Article Text


Concise report
Phosphodiesterase-5 inhibitors for the treatment of secondary Raynaud's phenomenon: systematic review and meta-analysis of randomised trials
  1. Matthieu Roustit1,2,
  2. Sophie Blaise1,3,
  3. Yannick Allanore4,
  4. Patrick H Carpentier3,
  5. Evren Caglayan5,
  6. Jean-Luc Cracowski1,2
  1. 1Inserm U1042, Université Joseph Fourier, Grenoble, France
  2. 2Clinical Pharmacology Unit, Inserm CIC03, Grenoble University Hospital, Grenoble, France
  3. 3Vascular Medicine Department, Grenoble University Hospital, Grenoble, France
  4. 4Inserm U1016, Service de Rhumatologie A, Hôpital Cochin, Université Paris Descartes, Paris, France
  5. 5Klinik III für Innere Medizin, Uniklinik Köln, Köln, Germany
  1. Correspondence to Dr Matthieu Roustit, Unité de Pharmacologie Clinique, Centre d'Investigation Clinique de Grenoble—Inserm CIC03, CHU de Grenoble, 38043 Grenoble Cedex 09, France; MRoustit{at}


Introduction Recent controlled trials have assessed the efficacy of phospodiesterase-5 (PDE-5) inhibitors in secondary Raynaud's phenomenon (RP). However, the conclusions are conflicting, and whether these drugs are effective remains unclear. The objective of this meta-analysis was to determine the efficacy of PDE-5 inhibitors on Raynaud's Condition Score (RCS) and frequency and duration of attacks.

Methods A systematic review of articles was performed (sources included Medline, Embase, Web of Science, the Cochrane Central Register of Controlled Trials). Only double-blind, randomised controlled trials (RCTs) were included. Studies were selected independently by two authors using predefined data fields, including study quality indicators.

Results Six RCTs were included (one with sildenafil, one with modified-release sildenafil, three with tadalafil and one with vardenafil). PDE-5 inhibitors significantly decreased mean RCS by −0.46 (−0.74 to −0.17) (p=0.002), the daily frequency of ischaemic attacks by −0.49 (−0.71 to −0.28) (p<0.0001), and daily duration of RP attacks by −14.62 (−20.25 to −9.00) min (p<0.0001).

Conclusions PDE-5 inhibitors appear to have significant but moderate efficacy in secondary RP. A further large RCT is needed.

  • Systemic Sclerosis
  • Cardiovascular Disease
  • Outcomes research

Statistics from


Raynaud's phenomenon (RP) is characterised by transient ischaemia in the extremities in response to cold or emotions. It can be primary (idiopathic) or secondary to a connective tissue disease, especially systemic sclerosis (SSc).1 While primary RP has in most cases a benign course, secondary RP is associated with significant morbidity (eg, ulcerations and gangrene), especially where functional and structural microvascular abnormalities combine.1

Calcium channel blockers (CCBs) are recommended as the first-line treatment.1 However, a meta-analysis showed only moderate efficacy of CCBs in reducing the frequency of SSc-related RP, with an average decrease of 4.1 attacks per week.2 Intravenous iloprost reduces the frequency and severity of attacks and is recommended for severe SSc-related RP with active digital ulcers.3 Bosentan failed to show efficacy on RP despite some effect in preventing digital ulcers.3

Nitric oxide is a potent vasodilator, as it generates cyclic guanosine-5-monophosphate (cGMP). As cGMP is principally hydrolysed by phosphodiesterase-5A (PDE-5A), PDE-5 inhibitors potentiate the effect of nitric oxide by decreasing cGMP metabolism, leading to enhanced vasodilatation and increased blood flow.

Three PDE-5 inhibitors (sildenafil, tadalafil and vardenafil) have been marketed for erectile dysfunction, and, more recently, pulmonary hypertension has become another indication for sildenafil and tadalafil. In the last 10 years, case reports and small series have described successful treatment of RP with sildenafil or tadalafil.4 More recently, controlled studies have been published, but their conclusions are conflicting and the efficacy of PDE-5 inhibitors remains unclear.5 This meta-analysis aimed to determine whether PDE-5 inhibitors can decrease the Raynaud's Condition Score (RCS) and the frequency and duration of ischaemic attacks in secondary RP.


This systematic review followed the PRISMA statement guidelines.6 The following databases were consulted: Medline (1950–2012), Embase (1966–2012), Web of Science (1975–2012) and the Cochrane Central Register of Controlled Trials (1975–2012). Additional sources included: databases of abstracts of the European League Against Rheumatisms (2002–2012) and the American College of Rheumatology (2006–2011); and the registry. The search terms were: Raynaud; sildenafil; tadalafil; vardenafil; phosphodiesterase inhibitors; PDE5 inhibitors; PDE-5 inhibitors. No language, publication date, or publication status restrictions were imposed. The last search was run on 12 October 2012.

Two reviewers (MR, J-LC) independently selected studies using a standardised form. Eligibility criteria included parallel or cross-over double-blind, randomised controlled trials (RCTs) assessing the efficacy of PDE-5 inhibitors (sildenafil, tadalafil or vardenafil) on secondary RP. Outcomes were the RCS and the frequency and duration of ischaemic attacks.

Data were extracted by one author (MR) using RevMan (V.5.1, The Cochrane Collaboration, 2011) and checked by another author (SB). From each included study we extracted: (1) patient characteristics (aetiology, concomitant treatments); (2) intervention type (studied drug and comparator, treatment duration, dose regimen); (3) outcomes (RCS, frequency and duration of attacks). The risk of bias was rated as low, unclear or high for the following items: randomisation; blinding; incomplete outcome data; selective reporting.

All analyses were performed using the generic inverse variance method (RevMan 5.1). For cross-over trials, paired SE of the mean difference was approximated from the t statistic.7 In one study it was not available,8 and SE was estimated from the lowest correlation between treatment and placebo calculated from all other cross-over studies,9–11 as previously described.7 In one parallel study,12 SD was not available and was imputed from the mean of the known SDs. Heterogeneity (χ2 test) and inconsistency (I2) were tested. The risk of publication bias was assessed by plotting mean difference by the inverse of its SE to obtain funnel plots. Symmetry of plots was assessed both visually and formally (Egger's method, ie, linear regression of normalised effect estimate against precision). All tests and CIs were two-sided. p<0.05 was considered significant.


Of 142 screened citations, 22 were assessed for eligibility, and six trials including 244 patients were finally selected.8–13 The main reasons for exclusion were open-label design (N=8), case reports or retrospective series (N=4), different outcomes from those reviewed (N=2), review article (N=1) and redundancy (N=1). One study available as an abstract was eligible but not included because of incomplete data (authors were contacted).

Study, patients and treatment characteristics are summarised in table 1. All the cross-over studies used a 1-week9–11 or 2-week8 wash-out period. Most patients (91.8%) had SSc-related RP, and 4.9% had mixed or undifferentiated connective tissue disease. Data from patients with primary RP (3.3%) were not included. All but one study9 included patients with digital ulcers. All studies specified discontinuation of other vasodilators for RP before the start of the protocol, except two trials assessing the efficacy of tadalafil as add-on therapy (all patients received CCBs).11 ,13 In another study, seven patients were taking CCBs, and eight patients were taking ACE inhibitors for other indications.8

Table 1

Characteristics of the double-blind, randomised controlled studies assessing the efficacy of PDE-5 inhibitors in RP

The risk of bias was low for all items except in two studies,12 ,13 where unbalanced baseline characteristics for one outcome might reflect inadequate randomisation (unclear risk). One had an unclear risk of attrition bias because of a lack of data.13 Blinding was compromised in one cross-over study, as all patients correctly identified whether they were on sildenafil or placebo10 (high risk of detection bias as outcomes were patient-reported). Moreover, the risk of carry-over or period effect was unclear for this study. Of note, despite elevated attrition in the sildenafil group in one study, intent-to-treat and per-protocol populations are well described and both analyses provided consistent results (although not detailed).12

The meta-analysis of selected trials shows that PDE-5 inhibitors significantly decrease RCS by −0.46 (95% CI −0.74 to −0.17) (p=0.002), daily frequency of ischaemic attacks by −0.49 (95% CI −0.71 to −0.28) (p<0.0001), and daily duration of RP attacks by −14.62 (95% CI −20.25 to −9.00) min (p<0.0001) (figure 1). There was mild to moderate heterogeneity, and visual and formal analyses of funnel plots did not suggest any risk of publication bias.

Figure 1

Forest plots of phosphodiesterase-5 (PDE-5) inhibitors vs placebo for (A) Raynaud's Condition Score, (B) daily frequency of Raynaud's phenomenon (RP) attacks and (C) cumulative daily duration of RP attacks. IV, inverse variance.

Sensitivity analyses were performed removing successively each study presenting a high or unclear risk of bias. Another sensitivity analysis assessed the impact of the assumed correlation for Schiopu et al8 by repeating the analysis assuming no correlation (ie, equivalent to a parallel design). None of these analyses significantly changed the results for any outcome.


Six trials assessing the efficacy of PDE-5 inhibitors in RP fulfilled inclusion criteria. Overall, the treatment shows significant benefit on RCS and frequency and duration of RP attacks. PDE-5 inhibitors reduced the frequency of RP attacks by ∼0.5/day compared with placebo, which is a comparable reduction to that found by another meta-analysis assessing the efficacy of CCBs in SSc-related RP (∼0.6/day).2 The benefit on RCS is significant but below the minimally important difference (estimated to be ∼1.4–1.5).14

Of interest, some studies suggested a beneficial effect of PDE-5 inhibitors on digital ischaemic ulceration (assessed as a secondary outcome measure).10 ,11 ,13 However, data are scarce, and larger RCTs are ongoing to address this issue ( identifier NCT01295736). In the same way, more data are needed to perform subgroup analyses between patients with limited cutaneous SSc (lcSSc), diffuse cutaneous SSc (dcSSc) and mixed/undifferentiated connective tissue disease. Only one study provides details showing a higher reduction in RCS for patients with lcSSc.9

We included only double-blind RCTs in order to limit heterogeneity and increase the reliability of the estimated treatment effect (substantial placebo effects have been observed in drug trials for RP). We excluded patients with primary RP to limit heterogeneity, and because patients with secondary RP require highly effective therapy.

Only half of the studies prespecified minimal RP attack frequency as an inclusion criterion. However, it was far beyond the mean baseline frequency (which was quite homogeneous between studies) and thus unlikely to affect the homogeneity of the population. In the same way, when sufficient description of outcome measures was provided, it appears that attacks were recorded with similar methods and averaged over 1-week periods.

Another source of heterogeneity could be the differences in treatments. Indeed, we included trials assessing the efficacy of sildenafil, tadalafil and vardenafil. One of the main pharmacological differences between these drugs is their half-lives, ranging from 3–5 h for sildenafil and vardenafil to ∼18 h for tadalafil. Consequently, dose regimens were different between studies, with two administrations/day for sildenafil and vardenafil9 ,10 and one each day8 or on alternate days11 ,13 for tadalafil. Modified-release sildenafil was used in one study and administered once a day.12

Although most of the selected studies had a cross-over design, the risk of carry-over was limited by a 1- or 2-week washout period, which seemed adequate. Moreover, carry-over was tested and was not significant in three cross-over studies,8 ,9 ,11 despite a trend in one of these.9 The fourth cross-over study did not report carry-over testing,10 and sensitivity analysis omitting this trial did not change the results.

In conclusion, this meta-analysis shows that PDE-5 inhibitors significantly improve RCS and frequency and duration of RP attacks compared with placebo in secondary RP. However, this effect is moderate. More data are needed to clarify the class effect of PDE-5 inhibitors in secondary RP.


We thank Dr Vikas Agarwal who kindly provided unpublished data, as well as Alison Foote for correction of English language usage.


View Abstract


  • Handling editor Tore K Kvien

  • Contributors MR and J-LC performed the study search. MR extracted the data. SB and EC checked the data. MR performed data analysis. All authors wrote the manuscript.

  • Competing interests MR and J-LC have received research grants from Pfizer, Actelion, GlaxoSmithKline and Bioprojet for other studies. YA received research grants and honoraria from Pfizer and Actelion for other studies.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.