shown. Dots indicate times at which data were censored.

Table 2. HRs for the duration of the first course of IFN treatment*
Risk factorHR (95% CI)
  • *

    Hazard ratios (HRs) and 95% confidence intervals (95% CIs) were determined by Cox regression for 1 risk factor at a time (bivariate) and by multiple Cox regression with all 3 risk factors together. IFN = interferon.

Bivariate risk factor 
 Sex1.44 (1.01–2.0)
 Ethnicity 
  German–Turkish1.24 (0.58–2.8)
  Other–German1.17 (0.60–2.3)
 HLA–B511.04 (0.75–1.5)
Multiple Cox regression risk factor 
 Sex1.54 (1.06–2.1)
 Ethnicity
  German–Turkish1.26 (0.57–2.9)
  Other–German1.51 (0.74–3.2)
 HLA–B511.01 (0.70–1.5)
Figure 3.

Kaplan-Meier curves for stratified data. A–C, Probability that the patients with Behcet's disease (BD) were still receiving the first course of interferon alfa-2a (IFN alfa-2a) treatment at the indicated time points, stratified by sex (A), HLA–B51 status (B), and ethnicity (C). D, Probability that the patients with BD were free of a uveitis relapse at the indicated time points after discontinuation of the first course of IFN alfa-2a, stratified by sex.

Relapse-free intervals.

As shown in Figure 2C, 18.7 months after cessation of the first IFNα course, ocular disease was in remission in an estimated 75% of the patients. After 45.9 months, disease was in remission in an estimated 50% of the patients, and after 107.2 months, an estimated 25% of the patients still had not experienced a relapse. Figure 2D demonstrates that 8.4 months after cessation of a second course of IFNα, disease was in remission in an estimated 75% of the patients. After 27.7 months, the probability of being relapse free was 50%. Women tended to have a lower relapse rate than men. As shown in Figure 3D, 50% of the male patients were estimated to be relapse free 45.2 months after cessation of the first course of IFN treatment. However, 50% of the female patients were estimated to still be relapse free at 107.2 months.

Visual acuity.

From baseline to end of followup, visual acuity either remained stable or improved in at least 1 eye in 48 patients (90.6%) and worsened in at least 1 eye in 5 patients (9.4%). Altogether, 91 of the 96 affected eyes (94.8%; 43 right and 48 left eyes) demonstrated improvement or maintenance of visual acuity. A worsening of visual acuity was registered in only 5 eyes (5.2%; 4 right eyes and 1 left eye) (Figure 4). No correlation between the proportion of eyes in which disease worsened and the duration of followup was detected.

Figure 4.

Comparison of visual acuity (VA) during remission of ocular disease (baseline) with visual acuity at the end of the followup period (final visual acuity) in the right eyes (A) and left eyes (B) of patients with Behçet's disease treated with interferon. The gray lines border the area of ±0.3 logarithm of the minimum angle of resolution (logMAR), in which development of visual acuity is considered to be stable. Triangles represent individual eyes.

Median (mean) visual acuity of all affected eyes was 0.30 (0.53) logMAR (Snellen: 20/40 for median, 20/68 for mean) at the initiation of IFNα treatment, improved to 0.09 (0.26) logMAR (Snellen: 20/25 for median, 20/36 for mean) at baseline, and was 0.07 (0.28) logMAR (Snellen: 20/23 for median, 20/38 for mean) at the end of the followup period. A visual acuity of 20/200 (1.0 logMAR) or worse was present in 25 eyes (26.0%) before IFN treatment, in 8 eyes (8.3%) at baseline, and in 12 eyes (12.5%) at the end of the followup period. Due to pre-existing irreversible retinal damage, 7 of these 12 eyes had a visual acuity of 20/200 or worse at baseline. In 3 eyes, a cataract or secondary cataract had developed, for which no surgery was performed during followup. The remaining 2 eyes belonged to patients who were receiving IFNα treatment at the end of the followup period due to a recent relapse of uveitis. However, not a single eye with a good initial visual acuity developed irreversible loss of vision during the followup period.

BD activity score.

The geometric mean of the BD activity score decreased from 11 (range 4–21) in the first month to 5 (range 3–12) after 1 year (±1 month). BD activity score fell by 30% (CV 40% [95% CI 22–37%]) from the geometric mean of the first 3 months (9.5) to the geometric mean of months 4–9 (6.8). It fell by a further 16% (CV 36% [95% CI 8–24%]) to the geometric mean of months 10–24 (5.7).

Adverse events.

Adverse events related to IFNα treatment occurred in the following frequencies: flu-like symptoms (100%; n = 53), mild leukopenia (>2,000/μl) (100%; n = 53), fibromyalgia (9.4%; n = 5), depression (7.5%; n = 4), hair loss (7.5%; n = 4), thrombocytopenia (>100,000/μl) (5.7%; n = 3), headache (3.8%; n = 2), increase in liver enzyme levels (<2 times the upper limit of normal) (3.8%; n = 2), weight loss (3.8%; n = 2), aggressiveness (3.8%; n = 2), psychosis (3.8%; n = 2), thyroiditis (3.8%; n = 2), fever (3.8%; n = 2), psoriasis (1.9%; n = 1), fatigue (1.9%; n = 1), loss of appetite (1.9%; n = 1), nausea (1.9%; n = 1), and myalgia/arthralgia (1.9%; n = 1). IFN was discontinued due to adverse events in 4 patients with BD in remission (3 due to depression and 1 due to myalgia and arthralgia), necessitating no substitute treatment. In 1 patient, IFN was discontinued because of aggressiveness, and azathioprine was used as a substitute. A total of 5 patients (9.4%) discontinued IFN treatment because of adverse events.

DISCUSSION

Although BD represents a multisystemic disorder, for many patients, the ocular involvement impairs quality of life more than most other manifestations of the disease. This can be explained by the poor visual prognosis, especially if the inflammation affects the posterior eye segment. Undoubtedly, the introduction of modern immunosuppressive agents has led to an improvement in visual prognosis in BD. However, a substantial proportion of patients with BD cannot be treated sufficiently with these drugs.

Since 1986, when Tsambaos et al (22) reported for the first time the successful treatment of 3 patients with predominantly mucocutaneous lesions, many cases, case series, and small open studies have been published demonstrating excellent efficacy of IFNα in the treatment of BD and especially of acute episodes of severe uveitis. So far, 1 randomized study of IFNα treatment in BD has been conducted (23). In that placebo-controlled, double-blind study that included 50 patients without retinal vasculitis, a significant superiority of IFN over placebo was shown for mucocutaneous and skin lesions, as well as for arthritis. The frequency of ocular attacks was reduced (23).

Only scarce information is available concerning the beneficial effect of IFNα on long-term visual prognosis in BD. Recently, 2 studies that address this topic have been published (24, 25). In a retrospective study by Krause et al (24), 45 patients who received IFNα for ocular BD were followed up for a mean of 6.67 years (range 0.3–22.3 years). Nine patients (20%) were able to stop IFNα after a mean treatment period of 33 months (range 8–67 months). Those patients were then relapse free for a mean of 37 months (range 2–113 months). Ninety-one percent of the affected eyes showed improvement in or stabilization of visual acuity during followup (24). Gueudry et al (25) reported on a series of 32 patients treated with IFNα for ocular BD who completed a mean followup of 66.4 months (range 16.1–129.2 months). Disease responded to IFN in 28 patients (87.5%). In 19 of these 28 patients (68%), IFN could be discontinued after a mean period of 32 months. Thirteen of the 19 patients (68%) were then free of relapses for a mean of 43 months (range 11–84 months). After 2 years of treatment, visual acuity was stabilized or improved in 87.5% of eyes (25).

We conducted a retrospective study of 53 patients with BD who were treated with IFNα for sight-threatening uveitis. To our knowledge, this represents the largest case series so far that has been analyzed for long-term effects of IFNα therapy. With a median followup period of 6.0 years, the observation period in our series was comparable with those reported by Krause et al (24) (mean 6.67 years) and Gueudry et al (25) (mean 66.4 months). When visual acuity at baseline was compared with visual acuity at the end of the followup period, 94.8% of the affected eyes in our series revealed stabilization or improvement in visual acuity based on the SUN recommendations. This rate was very similar to the rates reported by Krause et al (24) (91%) and Gueudry et al (25) (87.5%).

In this study, baseline visual acuity was defined as the visual acuity measured after remission of intraocular inflammation was achieved and not, as sometimes seen, as transiently impaired visual acuity during the acute phase of inflammation when IFN therapy has been initiated. A final visual acuity of 20/200 (1.0 logMAR) or less was present in 12 eyes (12.5%). In 10 eyes, this was due to preexisting irreversible retinal damage or to reversible visual impairment caused by cataract formation during followup. Thus, in only 2 eyes (2.1%) can the final absence of useful vision be related to an unfavorable disease course during followup despite IFNα therapy.

In our series, disease responded to IFNα therapy in 98.1% of the patients, whereas Gueudry et al (25) reported a response in only 87.5% of their patients. Moreover, we were able to discontinue IFNα after the first course in 88.7% of responders, compared with 68% of responders for Gueudry et al (25) and 20% of all patients for Krause et al (24). However, with a median duration of 22.4 months, the initial IFN course was shorter in our study than in the studies by Krause et al (24) (mean duration 33 months) and Gueudry et al (25) (mean duration 32 months). The median relapse-free period after cessation of the initial IFNα course was estimated to be 45.9 months in our patients compared with a mean of 37 months and 43 months in the series by Krause et al (24) and Gueudry et al (25), respectively. Using a data table and a Kaplan-Meier curve shown in the articles by Krause et al and Gueudry et al, we estimated the median relapse-free periods in those studies and concluded that the median relapse-free period was shorter in one series (24) and longer in the other (25) than the median reported in our series.

The higher number of responders, as well as the higher rate of patients who were able to discontinue IFN therapy despite a shorter treatment duration, in our cohort of patients than in the studies by Krause et al (24) and Gueudry et al (25) could be related to different treatment regimens. Whereas Krause et al and Gueudry et al used IFN alfa-2a at an initial dosage of 3–6 million IU 3 times per week, we used a higher initial dosage of 6 million IU every day. In addition, we reduced the prednisolone dose very quickly to a maximum of 10 mg, whereas both Krause et al and Gueudry et al used higher doses of corticosteroids in parallel during the first few weeks of IFNα treatment. In 2004, Kötter et al (26), who performed a review of the literature concerning IFN therapy for BD, indicated that long-term remissions of ocular disease are more likely associated with high initial doses of IFN than with long treatment durations. It also has been postulated that the effect of IFNα could be antagonized by high doses of corticosteroids (7).

We focused for the first time on long-term followup with regard to relapses after cessation of the initial IFNα course. Of the 47 patients in whom IFNα treatment could be discontinued during remission of ocular disease, 20 patients (42.6%) needed a second treatment course due to a relapse in at least 1 eye. Seventeen patients again received IFN alfa-2a. With a median duration of 17.0 months, this second IFN course was clearly shorter than the initial one. Of the 17 patients taking IFN, 11 (64.7%) were again able to stop taking IFNα while their disease was in remission, followed by a second relapse-free period with a median of 27.7 months. This means that if a patient with ocular BD experiences a relapse after cessation of the first IFNα treatment course, reinstituting this therapy is highly recommended because there is a >50% chance that the patient will be able to stop taking this second course of IFN when their disease is in remission and that disease will remain in remission for >2 years. We decided against continuing conventional immunosuppressives (e.g., azathioprine) for maintenance of remission after IFN treatment was completed because almost 60% of patients would probably never experience a relapse and would be unnecessarily exposed to potential treatment side effects.

In a few of our patients, we were unable to discontinue IFNα treatment, but uveitis remained in remission when the patients received a maintenance dose of IFN. At the end of the followup period, 4 patients (7.5%) were still taking the initial course of IFN, 2 of them for >5 years (IFNα dosage 3 million IU 2–3 times per week). Gueudry et al (25) reported that IFN treatment was still ongoing after 5 years in 6 patients in their series. Thus, it appears to be possible for patients to continue IFN therapy for a very long time if this is the only way to maintain remission and if no problems arise in terms of tolerability and side effects, which must be evaluated regularly in an interdisciplinary cooperation between the ophthalmologist and the rheumatologist.

Whether factors such as sex, ethnicity, or HLA–B51 positivity are predictive of the course of BD has been discussed often. In our study, female sex represented a positive prognostic factor. In women, the duration of IFN therapy was shorter and the relapse rate after cessation of IFN therapy was lower than that in men. In addition, the duration of IFN therapy seemed to be longer in patients of Turkish origin than patients of German or other origin. However, this may be due to the higher number of men in the Turkish patient group (men tending to have longer treatment durations and more relapses than women). In contrast, such differences could not be found for HLA–B51 status, which may have been confounded by the low number of HLA–B51–negative patients in our cohort. Consistent with our results, Sakamoto et al (27) described female sex as a good prognostic factor for maintenance of vision in an analysis of prognostic factors related to vision in a cohort of 52 patients (27). Female sex is associated with a good prognosis for almost all disease manifestations of BD. In contrast, young male patients have the most unfavorable overall prognosis (28).

With regard to HLA–B51 as a possible prognostic factor, Krause et al (29) recently demonstrated an association of HLA–B51 with the occurrence of uveitis, but not with its severity. Lewis et al (30) did not find any association between ethnic origin and disease expression in a meta-analysis. This supports our view that the apparently longer treatment duration in Turkish versus German patients was probably due to a confounder. Due to the limited number of patients, we were able to analyze these factors for the first course of IFNα treatment only.

The extraocular manifestations, which were assessed by the BDCAF, also improved but did not remit completely, which was expressed by the score not reaching 0. This was mainly due to recurrent oral aphthous ulcers and skin lesions (mainly papulopustular), whereas arthritis entered remission in all patients receiving IFN treatment (data not shown). This is consistent with the results of our pilot study and of the meta-analysis from 2004 (13, 26).

Adverse events were frequent, but mostly mild and dose dependent. They were consistent with the percentages and types of adverse events reported for IFNα treatment of BD and other diseases (such as hematologic and infectious diseases). Only 5 (9.4%) of the patients discontinued IFNα prematurely due to adverse events; in 4 of these patients, disease was in remission when IFNα was discontinued, and no other therapy was needed (26, 31). According to the internal treatment algorithm that has been used in the uveitis center since 1990, TNF antagonists, which became available for the treatment of ocular BD beginning in 1998, were only used in a minority of the patients with BD. TNF antagonists were used in those patients in whom IFN was not efficacious or who had contraindications for IFN, since IFN was effective in most patients and could even be discontinued without relapses in a large proportion.

Our study was limited by its retrospective nature. However, this was counterbalanced by the fact that all patients had to fulfill the same eligibility criteria and were treated according to a uniform and locally standardized IFNα regimen.

In conclusion, IFNα improves the long-term visual prognosis in patients with severe ocular BD in a manner that has not been observed with any other agent. This is a result of the capability of IFNα to induce and maintain very long-lasting remissions in a high percentage of patients, even after its discontinuation and repeated treatment courses. Such treatment-free long-term remissions have not been observed for cyclosporine or azathioprine, which are the standard immunosuppressive agents used in ocular BD with posterior uveitis or retinal vasculitis, nor for the TNF antagonists. In the case series examining TNF antagonists that have been published to date, relapses occurred ∼8–12 weeks after their discontinuation (32–34). Our findings also indicate that men treated with IFN appear to have more relapses and longer treatment durations than women, but no influence of HLA–B51 positivity or ethnicity on response to treatment was found.

AUTHOR CONTRIBUTIONS

All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Dr. Deuter had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study conception and design. Deuter, Zierhut, Vonthein, Kötter.

Acquisition of data. Möhle, Stübiger.

Analysis and interpretation of data. Deuter, Vonthein, Kötter.

Acknowledgements

We thank Professor Hasan Yazici for his valuable critical comments, fruitful discussions, and knowledgeable advice during data analysis and writing of the final manuscript.

Ancillary