In Raynaud phenomenon, on-demand sildenafil did not reduce disability or frequency or duration of attacks

Abstract

Question In moderate-to-severe primary or secondary Raynaud phenomenon (RP), does on-demand sildenafil reduce disability and frequency and duration of attacks compared with placebo? Methods Design Series of randomized, placebo-controlled, multiple crossover n-of-1 (single patient) trials. ClinicalTrials.gov NCT02050360. Allocation Concealed.* Blinding Blinded* (patients and investigators). Follow-up period Mean 61 days of observation. Setting Grenoble Alpes University Hospital, Grenoble, France. Patients 41 outpatients 18 years of age (mean age 47 y and 74% women in patients analyzed) who had primary or secondary RP (LeRoy and Medsger criteria) with 7 attacks/wk on 5 d/wk in the past 2 weeks. Exclusion criteria included secondary RP diagnosed in the past 2 months, pulmonary hypertension, hypertension or coronary disease, unstable diabetes, history of nonarteritic anterior ischemic optic neuropathy, use of phosphodiesterase-5 inhibitors for pulmonary hypertension or erectile dysfunction, or use of nitrates or potent cytochrome P450 3A inhibitors. Interventions On-demand sildenafil, two 40-mg doses/d taken 4 hours apart and 90 minutes before a situation expected to trigger an attack or 5 minutes after the start of an attack; on-demand sildenafil, two 80 mg doses/d; or placebo. A treatment block included all 3 treatments, each given for 1 week in randomized order, with a 48-hour washout period between treatments. Each patient received 2 to 5 treatment blocks. Outcomes Outcomes included Raynaud s Condition Score (RCS) and frequency and duration of RP attacks over 24 hours. Patient follow-up 93% completed 2 blocks of treatment and were included in the analysis. Main results 68% of patients had primary RP and 32% had secondary RP. The main results for data aggregated across n-of-1 trials using a Bayesian analysis are reported in the Table. Conclusion In Raynaud phenomenon, aggregated results showed that on-demand sildenafil did not reduce disability or frequency or duration of attacks compared with placebo. Aggregated data for on-demand sildenafil vs placebo in Raynaud phenomenon (RP) Outcomes Sildenafil, 40 mg, vs placebo at a mean 61 d Absolute difference (95% CrI) aRV (CrI) Probability of superiority Mean RCS (MCID =1.5) 0.14 (0.33 to 0.07) 0.92 (0.81 to 1.04) 91% RP attacks/d 0.1 (0.2 to 0.05) 0.91 (0.80 to 1.04) 93% Duration of RP attacks, min/d 4.4 (9.2 to 0) 0.90 (0.79 to 1.00) 97% Sildenafil, 80 mg, vs placebo at a mean 61 d Mean RCS (MCID = 1.5) 0.05 (0.21 to 0.17) 0.97 (0.88 to 1.10) 63% RP attacks/d 0.1 (0.2 to 0.05) 0.93 (0.83 to 1.04) 91% Duration of RP attacks, min/d 3.9 (8.3 to 0.9) 0.91 (0.81 to 1.02) 93% aRV = adjusted relative variation; CrI = credible interval; MCID = minimal clinically important difference; RCS = Raynaud s Condition Score (lower is better). aRVs from individual n-of-1 trials were aggregated and analysed using Bayesian methods. Analyses were adjusted for sex, age, secondary RP, calcium-channel blocker use, and daily outdoor temperature and humidity. Baseline values: mean RCS 2.9, mean attacks/d 1.8, mean attack duration 67 min/d. aRV <1 indicates benefit with sildenafil. An aRV of 0.90 indicates a 10% relative reduction for the measured outcome with sildenafil vs placebo. Probability that sildenafil will be more effective than placebo. Visual analogue scale score (score range 0 = no disability to 10 = maximal disability; MCID about 1.5 points). Commentary Roustit and colleagues presented results for an elegant series of n-of-1 randomized controlled trials (RCTs) designed to ascertain treatment response in individual patients. The approach typically involves repeatedly crossing a patient over between treatments until patients and clinicians convincingly establish the magnitude, if any, of treatment effects. Roustit and colleagues evaluated on-demand sildenafil for RP. Their findings were limited by inclusion of patients with primary RP (68% of patients), which is typically less severe than secondary RP (associated with mixed connective tissue disease), probably has a different pathogenesis, and has limited evidence to suggest a role for sildenafil (1, 2). The results showed how alternative presentations may convey different impressions. Roustit and colleagues used Bayesian analysis, which is less commonly used than frequentist analysis and involves specifying prior probabilities and labeling results with credible intervals (CrIs) rather than confidence intervals. In the analysis, all CrIs included no effect, yet the probability of benefit was >90% for all but 1 outcome. This suggests that on-demand sildenafil probably has an effect on RP attacks, but some doubt remains. If the drug has an effect, group data suggest that the magnitude is small: <10% relative reduction in RCS or duration or frequency of RP attacks (i.e., 1 less RP attack in any 10-day period in patients with a mean attack frequency of 1/d). N-of-1 RCTs are suited to address variability in response across patients. Although the group effect of sildenafil was small (if it exists), some patients may have substantial benefit. Unfortunately, this is unlikely. The largest individual effect for RP attack frequency was about a 20% relative reduction (i.e., about 1 less attack every 5 days in patients with a mean attack frequency of 1/d). The absolute effect could be larger in patients with more frequent attacks, but wide CrIs in effect estimates for individual patients mean that even modest effects are uncertain. Sildenafil also increased headaches: About 50% of patients had headaches when they received sildenafil compared with only 12% in the placebo periods. This may be the final nail in the coffin for use of on-demand sildenafil in primary RP.