Intravenous Iloprost Infusion in Patients with Raynaud Phenomenon Secondary to Systemic Sclerosis: A Multicenter, Placebo-controlled, Double-Blind Study

Methods

Patients

Outpatients with Raynaud phenomenon secondary to systemic sclerosis with either diffuse cutaneous or limited cutaneous variants from 12 clinical centers were eligible to enter the study. The diagnosis of Raynaud phenomenon was based on a history of episodic digital pallor and cyanosis. The diagnosis of systemic sclerosis was made according to the classification criteria developed by the American Rheumatism Association [23].

Patients were considered to have sufficiently severe Raynaud phenomenon for study entry if they met either of the following criteria: a minimum of 8 Raynaud attacks per week documented by patient diary during the 2 weeks before the first day of infusion or one or more cutaneous ischemic finger lesions (ulcers, fissures, or paronychiae). The patient could be male or female and at least 18 years old. Women of childbearing age were required to practice a medically acceptable method of birth control and to have a negative pregnancy test 1 week before the start of infusion and on the morning of the first day of infusion. All the patients provided written informed consent. Patients were excluded if they met any of the following conditions: surgical sympathectomy done within the past 12 months; bleeding diathesis or a platelet disorder; stroke, transient cerebral ischemic attack, or myocardial infarction within the past 6 months; history of angina pectoris within the preceding 12 months; evidence within the preceding 4 weeks of the onset of renal disease manifested by new hypertension or abnormal urinalysis results, evidence of microangiopathic hemolytic anemia, or azotemia; hypertension not controlled by diuretics, clonidine, or angiotensin-converting-enzyme inhibitors; evidence of cancer within the past year; evidence of any uncontrolled cardiovascular, pulmonary, renal, hepatic, endocrine, neurologic, or gastrointestinal disease; smoking within the preceding 4 weeks; history of substance abuse (drug or alcohol); or participation in an investigational drug study within 4 weeks before pretreatment.

The patients were required to discontinue certain medications at least 14 days before the first day of the study. Among the disallowed medications were aspirin, dipyridamole, calcium channel blockers, and other vasodilators.

Treatment Regimen

Iloprost was supplied in ampules containing 0.1 mg of the active drug in 1 mL of a sterile pyrogen-free solution. Placebo ampules contained the same solution without iloprost. Infusion rates of 0.5 to 2.0 ng/kg per minute were used, based on studies showing that 0.5 ng/kg per minute was the usual threshold for the inhibition of platelet aggregation and that infusion rates greater than 2 ng/kg per minute frequently produced side effects. Patients were assigned randomly to one of two treatment groups, either a 5-day (6 hours per day) infusion of iloprost or a 5-day (6 hours per day) infusion of placebo. The treatment groups were stratified by clinical center and by the presence of digital ulcers at entry. On day 1 of the treatment, the infusion was administered through a peripheral vein, starting with 0.5 ng/kg per minute of iloprost or a similar volume of placebo. The infusion rate was increased every 30 to 40 minutes to a maximum rate of 2.0 ng/kg per minute by increments of 0.25 to 0.5 ng/kg per minute. If side effects occurred, the dose was reduced in a stepwise manner until a tolerated dose was determined. The highest infusion rate tolerated by the patient was maintained until the end of the 6-hour treatment period. On days 2 through 5, the infusion was started using the highest infusion rate tolerated on the previous day. The infusion rate could be increased incrementally as tolerated to a maximum dosage of 2.0 ng/kg per minute. If side effects occurred, the infusion rate was reduced in a stepwise manner as on day 1.

Study Design

The study consisted of three periods: a 14-day pretreatment period; a 5-day, double-blind treatment period; and a 9-week, double-blind follow-up period. Follow-up visits were scheduled at 3, 6, and 9 weeks after completion of the infusion. The study was conducted during the winter months. The same investigators who supervised the drug infusion in a blinded manner also supervised the clinical outcome variables.

Clinical Outcome Variables

Two weeks before the first day of infusion and immediately after the last day, patients were given diaries and instructed to record daily the total number of Raynaud attacks, an estimate of the duration of an average or typical attack, the duration of the longest attack, and a Raynaud severity score. A 10-point, patient-completed scale was used to determine the Raynaud severity score, in which 0 represented no attack and 10 represented very severe attacks. Patients were asked to consider in their Raynaud severity score the number and duration of attacks; symptoms, such as numbness, burning, and pain and tingling; hand disability; and influence of cold on daily activity. An attack was defined as an episode of pallor followed by cyanosis with or without associated pain.

The mean weekly number of Raynaud attacks was calculated by dividing the total number of attacks recorded during a given 3-week period by the number of diary days and multiplying the result by 7. The average Raynaud severity score was calculated for a given 3-week period as the arithmetic mean of the daily scores in that period. Means for these variables were also calculated for the entire post-treatment period (weeks 1 to 9).

The physician global assessment of Raynaud phenomenon severity was recorded 7 days before infusion of iloprost, on day 1 of the infusion, and during weeks 6 and 9 of the follow-up period. The study physician evaluated the severity of patients' Raynaud phenomenon using a scoring system in which 0 represented no attack, 1 represented a mild attack, 2 represented a moderate attack, and 3 represented a severe attack.

Physician assessment of overall treatment effect, compared with baseline, was recorded at weeks 6 and 9. The physician recorded the overall effect of treatment on the patient's Raynaud phenomenon using a scoring system in which 1 represented greatly improved condition, 2 represented improved condition, 3 represented the same condition, and 4 represented worsened condition.

Digital cutaneous lesions were recorded on days 1 and 5 of infusion and at weeks 3, 6, and 9 of the follow-up period. The number of lesions on each hand was counted; each lesion was classified as ulcer, fissure, or paronychia; and the location of each lesion was recorded. Ulcers were considered to be of ischemic origin, whereas fissures and paronychiae were considered to be less certainly attributable to ischemia. Photographs of the lesions were taken during baseline and at each 3-week follow-up visit. The number of lesions at baseline (day 1) and at 3-week intervals during the follow-up period was recorded. The number of new digital cutaneous lesions appearing during the follow-up period and at each 3-week interval was also counted. The protocol defined a healing response as a reduction in the number of finger lesions from baseline (day 1) of at least 50%. Data for cutaneous lesions were analyzed for the subset of patients who successfully completed 5 days of infusion and had at least 6 weeks of follow-up.

The Modified Stanford Health Assessment Questionnaire, an index of overall functional status that emphasizes hand function, was administered at baseline and at each follow-up visit [24-26].

Cold Challenge Test

Cold tolerance was assessed by measuring the change of digital skin temperature before, during, and after a standardized cold stimulus. During exposure, the time required to cool the finger to 18 °C and the change in finger temperature per unit of time during cooling were recorded. The response to rewarming was measured when the hand was again exposed to ambient room temperature [18].

Patients were studied in a quiet room with stable environmental conditions (20 to 22 °C). The patients laid in the supine position and thermistors (YSI, Yellow Springs, Ohio) were attached to the pads of their left or right middle fingers. After a 30-minute period of acclimatization, baseline finger skin temperature measurements were recorded until two successive 2-minute readings were obtained that differed by less than 2 °C.

The patients' hands were placed in a cold-air chamber (4 °C) for a maximum of 10 minutes. The times at which the finger temperatures reached 18 °C were recorded. If the finger temperatures did not reach 18 °C, the finger temperatures were recorded at 10-minute intervals after being placed in the chamber. Their hands were then withdrawn from the cold chamber and finger temperature measurements were made at 1, 2, 5, 7, and 10 minutes during passive recovery. All measurements were taken with patients resting in the supine position and with their hands placed in a horizontal position level with the mid-thorax. To be safe, patients who had ulcers on one hand used the opposite hand for the cold challenge. Patients with ulcers on both hands were excluded from this assessment.

Vital Signs Measurement

Measurements of vital signs included assessments of brachial blood pressure and heart rate while patients were in the supine position, temperature, and respiratory rate. Vital signs were measured frequently on each infusion day and at each visit during the follow-up period.

Statistical Analysis

All continuous variables are presented as the mean ±SD. Baseline characteristics and outcome measures were compared for the treatment groups by Wilcoxon rank-sum tests for continuous variables and by chi-square tests for noncontinuous variables. Analysis of variance was used for the analyses of Raynaud attack frequency, Raynaud severity score, and the Health Assessment Questionnaire. For any analysis of variance or covariance, type III sums of squares from the SAS procedure GLM were used, with the exception of the test for significance of common slope in analysis of covariance. In this case, type I sums of squares were used, with the slope term entered into the model according to the investigator and treatment. The physician global assessment and the physician assessment of the treatment response were assessed using the Mantel-Haenszel tests for ordered categorical data. For digital lesion healing, we compared the number of responders (≥ 50% reduction in lesion count) using uncorrected chi-square tests. All analyses were done using the intention-to-treat paradigm, except for the analysis of the subset of patients with digital lesions; patients (two, both receiving placebo) who received fewer than 4 days of treatment were excluded from this analysis.

Results

Of the 131 patients enrolled in the study, 64 were assigned randomly to receive iloprost and 67, to receive placebo. Of the 131 patients who entered the study, 126 successfully completed the 5-day infusion and 114 (87%) completed at least 6 weeks of follow-up. Fifty-six received iloprost and 58 received placebo.

Of the 17 patients withdrawn from the study, 8 received iloprost and 9 received placebo. In both groups, the most common reason for patient withdrawal was lack of clinical improvement (7 patients who received iloprost and 6 patients who received placebo). Other reasons for patient withdrawal were intercurrent illness (1 patient receiving iloprost and 2 patients receiving placebo) and noncompliance (1 patient who received placebo).

The demographic characteristics and the duration of symptoms of Raynaud phenomenon, skin ulcers or skin thickening (scleroderma), were similar for patients receiving iloprost and those receiving placebo (Table 1). Patients were classified according to whether their systemic sclerosis satisfied major or minor classification criteria. Fifty-three (82.8%) of the patients receiving iloprost met major criteria and 11 (17.2%) met minor criteria. Fifty-one (76.1%) of the patients receiving placebo met major criteria, whereas 16 (23.9%) met minor criteria. Nifedipine was the most frequently used drug and had been used by 26 patients receiving iloprost and by 15 patients receiving placebo (P = 0.15). Eighty percent of the patients in both treatment groups characterized their response to previous vasodilator treatment as fair or poor. During the course of the trial, the mean daily outdoor temperature reported by the local weather bureau was similar for the two treatment groups (P > 0.20 at weeks 4 to 6).

Of the 64 patients receiving iloprost, 62 completed the 5-day infusion period. Of the 67 patients receiving placebo, 64 completed the 5-day infusion period. The overall average intravenous dosage received by patients during the 5-day period ranged from 0.34 to 1.91 ng/kg per minute in patients given iloprost and from 1.40 to 1.95 ng/kg per minute in patients given placebo. Nineteen of the 64 (29.7%) patients receiving iloprost tolerated the maximum dose of 2 ng/kg per minute on at least 1 of the 5 infusion days, whereas 65 of the 67 (97%) patients given placebo received an infusion equivalent to 2 ng/kg per minute for at least 1 of the treatment days. Eleven (17.2%) patients given iloprost received an overall daily dose of less than 1 ng/kg per minute.

Raynaud Phenomenon Severity Score

The Raynaud severity score measured patients' overall perceptions of the severity of Raynaud phenomenon. A decrease from baseline measurements indicated improvement. The mean Raynaud severity scores for the patients receiving iloprost and those given placebo were similar at baseline (P > 0.20) (Table 2). Those receiving iloprost had a decreased Raynaud severity score at weeks 3, 6, and 9 (Table 2). The treatment difference was greater at weeks 1 to 3 (P = 0.006) and weeks 4 to 6 (P = 0.05) than at weeks 7 to 9 (P = 0.09). Comparison of the changes in the entire follow-up period (weeks 1 to 9) showed a greater mean decrease in the patients given iloprost ( −34.8%)than in those given placebo ( −19.7%)(P = 0.01). In the subset of patients with digital cutaneous lesions at baseline, the mean decrease in Raynaud severity score remained the same in those given iloprost ( −34.5%)but was less in patients given placebo ( −9.9%)(P = 0.01).

Frequency of Raynaud Phenomenon Attacks

During the baseline period, the mean weekly frequency of Raynaud phenomenon attacks per week was similar for the treatment groups (P > 0.20) (Table 3). In both treatment groups, the attack frequency during the follow-up period decreased from baseline values. The decrease was greater in patients receiving iloprost than in those given placebo at every point during follow-up (Table 3). The treatment difference was greater at weeks 1 to 3 (P = 0.002) and at weeks 7 to 9 (P = 0.011) than at weeks 4 to 6 (P = 0.057). The mean decrease in attack frequency during the entire follow-up period (weeks 1 to 9) was 39.1% in patients receiving iloprost and 22.2% in those given placebo (P = 0.005), with a treatment difference of 16.9%. The subset of patients with digital cutaneous lesions had a mean decrease during the entire follow-up period of 36% in those given iloprost and 14.1% in those given placebo (P = 0.064), with a treatment difference of 21.9%. The duration of Raynaud attacks was similar over time (P > 0.20).

Physician Assessment of Raynaud Phenomenon Severity

At baseline, the physician global assessment scores were similar for the patients receiving iloprost and for those given placebo (P > 0.20). Improvement in the mean physician global assessment score tended to be greater in the patients receiving iloprost than in those given placebo at weeks 6 and 9 (P = 0.07 at week 9).

At weeks 6 and 9 the physician provided an overall assessment of the treatment effect (Table 4). Most of the patients receiving iloprost, but not those given placebo, improved or greatly improved at both week 6 (52% compared with 27%; P = 0.008) and week 9 (61% compared with 27%; P < 0.001) (Table 4).

Health Assessment Questionnaire

The modified Health Assessment Questionnaire consisted of 20 questions divided into eight categories of activities that are common in daily life and are affected by arthritis. The baseline Health Assessment Questionnaire Disability Index (which combines all eight activity categories) was 0.74 in the patients given iloprost and 0.63 in those given placebo (P > 0.20). At each of the three post-treatment evaluations, a slight decrease from baseline in the mean Health Assessment Questionnaire Disability Index was noted in patients given iloprost, whereas a slight increase from the baseline mean was recorded in those given placebo (P = 0.07 at week 9). However, in the subset of patients with digital cutaneous lesions, at week 9 the patients given iloprost had a mean Health Assessment Questionnaire Disability Index that was unchanged from baseline, whereas in patients given placebo a slight increase (worsening) from baseline was noted (P = 0.01) (Table 5).

Digital Cutaneous Lesions

Seventy-three patients (35 receiving iloprost and 38 receiving placebo) had digital cutaneous lesions when they entered the study. The protocol definition of improvement of these lesions was at least a 50% reduction from baseline score in the total number of lesions during the 9-week follow-up period. Patients with no lesions at baseline were not included. The percentage of patients exhibiting at least a 50% reduction from baseline in the total number of lesions was greater in the patients given iloprost than in those given placebo at all times during the 9-week follow-up period. In patients given iloprost, 20%, 28.1%, and 25.7% of the patients had at least a 50% reduction in the number of lesions at weeks 3, 6, and 9, respectively. In contrast, in patients receiving placebo, 5.4%, 15.2%, and 18.4% experienced at least a 50% reduction in number of digital lesions. The difference between treatment groups was similar at day 5, week 3 (P = 0.06), week 6, or at week 9. Patients in both treatment groups received similar wound care, including antibiotics or debridement if necessary.

The percentage of patients with digital lesions that healed completely (no remaining original lesions and no new lesions) was greater in patients receiving iloprost than in those given placebo at all times during the 9-week period (P > 0.20 at week 9). At week 9, 16 of 64 patients (25%) receiving iloprost and 22 of 67 patients (32.8%) receiving placebo (P > 0.20) had new digital lesions. The subset of finger lesions classified as ischemic (that is, excluding fissures and paronychiae) that healed by week 9 was 15% for patients given iloprost and 25% for patients given placebo (P > 0.20).

Cold Challenge

Cutaneous finger temperature was recorded under ambient conditions and during a cold chamber test done at baseline (day −7)on days 1 and 5, and at weeks 3, 6, and 9. The only difference between the treatment group means for any of the temperature variables was at week 3 for the average rate of cooling during cold challenge. The mean of the average rate of decrease in digital temperature during cold challenge decreased from baseline by 0.14 ±1.10 °C per minute in patients receiving iloprost and increased by −0.2±0.95 °C per minute in those given placebo (P = 0.04). All other cold challenge measures were similar for the patients receiving iloprost and for those given placebo at any other time during the study. Other cold challenge measures included the time to reach a digital temperature of 18 °C, the proportion of patients who reached 18 °C during the cold challenge, and the percentage recovery in digital temperatures after cold challenge.

Side Effects

At least one side effect was reported by 59 of the 64 patients receiving iloprost (92%) and by 38 of 67 patients receiving placebo (57%) (P < 0.001) (Table 6). The side effects were reversible and controlled by a reduction in the dose rate of the iloprost. In 5 patients receiving iloprost and in 1 patient receiving placebo, the infusion therapy was discontinued for part of an infusion day because of side effects but was resumed the next day. One patient given iloprost required intravenous saline for treatment of vomiting and dehydration.

There was one death during the study. That patient received placebo during the study infusion with no complications. He completed 6 weeks of follow-up and was considered to have improved based on his report of less frequent episodes of Raynaud phenomenon and healing of cutaneous lesions. Two days after the week 6 visit, he experienced 1 hour of substernal chest pain and was admitted to the hospital and withdrawn from the study. He was discharged 5 days later with a diagnosis of acute pericarditis. Two days after his discharge, he had increasing shortness of breath and recurrent chest pain. He was found to be hypotensive and in severe heart failure. He died the next day. No patient died during the infusion period.

Discussion

This is the largest placebo-controlled study of iloprost infusion in patients with systemic sclerosis and the first multicenter study using a parallel design to be reported. We found that the frequency of Raynaud attacks and the patients' overall perceptions of the severity of their attacks was improved by iloprost, with clinical benefits persisting for as long as 9 weeks after treatment. The investigators' blinded overall ratings of the treatment effect were better in the patients who received iloprost compared with those given placebo. These findings suggest that a 5-day intravenous infusion of iloprost is effective for the short-term treatment of severe Raynaud phenomenon secondary to systemic sclerosis.

Iloprost has been administered to patients with Raynaud phenomenon in several studies conducted in Europe [15-17, 19-22]. Intravenous infusions of up to 3 ng/kg per minute for 5 to 8 hours daily on 3 consecutive days resulted in reductions in frequency, duration, or severity of Raynaud attacks. The beneficial effects appeared in patients with scleroderma as well as in those with other secondary forms of Raynaud phenomenon. The improvement of symptoms was reported to accompany increased digital cutaneous blood flow for at least 6 weeks after treatment [17].

These studies were done with small numbers of patients with either systemic sclerosis or other secondary forms of Raynaud phenomenon. Some of these studies were not placebo-controlled [19-21], whereas others used a cross-over design [15-17]. For our study, we chose a placebo-controlled design with a parallel treatment group rather than a cross-over design because previously published data with prostacyclin [11-14] and iloprost [15-18] in patients with Raynaud phenomenon indicated that vascular effects and clinical improvement observed may persist for 6 to 10 weeks after treatment.

In an initial clinical trial reported by five of us using iloprost for Raynaud phenomenon in a small number of patients with systemic sclerosis, we were unable to show differences between iloprost and placebo in the frequency or severity of Raynaud attacks [18]. We speculated that the reason for the lack of clinical efficacy in this study was related to unexpected marked improvement in the patients given placebo because of warm ambient temperatures that increased from less than 10 °C to more than 26.6 °C during the study. Our study was done in the winter months, during which patients receiving placebo had a 15% to 27% improvement in the mean weekly frequency of Raynaud attacks compared to a 34% to 45% improvement in patients given iloprost (see Table 3). The improvement in patients receiving placebo may have been secondary to regression to the mean or to the placebo effect.

Rademaker and others [20] compared intravenous iloprost to oral nifedipine in the treatment of patients with Raynaud phenomenon associated with systemic sclerosis. That study showed a reduction in the number, duration, and severity of Raynaud attacks and a reduction in the number of digital lesions in both the iloprost and nifedipine treatment arms, but there was no placebo group for comparison. Similarly, Torley and colleagues [19] compared low (0.5 ng/kg per minute) and maximal tolerated dosages (up to 2 ng/kg per minute) of iloprost in patients with connective tissue disease. They found that both regimens reduced the severity of Raynaud phenomenon and improved digital ulcer healing. That study did not include a placebo-control arm.

Five of the authors of this study previously reported, in a smaller group of patients with systemic sclerosis from two U.S. centers, a statistically significant difference between iloprost and placebo in the healing of ischemic digital ulcers [18]. In the present study, 73 patients had digital cutaneous lesions (ischemic ulcers, fissures, and paronychiae): Thirty-five were treated with iloprost and 38 with placebo. Although a larger percentage of patients receiving iloprost than those given placebo had a 50% or greater decrease in the number of such lesions, this difference was not significant. A similar result was noted for ischemic digital ulcers. Because the current study involved 12 centers, the multicenter design may have increased the variability in assessment of digital ulcer healing.

Data from the Health Assessment Questionnaire provided evidence that iloprost was beneficial to patients with digital lesions. The Health Assessment Questionnaire has been used to evaluate functional status and quality of life in patients with rheumatoid arthritis [25] and in patients with systemic sclerosis [26]. Many of the domains of the questionnaire were appropriate for our patients because they evaluated hand function. Statistically significant improvement after iloprost treatment was seen in the Health Assessment Questionnaire Disability Index only in the subset of patients with digital cutaneous lesions at baseline, suggesting that iloprost may have had a beneficial effect on these digital lesions that was not measured by our methods of visual scoring.

As in other therapeutic trials in patients with Raynaud phenomenon [6, 27, 28], we were unable to show any statistically significant change in response to a laboratory-based cold challenge despite the clinical benefit of reduced frequency of Raynaud attacks and Raynaud phenomenon severity score. We previously reported an improvement in cold tolerance measured by a decrease in the temperature at which patients experienced visible color changes of a Raynaud attack or plethysmographic evidence of loss of digital pulsatile flow [18]. Others have reported improvement in blood flow to the digits after iloprost infusion, but these studies have been uncontrolled or there were important differences in baseline digital blood flow levels [15, 20, 21]. The reasons that laboratory-based measurements of blood flow and cold tolerance do not always agree with the clinical outcomes in studies of Raynaud phenomenon is not clear but probably include acclimatization to the laboratory challenge [29], large interpatient variability in the measurements, or inadequacy of the laboratory challenge.

Several studies have suggested that a relatively short course of intravenous iloprost or prostacyclin has biological effects that last for several weeks [11-21]. Iloprost has a plasma half-life of approximately 10 to 30 minutes and exhibits first-order kinetics. Total clearance in systemic sclerosis is reduced compared with other vascular diseases [30]. No residual active drug is present to account for this prolonged effect. This delayed benefit may be due to prolonged alterations in vascular geometry or patency or function of endothelium or vascular smooth muscle.

Adverse reactions in our study are similar to those reported in other studies of iloprost. The side effects related to the pharmacologic action of iloprost were primarily limited to the period of the drug infusion and improved immediately with dose reduction. Appendix Table 1