Eisinger, Steven H. MD; Meldrum, Sean MS; Fiscella, Kevin MD, MPH; le Roux, Heleen D. MS; Guzick, David S. MD, PhD
Leiomyomata, or uterine fibroids, are the most common benign neoplasm of women in the reproductive age, occurring in 20–77% of all premenopausal women.1 Leiomyomata are particularly prevalent among black women.2 Surgical resection of leiomyomata is the most common indication for hysterectomy in the United States, accounting for about 30% of all hysterectomies (or 165,000–175,000 procedures per year) at an estimated expense of greater than one billion dollars a year.3 The availability of a safe and effective nonsurgical treatment of symptomatic leiomyomata would be of considerable clinical and public health importance.
Mifepristone, synthesized in France by Baulieu in 1980 from the precursor norethindrone, competitively binds and inhibits progesterone receptors.4 It is currently approved by the Food and Drug Administration for medical abortion. Although leiomyomata contain progesterone receptors,5,6 only a few investigators have studied the effect of mifepristone on leiomyomata. Murphy and colleagues7 demonstrated a clinically significant regression in the size of uterine leiomyomata using 5, 25, and 50 mg of mifepristone daily for 12 weeks of therapy. Shrinkage was comparable at the 25- and 50-mg doses, but less at the 5-mg dose. Significant shrinkage did occur at the 5-mg dose, however, and the end point of shrinkage was not adequately established, as the subjects were observed for only 12 weeks. Yang and colleagues8 studied 10 mg and 20 mg of mifepristone administered daily for 12 weeks in women with large leiomyomata. Reduction in volume was equivalent in the two groups and was comparable to the shrinkage found by Murphy. Zeng9 compared mifepristone (12.5 mg) with gonadotropin-releasing hormone (GnRH) agonists in the treatment of leiomyomata and found comparable shrinkage. Based on our own unpublished pilot data collected by the authors, 25 mg of mifepristone given daily for 4 months to 12 women with large leiomyomata induced approximately 40% shrinkage in uterine volume, universal amenorrhea, and resolution of symptoms. As in the Murphy et al studies, most subjects experienced significant hot flashes.
This study was undertaken to answer the following questions: Does 5 mg of mifepristone per day provide reduction in leiomyoma size and symptoms comparable to that of 10 mg of mifepristone per day? Does a lower dose ameliorate side effects? Are the benefits of mifepristone maintained over 6 months of treatment? Are these doses associated with significant adverse effects?
MATERIALS AND METHODS
This study was approved by the University of Rochester Research Subjects Review Board. Subjects were recruited by letters to community physicians and by advertisements in local newspapers. In addition, the study was described on the National Uterine Foundation for Fibroids Web site.10
Inclusion criteria were premenopausal status, active symptoms related to leiomyomata, satisfaction of criteria for hysterectomy or myomectomy according to The American College of Obstetricians and Gynecologists,3 a total uterine volume equal to or exceeding 300 cc by ultrasonography, good general health, agreement to use a nonhormonal method of contraception, and having a primary care provider who was informed of the subject's participation. In addition, all subjects agreed to keep monthly menstrual logs and complete a monthly symptom questionnaire and to undergo ultrasonography examinations and phlebotomy every 2 months and endometrial biopsy at study completion. Exclusion criteria included pregnancy or active attempts to become pregnant; low ovarian reserve (indicated by a follicle-stimulating hormone [FSH] level of greater than 11.60 mIU/ mL); breast-feeding; adnexal masses or tenderness indicating further evaluation; abnormal or unexplained vaginal bleeding; diagnosed or suspected uterine, cervical, or ovarian cancer; and significant mental disorder. Exclusionary health problems contraindicating mifepristone included adrenal disease; sickle cell anemia; severe liver, respiratory, or renal disease; and blood clotting defect. Exclusionary medications included current use of steroids, anticoagulants, herbals, or botanicals with possible hormonal effects; oral contraception or hormone replacement therapy within the last 3 months; and GnRH analogues or depomedroxyprogesterone within the last 6 months.
A preliminary power analysis suggested that a sample of 20 women in each group would provide more than 80% power to detect a 30% difference in change in uterine volume at a significance level of .05. Forty women who met the inclusion criteria and who provided informed consent were randomly assigned by the study pharmacist into two treatment groups of 20 each: One group received 5 mg of mifepristone per day, and the other group 10 mg per day. The study pharmacist used a table of randomly computed numbers that was generated before subjects were recruited in order to make group assignments. Neither the investigators nor the subjects were blinded to the treatment group assignments. Subjects were enrolled between October 2000 and April 2001.
For the first three quarters of the study, the mifepristone was supplied by Abortion Rights Mobilization (New York, NY), which provided mifepristone in 200-mg tablets manufactured for research purposes only. The tablets were reduced to capsules of the proper dosage using lactose for filler with an estimated dosage error of ±3% by the University of Rochester Investigational Drug Service. During the last quarter of the study, the mifepristone was supplied by Danco Laboratories (Philadelphia, PA), in powder form, and was converted to capsules of the proper dosage.
Ultrasonograms for uterine volume were performed on registration and repeated every 2 months until the completion of the 6th month. Ultrasonography examinations were performed using either an SDU 350A (Shimadzu, Columbia, MD) or an Adara Sonoline (Siemens Medical Solutions USA Inc., Malvern, PA), by either an advanced nurse practitioner with ultrasonography skills or a certified ultrasonography technician, using a standard protocol. Quality control was conducted by a senior ultrasonography technician, who verified measurements and the accuracy of the ultrasonographic technique.
Blood samples were obtained to establish baseline concentrations of hemoglobin, liver enzymes, and FSH and to rule out pregnancy, and were then repeated at the end of the 6 months. Hepatic enzymes were obtained after 1 month on mifepristone, and if persistent elevation of enzymes was detected, the subject was removed from the study.
Endometrial biopsy was obtained for subjects who met one or more of the following criteria on registration: bleeding episodes of greater than 10 days' duration, bleeding episodes occurring within 3 weeks of the onset of the last period, heavy bleeding, anemia (hematocrit less than 36% or hemoglobin less than 12.0), and an endometrial stripe of greater than 8 mm on registration ultrasonography. Endometrial biopsy was deferred for subjects who had undergone endometrial biopsy or dilation and curettage within the last 6 months (and the results were obtainable). All subjects received an endometrial biopsy at 6 months. A standard technique using the Pipelle curette and paracervical block with chloroprocaine (Nesacaine; AstraZeneca, Wilmington, DE) when appropriate was used. Endometrial biopsies were read by the hospital pathologist. A gynecologic pathologist provided consultation.
All subjects completed a monthly menstrual log and a monthly symptom questionnaire. In the menstrual log, women were asked to classify the amount of menstrual flow each day as none, spotting, light, normal, heavy, or very heavy. A menstrual blood loss index was calculated for each woman by assigning a value of 0–5 to each day's flow and summing the values for the month. Items on the symptom questionnaire included seven symptoms of leiomyomata: pelvic pain, pelvic pressure, bladder pressure, urinary frequency, low back pain, rectal pain, and pain with intercourse. In addition, there were nine possible side effects of mifepristone: hot flashes, headache, nausea, vomiting, mood swings, diarrhea, decreased libido, weakness, and fatigue. All symptoms were scored on a five-point Likert Scale (0 = no symptom, 1 = mild symptom, 2 = moderate symptom, 3 = severe symptom, 4 = very severe symptom). Other or unforeseen medical events were also solicited from subjects with open-ended questions.
Analyses of all other outcomes included all subjects enrolled in the trial. Statistical analyses were conducted using SAS 8 (SAS Institute Inc., Cary, NC). For within-group comparisons, paired t tests were used for continuous measures, and the McNemar test with exact confidence limits was used for binary measures. The nonparametric Wilcoxon rank sum test was also used to examine differences between distributions of ordinally scaled measures. Between-group comparisons were analyzed using two-sample t tests and Fisher exact tests for continuous and binary outcomes, respectively. Bimonthly measures of uterine volume and monthly measures of menstrual blood loss were analyzed using mixed effects models11 with random intercepts and unstructured covariance matrices for the random effects. All statistical tests were two tailed, with a significance level of .05.
Forty women enrolled in the trial. One subject taking 5 mg per day discontinued after 3 months for reasons unrelated to the study. A second subject, also taking 5 mg per day, completed the trial but failed to return for final measurements. Data from these subjects were carried forward using an intent-to-treat analysis.
The baseline characteristics of the two treatment groups are shown in Table 1. Few differences were apparent.
Baseline uterine volume measurements for nine subjects could not verified by a second, senior technician. Data from these subjects were excluded from the primary analyses of changes in uterine volume. However, results were unchanged when data from these nine subjects were included in secondary analyses. Results of the random effects analysis of uterine volumes are shown in Table 2. All subjects experienced a decrease in uterine volumes between each measurement point. Mean decreases in uterine volume were significantly different from zero at each measurement point for each treatment group. There were no statistically significant differences between the treatment groups at any single point in time.
All subjects were premenopausal and reported menstrual activity upon registration into the study. Figure 1 presents the mean menstrual blood loss index for subjects taking 5 and 10 mg of mifepristone. The highest mean menstrual blood loss indices were found at baseline for both the 5-mg (mean 21.8; 95% confidence interval [CI] 16.9, 26.6) and 10-mg (mean 22.3; 95% CI 17.4, 27.1) treatment groups. Mean menstrual blood loss index scores reached a nadir of 2.4 (95% CI −2.4, 7.2) in the 5-mg group after 3 months of therapy and then increased to a final mean of 10.8 (95% CI 5.8, 15.9). Mean menstrual blood loss index scores reached a nadir of 2.4 (95% CI −2.5, 7.2) in the 10-mg group after 4 months of therapy and then increased to a final mean of 5.9 (95% CI 1.0, 10.7). A statistically significant (P = .04) difference of 7.1 units of menstrual blood loss index was seen between the two treatment groups after 1 month of therapy, but not at any other time point.
Amenorrhea was frequent in both groups. Eighty-five percent (17 of 20) of women taking 5 mg per day were amenorrheic after 2 months of treatment; an equal number of women taking 10 mg per day were amenorrheic after 3 months of treatment. Amenorrhea became less common after that; by the end of the trial 11 of the 18 women (61%) in the 5 mg per day group and 13 of 20 women (65%) in the 10 mg per day group were amenorrheic. None of the differences between treatment groups were statistically significant.
The prevalence and severity of all symptoms of leiomyomata decreased from registration through 6 months in both treatment groups (Table 3). Statistically significant declines were reported in severity of pelvic pain, pelvic pressure, bladder pressure, urinary frequency, and lower back pain in both groups. Women in the 10 mg per day group also reported marginally significant (P = .06) decreases in rectal pain. Statistically significant differences between treatment groups were not observed.
As shown in Table 3, significant (P = .001) increases were seen in both the prevalence and the severity of hot flashes among women taking 10 mg per day of mifepristone. Nonsignificant changes in hot flashes were observed among women taking the lower dose. Both the prevalence and the severity of hot flashes were significantly higher (P < .05) in the 5-mg group at baseline. After 6 months of treatment the prevalence of hot flashes was statistically indistinguishable between the two groups (P = .80), although the mean severity of hot flashes was marginally (P = .07) higher in the 10-mg group. Most other previously identified side effects of mifepristone were found less often at the end of the trial than they were before treatment commenced.
The following medical events were also noted by subjects during the course of the trial: hair loss (two subjects), weight gain (two subjects), cholecystectomy, undiagnosed skin rash (which had been present before taking mifepristone), myasthenia gravis with thymoma, and change in thyroid replacement needs.
Thirty-six endometrial samples were obtained during the study. In three subjects, all taking 5 mg per day, the biopsy could not be performed for technical reasons such as cervical stenosis and anatomic distortion of the uterine cavity from large leiomyomata. One subject failed to appear for the final biopsy. Ten of 36 subjects (28%) had simple hyperplasia on biopsy, four on 5 mg (25%), and six on 10 mg (30%) (Figure 2). No malignancy, complex hyperplasia, or atypical hyperplasia was noted. Additional pathologic findings were 19 (58%) with proliferative endometrium, 12 (33%) with cystic glandular dilatation, six (17%) with tubal or epithelial metaplasia, one (3%) with an endometrial polyp, and one (3%) with a secretory endometrium. No statistical differences in numbers of pathologic findings were noted between the two dosage groups.
Hepatic enzymes were slightly elevated at the end of the study in three of 38 women (8%), two from the 5 mg per day group and one from the 10 mg per day group. Two subjects returned to normal 2 months after completion of treatment, and the third remained slightly elevated. All subjects were premenopausal by FSH at the beginning of the trial and remained so throughout. Seven women (35%) in each treatment group were anemic (defined as hemoglobin less than 12.0 g/dL) at baseline. One of these women did not complete the trial. Of the remaining 13, four (31%) remained anemic (three from the 10 mg per day group and one from the 5 mg per day group). Mean hemoglobin values among anemic subjects increased by 2.5 g/dL, from 9.9 to 12.3 (P < .001). There was no significant difference in resolution of anemia between the two treatment groups.
This study extends previous reports7–9 that mifepristone results in leiomyoma regression by showing that efficacy is retained at low doses without the same degree of hot flashes. The reduction in leiomyoma size and improvement in symptoms observed with 5 mg per day of mifepristone were comparable to those of 10 mg per day. Reduction in size was comparable to that reported using higher dosages of mifepristone7 and also to that reported with GnRH analogues.12 Adverse effects in both groups included hot flashes, mild elevation of hepatic enzymes in a few subjects, and endometrial hyperplasia. In a previous, unpublished study, the authors administered 25 mg of mifepristone daily to 12 women for 4 months. Eight women (66%) developed “severe” or “very severe” hot flashes, and only four (33%) remained asymptomatic. In contrast, only one of 19 women (5%) and two of 20 (10%) from the 5- and 10-mg cohorts, respectively, reported “severe” or “very severe” hot flashes after 4 months of therapy, whereas 12 women from each cohort (63% and 60%) remained asymptomatic. Comparison of the 5- and 10-mg groups is difficult because the baseline prevalence of hot flashes differed. This was the only difference observed between the two groups at baseline and appears to represent a chance finding. Interestingly, women reporting hot flashes did not significantly differ in age or FSH levels from women who did not report hot flashes.
Simple endometrial hyperplasia was observed in four subjects on 5 mg (25%) and six on 10 mg (30%), for an overall rate of 28%. No atypical hyperplasia was observed in any of the samples. Because endometrial biopsies were not conducted at study entry (except to rule out endometrial pathology in subjects with menstrual or sonographic abnormalities) and because there was no placebo control group, we cannot be certain that mifepristone causes endometrial hyperplasia. However, an association between mifepristone and endometrial hyperplasia has been previously reported. For example, in a study of mifepristone (50 mg per day for 6 months) for endometriosis, Murphy et al13 noted endometrial hyperplasia consistent with unopposed estrogen effect. Massive endometrial hyperplasia was noted after treatment of a woman with Cushing syndrome with large doses of mifepristone over 6 months.14 Presumably, the antiprogesterone effect of mifepristone results in unopposed endogenous estrogen acting on the endometrium.13 Clearly, more data regarding endometrial hyperplasia associated with use of low-dose mifepristone must be obtained before this medication can be recommended for long-term use.
Rates of amenorrhea and resolution of anemia are comparable to those associated with GnRH analogues. However, given the 28% prevalence of endometrial hyperplasia, the mechanism of amenorrhea is likely to differ from the hypoestrogenic state induced by GnRH analogues.
The other potential health risk that emerged in this study was elevation of hepatic enzymes in 8% of subjects. The mechanism and significance of this finding are uncertain. Pending further study, it seems prudent to periodically measure hepatic enzymes in women taking mifepristone and to withdraw the medication in those whose values become markedly elevated.
The limitations of this study should be noted. The small sample size of the study limited its power to detect differences in dichotomous outcomes between the treatment groups. The absence of placebo controls precludes definitive assessment of the association of mifepristone with hyperplasia or hepatic dysfunction. The absence of blinding introduces potential bias by subjects in reporting of subjective symptoms and observer bias in the assessment of outcomes. However, there was little reason for subjects to systematically expect one dose to be different than the other. Similarly, we had no a priori hypothesis regarding differences between the two groups. Last, the finding of comparable effects in leiomyoma size reduction and symptom relief in the two treatment groups suggests that there was little bias in reporting or assessment of outcomes.
Mifepristone offers promise for three clinical uses. First, mifepristone may represent a viable alternative to GnRH analogues for use in the preoperative application. It appears to yield comparable reductions in leiomyoma size, may have fewer side effects, and can be orally administered. Second, if the safety of long-term, low-dose mifepristone is established, perimenopausal women with large, symptomatic leiomyomata may be able to take this medication until menopause, when the leiomyomata typically regress. A major savings in cost and morbidity may be possible in view of the large number of hysterectomies done for leiomyomata in perimenopausal women. The third possible application occurs in younger women with large leiomyomata who wish to retain their fertility. They may also benefit from chronic low-dose mifepristone, until such a time as they wish to attempt conception. However, the long-term effects of mifepristone on fertility are not known.
In conclusion, daily administrations of 5 and 10 mg of mifepristone show comparable effects on leiomyoma size and symptoms among premenopausal women. Further investigation is warranted to determine whether longer courses of therapy or larger cohorts of study subjects will uncover less common risks associated with therapy, whether still lower doses are equally as effective as the doses used here, whether leiomyomata will regrow after cessation of therapy, and whether these findings can be reproduced in a placebo-controlled randomized clinical trial.