Women aged 33–50 years with regular menses and one or more leiomyomata more than 2 cm in diameter were considered for enrollment. Eligible subjects were healthy, nonpregnant, and desired hysterectomy for symptomatic leiomyomata as defined by the American College of Obstetricians and Gynecologists Practice Bulletin.9
Additional inclusion criteria were a hemoglobin more than 10 g/dL, ovulatory cycles every 24–35 days, and the current use of nonhormonal contraception. In this study, we limited our subjects to a body mass index less than 33 kg/m2
to reduce surgical risk. Exclusion criteria were an inability to complete the study requirements, prior uterine artery embolization, menopausal status (follicle-stimulating hormone [FSH] more than 20 milli–International Units/mL), cervical dysplasia, adnexal mass, genetic cause or rapid growth of leiomyomata (doubled size within 6 months), unexplained vaginal bleeding, and the use of glucocorticoids, progestins, or agents (prescribed or herbal) that alter ovarian or hepatic function.
The Institutional Review Board of the National Institute of Child Health and Human Development approved this protocol (02-CH-0287). After obtaining informed oral and written consent, eligible women underwent a history, physical examination, pelvic imaging, and routine laboratory analysis. Potential subjects were instructed to detect their luteinizing hormone (LH) surge using a home urinary kit (Ovuquick; Quidel, San Diego, CA) and to document vaginal bleeding or spotting and any constitutional symptoms on a menstrual calendar.
The Pharmaceutical Development Service at the National Institutes of Health Clinical Center randomly assigned subjects to computer-generated blocks of six to receive CDB-2914 at a dose of 10 (T1) or 20 mg (T2) or placebo (PLC). Allocation concealment was assured by this Service. The intended target enrollment was 36 subjects. However, due to slow subject recruitment, the study was terminated after 22 subjects were enrolled. The blocks were adjusted for two dropouts occurring within 1 month of study drug initiation. HRA Pharma (Paris, France) supplied CDB-2914 as a micronized crystalline powder. To blind both patients and health-care providers, gelatin capsules containing either CDB-2914 in doses of 10 mg and 20 mg or inert material (PLC) were prepared as previously reported.10
Subjects had a progesterone (P4) level drawn 1 week after a documented LH surge to confirm ovulation. Women not exhibiting an LH surge or luteal phase P4 were excluded. Blood was obtained for complete blood count, electrolytes, blood urea nitrogen, creatinine, liver function, thyroid-stimulating hormone, free T4, follicle-stimulating hormone (FSH), and β-hCG. Partway through the study, baseline serum prolactin was obtained after some subjects were found to have elevated values during treatment. Subjects with hemoglobin less than 10 g/dL received iron supplementation until it exceeded that threshold. To determine leiomyoma number, location and size, eligible subjects underwent noncontrast enhanced pelvic T1- and T2-weighted spin echo magnetic resonance (MR) with images at a slice thickness of 6 mm (Achieva 1.5; Philips Medical Systems, Bothell, WA; or Signa HD 1.5, GE Healthcare, Buckinghamshire, UK).
On day 1 or 2 of menstrual bleeding, subjects underwent a urine pregnancy test before initiating the study agent. If negative, the subject began the study agent once daily while fasting and continued for three menstrual cycles or the equivalent (90–102 days), if amenorrheic. Subjects returned every 2–4 weeks during treatment to undergo endocrine evaluation, including measurements of adrenocorticotropic hormone (ACTH), cortisol, prolactin, LH, FSH, P4, and E2. At roughly 30-day intervals, safety monitoring blood work, including a complete blood count, hepatic panel, electrolytes, blood urea nitrogen, creatinine, and glucose, was obtained. Subjects provided 24-hour urine specimens for cortisol and creatinine measurements at three intervals throughout the treatment period (days 20–30, days 50–60, and days 80–90). Pelvic MR imaging was repeated within 2 weeks of surgery to document leiomyoma size, location, and number.
Patients were admitted for hysterectomy after the LH surge in the third treatment cycle or in the follicular phase of the fourth cycle; if anovulatory, hysterectomy was performed at 90–102 days of treatment. Patients returned the menstrual cycle charts and symptom records upon admission for surgery and completed a second Uterine Fibroid Symptom (UFS) and/or Short Form-36 (SF-36) quality-of-life questionnaire. The study drug was discontinued immediately before surgery. An endometrial biopsy was obtained at surgery and a hysterectomy was performed; salpingo-oophorectomy was performed if indicated. To assess endometrial pathology, both the biopsy and hysterectomy specimens were evaluated.
Subjects returned for a postoperative evaluation at 4–6 weeks, and those undergoing salpingo-oophorectomy were offered hormone therapy. Subjects with intact ovaries who were anovulatory during treatment were monitored for a urinary LH surge and returned 5–7 days later for a serum P4 value to document normal corpus luteum function. Subjects not demonstrating an LH surge during the posttreatment period were monitored until a luteal phase progesterone was documented or another reason for anovulation was determined.
The primary outcome was leiomyoma volume as determined by MR imaging. After both baseline and presurgical MR imaging was performed, a senior radiologist (A.P.), unaware of treatment allocation, identified and mapped the leiomyomata seen at baseline assessment, noting the location and three-dimensional diameters. Preoperative images were compared with baseline in order to correlate with leiomyoma location. Once the corresponding leiomyomata were identified, three-dimensional diameters of all leiomyomata were recorded. To calculate the volume of each leiomyoma, the formula for an ellipsoid was employed (π/6×d1×d2×d3) using the three-dimensional measurements. To avoid multiple observations from individual subjects and to assess the overall leiomyoma burden, the calculated leiomyoma volumes were summed to yield a total leiomyoma volume for each subject.
Secondary outcomes included assessment of treatment-dependent inhibition of ovulation, alteration in menstrual function, and change in leiomyoma-related symptoms. Presumptive evidence of an ovulatory event was a serum progesterone concentration more than 3 ng/mL. Menstrual data collected during each of the three treatment cycles were scored as 0 if there was no menses and 1 if some menstrual bleeding was present during the cycle. For amenorrheic subjects, treatment cycles were assigned by 30-day intervals. Hemoglobin and hematocrit changes over the course of therapy were analyzed.
Symptom scores were calculated based on calendar logs for headache, joint, calf, breast, abdominal, and pelvic pain, fatigue, appetite changes, nausea, vomiting, diarrhea, and skin reaction; symptom intensity in each cycle was determined by the following algorithm: symptom intensity equals the number of days the symptom was experienced during the cycle divided by the number of days in the cycle. The adjusted intensity equals (treatment symptom intensity minus baseline symptom intensity) times 100. The adjusted symptom intensity was used to evaluate treatment-related effects on symptoms.
All serum assays were performed in the Clinical Center Department of Laboratory Medicine using commercially available assays (LH, FSH, and prolactin: Abbott Diagnostics, Abbott Park, IL; E2, P4, and cortisol: Beckman-Coulter, Fullerton, CA; ACTH: Nichols Institute Diagnostics, San Clemente, CA). The urinary cortisol assessments were performed using Liquid Chromatography-Tandem Mass Spectrometry with an intraassay variation of less than 5% and an interassay variation of less than 10% (Mayo Medical Labs, Rochester, MN).
We quantified subjective symptoms by previously validated quality-of-life measurement instruments: the SF-36 version 2 for general quality-of-life measures and the leiomyoma-specific UFS questionnaire. 11,12
The SF-36 was incorporated from the study onset; however, the UFS questionnaire was incorporated mid way through the study recruitment when it became available. Both quality-of-life questionnaires were administered before study drug start and upon completion of the trial before surgery.
The SF-36 scores range from 0–100 (higher scores reflect higher functioning) and have been used previously to examine quality of life in women with leiomyomata.13,14
It has eight subscales compiled to global scores for mental and physical well-being. The UFS score also ranges from 0–100 and includes eight leiomyoma-specific quality-of-life subscales (symptom severity, concern, activities, energy and mood, control, self-consciousness, sexual functioning) compiled into an overall health-related quality-of-life score.11
The symptom severity subscale contains questions related to leiomyoma-associated bulk symptoms. The concern subscale focuses largely on concern surrounding the inconvenience of carrying sanitary protection, soiling garments, and anxiety about the unpredictability of menses. The activities subscale focuses on the effect of leiomyomata on the subject’s ability to travel, exercise, and plan and carrying out typical daily activities. Energy and mood subscale assesses the effect of fatigue on activity and moodrelated symptoms. The control subscale assesses emotions related to future uncertainty and control over one’s health. The self-consciousness subscale assesses the effect of leiomyomata on physical appearance, including stomach appearance and bloating. Finally, the sexual functioning subscale addresses attractiveness and whether one avoids sexual relations due to leiomyomata. For the UFS, higher health-related quality-of-life scores reflect superior quality of life, whereas higher scores in the subscales reflect increasing intensity of the respective index.
Data entry and data management were performed using the Clinical Trials Database at the National Institute of Child Health and Human Development. Imaging and hormonal data were analyzed using Stata 10.0 (Stata; College Station, TX) and StatXact 8 (Cytel Software Corp., Cambridge, MA). The variables determining the primary endpoint, total leiomyoma volume at baseline and presurgery, were analyzed by the Shapiro-Wilk test for normality. Because of severe nonnormality they were log-transformed. Other variables not expected to meet parametric distributional assumptions, such as those on an ordinal scale, were analyzed using exact Kruskal-Wallis nonparametric statistical methods. Descriptive statistics, including means, standard deviations, medians, frequencies, and proportions, were determined.
For each assessment, the treatment groups were compared first to each other to establish whether these groups behaved similarly. If similar behavior was demonstrated, the treatment groups were collapsed into a single group and compared with the PLC group. Confirmatory analyses compared the primary outcome in the three groups using the Jonckheere-Terpstra nonparametric test for trend.15
Exact two-tailed P
values were reported and α = 0.05 was considered significant. The SF-36 and UFS questionnaires were scored according to previously published and validated methods.11,12
Changes in the quality-of-life scores were evaluated using the Wilcoxon rank sum test on the difference between pretreatment and posttreatment scores for each woman as determined by SPSS 13.0 for Windows (SPSS; Chicago, IL). Dropouts were excluded from the primary analysis; however, a secondary analysis was performed to evaluate whether significant differences between dropouts and completers were present.