Recurrent digital ulcers are a manifestation of vascular disease in patients with systemic sclerosis (SSc; scleroderma) and lead to pain, impaired function, and tissue loss. We investigated whether treatment with the endothelin receptor antagonist, bosentan, decreased the development of new digital ulcers in patients with SSc.
This was a randomized, prospective, placebo-controlled, double-blind study of 122 patients at 17 centers in Europe and North America, evaluating the effect of treatment on prevention of digital ulcers. The primary outcome variable was the number of new digital ulcers developing during the 16-week study period. Secondary assessments included healing of existing digital ulcers and evaluation of hand function using the Scleroderma Health Assessment Questionnaire.
Patients receiving bosentan had a 48% reduction in the mean number of new ulcers during the treatment period (1.4 versus 2.7 new ulcers; P = 0.0083). Patients who had digital ulcers at the time of entry in the study were at higher risk for the development of new ulcers; in this subgroup the mean number of new ulcers was reduced from 3.6 to 1.8 (P = 0.0075). In patients receiving bosentan, a statistically significant improvement in hand function was observed. There was no difference between treatment groups in the healing of existing ulcers. Serum transaminase levels were elevated to >3-fold the upper limit of normal in bosentan-treated patients; this elevation is comparable with that observed in previous studies of this agent. Other side effects were similar in the 2 treatment groups.
Endothelins may play an important role in the pathogenesis of vascular disease in patients with SSc. Treatment with the endothelin receptor antagonist bosentan may be effective in preventing new digital ulcers and improving hand function in patients with SSc.
Systemic sclerosis (SSc; scleroderma) is an autoimmune connective tissue disease characterized by cutaneous and visceral fibrosis and widespread vascular pathology (1). The most common and evident expression of the vascular involvement is Raynaud's phenomenon, an abnormal reactivity of blood vessels to cold and other stimuli (2). Excessive vasoconstrictive responses lead to pallor and cyanosis of distal extremities, particularly the digits, with complications including digital ulceration and infarction. These complications are much more common in Raynaud's phenomenon associated with SSc than in idiopathic Raynaud's, which occurs in ∼10% of the adult population. In SSc, not only is a functional abnormality, i.e., vasoconstriction, more likely, but structural changes in the blood vessel, including intimal proliferation and obstruction, are also more likely to occur. Ulcers on the fingertips and over the interphalangeal joints cause pain, limit function, heal with scarring and digital resorption, and, when infected, can lead to osteomyelitis or other serious soft tissue infections. Although treatment with a variety of vasodilators, including calcium channel blockers, angiotensin-converting enzyme (ACE) inhibitors, and α-adrenergic inhibitors (3), has been shown to be effective in treating Raynaud's phenomenon, digital ulceration remains a serious complication for many patients, and effective therapy remains elusive.
The underlying physiologic basis for abnormal vascular responses in SSc is unknown. The initial events are thought to involve endothelial cell injury (4), with subsequent loss of normal vasodilatory mediators including prostacyclin and nitric oxide (5–7). As a result, abnormal responses to vasoconstrictive mediators, including catecholamines, may occur.
Endothelial injury also leads to increased release of endothelin 1 (ET-1) a 21–amino acid peptide released by endothelial cells (8, 9). ET-1 was originally identified by virtue of its action in stimulating vascular smooth muscle contraction, and it is the most potent naturally occurring vasoconstrictive mediator (10, 11). It binds to 2 cognate receptors, ETA and ETB, which are variably expressed on endothelial cells, smooth muscle cells, fibroblasts, and other cells throughout the body (11). ET-1 is found in increased levels in the serum of patients with SSc, suggesting that it plays a role in the pathogenesis of SSc vascular disease (12, 13). In addition, ET-1 has been shown to stimulate fibroblast and smooth muscle cell proliferation and stimulate fibroblast matrix biosynthesis (14–16). ET-1 may thereby also play a role in the genesis of the structural vascular lesion in SSc.
Bosentan, an oral endothelin dual receptor antagonist, was recently shown to be effective in the treatment of idiopathic pulmonary arterial hypertension as well as pulmonary hypertension in the setting of SSc (17, 18). Like Raynaud's phenomenon in SSc, pulmonary hypertension in these patients is characterized by both functional and structural abnormalities, and the structural lesion in pulmonary hypertension resembles the vascular lesion in SSc-associated Raynaud's. This study was designed to examine whether bosentan may be effective in preventing digital ulcers in patients with SSc.
PATIENTS AND METHODS
The RAPIDS-1 study (RAndomized Placebo-controlled study on prevention of Ischemic Digital ulcers in Scleroderma) was a double-blind, placebo-controlled study evaluating bosentan treatment for digital ulcers. The study was conducted at centers in the US, Canada, the UK, and Europe. Patients were randomized into 2 parallel groups using a 2:1 bosentan-to-placebo ratio and were treated for 16 weeks. The unbalanced 2:1 randomization ratio was chosen to increase the number of patients receiving treatment, for safety evaluation and to improve recruitment. The study protocol was initiated in the fall of 2001, and all enrolled patients completed the study by May 1, 2002; this time frame was chosen to incorporate the cold winter months and thus maximize the likelihood of digital ulcers developing during the study.
Patients had limited or diffuse SSc, as defined by LeRoy et al (19), and satisfied the American College of Rheumatology (formerly, the American Rheumatism Association) criteria for the classification of SSc (20). For entry into the study, all patients had to have a history of a documented digital ulcer within the previous 12 months. Patients entering the study were allowed to continue treatment with oral vasodilating drugs and other oral medications for Raynaud's, including ACE inhibitors. Patients who were receiving antibiotics for infected ulcers were excluded. No treatment with parenteral prostanoids within the previous 3 months was allowed, nor could any patients have received previous treatment with bosentan. Patients had to be older than age 18 years and have a body weight >40 kg, both as an index of nutritional status and because the dosing regimen of bosentan was developed for patients weighing >40 kg. Patients with active renal crisis or advanced cardiopulmonary disease or who were otherwise clinically unstable were excluded. Because of the known effects of bosentan on liver function tests, transaminase values had to be <3-fold the upper limit of normal at the baseline evaluation.
Treatment and evaluation protocol.
Following the baseline evaluation, patients were randomized to receive either bosentan or placebo treatment. Subjects received 62.5 mg of bosentan twice daily for 4 weeks, or identical placebo tablets. For the next 12 weeks, patients received 125 mg of bosentan twice daily or identical placebo pills. At each monthly evaluation visit, the number of new and existing digital ulcers was scored. If new ulcers developed between scheduled visits, patients were asked to come in for ulcer documentation. Ulcers were defined as loss of surface epithelialization and did not include fissures or cracks in the skin or areas of calcium extrusion from calcinosis cutis. Only ulcers at or distal to the proximal interphalangeal joint were scored. The longest diameter of ulcers was measured, and ulcers were photographed using a digital camera. However, inequalities of lighting and shot angles made photographic data difficult to evaluate, and only direct ulcer measurements are reported. Ulcer healing was considered to be complete when total reepithelialization of all ulcers was observed. Partial healing was considered to be a >50% reduction in surface area, calculated as the sum of the squares of all observed ulcer diameters at a visit. Liver function tests were administered monthly, and treatment was temporarily discontinued if levels of transaminases were elevated to >3-fold the upper limit of normal. If these levels did not return to the normal range, active treatment or placebo was discontinued. If antibiotics were required for ulcer infection that developed during the course of the study, this was scored as an adverse event, but patients remained in the study.
Function was evaluated using the modified Scleroderma Health Assessment Questionnaire (SHAQ) (21, 22). The modified HAQ excludes questions related to use of assistive devices and includes a visual analog scale (VAS) for disability related to specific organ systems. Individual questions were scored on a 0–3 scale (0 = no disability, 3 = severe disability). Patients used a VAS (range 0–10 [10 = most severe]) to rate the severity of pain, the degree to which Raynaud's affected their activity, the degree to which ulcers affected their activity, and the severity of scleroderma lung, gastrointestinal, and overall disease.
All subjects gave their informed consent to participate in the study, and the study was approved by the Institutional Review Board for Human Studies at each institution, in accordance with the Helsinki agreement (European centers) and human studies guidelines of the National Institutes of Health (US centers).
The a priori primary outcome variable for the study was the number of new digital ulcers developing during the 16-week study treatment period. Secondary a priori outcome variables included time to complete or partial healing of baseline digital ulcers, time to onset of new ulcers, and SHAQ (23) subsets for dressing, hygiene, grip, and hand function, and assessment of safety and tolerability.
Efficacy analyses were conducted at the 0.05 2-sided nominal Type I error, and were based on the intent-to-treat (ITT) population. All randomized patients were included in the ITT population, except for 1 patient in the bosentan group, who withdrew on day 22, and for whom no post-baseline efficacy assessment was available. This patient was included in the safety evaluation because he received study drug and was assessed for safety during the treatment period.
Treatment comparisons for the different efficacy end points were performed with Poisson regression analysis, adjusted for overdispersion. SHAQ scores were analyzed by the Mann-Whitney U test. Time-to-event end points were plotted using life-table methods, and treatment comparisons were carried out by log rank test. In the life-table display, time events were grouped into intervals corresponding to the planned visit (i.e., weeks 4, 8, 12, and 16 [or within 2 weeks of these time points]), thus enlarging the first and last intervals to accommodate for the entire study period (including, respectively, week 1 and week 19). Exploratory subgroup analyses were carried out on the primary outcome, examining the association between treatment effect and type of SSc (limited versus diffuse).
Of the 122 patients randomized during the course of the study, 79 were randomized to receive bosentan, and 43 were randomized to receive placebo in a 2:1 intended ratio. Demographic characteristics of the population are shown in Table 1 and reflect the predominant number of European centers. Thirty-eight percent of patients had diffuse scleroderma, and 62% had limited scleroderma. Overall, 79% of patients were female, the mean age was 51.8 years, and the mean disease duration (from time of diagnosis) was 109.6 months (9.1 years). Patients randomized to bosentan and those assigned to placebo were similar in terms of sex distribution, age, and disease duration. Overall, 62.3% of patients had limited cutaneous SSc; this subset was slightly overrepresented in the treatment group compared with the placebo group (67.1% versus 53.5%; P not significant [NS]).
Table 1. Baseline characteristics of the 122 patients with SSc*
Total (n = 122)
Placebo (n = 43)
Bosentan (n = 79)
SSc = systemic sclerosis.
Age, mean ± SD years
51.8 ± 12.9
48.0 ± 12.2
53.9 ± 12.8
Weight, mean ± SD kg
66.2 ± 12.7
64.5 ± 11.3
67.2 ± 13.4
Ethnicity, no. (%)
Disease duration, mean ± SD months
91.5 ± 72.3
119.5 ± 81.9
Diffuse SSc, no. (%)
Limited SSc, no. (%)
The baseline disease characteristics of the study subjects are shown in Table 2. Sixty-three percent of patients had a digital ulcer at entry (67.1% of patients in the bosentan group and 55.8% of those in the placebo group). The mean number of ulcers at entry was similar in the bosentan and placebo groups and in the subsets of patients with limited SSc and diffuse SSc. The total surface area of ulcers was also comparable between groups. Scores for patient's assessment of ulcer severity, Raynaud's severity, ulcer pain, and SSc severity at baseline were similar in the active-treatment and placebo groups and ranged from 5.3–6.6 on a VAS. The overall disability index, using the SHAQ, and the subindex for hand function were also similar in placebo- and bosentan-treated patients.
Table 2. Baseline disease characteristics in the 2 study groups*
Placebo (n = 43)
Bosentan (n = 79)
SSc = systemic sclerosis; SHAQ = Sclerosis Health Assessment Questionnaire.
In the placebo group, 20 patients had diffuse disease; in the bosentan group, 26 patients had diffuse disease.
In the placebo group, 23 patients had limited disease; in the bosentan group, 53 patients had limited disease.
In the placebo group, the minimum and maximum surface diameters were 0.0 mm and 564.0 mm, respectively. In the bosentan group, the minimum and maximum surface diameters were 0.0 mm and 784.0 mm, respectively.
Composite of the individual scores for dressing, hygiene, and grip.
No. (%) of patients with at least 1 digital ulcer at entry
There were no significant differences between groups in the use of concurrent therapies: 58.1% of patients in the placebo group and 41.8% of those in the bosentan group were receiving calcium channel blockers (9 different agents), and 25.6% and 25.3%, respectively, were receiving ACE inhibitors or angiotensin II receptor antagonists. Similar proportions of patients in both groups were receiving vasodilators, pentoxyfylline, nonsteroidal antiinflammatory drugs, corticosteroids, and antirheumatic drugs (data not shown). A total of 19 patients (6 of 43 in the placebo group [14%] and 13 of 79 in the bosentan group [16.5%]) discontinued therapy; thus, 37 patients in the placebo group and 66 in the bosentan group completed the study. Of the 6 patients in the placebo group who withdrew, 1 withdrew due to nausea, 1 due to infected ulcers, 1 for failure of therapy, and 1 due to headaches and diarrhea. One patient in the placebo group was lost to followup, and 1 was withdrawn due to a protocol violation (use of another investigational drug). Five patients in the bosentan group withdrew because of abnormal results of liver function tests, 1 withdrew because of increased dyspnea, 1 due to irritability, 2 because of increased esophageal reflux, and 1 withdrew consent after being diagnosed with hepatitis C virus infection. One patient in the bosentan group was lost to followup (unable to return for followup visits), 1 was withdrawn for protocol violation (ineligible due to elevated transaminase levels at baseline), and 1 withdrew consent.
Development of new digital ulcers.
The a priori primary outcome variable was the total number of new digital ulcers occurring during the treatment period. The overall number of new ulcers was significantly reduced in patients receiving bosentan. Among patients in the bosentan group, a mean of 1.4 new ulcers per patient developed, compared with 2.7 ulcers per patient among those receiving placebo (P = 0.0083) (Figure 1). A separate analysis was performed for patients who had digital ulcers at baseline; this group was thought to be at greater risk for the development of new ulcers. Indeed, among bosentan-treated individuals in this subgroup, an average of 1.8 new ulcers per patient developed during the course of the study compared with 3.6 ulcers per patient in the placebo group, a reduction of 50% (P = 0.0075). The total number of ulcers was reduced both in patients with diffuse SSc and in those with limited SSc, although only the reduction in the group with diffuse SSc achieved statistical significance.
New digital ulcers developed in 45 (58%) of 78 patients receiving bosentan and in 26 (61%) of 43 patients receiving placebo (Figure 2A). However, large numbers of new ulcers were much less likely to develop in patients who were treated with bosentan. The proportion of patients in whom ≥4 ulcers, ≥7 ulcers, and ≥10 ulcers developed was notably reduced in the bosentan group (Figure 2A.) Among subjects who had ulcers at baseline, ≥4 new ulcers developed in 10 (42%) of 24 patients in the placebo group, compared with 7 (13%) of 52 patients in the bosentan group.
The effect of bosentan on digital ulcers was examined in the subgroups of patients with diffuse SSc and limited SSc (Figure 2B). The reduction in the development of multiple ulcers observed with bosentan was evident in both groups of patients but was even more dramatic in the subset with diffuse scleroderma. Among patients with diffuse scleroderma who had digital ulcers at baseline, ≥4 new ulcers developed in 50% and ≥7 new ulcers developed in 20% of the placebo group, compared with 11% and 0%, respectively, of patients in the bosentan group (Figure 2B).
New digital ulcers developed in ∼25% of patients within the first 4 weeks of treatment, and in ∼40% within the first 8 weeks of treatment. Bosentan had no apparent effect on the time to development of the first digital ulcer (Figure 3). Using a Kaplan-Meier–type estimate for risk of onset of subsequent ulcers, the protective effect of bosentan persisted throughout the study but was most evident after 8 weeks of treatment (Figure 4).
There was no difference between placebo- and bosentan-treated patients in the time to complete or partial healing of ulcers (Figure 5). Indeed, a slight trend toward slower healing was observed in the bosentan-treated patients (P NS). In both groups, ∼50% of patients showed partial or complete ulcer healing within 8 weeks. Nonetheless, 10–20% of patients did not show healing of the ulcers during the course of the study.
There was a trend toward an improvement of the HAQ score in bosentan-treated patients compared with placebo-treated patients. There was a small decrease in HAQ scores in the bosentan group and a slight increase in the overall HAQ score in placebo-treated patients (data not shown); this difference did not achieve statistical significance. The HAQ includes evaluation of both upper extremity and lower extremity function. The questions in the HAQ that specifically address hand function were separately evaluated, as was a composite score for hand function overall. There was a statistically significant improvement in scores for dressing and grooming and personal hygiene, as well as a strong trend for improvement in grip ability, in patients receiving bosentan compared with those receiving placebo (Figure 6). The overall score for hand function deteriorated in placebo-treated patients and improved in bosentan-treated patients, and the difference was statistically significant (P < 0.005).
Five patients (3 in the placebo group and 2 in the bosentan group) experienced serious adverse events during the study. Dyspnea, palpitations, and ventricular tachycardia developed in 1 bosentan-treated patient, who required 2 hospitalizations. Ventricular tachycardia developed in a second bosentan-treated patient, who also required hospitalization. In both patients, the arrhythmia resolved without sequelae. One patient had continuing dyspnea at the end of the study. Among placebo patients, 1 had esophagitis and vomiting that were serious enough to require hospitalization, 1 developed severe and persistent digital ischemia, and 1 developed high-altitude pulmonary edema, high-altitude syndrome, and subsequent recurrent dyspnea.
Table 3 summarizes other adverse events observed in patients in the study. Diarrhea was seen in 9% of bosentan-treated patients compared with 2% of patients in the placebo group. Arthralgia was reported by 16% of placebo-treated patients and 6% of bosentan-treated patients. Other side effects were infrequent, with comparable incidences in the treatment and placebo groups.
Table 3. Adverse events occurring in subjects in the 2 study groups*
Placebo (n = 43)
Bosentan (n = 79)
Values are the number (%). NOS = not otherwise specified.
Adverse events, frequency ≥5%
Liver function tests NOS abnormal
Upper respiratory tract infection NOS
Infected skin ulcer
Pain in limb
Edema lower limb
Esophageal reflux aggravated
Serious adverse events
High-altitude sickness, acute
There was a notable increase in elevated transaminase values in bosentan-treated patients, with 14% of patients showing elevations of up to 3-fold the upper limit of normal at some point during the study. This is comparable with the frequency of abnormal liver function tests seen in previous studies of bosentan in patients with pulmonary hypertension (18). Among the 11 patients with elevated transaminase levels, 3 had aspartate aminotransferase or alanine aminotransferase values >8-fold the upper limit of normal. In 5 patients (6% of the active-treatment group), abnormalities led to permanent discontinuation of the drug. In all patients in whom treatment was discontinued, transaminase values returned to normal.
Digital ulcers are a frequent complication in patients with scleroderma, with an estimated frequency of 30–50% (24). The etiology of the ulcers is multifactorial. Ischemia due to vascular disease (Raynaud's), sclerodactyly, dry skin, calcinosis, and local trauma may all contribute to ulcer initiation. Minor skin trauma, or spontaneous fissuring or cracking, may progress to ulcers in patients with severe Raynaud's in the setting of scleroderma. The failure of cuts to heal and the development of ulcers result from poor blood flow and tissue oxygenation as well as complicating factors such as infection, epidermal thinning in patients with long-standing disease, and skin being tightly stretched over joints, with associated contractures.
Whatever the etiology, ulcers are often extremely painful and cause significant impairment of hand function and activities of daily living. In some patients, particularly those with limited cutaneous scleroderma, digital ulcers represent the major limitation to function in winter months. In some cases, infection may supercede and lead to substantial loss of tissue. Over a period of years, healing of ulcers leads to residual scars or “pits,” with progressive digital shortening. Prevention of digital ulceration may thus improve patient function in the short term as well as prevent long-term changes in digits.
In the present study, fewer digital ulcers developed in patients treated with bosentan than in those receiving placebo. The reduction in the number of new ulcers was most significant in patients who had ulcers at baseline and in those with diffuse SSc. In addition, bosentan appeared to be most effective in patients who were most severely affected by ulcers; the proportion of patients with multiple digital ulcers was markedly reduced. This improvement is reflected, in part, in the improved hand function observed in the bosentan-treated group. It is unknown whether long-term functional improvement will be maintained, perhaps from decreased digital scarring and preservation of digital tissue.
Although new ulcers were prevented, no improvement in ulcer healing with bosentan was observed. It is unclear why a drug that presumably promotes vasodilation as an antagonist of ET-1 would not also promote ulcer healing. It is possible that the ulcers that were present at the time of study entry were chronic, and that other factors, including infection, lack of underlying granulation tissue, and poor epithelial function, masked any salutary effects of bosentan on blood flow. The expression of endothelin receptors on keratinocytes suggests that ET-1 could be a modulator of function and, therefore, receptor blockade might impair reepithelialization despite beneficial effects on other aspects of healing. The failure of patients receiving bosentan to show improvement in Raynaud's symptoms suggests that bosentan may have effects other than simple vasodilation. The balance of factors involved in healing may be very different from those in ulcer initiation and may vary among digital ulcers of different types. Followup studies that address the healing of ulcers may clarify this point.
Few drugs have previously been shown to affect digital ulcers in the setting of scleroderma. Results of previous studies suggest that intermittent intravenous iloprost improves healing of digital ulcers and prevents episodes of digital ischemia (25–27). In a recent study of pulmonary hypertension, fewer ulcers developed in patients with SSc who were receiving epoprostenol infusion than in subjects receiving placebo (28).
The present study is the first one designed specifically to look at prevention of ulcer formation. Furthermore, this study is among a very few to demonstrate clinical efficacy in SSc. The beneficial effect of bosentan in pulmonary hypertension and in preventing digital ulcers suggests that the drug promotes vascular function in divergent organ systems. There is an obvious need for caution in applying the results of this small preliminary study. Longer-term, larger studies will be required to assess whether the effects are sustained, in terms of both prevention of digital ulcers and improved patient function, and whether other organ systems affected by vascular disease, including the heart and intestinal tract, might also be beneficially affected.
We thank our many research and clinical trials coordinators and colleagues, including D. McCluskey, RN, D. Therrien, RN, Christine Scholz, MD, and PY Hatron, MD, and the many physicians who referred patients for this study.