{"paper_id":"e11419d7-12a9-4ab6-aa8c-66ce99dcbe94","body_text":"This study was designed as a parallel, double-blind, randomized, placebo-controlled trial with a 1:1 ratio.\nWomen between the ages of 20 and 50 with laparoscopic-proven diagnosis of endometriosis were eligible for the study. Subjects were excluded if they were pregnant, had known allergy to resveratrol or contraindications to monophasic contraceptive pills, according to World Health Organization eligibility criteria [ 26 ], or used GnRH agonists or danazol in the last month. The use of medroxiprogesterone acetate depot in the last 12 weeks and current use of Mirena® were added as ineligibility criteria after the trial started because both could introduce bias in the reporting of pain scores. Pregnancy was also an exclusion criteria because it is a contraindication for the use of COCs.\nThe trial took place at the Clinical Research Center of the Hospital de Clínicas de Porto Alegre, Brazil, between 18 June and 6 November 2015.\nAt the first consultation (day 1), pregnancy was ruled out with a negative urinary pregnancy test. After a standard questionnaire, subjects were randomized to receive the first part of the treatment: one pack of COC (levonorgestrel 0.15 mg/ethinyl estradiol 0.03 mg) to be taken daily for 21 days without pause and a coded bottle that contained 21 identical capsules of either 40 mg of resveratrol or placebo. Capsules were prepared by a local pharmacy. Patients started both medications (contraceptive pill and a capsule) on day 1. They were instructed to record any side effects and to use analgesics as needed. Specific analgesics were prescribed, and subjects were instructed to use them in the following order: dipyrone 500 mg up to 6/6 hours, ibuprofen 600 mg up to 8/8 hours, and codeine 30 mg up to 6/6 hours. Subjects were also instructed to report the use of any other medication and to return on day 21 and again on day 42 of treatment. Seven days after the first consultation, patients were contacted by telephone to determine if they had any early complications of therapy and to provide initial pain scores. On day 21, patients returned to receive the second installment of the treatment. On day 42, patients returned for the final visit. Compliance was verified by the presence of empty packs of contraceptive pills and the capsules in the coded bottle on days 21 and 42.\nPain was assessed using a visual analog scale (VAS) scale as previously described [ 27 ] and was the primary outcome. VAS is the most commonly used instrument for the assessment of pain in endometriosis [ 28 ]. Briefly, a standard question was read by one of the investigators, and a 10-cm slide ruler, marked at regular intervals in millimeters, was presented to the patient. A scale of 0 to 10 was printed on the back of the ruler, facing the investigator, and a pain intensity drawing without numbers was facing the patient. Subjects were asked to move the slide somewhere along the scale to described their average pain in the last 7 days. Pain scale varied from 0 (no pain) to 10 (worst imaginable pain). The pain scores were evaluated on days 1, 7, 21, and 42, using the same standard question. The follow-up period of 42 days was based on two consecutives cycles of COCs. Reduction of levels of pain caused by endometriosis has been observed after the first month of treatment by many authors using different hormonal treatments [ 29–34 ].\nSerum levels of CA-125 and prolactin were measured by enzyme-linked immunosorbent assay at a local endocrine laboratory using electrochemiluminescence immunoassay (Elecsys 2010, Modular Analytics E170, Roche Diagnostic GmbH, D-68305, Indianapolis, IN), with a sensitivity of 0.6 U/mL and 0.047 ng/mL for CA-125 and prolactin, respectively. Repeatability and intermedia precision in both methods were <2 for CA-125 and prolactin, respectively. Serum levels were measured on days 1 and 42.\nThe use of analgesics and side effects of COCs plus placebo or resveratrol were compared in both groups after entering the trial.\nReduction of pain after 7, 21, and 42 days was compared with baseline pain scores in each group.\nA sample size was calculated for a superiority trial of resveratrol over placebo for a final difference of pain score between groups, based on data derived from Maia  et al . [ 23 ]. A sample size of at least 21 patients per arm would be necessary to have a 90% chance of detecting, as significant at the 1% level, a decrease in the median pain levels from 5 in the placebo group to a median pain level of 2 in the resveratrol group, after 42 days of treatment. A standard deviation of 2.5 was obtained from a pilot study with 10 cases. Due to the nature of resveratrol, a nutritional supplement available as an over-the-counter drug, no stopping guidelines were established.\nParticipants were randomly assigned to resveratrol (treatment A) or placebo (treatment B) according to a computer-generated randomization list ( www.randomization.com ). Subjects were allocated in blocks of four, as suggested in the literature [ 35 ].\nAllocation sequence was concealed using coded, numerically sequenced, opaque-sealed envelopes, which were opened after patients were included in the study. Two investigators, blinded to treatments, were responsible for assignment, enrollment, and follow-up of the subjects.\nNormal distribution of data was verified with a Kolmogorov-Smirnov normality test. If normal distribution was not present, data were transformed into a logarithmic scale for CA-125 and prolactin levels. Direct comparison between pain scores in both groups at day 1 and day 42 was performed using a Mann-Whitney  U  test and unpaired  t  test with Welch’s correction, according to data distribution. Data on CA-125 and prolactin were analyzed using repeated-measures, two-way analysis of variance (RM-2 Way ANOVA), because we compared the effect of resveratrol and placebo on the plasma concentration of CA-125 and prolactin on days 1 and 42.\nGeneralized estimating equations (GEEs) were used for further statistical analysis, having the patients as the subject variable, time (visits), as within-subject variable, pain scores (VAS) as the dependent variable, treatment type (placebo or resveratrol), use of medication for pain (yes/no), and time were the factors analyzed (visits on days 0, 7, 21, and 42). The analysis used a linear-scale response for pain score and a factorial model analysis for the analyzed factors. A maximum-likelihood estimate, where estimates and standard errors are based on the likelihood function given the observed data, was used in the analysis, as suggested in the literature [ 36 ]; correction for multiple comparisons was made using a Bonferroni test. GEE is recommended for missing data in clinical trials, because calculation with missing data is considered in the GEE algorithm.\nSensitivity analysis was performed on data with missing values,  i.e. , intention to treat (ITT), and compared with data without missing values [per protocol (PP)]. In PP analysis, only patients who complied with trial protocol were considered in the analysis. In ITT analysis, patients who were lost on follow-up had their last registry on pain values or plasma level measurements repeated in the follow-up consultation. If no systematic differences were observed between participants with complete data and those with missing information, data were considered to be missing at random. Data were analyzed using GraphPad Prism 6.0 for Mac (GraphPad Software, La Jolla, CA) and SPSS version 23 for Mac (IBM Software Group, Chicago, IL).\nThis study was approved by the Ethics Committee of the Hospital de Clínicas de Porto Alegre (no. 14-626) and was registered at ClinicalTrials.gov (no.  NCT02475564 ).\n\nTwo hundred and forty-three women with diagnosis of endometriosis were screened for acceptance into the study between 18 June and 6 November 2015. Of these, 199 were excluded for not meeting the inclusion criteria. Forty-four subjects were randomly assigned to one of the two treatment regimens. Forty-three women received treatment as allocated. Five women were removed from the PP analysis. One discontinued the study after signing the informed consent, one was removed from the study for intense abnormal uterine bleeding that stopped after medical intervention, and three were lost to follow-up ( Fig. 1 ).\nFlow diagram of study population randomized to two treatment groups.\nFor each group, 22 subjects were analyzed as ITT. Outcomes PP were analyzed in 21 and 18 subjects in the resveratrol and placebo arms, respectively.\nBaseline data of the studied population are depicted in  Table 1 .\nDemographic and Clinical Characteristic of the Studied Population\nRaw values are median (range). Analysis after transformation using  Y  = log( Y ) was performed, and no difference was found using Student  t  test. Transformed data are shown in  Figure 2 .\nn = 21 for serum levels; one case of the placebo group dropped out before collecting blood.\nMann-Whitney  U  test.\nFisher’s exact test.\nStudent  t  test.\nAfter 7 days of treatment, pain scores were significantly reduced in the resveratrol group, compared with the placebo group. After 42 days of treatment, however, no difference was observed in pain scores between treatment groups. The median difference between groups was 0.75 [95% confidence interval (CI): –1.6 to 2.3;  Table 2 ; ITT results;  P  = 0.7, Mann-Whitney  U  test]. Subgroup analysis was performed on subjects that followed protocol and did not use any medication for pain. Final pain scores were not significantly different between groups by PP analysis or by use or not of analgesics ( Table 2 ). Similar results were found using a Student  t  test after data normalization (data not shown).\nOutcomes Measured in Both Groups\nValues are median (range).\nMann-Whitney  U  test.\nUnpaired  t  test with Welch’s correction.\nFurther analysis using GEEs and the possible confounding factor (use of medication, time to follow-up) revealed no difference between groups [ Fig. 2(a) ; all analyses were made using GEEs]. Further details are provided in  Supplemental Table 1 .\n(a) Variation of pain scores between groups and compared with baseline; data were analyzed using GEEs. (b) CA-125 and (c) prolactin levels before and after treatment with monophasic contraceptive pill with or without resveratrol. Data were transformed into  Y  = log( Y ) for achieving normal distribution and analyzed using RM-2 Way ANOVA. Bars represent means for each variable studied.\nCA-125 levels were reduced after 42 days of treatment in placebo ( P  = 0.01; RM-2 Way ANOVA) and the resveratrol group ( P  = 0.02; RM-2 Way ANOVA). Prolactin levels did not vary by treatment type or over time [ Fig. 2(b)  and  2(c) ].\nIn both arms, subjects used a similar amount of pain medication. Reported side effects were comparable in both groups and were mild ( Table 2 ).\n\nThe use of complementary and alternative medicine has been widely studied for the treatment of endometriosis [ 37 ]. Plant-based products including genistein, green tea, and resveratrol have each been reported to exhibit efficacy against endometriosis in animal studies [ 20, 38–41 ]; however, no proper clinical trials have been reported for these compounds for the treatment of endometriosis in humans.\nThis study investigated whether 40 mg/d of resveratrol with a COC would reduce pain scores after 42 days of use in women with endometriosis, compared with a placebo arm with a COC alone. No difference was found between treatments when pain scores were compared between groups after treatment. These results were the same when data were analyzed using ITT or PP analysis ( Table 2 ). Likewise, no difference was found between groups when the use of analgesics and differences in pain score over time were considered ( Supplemental Table 1 ).\nThis is a randomized, double-blind, controlled comparison of resveratrol vs placebo for the treatment of endometriosis-related pain. A smaller nonrandomized, open-label study by Maia  et al . [ 23 ] reported reduced pain in response to resveratrol. Possible explanations for these differences in reported outcomes could be related to methodology, including length of treatment, pain score scales, or statistical analysis. For practical reasons, we used a more commonly available birth control pill with 21 days of active medication for two cycles (extending the treatment to 42 days) compared with their 56 days of treatment. Because  in vivo  and  in vitro  studies have shown greater benefit at higher doses [ 20, 41 ], we used a higher dose of resveratrol compared with the prior study.\nThis double-blinded, controlled, randomized clinical trial followed the CONSORT guidelines [ 42 ]; Maia  et al.  used a before-and-after treatment without randomization. We used a more commonly used linear pain scale (VAS) [ 28 ], whereas Maia  et al.  used a three-point scale [ 23 ]. This study used levonorgestrel as progestin, whereas drosperinone was used in the previous study; however, a difference in progestins is unlikely to be responsible for the discrepant results [ 43, 44 ].\nSecondary outcomes were also examined in our study. CA-125 and prolactin levels have been shown to decrease after treatment with resveratrol [ 17, 45 ]. CA-125, but not prolactin plasma levels, were significantly reduced after treatment in both groups in our study ( Fig. 2 ). CA-125 results are similar to those reported by de Sá Rosa e Silva  et al.  [ 46 ]. They reported that intrauterine devices with levonorgestrel reduce CA-125 plasma levels after 6 months of treatment. Although we see a reduction in CA-125, we did not find the same changes in prolactin as previously reported.\nThis trial has several strengths, including the controlled, randomized, double-blind study design. Less than 20% of the study population was lost to follow-up (only five cases out of 44). PP analysis was done with 39 subjects (88%). A more robust statistical analysis (GEE) was used, and similar results were observed in the PP, ITT, and sensitivity analysis ( Table 2 ). In addition to the subjective measure of pain, nonsubjective results, including CA-125 and prolactin, were measured.\nWe documented activity of the treatment in both groups, with a reduction found in CA-125 levels, in both the placebo and resveratrol group ( Fig. 2 ). In addition, a final median difference of 0.75 (95% CI: –1.6 to 2.3;  Table 2 ; ITT results) was noted between groups. Based on this small median difference in outcome, we estimate that 3285 subjects would be required in each arm to detect a significant difference.\nThere are few limitations to this study. We treated subjects for 42 days. A longer course of treatment might give different results. However, a plateau was reached between 7 and 42 days of treatment, suggesting that longer treatment would be unlikely to change the final outcome ( Fig. 2 ). Many other authors using different hormonal treatments have observed significant reduction of pain scores, compared with baseline, in treatments that lasted less than 6 months [ 29–34 ]. Petta  et al . [ 34 ] found that pain scores were significantly reduced after only 30 days of treatment with the intrauterine device containing levonorgestrel [ 34 ].\nPain scores are prone to recall bias. Although we used a standard sentence for every subject to recall the mean levels of pain in the preceding week, it is possible that subjects graded their pain score according to the highest pain that they experienced; the randomization, however, should have balanced this potential bias. The registration of use of pain medications (dipyrone, ibuprofen, and codeine) may also be prone to bias. However, the randomization and the similarity of the amount of pain medication used between groups make this unlikely. A higher mean use of dipyrone was observed in the resveratrol group, but it did not reach statistical significance. Although prior laparoscopy was required to confirm a history of endometriosis, we did not examine the stage of disease between groups, as suggested by the American Society of Reproductive Medicine [ 47 ]. It is well established, however, that stage of endometriosis does not correlate well with the experience of pain [ 48 ].\nOur studied population is likely to have external validity; women with pelvic pain were recruited from social media and not from a single outpatient clinic.\nIn conclusion, daily use of 40 mg of resveratrol, combined with continuous use of a COC, was not superior to a COC alone for the treatment of pain in women with endometriosis.","source_license":"CC0","license_restricted":false}