Thalidomide in the Treatment of the Mucocutaneous Lesions of the Behcet Syndrome

Methods

Patients

The multidisciplinary Behcet's Syndrome Research Center at the Cerrahpasa Medical Faculty at the University of Istanbul in Turkey was established 20 years ago and now has approximately 3000 registered patients. Consecutive patients attending this center were recruited into our study between October 1993 and April 1996.

Only male patients who were 18 to 35 years of age and met the international criteria [16] were enrolled. The presence of active mucocutaneous disease, defined as the occurrence of at least two episodes of oral or genital ulceration within 3 months before study entry, was required for inclusion.

Exclusion criteria were the presence of moderate or severe eye disease resulting in any decrease of visual acuity, any organ involvement requiring immunosuppressive therapy, previous use of immunosuppressive therapy, and the presence of clinical neuropathy. The institutional ethics committee of the Cerrahpasa Medical Faculty approved the study protocol and the informed consent form.

Randomization

A simple, computer-generated, random-number list was prepared by a person not involved in the trial. The code was kept in an opaque, sealed envelope by the senior author of the study and was opened only after all data had been entered into a computer for analysis. Consecutive patients who attended the dedicated multispecialty clinic and were considered eligible for inclusion by the other physicians in the clinic were screened by the same rheumatologist for the inclusion and exclusion criteria. Before study entry, each candidate was informed about the design, purpose, and duration of the study and about the potential hazards and benefits of thalidomide. Patients who gave written consent and had normal results on a neurologic evaluation were enrolled.

At each visit, patients received three bottles containing either 100-mg thalidomide tablets [Grunenthal GmbH, Aachen, Germany] or placebo tablets identical in appearance to the thalidomide tablets. From each bottle, as instructed on the affixed labels, the patients took one tablet in the morning and two tablets in the evening each day. This resulted in three treatment groups: 1) patients receiving thalidomide, 300 mg/d [each bottle contained 100-mg thalidomide tablets]; 2) patients receiving thalidomide, 100 mg/d [the morning bottle and one of the evening bottles contained placebo tablets and the second evening bottle contained 100-mg thalidomide tablets]; and 3) patients receiving placebo (each bottle contained placebo tablets). All physicians involved in the study, including the rheumatologist who assessed the side effects, were blinded to study group assignments.

Patients received the assigned study medication until one of the following occurred: emergence of active systemic disease requiring immunosuppressive therapy, emergence of polyneuropathy, noncompliance with the study medication, or any request for withdrawal from the study. At each visit, unused tablets were collected and counted to determine compliance with treatment. Patients were permitted to use topical lidocaine (Calgel, Glaxo-Wellcome, Turkey) for pain relief when needed.

Follow-up and Data Collection

After enrollment (week 0), patients were seen every fourth week and whenever they needed to visit the clinic for a problem. An additional visit was held 4 weeks after treatment ended (week 28). For most visits, all patients were seen by the same dermatologist, ophthalmologist, rheumatologist, and neurologist. For data collection, only the number of new lesions was considered. Oral ulcers were divided into minor (<5 mm in diameter), major (>5 mm in diameter), and herpetiform (occurring in crops and with sizes ≤ 1 mm). For the number of follicular lesions, the following scoring system was used: no lesions = 0; 1 to 5 lesions = 3; 6 to 15 lesions = 10; and more than 15 lesions = 16. A separate record was made for lesions that could have occurred and healed between two consecutive visits. These data were collected through direct questioning of patients at each visit. A detailed ophthalmologic examination that consisted of a slitlamp examination, tri-mirror fundus-lens ophthalmoscopy, and measurement of visual acuity using a 10-scale chart (best vision, 10/10) was done at each visit and whenever a patient returned to the clinic with an eye problem. Eye activation was defined as the emergence of any of the following: cells or protein in the anterior chamber or vitreous, macular edema, or vasculitic infiltrations in the retina. Arthritis was defined as the presence of pain and swelling in a joint or the patient's verbal description of these symptoms. At each visit, all patients underwent neurologic examination that emphasized sensory findings. Electroneuromyography was offered to every third patient.

Side effects were monitored by asking patients a predefined list of questions at each visit. The questions were about sedation, fatigue, constipation, sleeplessness, sensation of numbness or tingling at the extremities, skin rashes, and headache. Any additional symptoms were also noted.

Outcome Measures

The primary outcome measure, complete response, was defined as the absence of any oral or genital ulcer of any size during the 24-week treatment period. Other measures were 1) changes in the number of mucocutaneous lesions and 2) response of eye disease to treatment, defined as the absence of uveitis activations and any decrease in visual acuity in either eye. All outcome measures except for eye involvement were based on the combination of the data obtained by the physicians at clinic visits and the data reported by the patients on the occurrence of such lesions between the visits.

Statistical Analysis

Assuming a 50% response rate in either thalidomide group and a 15% response rate in the placebo group, we chose a target sample size of 33 patients in each group. This provides a power of 80% and an α level of 0.05. All analyses were done by the intention-to-treat principle. Thus, patients who had at least one visit after the beginning of treatment were analyzed for the outcome measures, and the dropouts were considered nonresponders.

Differences in all noncontinuous outcome measures (complete response rate, proportions of eye activations, and decreases in visual acuities) among the three groups, as well as pairwise comparisons, were analyzed by using the two-tailed Fisher exact test. A Kaplan-Meier plot was developed for the time-dependent distribution of the primary outcome measure. Time zero was the beginning of treatment (week 0), at which point all patients were considered responders. Continuous variables were compared by using the Kruskal-Wallis and Wilcoxon rank-sum tests. Data were analyzed by using SPS software, version 6.1 for Windows (SPS, Inc., Chicago, Illinois).

Role of Study Sponsor

The biometrics department of the sponsor, Grunenthal GmbH, assembled and analyzed the collected data in Aachen, Germany. The investigators verified, reanalyzed, and interpreted the data at Cerrahpasa Medical Faculty, Istanbul, independently of the sponsor. The investigators had complete freedom over the content of the paper, but the final version was discussed in a consultative capacity with the sponsor before it was submitted to the journal of the investigators' choice. One of the authors is a Grunenthal employee and participated in the development of the study protocol.

Results

Patients

When the medication arrived from the manufacturer, one package was missing for each study group. Of the 101 patients eligible for the study, 5 declined to participate after reading the consent form. Ninety-six patients, divided equally among the groups, entered the study. The groups were not significantly different at baseline (Table 1). Twelve patients withdrew from the study prematurely, 7 (3 in the placebo group, 2 in the 100-mg/d thalidomide group, and 2 in the 300-mg/d thalidomide group) because of disease activity and 4 (1 in the 100-mg/d group and 3 in the 300-mg/d group) because of side effects. One placebo recipient left the study for an unknown reason in the fifth week. Three patients, one from each group, received short courses of corticosteroid therapy during the trial. All 3 were considered nonresponders for the primary outcome measure.

We used data on 95 patients for statistical analyses because 1 patient from the 300-mg/d group, who withdrew from the study because of excessive sedation at the first visit, had no lesions to measure during the trial up to that point. Compliance rates for thalidomide or placebo use, as calculated from the returned pill counts, were similar in all three groups (placebo group, 92.5%; 100 mg/d group, 92.4%; and 300 mg/d group, 91.3%).

Treatment Outcomes

Complete Response

According to the data obtained by the physicians at clinic visits and data collected from patient reports on the occurrence of ulcers between visits, no placebo recipient had persistent absence of any oral or genital ulceration. Two patients in the 100-mg/d group and 5 patients in the 300-mg/d group had persistent absence (P = 0.031; Table 2). Paired comparisons showed a significant difference between the placebo and 300-mg/d groups (P = 0.024). According to a survival analysis, most treatment failures occurred within the first 4 weeks of the trial (Figure 2).

Frequency of Mucocutaneous Lesions

Major or herpetiform oral ulcers occurred in a few patients and as a single-time event; thus, a formal analysis could not be done. The mean numbers of minor oral ulcers were significantly lower at week 4 in both thalidomide groups than in the placebo group (P < 0.001); this decrease persisted thereafter with no significant changes between the two thalidomide groups. The suppressive effect of thalidomide in both groups reached statistical significance for genital ulcers (P < 0.001) and follicular lesions (P = 0.008) at week 8; again, the two groups did not differ significantly. In contrast, the evaluation done 4 weeks after treatment ended showed increased numbers of mucocutaneous lesions in both thalidomide groups; thus, the difference compared with the placebo group was statistically significant only for follicular lesions (P = 0.03).

Thalidomide treatment resulted in significant increases in the frequency of erythema nodosum lesions compared with placebo during the first 8 weeks of the trial (for week 4, P = 0.046; for week 8, P = 0.03). The increase in the frequency of erythema nodosum lesions was significantly higher in the 100-mg group than that in the 300-mg group at week 8 (P = 0.03). With the continuation of treatment, however, the frequency of erythema nodosum lesions was similar in all three groups. Figure 3 shows the unadjusted differences in the frequency of mucocutaneous lesions with multiple group comparisons at each time point.

Eye Disease

Patients from both thalidomide groups had fewer activations (P = 0.041) and decreases in visual acuity (P = 0.045) in either eye than did patients in the placebo group (Table 2).

Other Symptoms

Arthritis was detected in 4 patients in the placebo group (11 attacks), 5 patients in the 100-mg/d group (7 attacks), and 2 patients in the 300-mg/d group (4 attacks). Superficial thrombophlebitis occurred more frequently in the thalidomide groups (10 patients in the 100 mg/d group and 2 patients in the 300-mg/d group) than in the placebo group (3 patients). Epididymitis was diagnosed in 6 patients (2 in the placebo group, 1 in the 100-mg/d group, and 3 in the 300-mg/d group). Other lesions were extragenital ulceration (1 patient in the 100-mg/d group), leukocytoclastic purpura (1 patient in the 100-mg/d group and 2 in the 300-mg/d group), myositis (1 patient in the 300-mg/d group), and eruptions similar to those seen with the Sweet syndrome (1 patient in the 300-mg/d group).

Side Effects

Three patients in the thalidomide groups (1 receiving 100 mg/d and 2 receiving 300 mg/d) discontinued treatment because of severe sedation. One patient in the 300-mg/d group was withdrawn from the study because polyneuropathy developed in the fifth week. This patient had no symptoms and had a normal electroneuromyogram when reevaluated 3 years after withdrawal. Polyneuropathy developed in 3 patients after the trial ended. One of the 3 (from the 100-mg/d group) developed painless paresthesias 1 year later. Physical examination revealed a 25% reduction in the vibration sensation of the lower extremities; this was confirmed by electroneuromyography. A baseline electroneuromyogram was not available for this patient. The other 2 patients were in the 300-mg/d group. Both reported transient paresthesias but had normal results on examinations and electroneuromyography during treatment. One of the 2 patients developed mild axonal polyneuropathy 3 months after the trial ended, and the other had signs of mild sensorial polyneuropathy in the presence of central nervous system involvement 1 year after the trial ended.

Thirty-seven additional patients (12 in the placebo group, 12 in the 100-mg/d group, and 13 in the 300-mg/d group) had electroneuromyography. Such changes as minimal tremor activity or minimal neurogenic motor unit potentials were detected in 7 patients (1 in the placebo group, 4 in the 100-mg/d group [in 2 of the 4, changes were detected before treatment began], and 2 in the 300-mg/d group). These changes were not accompanied by symptoms or signs of polyneuropathy, and the neurologist did not think that these changes were specific enough to warrant an alteration in treatment. During the study, electroneuromyography was repeated in 3 patients in the 100-mg/d group; results were normal in each case.

Five thalidomide recipients (2 in the 100-mg/d group and 3 in the 300-mg/d group) reported decreases in libido, and 4 (2 each in the 100-mg/d and 300-mg/d groups) reported weight gain. The distribution of other side effects, determined according to positive answers to the predefined set of questions, is shown in Table 3.

Discussion

Our controlled trial shows that treatment with thalidomide not only is effective for the oral ulcers, genital ulcers, and follicular lesions of the Behcet syndrome but can also prevent the development of new oral and genital ulcers. This beneficial effect was sustained throughout the trial but rapidly diminished once treatment ended. The efficacy of thalidomide, 100 mg/d, seemed to be similar to that of thalidomide, 300 mg/d.

We believe that the strict nature of our criteria explains the relatively small percentage of patients who had a complete response as we defined it. First, to yield results more applicable to clinical practice, we combined data obtained at clinical visits with those collected from patient reports on the occurrence of ulcers between visits. If we consider only the data obtained at the fixed physician evaluations, the complete response rate increases to 30% in both thalidomide groups but remains 0% in the placebo group (Table 2). Second, we studied only relatively young male patients, a group in whom the disease course is more severe [15]. Thus, the complete response rate achieved in our study can be considered to have been achieved under worst-case conditions.

A wide range of thalidomide dosages has been used for patients with the Behcet syndrome; in the earlier experience, higher dosages, such as 400 mg/d, were used [9]. We found that 100 mg of thalidomide per day had a beneficial effect similar to that of 300 mg/d, but symptoms recurred after treatment with both dosages was discontinued. One placebo-controlled, crossover trial in patients with primarily idiopathic oral ulceration reported success with 100 mg of thalidomide per day, but ulcers recurred shortly after treatment was discontinued [13]. Another study used four sequential, weekly, on-and-off therapeutic cycles of 200 mg of thalidomide per day in four patients with the Behcet syndrome. In all three patients who completed the study, lesions recurred within 1 to 2 days after thalidomide therapy ended [11]. Thus, thalidomide cannot be considered a true disease-modifying agent for the Behcet syndrome.

Our study has some limitations. We studied only patients with primarily mucocutaneous manifestations and excluded those with more extensive systemic disease. In this regard, our data on the beneficial effect of thalidomide for eye disease and perhaps for the prevention of arthritis attacks should be viewed with caution and must be tested by further studies. All previous descriptions of the effect of thalidomide on eye disease and joint involvement were contained in retrospective reports and gave contradictory results [9, 11, 17].

Another limitation of our study is that it was restricted to men. We excluded women not only because of the distinctly more severe form of disease in men but also because of the risk for teratogenicity. A single 100-mg dose of thalidomide can cause malformations in a fetus if taken during the vulnerable period in pregnancy [18]. We believe that thalidomide should be used in a female patient of childbearing age only after a critical risk–benefit appraisal.

Treatment with either thalidomide dosage exacerbated erythema nodosum lesions during the first 8 weeks of treatment. The frequency of superficial thrombophlebitis (another nodular lesion occurring with the Behcet syndrome that is often clinically misdiagnosed as erythema nodosum) was also higher in the 100-mg group. We have no explanation for the exacerbation of erythema nodosum with the initiation of thalidomide treatment, which has previously been reported [9, 19]. If thalidomide indeed induces these transient eruptions, the cause of aphthous ulcers and erythema nodosum in the Behcet syndrome may not have a putative common denominator, such as streptococcal antigens [20].

After teratogenicity, polyneuropathy is the second most serious complication of thalidomide. In previous reports, the frequency of this syndrome varied from 14% to 50% in patients with idiopathic oral ulceration or the Behcet syndrome [21]. Although controversy surrounds the cause of polyneuropathy, the condition seems to be dose related and can be irreversible if not diagnosed early [22, 23]. The frequency of polyneuropathy, including that occurring in cases diagnosed after discontinuation of treatment, was 6.3% in our study. The rate of thalidomide-related polyneuropathy may be higher in women and elderly patients [24], two groups not included in our trial. The development of polyneuropathy in an early phase of the trial and 1 year after thalidomide therapy was discontinued suggests that the duration of treatment might not be important. Regular monitoring of sensory nerve action potentials with comparisons to baseline values has been suggested as useful for detecting early poly-neuropathy [21]. In one uncontrolled study, 50 mg of thalidomide was given three times a week in 43 patients with the Behcet syndrome for a mean of 17 months [12]. The drug was effective at this low dose, and no polyneuropathy was seen. Thus, further studies using similar low doses may be warranted for a better understanding of the efficacy-safety profile of thalidomide, especially from the standpoint of polyneuropathy.

For now, thalidomide seems to be a powerful and rapidly acting remedy for symptoms in a condition whose severity abates with time. In daily practice, the drug should be reserved for patients with severe oral or genital ulceration resistant to other therapies. All patients receiving thalidomide should be closely monitored for potential toxicity. A clinical guideline was recently proposed for the relatively safe use of thalidomide [25].

Dr. Mat: Department of Dermatology, Cerrahpasa Medical Faculty, University of Istanbul, 34303 Aksaray, Istanbul, Turkey.

Drs. Saip and Siva: Department of Neurology, Cerrahpasa Medical Faculty, University of Istanbul, 34303 Aksaray, Istanbul, Turkey.

Dr. Ozyazgan: Department of Ophthalmology, Cerrahpasa Medical Faculty, University of Istanbul, 34303 Aksaray, Istanbul, Turkey.

Dr. Zwingenberger: Medical Scientific Division, Grunenthal GmbH, D-52220, Stolberg, Germany.