To assess the efficacy of therapies in healing and preventing digital ulcers (DUs) in systemic sclerosis (SSc; scleroderma).
To assess the efficacy of therapies in healing and preventing digital ulcers (DUs) in systemic sclerosis (SSc; scleroderma).
Medline and EMBASE databases, and American College of Rheumatology and European League Against Rheumatism abstracts, were searched. Randomized controlled trials (RCTs) with outcomes investigating healing or prevention of DUs in SSc and comparing a pharmacologic therapy with placebo or an active agent were included. The pooled risk ratios (RRs) using the fixed-effects model were calculated and heterogeneity was tested using the I2 statistic.
Sixty studies were found; 19 were not randomized, and 10 did not give DU quantitative data or no comparison of a different drug, leaving 31 RCTs with a total of 1,989 patients. Quality was 3 of 5 or less for 11 trials. DUs were not the primary outcome in many RCTs. Phosphodiesterase type 5 (PDE-5) inhibitors were significant for DU healing (RR 3.28 [95% confidence interval (95% CI) 1.32, 8.13], P = 0.01). Two large bosentan trials were significant for mean number of new DUs (standardized mean difference [SMD] −0.34 [95% CI −0.57, −0.11], P = 0.004). Oral prostacyclins were not statistically different from placebo, but intravenous (IV) iloprost prevented new DUs (SMD 0.77 [95% CI −1.46, −0.08], P = 0.03). Single trials for atorvastatin and vitamin E were positive in the prevention and healing of DU, respectively. There were many negative trials: antiplatelet therapy, oral N-acetylcysteine, heparin, dimethyl sulfoxide, ketanserin, prazosin, prostaglandin E1, cyclofenil, quinapril, and topical nitroglycerin formulation.
Small sample sizes, few comparative trials, and heterogeneity limits the conclusions. The results suggest a role for PDE-5 inhibitors in the healing of DUs; bosentan and IV iloprost may prevent new DUs.
Systemic sclerosis (SSc; scleroderma) is a connective tissue disease with fibrosis and vascular abnormalities in the majority of patients (). Raynaud's phenomenon (RP) occurs in more than 90% of patients with SSc and results from an excessive, vasoconstrictive response to low temperatures and other stimuli, as well as fixed arterial defects due to hypertrophy and occlusion of the vasculature. The characteristic pallor and cyanosis of RP most frequently affects the distal extremities (fingers and toes). In its most severe form, RP can lead to digital ulcers (DUs) (). DUs occur in 15% of SSc patients at any given visit and in 50% of SSc patients ever. They typically present with painful ulcers on fingertips, but can also have skin break down at skin creases, over the proximal interphalangeal joints, or in association with calcinosis (). Complications of DUs include local infection, osteomyelitis, gangrene, or amputation. Disability and reduced hand function may occur, resulting from fibrosis and tissue loss ().
Given that DUs carry high clinical impact, it is very important for health care providers who treat DUs in SSc to have knowledge of potential healing and prevention modalities. The European League Against Rheumatism (EULAR) has published evidence-based guidelines for treatment of RP and DUs, but relative benefits of the treatment strategies described have not been meta-analyzed (). The guidelines did not include more recent studies of phosphodiesterase type 5 (PDE-5) inhibitors. Several trials have investigated the effects of pharmacologic agents on healing and preventing DUs in SSc, often in RP trials where DU treatment was not the primary outcome. A meta-analysis to estimate the effects of drugs for both healing and prevention was performed in order to summarize the clinical trials for SSc DUs.
A comprehensive literature search of the Medline, PubMed, and EMBase databases from 1980 to February 29, 2012, and of the American College of Rheumatology and EULAR abstracts, was conducted using the following search terms: Digital Ulcers (DU) and Raynaud's Phenomenon (RP)/ or Scleroderma/ or Systemic Sclerosis (SSc). Furthermore, key references were hand searched to identify other potentially relevant studies. Studies met inclusion criteria if they were randomized controlled trials (RCTs); had primary, secondary, or exploratory outcomes investigating healing or prevention of DUs in SSc; and compared a pharmacologic therapy with placebo or another active agent. Studies were excluded if there were insufficient data on DUs, including lack of quantitative data. Recent abstracts that did not have full manuscripts published were included if there were sufficient data obtained from the registration of the clinical trial protocol and published abstract(s). Full text articles were retrieved if review of titles and abstracts did not obviously exclude the studies. Two reviewers (TT and JS) separately identified papers for inclusion into the meta-analysis. If there was disagreement, the full papers were read and agreement was achieved by consensus of authors.
The quality of the studies was assessed using the methods described by Jadad et al (). Quality score was based on randomization, blinding, statistical methods, intent-to-treat analysis, and method of randomization. If an article did not explain/discuss a quality item, it was assumed a “0” (zero) score for that item.
Data extraction was performed using a standardized data extraction form. The following data were extracted from each study: first author, year of publication, pharmacologic agent(s) studied, study design (crossover or parallel), total sample size and number with ulcers, specific clinical outcome relating to DUs in SSc, and primary study outcome, noting that many studies were trials of RP where DU outcome measures were secondary or exploratory. Where possible, data on the following outcomes were extracted: the number of new DUs in each group overall and/or the number of patients with new DUs, number of patients with DUs at the beginning and end of the study, improvement in or full or partial healing of DUs, adverse events, and functional assessment using standardized outcomes, such as the Scleroderma Health Assessment Questionnaire ().
Raw numerical data, such as event rates, and pre- and post-number of DUs for within- and between-groups differences were extracted from studies, and the risk ratio (RR) and standardized mean difference (SMD) were calculated where appropriate. RR was done for proportions (such as the proportion of patients with ulcers in 2 groups at the study beginning and end) and the SMD was performed when there were numerical data (such as the number of ulcers that healed in each group or the number of patients who became ulcer free in each group). Studies within the same class of drug were combined if similar outcomes were provided. The random-effects model, as described by DerSimonian and Laird (), was used to calculate the pooled RRs or differences between groups. In this model, the weight of the study is proportional to the inverse of the within-study variance. The I2 statistic was used to examine the between-study heterogeneity (). Values <30% indicate mild heterogeneity, whereas values >50% indicate notable heterogeneity. All statistical analyses were completed using Review Manager software, version 5.0.11.
The electronic database search and initial screening for eligibility identified 60 studies for full text review (see Figure 1 for full search strategy). There were 31 RCTs ([8-37]) that met inclusion criteria for the meta-analysis, with a total of 1,989 participants enrolled (Table 1), but all did not have baseline DUs and, in some trials of RP in SSc, only a few patients had an ulcer even throughout the trial. All patients had a diagnosis of SSc, and 811 patients had a baseline DU (Table 1). Most trials studied RP, with secondary or exploratory outcomes documenting either DU healing or prevention of new DUs. The trial duration ranged from 4 weeks to 3 years (Table 1). Included trials had a mean quality score of 3.65 out of a maximum score of 5 (range 1–5) (Table 1) (). One trial contained SSc and other patients with other autoimmune diseases and RP ().
|Author, year (ref.)||Comparison||Clinical outcome of DUs||Number of participants||Number of participants with baseline DUs||Study quality (Jadad score)||Trial duration|
|Blom-Bulow et al, 1981 ()||Cyclofenil vs. placebo||Healing||Total 27; placebo 27; cyclofenil 27||14||4||12 months (2 × 6 months)|
|Roald and Seem, 1984 ()||Ketanserin vs. placebo||Healing||Total 10; placebo 10; ketanserin 10||5||3||9 weeks|
|Beckett et al, 1984 ()||Dipyridamole and aspirin vs. placebo||Healing||Total 28; placebo 28; ASA and dipyridamole 28||10||5||1–2 years|
|Belch et al, 1985 ()||CL 115,347 PGE2 analog vs. placebo||Severity||Total 29; placebo 14; PGE2 15||Total 8; placebo 3; PGE2 5||4||6 weeks|
|Lukac et al, 1985 ()||Ketanserin vs. placebo||Healing||Total 15; placebo 7; ketanserin 8||Total 7; placebo 3; ketanserin 4||4||3 months|
|Mohrland et al, 1985 ()||PGE1 vs. placebo||Healing, new||Total 55; placebo 28; PGE1 27||Total 14; placebo 10; PGE1 4||3||4 weeks|
|Williams et al, 1985 ()||Topical DMSO vs. placebo||50% healing||Total 84; placebo 31; 2% DMSO 25; 70% DMSO 28||84||3||12 weeks|
|Wollersheim et al, 1986 ()||Prazosin vs. placebo||Healing||Total 24; placebo 24; prazosin 24||9||2||8 weeks (2 weeks × 4)|
|Meyrick Thomas et al, 1987 ()||Nifedipine vs. placebo||New||Total 10; placebo 10; nifedipine 10||Not given||4||20 weeks (6 weeks in each arm, 4 week washout × 2)|
|McHugh et al, 1988 ()||IV iloprost vs. placebo||Healing||Total 20; placebo 9; IV iloprost 11||8/29 at some stage||3||7 weeks (6 hours × 3 days in each arm, 6 week washout)|
|Rademaker et al, 1989 ()||IV iloprost vs. oral nifedipine||Healing (change in mean number of DUs)||Total 23; placebo 23; IV iloprost 12; nifedipine 11||Not given||4||16 weeks (iloprost 8 hours × 3 days then 8 hours × 1 at 8 weeks, oral nifedipine 4–16 weeks)|
|Wigley et al, 1992 ()||IV iloprost vs. placebo||Healing, new||Total 35; placebo 35; IV iloprost 35||Total 11; placebo 4; IV iloprost 7||4||10 weeks (iloprost 6 hours × 5 days, 2 week outpatient washout)|
|Lau et al, 1993 ()||Oral cicaprost vs. placebo||Healing||Total 49; placebo 16; cicaprost 33||Total 5; placebo 2; cicaprost 3||3||4 months|
|Wigley et al, 1994 ()||IV iloprost vs. placebo||Healing, new||Total 131; placebo 67; IV iloprost 64||Total 73; placebo 38; IV iloprost 35||4||12 weeks (14 day pretreatment, 5 day treatment, 9 week followup)|
|Black et al, 1998 ()||Oral cicaprost vs. placebo||Healing||Total 103; placebo 35; oral iloprost 68||Total 28; placebo 9; oral iloprost 19||5||6–12 weeks|
|Vayssairat, 1999 ()||Beraprost vs. placebo||New||Total 107; placebo 52; beraprost 55||0 (must heal more than 1 month before inclusion, but history of DU in last 3 years)||5||179–184 days of treatment, 5-month surveillance|
|Denton et al, 2000 ()||Heparin vs. placebo||Healing||Total 30; placebo 14; heparin 16||Not given (baseline mean 2.0 heparin, control 1.0)||2||24 weeks|
|Scorza et al, 2001 ()||Cyclic IV iloprost vs. nifedipine||Improvement||Total 46; nifedipine 17; IV iloprost 29||Total 17; IV iloprost 14; nifedipine 3||3||12 months (IV iloprost 8 hours × 5 days then 8 hours × 1 every 6 weeks, nifedipine × 12 months)|
|Korn et al, 2004 ()||Oral bosentan vs. placebo||Healing (total population, if baseline); new||Total 122; placebo 43; bosentan 79||Total 77; placebo 24; bosentan 53||4||16 weeks|
|Marasini et al, 2004 ()||IV iloprost vs. alprostadil||Improvement and healing||Total 21; IV iloprost 11; alprostadil 10||Total 7; IV iloprost 5; alprostadil 2||1||60 days (treatment loading × 5 days, maintenance × 2 every 30 days, alprostadil 3 hours/day, iloprost 6 hours/day)|
|Fries et al, 2005 ()||Sildenafil vs. placebo||Improvement and healing||Total 16; placebo 16; sildenafil 16||6||4||9 weeks (treatment 4 weeks in each arm, 1 week washout)|
|Gliddon et al, 2007 ()||Quinapril vs. placebo||New||Total 210; placebo 106; quinapril 104||50||5||2–3 years|
|Abou-Raya et al, 2008 ()||Atorvastatin vs. placebo||New, healing (decrease in mean no. of DUs)||Total 84; placebo 28; atorvastatin 56||Not given (baseline mean 3.4 placebo, 3.3 atorvastatin)||4||4 months|
|Chung et al, 2009 ()||Nitroglycerin topical formulation vs. placebo||New||Total 219; placebo 108; nitroglycerin 111||Not given||4||4 weeks|
|Fiori et al, 2009 ()||Vitamin E vs. placebo||Mean time to healing||Total 27; placebo 12; vitamin E 15||27||1||Unclear, 20 weeks (treated twice a week)|
|Herrick et al, 2009 (), 2011 ()a||Sildenafil vs. placebo||Healing||Total 45; placebo 25; sildenafil 20||Total 8; placebo 3; sildenafil 5||3||28 days|
|Shenoy et al, 2010 ()||Tadalafil vs. placebo||Healing, new||Total 24; placebo 24; tadalafil 24||Total 7; placebo 5; tadalafil 2||5||13 weeks (6 week treatment in each arm, 1 week washout)|
|Correa et al, 2010 ()b||Oral NAC vs. placebo||New||Total 42; placebo 21; oral NAC 21||Not given||3||4 weeks (3 times daily)|
|Matucci-Cerinic et al, 2011 ()||Bosentan vs. placebo||Healing, new||Total 188; placebo 90; bosentan 98||188||5||32 weeks (24 week treatment, 8 week followup)|
|Seibold et al, 2011 ()b||Treprostinil vs. placebo||Healing, new; time to healing||Total 148; placebo 76; treprostinil 72||148||4||20 weeks (primary end point 16 weeks)|
|Bali et al, 2011 ()||IV iloprost vs. placebo||Healing, new||Total 17; placebo 7; IV iloprost 7||Not given||5||4 months (once monthly)|
Statistical analyses for relevant outcomes pertaining to the healing or prevention of DUs are shown for all 31 included studies (Table 2). However, there were only 10 studies ([19, 21, 22, 25, 27, 32-37]) whose outcomes could be combined in forest plots (Figures 2, 3, and 4), displaying pooled RRs or SMDs.
|Author, year (ref.)||Comparison||Statistic (95% CI)||P|
|Blom-Bulow et al, 1981 ()||Cyclofenil vs. placebo||DU healing: RR 2.00 (0.62, 6.45)||0.25|
|Roald and Seem, 1984 ()||Ketanserin vs. placebo||N/A||0.14a|
|Beckett et al, 1984 ()||Dipyridamole and aspirin vs. placebo||DU healing: RR 0.71 (0.38, 1.33)||0.29|
|Belch et al, 1985 ()||CL 115,347 PGE2 analog vs. placebo||DU severity: SMD 0.81 (−0.72, 2.35)||0.3|
|Lukac et al, 1985 ()||Ketanserin vs. placebo||DU healing: RR 5.60 (0.39, 79.70)||0.2|
|Mohrland et al, 1985 ()||PGE1 vs. placebo||DU healing: RR 0.89 (0.27, 2.90) New DU: RR 0.31 (0.03, 2.78)||DU healing: 0.85 New DU: 0.29|
|Williams et al, 1985 ()||Topical DMSO vs. placebo||50% DU healing: RR 1.07 (0.62, 1.85)||0.8|
|Wollersheim et al, 1986 ()||Prazosin vs. placebo||DU healing: RR 15 (0.98, 228.9)||0.05|
|Meyrick Thomas et al, 1987 ()||Nifedipine vs. placebo||New DU: RR 0.50 (0.18, 1.40) SMD −0.62 (−0.57, 1.33)||0.19 (RR) 0.20 (SMD)|
|McHugh et al, 1988 ()||IV iloprost vs. placebo||Insufficient data (nonsignificant difference)||P not given|
|Rademaker et al, 1989 ()||IV iloprost vs. oral nifedipine||DU healing: SDM 0.00 (−0.82, 0.82)||1|
|Wigley et al, 1992 ()||IV iloprost vs. placebo||DU healing: RR 3.00 (0.76, 11.81) New DU: RR 1.18 (0.30, 4.72) SMD: −0.77 (−1.46, −0.08)||DU healing: 0.12 New DU: 0.81 (RR), 0.03 (SMD)|
|Lau et al, 1993 ()||Oral cicaprost vs. placebo||Insufficient data (nonsignificant difference)||Not given|
|Wigley et al, 1994 ()||IV iloprost vs. placebo||DU healing: RR 1.40 (0.58, 3.35) New DU: RR 0.76 (0.44, 1.31)||DU healing: 0.45 New DU: 0.33 (RR) 0.03 (SMD)|
|Black et al, 1998 ()||Oral iloprost vs. placebo||DU healing: RR 1.87 (0.25, 13.78)||0.54|
|Vayssairat, 1999 ()||Beraprost vs. placebo||New DU: SSc subgroup: RR 0.66 (0.37, 1.18) Total population: RR 0.82 (0.57, 1.18)||SSc subgroup: 0.16 Total population: 0.28|
|Denton et al, 2000 ()||Heparin vs. placebo||DU healing: SMD 0.38 (−0.34, 1.11)||0.3|
|Scorza et al, 2001 ()||Cyclic IV iloprost vs. nifedipine||DU improvement: RR 0.95 (0.62, 1.47)||0.83|
|Korn et al, 2004 ()||Oral bosentan vs. placebo||DU healing: RR 0.94 (0.70, 1.26) New DU: total population: RR 0.95 (0.70, 1.30) SMD −0.42 (−0.80, −0.04) If baseline DU: RR 0.32 (0.14, 0.75) SMD −0.57 (−1.06, −0.08)||DU healing: 0.68 (RR) New DU: total population 0.76 (RR) 0.03 (SMD) If baseline DU: 0.008 (RR) 0.02 (SMD)|
|Marasini et al, 2004 ()||IV iloprost vs. alprostadil||DU healing: RR 2.50 (0.17, 37.26) DU improvement: RR 0.95 (0.62,1.47)||DU healing: 0.51 DU improvement: 0.83|
|Fries et al, 2005 ()||Sildenafil vs. placebo||DU healing: RR 5.00 (0.29, 86.43) DU improvement: RR 13.00 (0.89, 89.39)||DU healing: 0.27 DU improvement: 0.06|
|Gliddon et al, 2007 ()||Quinapril vs. placebo||New DU: RR −0.08 (−0.23, 0.06)a||Not given|
|Abou-Raya et al, 2008 ()||Atorvastatin vs. placebo||New DU: RR −0.85 (−1.32, −0.38) Healing DU: RR 0.30 (−0.15, 0.76)||New DU: 0.0004 Healing DU: 0.20|
|Chung et al, 2009 ()||Nitroglycerin topical formulation vs. placebo||New DU: RR 0.69 (0.32, 1.50)||0.35|
|Fiori et al, 2009 ()||Vitamin E vs. placebo||Mean time of DU healing: SMD −2.37 (−3.39, −1.35)||< 0.00001|
|Herrick et al, 2009 (), 2011 ()b||Sildenafil vs. placebo||DU healing: RR 3.33 (0.21, 52.68)||0.39|
|Shenoy et al, 2010 ()||Tadalafil vs. placebo||DU healing: RR 2.67 (0.80, 8.86) New DU: RR 0.20 (0.03, 1.59)||DU healing: 0.11 New DU: 0.13|
|Correa et al, 2010 ()b||Oral NAC vs. placebo||Insufficient data (nonsignificant difference)||Not given|
|Matucci-Cerinic et al, 2011 ()||Bosentan vs. placebo||DU healing: RR 0.94 (0.65, 1.35) SMD 0.19 (−0.10, 0.47) New DU: RR 0.93 (0.76, 1.13)||DU healing: 0.73 (RR) 0.20 (SMD) New DU: 0.45 (RR)|
|Seibold et al, 2011 ()b||Treprostinil vs. placebo||DU healing: RR 1.21 (0.84, 1.73) New DU: RR 0.95 (0.71, 1.26) Time to DU healing: SMD −0.19 (−0.51, 0.14)||DU healing: 0.30 New DU: 0.72 Time to DU healing: 0.26|
|Bali et al, 2011 ()||IV iloprost vs. placebo||Insufficient data (nonsignificant difference)||Not given|
We analyzed 3 RCTs comparing various PDE-5 inhibitors to placebo treatment on DU improvement (defined or not within each paper) and healing. The first trial compared sildenafil (50 mg) to placebo (n = 16) in a crossover trial (); the second compared modified-release sildenafil (100 mg initially then increased to 200 mg once daily) versus placebo (n = 45) ([32, 33]); and the third studied tadalafil (20 mg) and placebo (n = 24) in a crossover trial () (Table 1). One publication did not provide DU data, but the data were obtained from a presented abstract. Although the 3 trials were individually underpowered to detect a statistically significant benefit, the pooled effect showed a definite benefit of PDE-5 inhibitors on the number of patients with both DU healing (RR 3.28 [95% confidence interval (95% CI) 1.32, 8.13], P = 0.01) and DU improvement (RR 4.29 [95% CI 1.73, 10.66], P = 0.002) (Figure 2). There were significant serious adverse events with PDE-5 inhibitors, and a total of 7 patients from the 3 trials withdrew as a result of potential treatment-related adverse events, including headaches, myalgias, nonpainful erections, allergic reaction, chest pain, palpitations, and facial edema ([27, 32-34]).
There were 2 RCTs comparing the effect of bosentan to placebo on DU healing and prevention ([25, 35]). Both trials began with 62.5 mg of bosentan or matched placebo twice daily for 4 weeks, but the duration of the subsequent increased dosage (125 mg twice daily) was 12 weeks in the first trial (n = 122) () and 20 weeks in the second trial (n = 188) (). There was no statistically significant difference in DU healing for both trials. A meta-analysis of these 2 trials showed that bosentan was successful in DU prevention, with a statistically significant reduction in the mean number of new DUs per patient (SMD −0.34 [95% CI −0.57, −0.11], P = 0.004) (Figure 3). The effect seemed similar for patients who already had baseline DUs (SMD −0.36 [95% CI −0.61, −0.11], P = 0.005) compared to the overall group, but the second trial had all patients enrolled with a baseline DU and approximately half of the first trial enrolled patients with a current DU. The reduction for new DUs in patients without a current DU was smaller (data not shown). The reduction in the number of patients with a new DU was not statistically significant. The proportion of patients experiencing at least 1 adverse event was similar in both bosentan and treatment groups (). The most common significant adverse events within those treated with bosentan included abnormal liver enzymes and edema/fluid retention ([25, 35]).
There were 11 RCTs with a total of 700 patients comparing iloprost to placebo or another active pharmacologic agent, but the trials were for treatment of RP, so many did not have a DU during the trials ([17-22, 24, 26, 36, 37]). The active treatment in most trials was iloprost; 4 studies compared intravenous (IV) iloprost to placebo ([17, 19, 37]). One study compared oral iloprost to placebo ([21, 36]), 2 studies compared IV iloprost to nifedipine ([18, 24]), and 1 study compared IV iloprost to alprostadil (). Three other studies compared beraprost (), cicaprost (), and treprostinil () to placebo. None of the studies showed a statistically significant difference in DU healing or improvement individually or pooled in forest plots (Table 2 and Figure 4). Only IV iloprost prevented new DUs (SMD −0.77 [95% CI −1.46, −0.08], P = 0.03) (Table 2). There were insufficient data in 3 trials to extract DU data ([17, 20]). The 2 trials comparing IV iloprost to nifedipine ([18, 24]) could not be combined into a forest plot due to heterogeneity in clinical outcomes, as well as study duration (16 weeks  and 12 months ) (Table 1).
In our meta-analysis, there was only 1 study comparing nifedipine to placebo () that found no statistically significant difference in the prevention of new DUs, but the data were from a trial of RP and few ulcers occurred (which means the trial is underpowered) (Table 1). There were 2 other studies ([18, 24]) that compared 2 active pharmacologic treatments against each other, namely IV iloprost versus nifedipine, but did not compare CCBs to placebo (Table 1).
Single trials investigated DUs in patients with RP using cyclofenil (), aspirin (), dipyridamole (), prostaglandin E2 (PGE2) analog (), PGE1 (), low molecular weight heparin (), prazosin (), ketanserin (2 trials, but only 1  had data to extract on both active and placebo groups for DUs [[12, 38]]), topical vitamin E (), a topical nitroglycerin formulation (), quinapril (), atorvastatin (), oral N-acetylcysteine (), and topical dimethyl sulfoxide () were all compared to placebo. There were 2 trials, vitamin E () and atorvastatin (), that showed statistically significant differences in clinical outcomes in comparison to placebo. The mean time to DU healing with vitamin E gel was significantly less than placebo (P < 0.00001) (Table 2). Although there were no statistically significant differences in DU healing, atorvastatin treatment did prevent new DUs in comparison to placebo, with mean ± SD 1.6 ± 0.9 DUs in atorvastatin versus 2.5 ± 1.3 (P < 0.0004) in 56 treated with active and 28 treated with placebo, with an SMD (in fewer new DUs) of 0.85 (95% CI [of reduced number of DUs] 1.32, 0.38) favoring atorvastatin. Funnel plots did not suggest publication bias (see Supplementary Figure 1, available in the online version of this article at http://onlinelibrary.wiley.com/doi/10.1002/acr.22018/abstract).
As DU complications in SSc are common (), several treatments for healing and prevention of both RP and DUs have been explored, including CCBs, antiplatelet and anticoagulant therapies, ERAs, PDE-5 inhibitors, and statins (). This meta-analysis of RCTs summarizes the treatment effect of various pharmacologic agents in the healing and prevention of DUs in RP.
CCBs, which have a direct effect on vascular smooth muscles and inhibit platelet activation, are considered first-line treatment for RP in SSc patients (). They have been tested in several small trials with a significant clinical improvement in the frequency and severity of RP attacks. However, data were insufficient to support a role in reducing new DUs (), given the small sample size in a single study, so larger RCTs are required to estimate the role of CCBs on DUs. Years ago, a trial was performed comparing nifedipine to placebo in RP, including SSc patients, which helped to provide the standard of care in RP treatment with nifedipine; but there was not extractable data on digital ulcers in this landmark trial (). The EULAR Scleroderma Trial and Research SSc guidelines suggest that nifedipine be used as a first-line treatment in RP for SSc patients () and experts have suggested that nifedipine be used as initial treatment of DUs ().
There was a strong statistically significant difference (P < 0.002) in the healing or improvement of DUs with PDE-5 inhibitors compared to placebo in 3 RCTs ([27, 32-34]). These are small studies where RP was the primary outcome, so results can only be interpreted with caution. PDE-5 inhibitors are approved for pulmonary arterial hypertension (PAH) and modulate cGMP (). Randomization was not stratified according to baseline ulcer presence or absence, which is a potential bias. There is a completed registered trial of sildenafil in DUs, but the results were not available at the time of the search for this meta-analysis. The EULAR Scleroderma Trial and Research guidelines do not include PDE-5 inhibitors for DU treatment (), but PDE-5 inhibitors were frequently chosen by SSc experts in the treatment and prevention of DUs after failure of CCBs ().
Bosentan, an endothelin receptor antagonist, reduces the number of new DUs in SSc patients, especially if there is a baseline DU at trial entry. Elevated levels of endothelin have been found in the serum of patients with SSc and may be a marker of severe vascular injury (). Bosentan has relatively good tolerability and is given orally (). Previous smaller, uncontrolled studies suggested some benefit in the number of healed DUs (), but the 2 RCTs did not. There is no effect on healing of DUs. A factor-limiting use of bosentan for DU prevention is access, as it is costly and not approved in North America for this indication, although it is approved in Europe.
Iloprost, a prostacyclin analog given intravenously, is a potent vasodilator with platelet inhibitor properties, as well as efficacy in severe RP in SSc (). The standard administration of IV iloprost is for 6-hour infusions of 0.5–2 ng/kg/minute daily for 5 days ([43, 47, 48]). Our meta-analysis also examined oral prostacyclins (including oral iloprost, beraprost, cicaprost, and treprostinil; the latter had not been published at time of search). Only IV iloprost was found to have a statistically significant effect in preventing new DUs (), but many of the trials were likely underpowered for DUs.
Overall, there are many limitations to this meta-analysis. Given the small sample size in many studies, small treatment effects, and inclusion of many RP trials that studied DU less systematically, a large trial (yet to be performed) or a missed trial could have changed the results. There could also be publication bias where studies with positive findings are more likely to be published (however, the funnel plots did not demonstrate publication bias). The definitions of DU healing, partial healing, and number of DUs (in total, per person, or number of people with DUs) and variable length of the trials all could affect results. For instance, a short trial may have had less opportunity for healing, whereas a longer trial could have more healing and new ulcers. Examining time to healing and time to ulcer development may be better outcomes in future trials. Additionally, the small number of studies for each drug class and heterogeneity of the studies limit conclusions.
Furthermore, we had to exclude 10 RCTs: 8 did not include DU subanalyses and 2 compared the same pharmacologic agent at different doses. Of the 31 RCTs meeting inclusion criteria, 5 had insufficient data to analyze. Most drug classes were only represented by a single trial and were unable to be pooled. This may contribute to the lack of statistical significance in clinical outcomes for DU healing or prevention with various treatments ([8-16, 23, 26, 28, 30]). Although topical vitamin E () and atorvastatin () showed statistically significant benefits, each had a single, small RCT, so results may not be reproducible or generalizable. The 2 trials comparing IV iloprost to oral nifedipine could not be combined because of different definitions for DU healing: Rademaker et al () used the change in mean number of DUs per patient, whereas Scorza et al () used the number of people with DU improvement. All results were also re-analyzed using a fixed-effects model that yielded similar results (data not shown).
It could be beneficial to examine the role of combination therapies in DUs, such as sildenafil and bosentan, which have been shown to potentially be of benefit in PAH, and in a case report to have had a possible synergistic effect (). DU end points are not standardized and are somewhat subjective (i.e., when to classify an ulcer as partially healed versus fully healed; the depth and size are difficult to measure). The data may be helpful in designing future (better) studies and informing clinicians about both the lack of data and positive, but potentially flawed, results. For instance, PDE-5 inhibitors may be important in SSc DU treatment, and statins may be a class worth exploring further for the prevention of SSc-associated DUs.
In summary, our meta-analysis of randomized trials for the pharmacologic management of DUs in SSc patients (mostly subanalyes of RP trials) identified 31 RCTs, with 10 RCTs whose outcomes could be combined in forest plots. In comparison to placebo, statistically significant improvements in DU healing and/or improvement have been shown with PDE-5 inhibitors and vitamin E, whereas bosentan and IV iloprost prevented new DUs. There was a great degree of heterogeneity among the trials, and several were underpowered, as DU outcomes were not part of the primary analyses within most trials. Standardized outcome measures and larger, multicenter, randomized trials of high quality may be helpful in defining healing and prevention strategies for DUs in SSc with several potentially effective classes of drugs.
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 submitted for publication. Dr. Pope 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. Tingey, Shu, Smuczek, Pope.
Acquisition of data. Tingey, Shu, Smuczek, Pope.
Analysis and interpretation of data. Tingey, Shu, Smuczek, Pope.