This publication is provided for historical reference only and the information may be out of date.
This chapter presents the results of our evidence review for the first four key questions (KQs): KQ 1, incidence and prevalence of uterine fibroids; KQ 2, outcomes of interventions intended to relieve symptoms of uterine fibroids; KQ 3, outcomes of interventions for uterine fibroids for reasons other than symptom relief (enhancing fertility, reducing further growth, or other reasons); and KQ 4, costs. As explained in Chapter 2, this review is an update of an earlier systematic review with a publications cutoff date of 2000. For our searches, therefore, we did not include any citations published before February 2000.
KQ 5, on modifiers of outcomes, KQ 6, on comparisons of interventions, and KQ 7, on the geographic variation in treatment in the United States, are derivative of these first four questions. We did not do systematic literature searches for them but instead relied on the systematic searches for the primary questions. For that reason, we discuss KQs 5, 6, and 7 in Chapter 4 of this report.
Appendix C * provides the detailed evidence tables for KQs 1, 2, 3, and 4. Our summary tables below feature groups of studies addressing each treatment; they are organized alphabetically by author, unless otherwise stated.
KQ 1: Incidence and Prevalence of Uterine Fibroids
KQ 1 refers to the incidence and prevalence of uterine fibroids, as estimated in representative U.S. populations through use of diagnostic imaging or histology to document uniformly the presence or absence of fibroids. The prior systematic review estimated that the cumulative risk of diagnosis for fibroids between the ages of 25 and 44 was approximately 30 percent.30
The evidence concerning prevalence of uterine fibroids in women since 2001 is limited to two articles that meet our inclusion criteria, both of fair quality (Table 4 and Evidence Table 1).2, 41 One study used a combination of medical records and self-report for the 16 percent of its sample that was postmenopausal and ultrasound for the 84 percent of the sample that was premenopausal.2 The other study relied on self-reports of ultrasound- or hysterectomy-confirmed diagnosis of fibroids of premenopausal women without a prior diagnosis of uterine fibroids among U.S. black women.41
A prospective cohort study conducted in the Washington, DC, metropolitan area randomly selected 1,364 subjects between the ages of 35 to 49 years from a prepaid health plan for ultrasound examination to detect uterine fibroids.2 Of this sample, 38 percent of the women were white and 62 percent were black. The two groups were similar in age but, compared with the white women, the black women were less educated, had more children, and had a higher body mass index (BMI). Black women were more likely to have been previously diagnosed with uterine fibroids (45 percent) than white women (21 percent) and, in those not previously diagnosed, to show ultrasound evidence of uterine fibroids (59 percent vs. 43 percent, respectively). Overall, black women were significantly more likely to have uterine fibroids with an odds ratio (OR) of 2.9 (95% confidence interval [CI], 2.5–3.4; P < 0.001). The authors reported that the importance of race changed little after adjusting for BMI and parity (OR, 2.7; 95% CI, 2.3–3.2; P < 0.001). In both groups, prevalence increased with age. Estimated cumulative incidence of fibroids by age 50 was more than 80 percent for black women and nearly 70 percent for white women.
A second study examined the incidence of uterine fibroids and factors that affect them in black women.41 The study is a prospective, ongoing cohort study of U.S. black women with data reported from 1997 to 2001. The sample for this study was limited to premenopausal women with intact uteri and no reported diagnosis of fibroids before 1997. The study found uterine fibroids in 2,279 women in 76,711 documented person-years (2.97 percent). Factors that affected the prevalence of uterine fibroids included age at first birth, years since last birth, and younger age at menarche. Women who were parous had an incidence risk ratio of 0.7 (95% CI, 0.6–0.8) relative to nulliparous women. Women who had a child less than 5 years of age were less likely to have uterine fibroids than those who had had a child 5 to 9 years previously (multivariate incidence rate ratio [IRR], 2.0; 95% CI, 1.6–2.5). Finally, women who were older at menarche were less likely to have uterine fibroids than women who experienced onset of menses at 12 to 13 years (IRR, 0.8; 95% CI, 0.7–0.9). The current use of progestin-only injectables as birth control was associated with a 40 percent reduction in risk (95% CI, 0.4–0.9).
KQ 1 also asks about the incidence, type, and severity of symptoms. We found no direct evidence based on prospective observational studies of representative U.S. populations.
KQ 2: Outcomes of Interventions for Relief of Symptoms Related to Uterine Fibroids
We document here our findings about outcomes of interventions for women with symptomatic fibroids. Symptoms can include anemia, problematic bleeding patterns, bulk symptoms (low back pain, urinary frequency, and constipation), pain, and dyspareunia (pain during or after sexual intercourse). We initially considered the following treatment approaches or combinations of treatment approaches:
Expectant management without intervention;
Medical management (including oral contraceptives, menopausal hormone therapy, GnRH [gonadotropin-releasing hormone] agonist therapy, antiprogestins, progesterone-containing intrauterine devices [IUDs], and nonsteroidal anti-inflammatory drugs [NSAIDs]), referred to henceforth as pharmaceutical management;
Uterine artery embolization (UAE);
Endometrial ablation with or without myomectomy;
In situ destructive techniques, specifically by focused ultrasound guided by magnetic resonance imaging (MRI) and cryotherapy;
Myomectomy by abdominal, laparoscopic, or hysteroscopic techniques;
Hysterectomy by abdominal, laparoscopic, or vaginal techniques; and
Complementary and alternative therapies including acupuncture.
KQ 2 distinguishes between short- and long-term outcomes. Most studies in this literature, however, limit themselves to the postoperative period. We do not report short- and long-term outcomes separately for each intervention, but we do call attention to longer-term outcomes whenever reported.
Expectant Management: Overview and Nomenclature
We did not identify any studies that specifically focused on documenting the natural history of uterine fibroids, course of fibroid symptoms, or clinical care received for fibroids over time in a cohort of women with known baseline fibroid status. No studies focused on either outcomes, such as anemia, bleeding patterns, pain, and health-related quality of life, or modifiers of outcomes of expectant management per se as the primary topic of their research. However, RCTs that include a no-treatment comparison group may provide a glimpse of anticipated outcomes in the absence of intervention. With caveats about the limitations of such data, we summarize in this section information about the outcomes of women in trial groups that received no treatment, placebo treatment, or minimal intervention such as multivitamin use (Appendix C *, Evidence Table 2).
Thirteen studies included groups that received no treatment or only minimal intervention.42–57 Five studies did not include symptoms or fibroid size; instead, they used the comparison group to assess characteristics of specimens of surgical tissue as they related to the anticipated effects of the treatment drug on the fibroids42, 43, 45 or to examine other aspects of treatment response such as changes in bone marrow density49or hemoglobin47in response to medical treatment. Information on these trials can be found in Appendix C *, Evidence Table 2.
Eight studies with nine publications provided information about uterine size, fibroid size, or participants' symptoms for the no-treatment comparison groups at baseline and at the end of followup.44, 46, 48, 50, 51, 55, 56, 58 All but one48 were conducted in Italy by two inter-related groups of investigators whose work on medical management and preoperative management of fibroids was featured in detail in the section on outcomes of pharmaceutical management. In the seven Italian study groups, women were followed for a range of 2 months to 12 months without treatment, with the median duration being 6 months across the studies. The majority of studies used no treatment or placebo; some used a calcium, iron, or multivitamin tablet as the placebo. We do not discuss hemoglobin changes for the studies in which women received iron or multivitamins.
Expectant Management: Results
Shorter treatment spans, of 2 and 3 months, were associated with preoperative studies done most often in premenopausal participants. For those reasons they may be the least informative—all women had symptoms enough to warrant surgery and the followup is extremely brief. In no case was a significant change in uterine size documented. Most studies documented almost identical fibroid volume;46, 55, 59 one study, which did not note masking of the individuals conducting the ultrasounds, reported an increase in fibroid volume of 11 percent over 3 months. The longest followup for symptoms was in a group of women using a multivitamin placebo for 6 months. Compared to baseline values, their severity of bleeding, length of bleeding with menses, and hemoglobin levels were unchanged; 72 percent had no change in fibroid size; 24 percent had increases in fibroid size; and 3 percent had a decrease;51 the increase was not statistically significant.44 One other study reported a nonsignificant increase in menorrhagia, pelvic pain, and pressure among women receiving iron tablets only for 2 months.59
Longer studies were generally done among postmenopausal women to determine whether a specific medication influenced fibroid size or symptoms. Overall, these untreated comparison groups were the most likely to have less severe presentations and perhaps be more representative. However, they can shed light only on postmenopausal management. In these groups, observed for 12 months, the investigators saw no trend for fibroid growth; they did not, however, document any significant decrease in fibroid or uterine size. Fibroids were consistently reported to be unchanged;50, 56, 58 one study noted that 2 of 35 women had a “mild reduction in uterine and fibroid size,”56 (p40) suggesting that fibroid involution (regression in size) may not be marked during menopause.
The last study group was a medical record control group matched to participants in a study of complementary and alternative medicine treatments. The study was conducted in a U.S. academic center. Symptoms at clinical encounters and available radiologic studies were provided for 6 months of followup. Within a group of 37 women (who may have received other clinical care), none had documented worsening of symptoms, three had reduced size or reduced growth of fibroids documented, 20 had no change in fibroid size, and four had documented growth of more than 1 centimeter (cm) per month in diameter of a fibroid.48
The size of the comparison groups from these trials is small, from 22 to 60 women, and the time frames are very brief. They offer an initial impression that fibroids may not have a continuous, slow-growth pattern before menopause and that, after menopause, decreases in size may not be as profound as clinical wisdom suggests. However, the total picture provided is insufficient to project what the course of watchful waiting might be for an individual woman with fibroids. Because these studies were not designed for this purpose, the overall quality of the research is too poor to inform the choice of expectant management over other intervention options.
Pharmaceutical Management: Overview and Nomenclature
The etiology of uterine fibroids is not well understood. Pharmaceutical management of fibroids is most commonly done as an adjunctive treatment before surgery. Few medications serve as permanent alternatives to surgery. KQ 2b asks about outcomes from GnRH agonist therapy, menopausal hormone therapy, antiprogestins, oral contraceptives, progesterone-containing IUDs, and NSAIDs among possible medical treatments for uterine fibroids. We did not find any new studies since February 2000 on oral contraceptives, progesterone-containing IUDs, or NSAIDS. The majority of our included studies examined the effect of GnRH agonists on uterine fibroids (Appendix C *, Evidence Table 3). Some studies also reported on progestin, estrogen receptor antagonists and modulators, and antiprogestin. We also report on studies that examined the effects of tibolone as adjuvant therapy to GnRH on uterine fibroid growth. For convenience and consistency, we briefly list and define medications evaluated in the studies reviewed below.
GnRH Agonists and Other Adjuvant Therapies. GnRH agonists are often used as preoperative adjunctive therapy to surgery. They cause down-regulation of estrogen receptors, which decreases fibroid growth. GnRH agonist therapy also helps to optimize hematocrit levels that may have declined secondary to menorrhagia from fibroids. Low hematocrit levels can pose a risk for surgical complications. Studies in this review examined leuprolide acetate, triptorelin, and goserelin. One study also reviewed the effect of ipriflavone as adjuvant therapy to prevent osteoporotic side effects of GnRH agonists.60
Leuprolide or Leuprolide Acetate. Leuprolide is a potent inhibitor of gonadotropin secretion. Trade names for use with uterine fibroids include Eligard®, Lupron Depot-Ped®, Lupron Depot®, Lupron®, and Viadur®. Leuprolide is often used as an alternative to surgery for fibroids or for preoperative adjunctive therapy. Its potent effect on reducing estrogen activity in the uterus can decrease fibroid size and symptoms including menorrhagia. The majority of studies (13, in 15 articles) evaluated a GnRH agonist treatment of uterine fibroids; of these, 10 evaluated leuprolide as the primary intervention.42–45, 49, 51–55, 59–62
Triptorelin. Triptorelin (trade names Decapeptyl® and Gonapeptyl®) is generally used in the United States to treat men for advanced prostate carcinoma. Its activity on fibroids and use for fibroid management is similar to that for leuprolide. Two studies from Italy examined the effect of triptorelin on fibroids.63, 64
Goserelin. Goserelin (Zoladex®) is also a potent inhibitor of gonadotropin secretion. In one study, goserelin was used in one treatment arm of a five-arm study to evaluate fulvestrant (a drug that blocks estrogen in the treatment of breast cancer [see below]).46
Ipriflavone. Ipriflavone is a synthetic isoflavone in the herb category of natural products with a structure similar to that for estrogen. It has gained acceptance as an alternative medication for treatment of osteoporosis. One study uses ipriflavone as adjuvant therapy to prevent osteoporotic side effects of GnRH agonists.60
Tibolone. Tibolone is an estrogen and progestin combination therapy used for several purposes: to prevent postmenopausal osteoporosis, to treat symptoms such as hot flashes and associated sweating resulting from menopause (surgical or natural), and to improve bone mineral density (BMD) in patients with established postmenopausal osteoporosis. It is used in these studies as adjunctive therapy to GnRH agonists to prevent negative side effects of GnRH agonists or to study whether the addition of tibolone as add-back therapy alters the effect of GnRH on fibroids. Currently this drug is not available in the United States. International brand names include Climatix®, Livial®, Tibial®, and Tibofem®.
Progestin. Lynestrenol, a progestin, known by international brand names including Endometril®, Exluton®, Linestrenol®, and Orgametril®, is used to treat endometriosis, prevent pregnancy, and treat symptomatic fibroids in countries other than the United States, where it is not currently available. One study compared lynestrenol to leuprolide in an assessment of preoperative treatment of fibroids.62
Antiprogestin. Mifepristone (Mifeprex ®; also known as RU-486) is a synthetic steroid that competitively binds to the intracellular progesterone receptor, thereby blocking the effects of progesterone and causing significant shrinkage in fibroids. One study (two articles) evaluated mifepristone as an alternative medical treatment for fibroids.65, 66
Estrogen Receptor Antagonists and Modulators. Raloxifene hydrochloride (Evista ® ) is a selective estrogen receptor modulator (SERM) that has been reported to cause a significant reduction in fibroid size. One study (three articles) evaluated raloxifene.50, 56, 67
Fulvestrant (Faslodex ® ) is an estrogen receptor antagonist used to prevent fibroid growth; it is also used for treatment of postmenopausal breast cancer. One study evaluated fulvestrant.46
Pharmaceutical Management: Results
The prior review on uterine fibroids attempted to identify the most appropriate candidates for GnRH agonists, to document the incidence of need for additional treatment following GnRH agonist therapy, and to estimate risks and benefits of pharmaceutical management.30 The review found some evidence that GnRH agonist therapy may be more effective in perimenopausal women than in premenopausal women, but it cautioned that additional studies were necessary. The review did not find sufficient evidence to draw conclusions on the proportion of women likely to experience recurrence of symptoms, the level of severity of symptoms, and the probability of success of alternative treatments after GnRH agonist therapy. The review found “good evidence based on randomized trials that use of GnRH agonists prior to myomectomy or hysterectomy reduces estimated blood loss and may facilitate certain surgical approaches (use of laparoscopic or vaginal approaches and use of transverse abdominal incisions as opposed to vertical incisions).”30 (p8) The review noted, however, that there are no long-term data on the clinical significance of these effects and that some studies suggest that fibroids are more difficult to separate from the uterus after GnRH agonist treatment because “pretreatment with GnRH agonists obliterates the cleavage plane between myometrium and fibroid.”30 (p92) The review found that hormone therapy and progestins were ineffective in alleviating fibroid symptoms or fibroid growth, but progestins, when used concurrently with GnRH agonists, were effective in eliminating hot flashes associated with GnRH agonist therapy.
Studies, Designs, and Populations. We identified 19 studies reported in 24 publications42–46, 49–56, 59–69 on outcomes of fibroids after medical interventions. The studies were predominantly of fair quality, unless otherwise stated. Some study populations are represented more than once in the total number of publications because the authors focused on individual outcomes in separate publications. A study on mifepristone65 reported on a 12-month extension of the study in a subsequent publication.66 Another set of authors reported the outcomes of their study in two publications.55, 59 In addition to their first publication on one study group,52 Palomba and colleagues focused on individual outcomes in three additional publications.51, 53, 54 In summary, the 24 publications reviewed in this section represent 19 studies of 19 distinct populations.
Thirteen studies were conducted as randomized controlled trials (RCTs).46, 49–56, 59–67 The majority (eight trials) had two arms, but four studies randomized participants into three arms;49, 55, 59, 61, 67 in addition, one study randomized subjects into five evaluable groups.46
The majority of the studies were undertaken in Italy at academic medical centers.42–45, 49–56, 59, 61, 63, 64, 67–69 One study was done in the United States,65, 66 one in Japan,60 and one in France.62 One was a multinational study.46
Fifteen studies evaluated a patient population in their premenopausal years.42–46, 51–55, 59–62, 64–69 Two studies specifically evaluated women in the perimenopausal years,49 and two evaluated women who were postmenopausal.50, 56 One study did not specifically state which group of women it was targeting.63
Most studies included information on changes in fibroids or uterus size. The six studies that also examined the effects of medical treatment on hemoglobin are discussed separately below.42, 55, 59, 61–64 Five studies reported on changes in symptoms.56, 58, 61, 63, 65 Measurement of symptoms varied from study to study; inconsistencies across the literature make comparisons of symptom relief challenging. Intraoperative outcomes generally included length of time of surgery or intraoperative blood loss.63, 64 Two studies evaluated the effects that medical management of fibroids has on metabolic measurement such as lipid profiles.51–54, 60
Outcomes of GnRH Agonists. Thirteen studies in 17 articles reported on outcomes after administration of GnRH agonists.42–45, 49, 51–55, 59–64, 69 Seven studies were RCTS;49, 51–55, 59, 60, 62–64 five were prospective cohorts with comparisons;42–45, 61 and one was a retrospective cohort with a comparison group.69 Four studies were of poor quality and the remainder were of fair quality.60–63
Outcomes. Six studies compared leuprolide alone to leuprolide with additional treatment to evaluate differences in effects of leuprolide on outcomes such as BMD, metabolic changes, symptoms, and overall tolerance.49, 51–55, 59–62 One study compared leuprolide alone to leuprolide plus raloxifene and evaluated BMD, uterine size, and metabolic differences.52
Two studies evaluated the effects of pharmaceutical management on BMD. A side effect of hypoestrogenism, from GnRH agonist administration, is bone loss, which may or may not be recoverable; generally, the recommended length of treatment with GnRH agonists is limited to 6 months to avoid bone loss. Two studies evaluated the protective impact that therapy additional to GnRH may have on bone loss. One study of GnRH and raloxifene51–54 studied the effect that adding raloxifene may have on BMD.51 The authors reported that BMD was significantly higher in the group that received raloxifene. The second study addressing this question reported on a three-arm RCT comparing (1) leuprolide plus tibolone for medical management of fibroids as an alternative to surgery, (2) hysterectomy and bilateral oophorectomy for surgical management of symptomatic fibroids, and (3) natural menopause.49 The authors reported that the two groups that underwent treatment of fibroids had comparable bone loss; both treated groups had greater bone loss than the natural menopause group. The rate of bone loss in the two groups treated for fibroids was 5.7 percent and 6.4 percent; comparisons between baseline and followup were statistically significant for both treated groups. The study provides little information on the effectiveness of the addition of tibolone to GnRH agonist treatment.
Two studies (five articles) that compared leuprolide plus supplemental therapy to leuprolide alone reported metabolic parameters as their outcomes.51–54, 60 One study evaluated leuprolide at a dose of 1.88 mg per month with supplemental ipriflavone for 6 months to the same dose of leuprolide for 6 months.60 The group treated with leuprolide plus ipriflavone had an 8.4 percent increase in low-density lipoprotein (LDL) levels when compared with baseline levels (P < 0.01). The group treated with leuprolide alone had a 22.4 percent increase in LDL levels (P < 0.01) when compared to baseline. After the full 6 months of treatment the increase in LDL was significantly less (P < 0.01) in the group that received supplemental treatment with ipriflavone.
The second study compared the effect of leuprolide (3.75 mg per month) administered with supplemental raloxifene with leuprolide plus placebo on serum levels of lipoproteins.51–54 After six cycles of treatment, total cholesterol, high-density lipoprotein (HDL), LDL, and total triglyceride levels were significantly increased (P < 0.05) in the placebo group when compared with baseline levels. The group that received raloxifene was reported to have minimal increase in LDL levels; this increase in LDL levels was significantly lower than in the leuprolide plus placebo group (P < 0.05). Similarly, levels of total cholesterol were also higher in both groups compared with baseline levels, but the increase in total cholesterol in the group that received supplemental raloxifene was significantly less than in the leuprolide plus placebo group.
One study measured a surrogate marker for estrogen activity in the uterus as a “quick score.” The authors found that the group treated with leuprolide had decreased estrogen receptors after 3 months of treatment compared with no treatment.43
Three studies compared triptorelin with no treatment.63, 64, 68 Two studies reported improvements in fibroid size, hemoglobin changes, and intraoperative outcomes for the triptorelin group;63, 64 the third study found significantly shorter operative times for the triptorelin group but no difference in hemorrhage, uterine perforation, length of stay, recurrence of fibroids, or abnormal uterine bleeding.68
Pharmaceutical treatment is generally intended to reduce fibroid size and stabilize hemoglobin levels before surgery. The following discussion presents the effects of pharmaceutical treatment on fibroid size and hemoglobin first, followed by studies on symptom control and other outcomes.
Fibroid and Uterine Size Outcomes. GnRH agonists were effective in decreasing overall uterine size when used as preoperative treatment or as an alternative to surgery in all eight studies that reported on uterine and fibroid size changes in response to GnRH agonists (see Table 5).42–45, 55, 61, 63, 64 Three studies reported GnRH agonist effects on fibroid size alone.60, 62, 69 Study groups receiving GnRH agonists alone had an average decrease in uterine size of 209.8 cm3 from an average starting size of 637 cm3. Mean decrease in fibroid size was 66 cm3 decreased from a mean starting size of 247 cm3. The addition of add-back therapy to GnRH agonists did not affect the extent of uterine or fibroid size decrease. In these groups, the mean decrease in uterine size was 111.6 cm3 and the mean decrease in fibroid size was 49 cm3.
Three studies reported fibroid or uterine size changes over time in women who received no treatment44, 63 or placebo treatment with iron only.59 All three studies reported an increase in uterine size ranging from 2 cm3 to 60.7 cm3 with an average increase of 23.6 cm3. The increase in size of individual fibroids was reported in only one study and that increase was very small at 1 cm3.59
Trials with comparative groups produced good evidence that administration of GnRH agonists with or without add-back therapy significantly decreases the overall size of the uterus and fibroids by as little as 22 percent to as much as 53 percent. The greatest decrease in uterine size was reported by Di Lieto and colleagues,61 who treated their study group with 4 months of leuprolide 3.75 mg subcutaneously. They reported an average baseline uterine size of 977.1 cm3 and an average decrease in size of 42 percent.
Hemoglobin Outcomes. Six studies (three of fair quality,42, 55, 59, 64 and three of poor quality61–63) in seven articles reported hemoglobin changes after GnRH agonist therapy, to assess if its use would improve anemia in women with fibroids (Table 6). The outcome reported in five studies was hemoglobin (grams/deciliter [g/dL]) measured before surgery (preoperatively).42, 55, 59, 61–63 All five studies reported an increase in hemoglobin when measured preoperatively, after the completion of GnRH agonist treatment ranging from 2 to 4 months. The reported increase in hemoglobin ranged from 0.9 g/dL to 5.2 g/dL. None of these five studies was designed to determine if GnRH agonist administration can improve anemia in women with symptomatic fibroids before surgery, so they provide only weak evidence to answer that question. Additionally, the results were statistically significant in only two of these studies.55, 59, 63 One study reported hemoglobin measurement only after surgery, and hence the result was a decrease in hemoglobin.64
Symptom Outcomes. Three studies on GnRH agonist therapy examined symptom outcomes.51–54, 56, 61 One study comparing leuprolide, leuprolide plus tibolone, and placebo reported significant differences in menorrhagia and pelvic pain at baseline, but no differences after treatment between the leuprolide-only group and the leuprolide plus tibolone group.61 The authors also reported a significant difference in the leuprolide and leuprolide plus tibolone groups, with the former group reporting increases in hot flash episodes, and the latter group reporting constant numbers of hot flashes. Another study of raloxifene versus placebo did not demonstrate any differences in amenorrhea or abnormal uterine bleeding at 3, 6, 9, or 12 months of treatment.56 A third study, comparing leuprolide plus raloxifene versus leuprolide plus placebo found no differences in menorrhagia, pelvic pain, pelvic pressure, urinary frequency, or constipation after treatment.51–54
Outcomes of Progestins. A single RCT of poor quality compared outcomes from 33 women receiving lynestrenol with 23 women receiving leuprolide acetate.62 Patients receiving leuprolide reported a significantly greater reduction in fibroid size than the group receiving lynestrenol, but the study found no differences in hemoglobin after 16 weeks of therapy and before surgery.
Outcomes of Antiprogestins. One fair-quality study compared the outcomes of 5 mg per day to 10 mg per day of mifepristone.65 The authors reported significant reductions in uterine volume compared with baseline values at 2, 4, and 6 months. They also reported significant reductions in menstrual blood loss from baseline values in both groups, but the differences between groups were not significant other than at a single time, 1 month after therapy. The authors noted that although all women reported menstrual activity on registration in the study, 61 percent and 65 percent, respectively, had amenorrhea by the end of the trial. A followup to the original study evaluated the development of endometrial hyperplasia after 18 months of treatment with mifepristone in 21 of the original 40 women in the study.66 The authors reported no hyperplasia at both 6 months and 12 months at the 5 mg dose, and a 25 percent rate at 6 months and 7.7 percent rate at 12 months at the 10 mg dose.66
Outcomes of Estrogen Receptor Modulators and Antagonists. Three studies (all of fair quality) evaluating the outcomes of the SERM raloxifene in comparison with a placebo were conducted in Italy by Palomba and colleagues.50, 56, 67 Two studies evaluated women who had undergone menopause within the previous 2 years.50, 56 Both reported that uterine size and fibroid size significantly decreased after treatment compared with baseline values. These significant differences did not extend to amenorrhea and abnormal uterine bleeding in the one study that also reported these outcomes.56 The study that evaluated premenopausal women reported that uterine and fibroid size increased after 3 months of treatment compared with baseline levels.67
A five-arm trial of poor quality compared three different doses of the estrogen receptor antagonist fulvestrant with goserelin and a placebo.46 Goserelin significantly reduced fibroid growth and endometrial thickness compared with placebo and fulvestrant, but fulvestrant did not significantly alter fibroid volume or endometrial thickness compared with placebo.
Uterine Artery Embolization: Overview and Nomenclature
This section presents the results of our literature searches and findings about outcomes of fibroids treated with uterine artery embolization (UAE), also known as uterine fibroid embolization. UAE blocks the blood vessels supplying the fibroids by injections of small particles into the arteries feeding the uterus. Because the procedure is minimally invasive, it is an option available to women who wish to avoid surgery, are poor surgical candidates, or wish to retain their uterus. The literature discussed in this section includes studies focusing on UAE only, with the exception of UAE compared with laparoscopic occlusion of the uterine arteries. Studies comparing UAE with myomectomy or hysterectomy are discussed in those respective sections below. For convenience and consistency, we have used uniform terminology and abbreviations to describe the different techniques used to treat uterine fibroids.
Laparoscopic Occlusion of the Uterine Arteries involves a laparoscopic procedure in which the clinician places clips over the uterine arteries at the level of the internal iliac artery. The collateral arteries between the uterus and the ovaries are also coagulated with bipolar forceps. UAE is a technique in which the clinician introduces tiny particles or microspheres into the arteries feeding the uterus. The procedure is based on the theory that occluding blood flow to the muscular portion of the uterus will produce infarction of the fibroids and control symptoms.
Given the relatively new nature of this procedure, very little information was available at the time of the prior review on uterine fibroids; the authors concluded that they could not make estimations of recurrence, persistence, or need for subsequent therapy.30
Studies and Designs. Thirty-one articles report on outcomes of UAE, comparisons of UAE with other procedures, modifiers of UAE outcomes, and related issues (Appendix C *, Evidence Table 4).70–100 The 31 publications represent 24 studies and 22 distinct study populations.
The UAE literature consists primarily of studies done at academic centers; at least two-thirds of the studies took place in this setting. One study was done in a community setting, and three combined data from both academic and community hospitals. The majority (13) of the studies was done in the United States; the remaining countries accounted for fewer studies: Canada, 3; United Kingdom, Netherlands, and Japan, 2 each; and Norway, 1. Finally, one study compiled data from studies done in both the United States and abroad.
Study Populations and Outcomes Measured. Twelve of the publications listed here represent five studies and three distinct populations. In the summary tables below, we elected to group articles primarily by study groups and secondarily in alphabetical order by author, owing to the multiplicity of papers from single studies, overlapping samples, and distinct differences in quality of studies across these study groups. We report on multiple studies from a common population source in Table 7 and on single studies from varied populations in Table 8.
One set of five publications, all by Pron and colleagues, on the Canadian Ontario Uterine Fibroid Embolization Trial focused on individual outcomes from the same sample in separate publications; we count all five as a single study, of fair quality.83–87
A second population served as the precursor to the FIBROID registry (specifically, the Uterine Artery Embolization Fibroid Registry for Outcomes Data [FIBROID], a U.S. multicenter prospective voluntary registry of patients undergoing uterine embolization for fibroids [www.fibroidregistry.org]). Spies and colleagues published short-term outcomes,92 a subanalysis,96 and long-term outcomes93 from a case series of 200 women. They subsequently closed enrollment of patients in that protocol, began a new protocol to coincide with participation with the FIBROID registry, and published one study presenting results from both populations.97 The studies published from this group are of fair to good quality.
The third, and largest, study population is from the FIBROID registry. Two publications reported on different samples based on eligibility for the outcome considered in the publication (N = 3,04182, 100 and N = 2,11295), although the articles do not specify whether these two samples overlap completely. We consider the two FIBROID registry papers as two separate studies but, for purposes of tabulating information from the same or similar sources, kept them in Table 7. We rated these studies to be of fair quality.
Outcomes and Modifiers. Among the 24 distinct UAE studies, 17 reported on outcomes or modifiers of UAE. Of these, three were retrospective case series, focusing on outcomes associated with failure or success of UAE.77, 78, 88 Twelve studies are prospective case series. Of these, nine reported on short- and/or long-term outcomes;79, 80, 82–87, 92, 93, 95–100 two reported on imaging modalities associated with UAE;71, 81 and one reported on use of a percutaneous closure device during UAE.72
Two cohort studies addressed pain in relation to the UAE procedure. One investigated a prospective sample to compare pain medications,89 and the other examined data for a retrospective sample comparing the use of embospheres and polyvinyl alcohol particles.91
Comparative Studies. Seven studies compared more than one type of procedure. Two were retrospective cohorts, comparing UAE and myomectomy.70, 90 Four prospective cohorts were identified; one compared UAE with myomectomy,73 two compared UAE with hysterectomy,75, 94 and one compared UAE and laparoscopic occlusion of the uterine arteries.74 The only RCT compared UAE with hysterectomy.76
UAE Outcomes. This literature comprises nine prospective case series studies (in 15 articles, one of good quality,80, 92, 93 four of fair quality,82–87, 95, 97, 100 and four of poor quality79, 80, 98, 99) and three retrospective case series (two of poor quality77, 78 and one of fair quality88) that described either short- or long-term outcomes (or both) (see Table 7 and Table 8).
Satisfaction. All studies reported high levels of satisfaction on the part of the women assessed, measured at various points in time and along varied scales. They reported a range from 87 percent to 97 percent satisfaction with outcomes.
Symptom Improvement. Studies reported high levels of symptom improvement, however longer-term studies appeared to indicate some decline in improvement in symptoms over time. One study found that, at 3 months, the great majority (93 percent) of women had improved symptoms; by 5 years, the proportion reporting improvement in symptoms had declined to 73 percent (of 143 women still in the sample).92, 93 Studies also reported some variability in which symptoms were improved: one study found that women reported statistically significant improvement in menorrhagia (83 percent), dysmenorrhea (77 percent), bulk symptoms (84 percent), and urinary symptoms (86 percent) at 3 months.86 Improvement in menorrhagia was not related to preprocedure uterine volume or amount of volume reduction. Overall life impact scores (representing the interference of symptoms with everyday or usual activities) were markedly improved after UAE. Before UAE, 72 percent reported impact scores of 7 to 10 (high interference with daily activities); after UAE, this figure dropped to 11 percent.
Pain. In one study, 70 percent of patients reported no pain and 4 percent reported ineffective pain management during the procedure; with respect to postoperative pain, 92 percent reported at least some pain (tolerable pain through unbearable pain) and 10 percent reported ineffective pain management after the procedure.83–87
Uterine Volume Reduction. Studies varied in their period of reporting for uterine volume reduction. Studies reported the following percentages of mean uterine volume reduction: at 3 months, 27 percent;83–87, 92, 93, 96 at 6 months, 52 percent80 and 58 percent;99 and at 12 months, 38 percent92, 93, 96 and 47 percent.79
Mean Recovery Time. Three studies, set in Canada, the United Kingdom, and the United States, were consistent in reporting a 13- to 14-day period for recovery.83–87, 98, 100 One U.S. study reported an 8-day period for recovery.92, 93, 96
Complications. Variations in the methods and timing of reporting make the summary evaluation of complication rates across all studies extremely challenging. The largest of these studies, the FIBROID registry, reported a major in-hospital complication rate (e.g., hospitalization, major therapy, unplanned increase in care, or permanent adverse sequelae) of 0.6 percent of the sample; the postdischarge major complication rate was 4.1 percent.100 The rates of minor complications (nominal or no therapy, no consequences) was 2.1 percent during the admission and 22 percent within 30 days of discharge.
Rate of Subsequent Interventions. As with complication rates, studies vary in the method and timing of reporting rates of subsequent interventions. The FIBROID registry reported that 141 women (9.5 percent of their sample) had experienced at least one gynecological procedure by 12 months; procedures included 49 hysterectomies, 25 myomectomies, and 21 repeat UAEs. The study with the longest period of measurement reported a 25 percent failure (no improvement in symptoms—menstrual bleeding, pain, pressure—or major intervention) by 60 months.92, 93, 96
Modifiers of UAE Outcomes. Demographic Variables and Uterine Characteristics. Nine studies examined modifiers of UAE outcomes,77–80, 83–88, 95, 96, 100 including two from the FIBROID registry trial.95, 100 These studies examined a variety of demographic characteristics including age, race, parity, menopausal status; uterine characteristics, including size and location of the dominant fibroid; health characteristics such as prior surgery and smoking; and UAE characteristics such as UAE particle type and load. Outcomes examined included volume reduction, treatment failure, treatment success, satisfaction, and complications.
Studies that examined age found no association between age and UAE failure80 or satisfaction with outcomes.96 One of two studies96, 100 examining race found it to be a significant predictor of outcomes; the study found that African-American women have a higher risk of adverse events following UAE.100 Parity and menopausal status were not significant predictors of UAE failure.80
Regarding uterine characteristics, four studies that examined baseline uterine characteristics found no relationship between size or volume and UAE failure,77, 80 satisfaction with outcomes,96 or development of intrauterine infection.88 Two of four studies examining fibroid size found no effect on outcomes (failure77 or complications78). Other studies on volume reduction reported conflicting results: one study reported that larger fibroid size predicted greater decrease in volume,83–87 whereas two others reported that size of the dominant fibroid at baseline predicted less volume reduction at both 395, 96 and 1296 months after therapy. Studies also found that adjusted for fibroid volume, submucosal dominant fibroids predicted greater volume reduction96 and improvement in symptoms95 than subserosal fibroids. The location of the fibroid did not predict intrauterine infections.88
UAE characteristics such as size of particles used, particle load, and post-UAE complication events did not predict treatment failure at 6 months80 or intrauterine infection88 in two studies; in a third study, the use of EmboGold® particles versus Embosphere® particles resulted in significantly higher risk of skin rash and slower return to usual activities with EmboGold®, but no difference in volume reduction, fibroid expulsion, or satisfaction.79
Modifiers of Pain Associated With UAE. Two studies addressed pain in relation to the UAE procedure.89, 91 A prospective case series evaluated the effectiveness of superior hypogastric nerve block (SHNB) in addition to conventional conscious sedation for pain control in 139 patients.89 The investigators contacted patients on the third and fifth day after their procedures to elicit pain scores (numeric rating scale from 0 to 10). The first 100 patients had received the standard pre- and postprocedural analgesia. However, after review with the institutional pain management clinic, clinicians had identified a potential for enhanced postprocedural pain and antiemetic treatment, and the last 39 patients received the different regimen, involving SHNB (see Evidence Table 4, Appendix C *), which added 8 minutes to 10 minutes to the procedure. All patients could be discharged home by 6 hours after the procedure and had mild pain or no pain at the time of discharge. Readmission rates did not differ significantly between the two regimens (6 percent for conventional vs. 2.6 percent for enhanced intervention). The mean peak pain score did differ significantly between groups; women receiving the enhanced SHNB protocol reported lower pain scores (5.7 ± 2.2 vs. 2.7 ± 2.5; P < 0.01).
Based on animal models, Ryu and colleagues had hypothesized that Embosphere® would be associated with less pain after UAE than polyvinyl alcohol particles. They compared 29 patients in an Embosphere group with 26 patients in a polyvinyl alcohol particles group in a retrospective analysis.91 They reported no difference between the groups either in the dosages administered through a patient-controlled analgesia pump that delivered morphine sulfate or in the mean subjective pain scores.
Use of Imaging Techniques in UAE. Two studies evaluated the role of imaging modalities in UAE, such as magnetic resonance imaging (MRI).71, 81 One study prospectively followed 111 patients to assess them for the presence of persistent contrast enhancement of fibroids on a routine 6-month follow-up MRI; the investigators specifically tested whether continued gadolinium enhancement (contrast material-enhanced MRI) of the fibroid after failed primary UAE would predict a subsequently successful repeated UAE.71 Clinical failure was reported in 11 patients (10 percent). Of these 11 patients, eight (73 percent) showed persistent gadolinium enhancement of their dominant treated fibroids on MRI. All eight women were offered a repeat UAE; six accepted. All six had complete resolution of clinical symptoms at 12 months' followup. Additionally, in all six patients, no contrast material enhancement was identified on follow-up MRI at 6 months. The three patients who did not show persistent enhancement demonstrated complete tumor necrosis and were not offered a repeat procedure. Of note, of the 100 clinically successful cases, four had some persistent enhancement of their dominant fibroid with complete necrosis of the remainder.
The second study sought to determine whether Doppler flow measurements are useful in predicting variables associated with UAE, including shrinkage of the uterus and fibroids, adenomyosis, and procedure failure.81 The investigators evaluated 188 women with Doppler sonography before and 6 months after the procedure. The specific factor analyzed in this study was peak systolic velocity (PSV), an indicator of blood flow. Pre-embolization PSV values were positively correlated with total uterine volume and the diameter of the largest fibroid; that is, stronger blood flow was positively correlated with larger uterine and fibroid volume. In addition, the authors noted a positive correlation between the particle load required to block the vessel and the pre-embolization PSV values (P = 0.009). Higher pre-embolization PSV was associated with greater reduction of the largest uterine fibroid (P = 0.0174) and reduction in uterine volume (P = 0.0440); however, pre-embolization PSV was a significant predictor of failure (P = 0.02). Finally, the authors did not report any association between baseline uterine size or factors related to the procedure and failure of embolization.
Effects of Operator Experience on UAE. One Canadian study examined the effects of the experience of interventional radiologists on procedure and fluoroscopy time through a multicenter prospective design.83 UAE was successful bilaterally (in both uterine arteries) in 97 percent of patients; 94 percent of the procedures were completed on the first attempt. The overall procedural complication rate was 5.3 percent (30 of 570 procedures). Of these 30 procedures with complications, the most common complications were related to angiography; three women required extra care or an extended hospital stay. The article does not provide information on whether complications were influenced by operator experience. The study also found that procedure time and fluoroscopy time differed significantly for early experience (the first 20 consecutive procedures) versus later experience (the next 20 consecutive procedures) (P < 0.001).
Evaluation of Devices Used in UAE. Previous studies have suggested that the use of suture-mediated closure devices (SMCDs) may be associated with a higher rate of complications in patients who are undergoing UAE than in patients who have peripheral vascular disease and/or are undergoing anticoagulation.101 One study assessed the safety and efficacy of SMCDs in UAE through a prospective case series involving attempts to use SMCDs in 328 of 342 consecutive patients.72 Device failure occurred in eight patients (2.4 percent; 99% CI, 0.2–4.5 percent). No long-term major complications occurred; however, the rate of minor complications, including thigh pain related to the puncture site and minor hematomas, was 22 percent (72 of 328 women; 99% CI, 16–28 percent).
Comparative Studies. Three studies compared UAE with myomectomy.70, 73, 90 Three studies compared UAE with hysterectomy.75, 76, 94 The results of these studies are reported in detail in the myomectomy and hysterectomy sections, respectively, later in this chapter.
A single study compared two different methods of UAE: 24 women undergoing UAE and 22 women undergoing laparoscopic occlusion of the uterine arteries; the project was done in a nonrandomized prospective cohort of women with symptomatic fibroids in Norway.74 The investigators assigned women to laparoscopic occlusion when the size of the uterus did not exceed the umbilical level and to embolization regardless of fibroid size. They reported no differences between the groups in bulk symptoms or initial pictorial blood loss assessment score. Both groups had a statistically significant decrease in the volume of the dominant fibroid and the uterus from baseline following the procedure, but the groups did not differ significantly from each other. Postoperative pain medication consumption was significantly greater in the UAE group. By the final followup at 6 months, 15 UAE and 14 laparoscopy patients reported a satisfactory reduction in their bleeding. Four hysteroscopies, one dilatation and curettage, and two hysterectomies were performed during the follow-up period.
Endometrial Ablation (With or Without Myomectomy)
The prior evidence review did not identify publications about use of endometrial ablation specifically for the management of uterine fibroids.30 A single study appearing since 2001 reported results on endometrial ablation in comparison with myomectomy.102 The results are reported in the section on myomectomy (Appendix C *, Evidence Table 5).
In Situ Destructive Techniques (MRI-Guided Focused Ultrasound): Overview and Nomenclature
One part of KQ 2 assesses outcomes of interventions to treat fibroids that use techniques to destroy them in situ. Methods previously explored in the research literature include cryoablation (which is freezing the fibroid tissue) and laser ablation (which burns the tissue to destroy the fibroid) via laparoscopy. Neither of these methods is currently available in clinical practice outside research settings. We did not identify any publications on these methods in the timeframe for this review. MRI-guided focused ultrasound, a new technique, is the only in situ destructive technique currently being used outside academic and specialty clinics. This method did not have eligible publications to include in the prior evidence review on management of fibroids. Our search identified two publications of fair quality from a single cohort.103, 104
In MRI-guided focused ultrasound, the clinician uses the MRI to guide the ultrasound energy (i.e., sound waves from the ultrasound) directly to the fibroid. The highly focused ultrasound beam (very different from ultrasound used for imaging studies) causes the temperature in the target tissue to rise. The clinician can monitor the thermal destruction of the fibroid during the procedure with the MRI and avoid damage to nearby tissue or structures. We describe both the conduct of the procedure and the findings of these studies in some detail because the technique is so new.
The treatment is conducted in an MRI suite using an imaging system that integrates real-time MRI and thermometry with an ultrasound unit specially designed to focus the ultrasound waves to create heat; a process like this, which is intended to disrupt biologic materials by use of sound wave energy, is also termed sonication. The woman receives light sedation and is positioned face down on a gantry over a contact pad required for the ultrasound. The gantry, which is a treatment table on a track, is moved to help position her correctly within the MRI machine. The MRI is used to image the fibroids, to finalize positioning of the patient, to help avoid exposing other organs such as the bladder and bowel to the ultrasound, and to define tightly the area of each fibroid for treatment. The clinician uses a “test dose” of ultrasound so that the MRI thermal measurements can confirm that the correct area will receive treatment and then begins the focused ultrasound heating of the target fibroid tissue. Thermal destruction is monitored in real time using MRI estimates of the tissue temperatures achieved. Each fibroid is treated separately, and total treatment times are generally longer than an hour for most women, with a 3-hour total treatment time limit.103–106 The U.S. Food and Drug Administration approved the treatment system in 2004.107
We identified two publications that present data from the same study population (Appendix C *, Evidence Table 6).103, 104 This work was undertaken to assess the safety and efficacy of this technique, within a collaborative network of sites including three U.S. centers, two European centers, and one Israeli center, all at academic institutions.
The research collaborative focused on documenting adverse events and identifying the proportion of women who had meaningful improvement in their symptoms as defined by use of the standardized and validated, disease-specific Uterine Fibroid Symptoms Quality-of-Life (UFS-QOL) scale32 in addition to the Medical Outcomes Trust Short Form-36 (SF-36).103, 104 The UFS-QOL questionnaire has eight symptom questions and 29 health-related quality-of-life questions (with six subscales), scored on a 100-point scale with higher numbers indicating more severe impairment.
The study population comprised 109 women who were premenopausal and who reported that they had completed childbearing. Eleven percent of the women were African American, African, or Afro-Caribbean. All participants scored above the mean for women with fibroids on the UFS-QOL, reflecting good representation of highly symptomatic women. Each woman was clinically considered a candidate for hysterectomy or myomectomy based on disease severity.
Fifty-five percent had one fibroid treated; the remainder had two or more treated. Overall, 22 percent of fibroids were submucosal, 57 percent intramural, and 21 percent subserosal. The average duration of time within the MRI scanner was 202 minutes (range, 90 to 370 minutes); a portion of this was ascribed to the time required to position the patient correctly. At some point during the procedure, 16 percent of women reported severe pain; 1 percent and 7 percent reported severe or moderate pain, respectively, immediately following the procedure. The majority of women reported mild (33 percent) or moderate (33 percent) pain during the treatment portion of the procedure and no (75 percent) or mild (18 percent) pain immediately afterwards. The single serious complication deemed related to the procedure was a sciatic nerve palsy that fully resolved by 12 months. The injury resulted in a change to pretreatment planning during the balance of the study. One woman had an abdominal skin burn that caused ulceration prior to healing and was traced to incomplete shaving of the abdomen. (Complete shaving in the path of the ultrasound beam is critical because it prevents air pockets that can result in local skin heating.) Six percent of women had febrile morbidity and all six received antibiotics as a precaution. Overall, participants returned to work an average of 1 day after the procedure.
At 6 months after treatment, 77 participants (70.6 percent) reported a 10-point or greater improvement on the UFS-QOL questionnaire. The mean decrease in the score of the women treated was 23.8 points; symptom relief was similar for “bulk symptoms” related to the size and position of fibroids, and for bleeding symptoms. This similarity in levels of improvement for both pressure and bleeding symptoms is notable in the context of a 13.5 percent decrease in the average total volume of fibroids treated; specifically, it suggests possible placebo effects at work because actual volume reduction was quite modest. By 12 months, 51 percent (42 of 82 women who could be evaluated) had sustained improvement of 10 points or more on the UFS-QOL. Scores on the SF-36, which was also administered, indicated improvements at 1, 3, and 6 months compared with baseline scores.
Failure, defined as worsening of symptoms by 6 months, was 11 percent (12 of 109 women); 10 women were classified as unchanged. By 1 year, 23 of 82 evaluable women (28 percent) had sought additional treatment including myomectomy, hysterectomy, or UAE.
The authors noted that the safety protocol requirements of this initial research were highly conservative and required that only a small proportion of the fibroid volume be treated. Clinical practice now successfully targets substantially larger proportions of the total fibroid volume. Future research may yield greater improvements in outcome, but these data do demonstrate the safety and preliminary efficacy of the procedure for improving symptoms.103, 104 The need for comparative trials and longer-term followup for this and all fibroid treatment modalities is highlighted in the discussion in Chapter 4.
Myomectomy: Overview and Nomenclature
This section presents the results of our literature search and findings about outcomes of surgical removal of fibroids, termed myomectomy. Myomectomy removes the fibroid(s) that can be surgically removed, repairs the defect in the uterine wall, and does not remove the uterus. For this reason, myomectomy is the surgical option available to women who wish to have future pregnancies or who wish to retain their uterus.
As we briefly describe below, the content of the literature spans the range of surgical approaches currently available in routine clinical practice. We found no publication that described outcomes of robotic surgery, which is becoming available at a limited number of highly specialized sites. Detailed information on all studies relating to myomectomy can be found in Evidence Table 7 in Appendix C *.
For convenience and consistency we have used uniform terminology to describe and discuss the different surgical techniques used to remove or destroy uterine fibroids. The groupings that follow—abdominal, laparoscopic, and hysteroscopic—are approximately in the order of “invasiveness” as reflected by the size of the surgical incision to be healed, the degree of disruption of nearby tissue, and, therefore, the amount of healing required after the procedure.
Abdominal Myomectomy and its Variants. Abdominal myomectomy per se is the removal of fibroids through an incision in the skin of the abdomen; this is also called a laparotomy incision. This includes midline incisions made along the line between the umbilicus and the pubic symphysis or “pelvic bone,” as well as incisions made lower on the abdomen at a right angle to that line. The surgeon operates with his or her hands and instruments in direct contact with the abdominal and pelvic organs.
Minilaparotomy is the removal of fibroids through an incision in the skin of the lower abdomen that is smaller than conventional abdominal myomectomy; it is intentionally made to be as small as possible while still allowing the surgery to be performed. Surgeons use several techniques to accomplish this, including raising the uterus through a small incision to operate on the uterus “exteriorized,” meaning elevated out of the pelvis through the skin incision, or raising each fibroid individually up to, or out of, the incision. In our classification of surgical methods, we use this term to refer to myomectomies done by minilaparotomy and accomplished with the surgeon's hands and conventional instruments in direct contact with the abdominal and pelvic organs.
Laparoscopically assisted myomectomy is the removal of fibroids assisted by use of a laparoscope and other instruments inserted through small incisions in the abdominal wall; generally, each incision is less than 1.0 to 1.5 centimeters in size. The laparoscope is attached to a video camera and the surgeon(s) conduct a portion of the procedure while watching the surgery progress on a display screen. In the majority of the cases described in this literature as laparoscopically assisted, the laparoscope was used to augment minilaparotomy to ensure that the skin incision size could be kept small and still allow surgeons to view and operate in areas of the uterus that were more difficult or not possible to reach through the minilaparotomy incision alone; this situation might occur, for instance, low on the posterior aspect of the uterus. Some publications also describe laparoscopic removal of the fibroid followed by closing the defect in the uterus after the fibroid is removed by working through a minilaparotomy incision, in which the surgeon raises the uterine defect up to the incision to close the incision in the uterus using traditional open surgical techniques.
Laparoscopic Myomectomy. Laparoscopic myomectomy is the removal of fibroids using a laparoscope and instruments inserted through “ports” in the abdominal and pelvic wall to accomplish the entire surgery. The surgeon's hands are not directly in contact with the fibroid(s) or pelvic organs during the surgery. The fibroid is removed and the uterine defect is repaired entirely through the laparoscope. During laparoscopic myomectomy, fibroids are generally morcellated (i.e., cut into smaller pieces) to remove them from the abdomen through small openings. This can be accomplished with laparoscopic instruments like scissors or various forms of scalpels or with a specialized device called a morcellator. For this review we have included in this category any myomectomy done via colpotomy, which involves a vaginal incision to remove the fibroid(s) intact or in large pieces from the pelvis.
Hysteroscopic Myomectomy. Hysteroscopic myomectomy is the removal of fibroids using a hysteroscope. This instrument is inserted via the vagina through the cervix, which is dilated to allow the hysteroscope to pass into the uterus. Most often, a camera is attached to the hysteroscope and used to view the procedure on a display screen, although the surgeon may also view the procedure directly through the hysteroscope. Instruments are passed through a single tube that houses the lens for the camera, allows fluid to flow into the uterus to distend it, and provides access for operative instruments, one at a time. The instruments are used to cut, burn, or “shave down” fibroids that can be seen and reached through this interior view of the uterus. As needed, pieces of fibroid are flushed out with the fluid or grasped and removed from the uterus using instruments.
Endometrial ablation, which is the permanent surgical scarring or removal of the lining of the uterus (i.e., the endometrium), is reported in this literature as a concurrent hysteroscopic procedure; in one publication, it is considered as a primary treatment for fibroids associated with abnormal bleeding. To accomplish it, the surgeon inserts the hysteroscope as for hysteroscopic myomectomy and then uses an instrument to resect or cauterize and destroy the endometrium so that it is scarred and unable to support growth of an endometrial lining. When all or a sufficient proportion of the interior of the uterus is ablated, future bleeding decreases or stops altogether. In the literature in this review, all procedures were done with roller-ball or loop ablation techniques. No publications that reported on thermal balloon or wire mesh systems, which are designed to treat the interior surface in a single round of heating, met inclusion criteria.
The prior evidence review identified 43 studies about myomectomy outcomes, overwhelmingly case series. All studies that included symptom outcomes reported improvements, although measures and follow-up timing were poorly described. Transfusion was the most common short-term complication reported (1.2 percent to 16 percent); uterine perforations and fluid and electrolyte disturbances after hysteroscopy were inconsistently noted.
In summary, our review yielded the following findings. Data were limited on the effect of myomectomy for long-term symptomatic relief. No data supported use of prophylactic myomectomy in women with asymptomatic fibroids. Clear data from multiple studies indicated that myomectomies do have a risk of complications, which appears to increase with increasing number of fibroids removed. Data were insufficient to allow estimation of the cumulative incidence of recurrent symptoms after conservative management of fibroids. Reported recurrence rates ranged up to 50 percent by 5 years after myomectomy, with up to 8 percent of patients undergoing hysterectomy. Data for direct comparison of the risks and benefits of myomectomy and hysterectomy were lacking. The report noted two modifiers of myomectomy outcomes: (1) risk of recurrence of symptoms and fibroids may be lower when only one fibroid is present and removed, and (2) myomectomy may be more effective in perimenopausal women than in premenopausal women. Overall, we judged the quality of the literature about myomectomy to be weak.30
Studies, Designs, and Populations. For this update, we identified 39 unique studies (with 44 publications) that reported on outcomes of myomectomy of any type, including comparisons of myomectomy with other treatments or procedures.70, 73, 90, 102, 108–148 Some publications dealt with use of GnRH agonist medications to reduce the size of fibroids prior to surgery, either hysterectomy or myomectomy; we summarized these findings above (in pharmaceutical interventions) and do not review them in detail here.
The overall quality of this literature was poor (21 publications) to fair (22 publications); with a single small RCT receiving a quality rating of good.146 Statistical weaknesses were most common. Six studies either provided an a priori calculation of statistical power and required study size or calculated power achieved. They also included multivariate analysis to adjust for potential confounders or to identify and assess effect modifiers as needed. Nine had either a power calculation or multivariable analysis; the remainder provided neither. Likewise, documentation of inclusion and exclusion criteria was weak, as was documentation of participant characteristics including key information about fibroids such as baseline number and size.
Some study populations are represented more than once in the total number of publications because the authors focused on individual outcomes in separate publications,108–110 published subanalyses,128, 129 followed up participants at a later time in order to report on later outcomes such as satisfaction with surgery or pregnancies,125, 126 or expanded on a case series by including the original participants in a larger series in a subsequent publication.119, 149
This literature base included 24 case series studies, which we have operationally defined as descriptive analyses of a sequence of participants having the same type of procedure without a comparison with another type of surgery or treatment. Eighteen of these studies are retrospective case series of a particular type of myomectomy: five report on abdominal myomectomy,90, 123, 133, 136, 144 one on minilaparotomy,122 six on laparoscopic myomectomy,116–119, 134, 147, 149 four on hysteroscopic myomectomy,102, 111, 124, 132 and one on myomectomy at the time of cesarean.114 Six studies are prospective case series: four of laparoscopic myomectomy115, 125, 127, 150 and two of hysteroscopic myomectomy (one with multiple reports).108–110, 130
Eight studies are cohort studies that compare outcomes across two or more types of myomectomy procedures. Of these studies, five involve retrospective cohorts: three of fair quality compared abdominal and laparoscopic myomectomy outcomes;128, 138, 139 one of poor quality compared abdominal myomectomy with UAE;70 and one of fair quality compared myomectomy to expectant management to examine the outcomes of assisted reproductive technology (ART) treatment.141 We identified three prospective cohort studies: one of fair quality compared abdominal and laparoscopic myomectomy;121 one of fair quality compared abdominal myomectomy and UAE;73 and one of poor quality compared laparoscopic myomectomy and expectant management before in vitro ART treatment.112
Eight studies were RCTs: one of good quality examined “chemically assisted dissection” with sodium-2-mercaptoethanesulfonate (mesna) to define tissue planes and facilitate fibroid resection during myomectomy;146 two trials, one poor and one fair, examined interventions to reduce blood loss at the time of myomectomy;145, 148 two trials, one fair and one poor, examined products applied at the time of myomectomy to reduce adhesion formation,142, 143 two fair-quality studies randomly allocated participants to abdominal or laparoscopic myomectomy;135, 137 and one, of fair quality, randomly allocated participants to three arms—abdominal myomectomy, minilaparotomy, and laparoscopically assisted minilaparotomy.113
The myomectomy literature is dominated by case series from large academic, tertiary care centers and internationally recognized fibroid surgery centers. Studies conducted in Europe outnumber those conducted in the United States or Canada by more than two to one. Among European studies, the majority were conducted in Italy or France. All but one of 11 North American studies were conducted in the United States, including one study that had a study site in Taiwan.
Regardless of country, the majority of studies were conducted in academic centers or specialty fibroid care facilities. Ten studies reflected care in community hospitals or clinics. Two studies relied on large databases, one in a national health care database in Norway and the other in a large private insurer database in the United States.
Outcomes Measured. For each type of surgical procedure, we combed the publications for the outcomes and complications summarized in the analytic framework presented in Chapter 1 (Figure 1). The majority of studies included perioperative outcomes. The clinical outcomes fairly uniformly included the number or size (or both) of fibroids removed, estimated blood loss or change in blood count (e.g., hemoglobin levels), transfusions (number needed or percentage of women receiving at least one), febrile morbidity, and complications. Virtually all studies used conventional clinical measures for these outcomes; some specified operational definitions or specifically timed measurements. Two measures assessed clinical processes including the length of the procedure and length of the entire hospital stay.
Retrospective studies, by definition, rely entirely on existing clinical or administrative data. Such use of clinical data means that measures such as intraoperative blood loss will be biased by documented phenomena such as digit and rounding preferences and by the a priori impressions of the surgical team about how the type of procedure relates to anticipated blood loss. Likewise, clinical practice routines play a large part in determining outcomes such as pain medication strategies or timing of discharge after the procedure. Of note, given the peculiarity of surgical studies, even RCTs are not insulated from these effects of practice patterns and assumptions about likely outcomes. Unless intraoperative details, such as blood volume in the suction canisters and on sponges, were recorded by an observer for whom the group allocation was unknown, and unless postoperative care was coordinated by an individual unaware of type of surgery, the influence of practice patterns on outcomes cannot be avoided. In this literature, such a high level of masking of assessors and care providers is understandably not achieved.
Fourteen studies included some level of detail about the degree to which myomectomy improved symptoms related to recurrence of fibroids or was followed by other surgical interventions after the index myomectomy. None of the identified studies of myomectomy outcomes made use of standardized and validated measures of menstrual bleeding, participant satisfaction, or health-related quality of life.
Eighteen studies provided data about pregnancy outcomes after myomectomy; a large proportion of these focused exclusively on ART outcomes.112, 115–117, 119, 123, 125–128, 130, 133, 134, 137, 139, 141, 147, 150–152 However, several of these did not meet inclusion criteria for our later discussion of pregnancy outcomes in KQ 3 because they did not track or report the proportion of the women in the original study group who attempted to conceive or because they did not provide denominator data among those who did conceive to allow calculation of the probability of conception, pregnancy loss, or live birth among participants.
Limitations of Study Quality for Reproductive Outcomes. The overall quality of these studies was poor to fair. Because so many of these studies appear low on most study design hierarchies and because quite a few do not meet reasonable quality criteria, we have included in this review even articles and studies that we graded as poor. Quality grading procedures, drawing largely on the original review,30 are explained in Chapter 2. Except for studies in which the myomectomy was done concurrent with evaluation and treatment for infertility, most of the case series and cohorts do not report an approach to data collection that would provide an accurate measure of the proportion of women in the studies who desired a future pregnancy and who attempted conception. Without this information, and without clearly specified lengths of followup, reports of pregnancy, miscarriage, and birth rates are flawed because rates require both an accurate denominator and unit of time over which the outcome was assessed. Likewise, simple proportions of women achieving pregnancy after myomectomy require at minimum an accurate denominator of women attempting to conceive.
Reports of the outcomes of pregnancies achieved can nonetheless be accurately summarized as descriptive data about the proportion of known pregnancies resulting in miscarriage, preterm birth, or cesarean birth and about the proportion associated with complications such as uterine rupture. Miscarriage data will underrepresent true reproductive inefficiency because some pregnancies will be lost before conventional pregnancy testing identifies the pregnancy. No studies of reproductive outcomes after myomectomy used daily urine or serum human chorionic gonadotropin testing to identify pregnancies close to the time of implantation and none, other than those among women receiving care for infertility, tested for pregnancy on a predetermined schedule. If women with one type of treatment for fibroids or without fibroids are differentially likely to conduct pregnancy testing earlier rather than later, the potential for bias caused by differences in very early loss rate is not negligible but cannot be assessed using outcomes reported in these studies.
Abdominal Myomectomy: Perioperative Outcomes. Thirteen publications (eight of poor quality, four of fair, and one of good) provided information about perioperative outcomes of abdominal myomectomy (Table 9).90, 113, 121, 122, 128, 133, 136–138, 140, 144, 146, 148 This category includes studies of abdominal surgery and those involving minilaparotomy or laparoscopically assisted myomectomies, as all involve some form of abdominal incision. Four studies presented RCT results.113, 137, 146, 148 One study was a prospective cohort;121 two were retrospective cohorts;128, 138 and six were retrospective case series.90, 122, 133, 136, 140, 144
One RCT evaluated use of mesna as a “chemical aid to dissection” of the myoma at the time of abdominal myomectomy.146 One RCT evaluated techniques for reducing blood loss at the time of myomectomy.148 In a total study population of 94 women, 31 had vaginal myomectomy. The authors did not provide data separately by myomectomy approach. We present surgical and trial outcomes here (with respect to results for abdominal myomectomy) because this was the only study in the review that included any women who had vaginal myomectomy, which is not a common technique in the United States.
One other trial compared abdominal myomectomy, minilaparotomy, and laparoscopically assisted myomectomy.113 Data from each arm of this trial and the results of comparisons across arms appear here because each participant had at least a minilaparotomy incision. One RCT compared abdominal with laparoscopic myomectomy;137 outcomes for the abdominal group are presented here, and more detailed consideration of direct comparisons are discussed in the next section on laparoscopic myomectomy.
One case series,122 one group within a cohort,121 and one arm of a clinical trial113 focused exclusively on outcomes of minilaparotomy. Finally, the largest study (N = 1,959), conducted using data in a large private insurance database in the United States, includes some outcomes data, which are presented here.140 As this work was focused predominantly on costs, and we review those results as part of KQ 4.
Excluding the large insurance database study, the remainder of the publications that include operative outcomes reported on study populations of small to modest size. Populations range from 41 in a clinical trial to 225 in a retrospective cohort formed by hospital record review.
Abdominal myomectomy is a major surgical procedure, as reflected in the data on perioperative outcomes and complications presented in Table 9. We summarize the clinical and utilization measures below.
Fibroids Removed. Seven studies reported some form of data on this outcome. With respect to the number of fibroids removed, the range over five studies was 1.2 to 9, and with respect to weight, the range (three studies) was 170 grams to 286 grams.
Blood Loss and Transfusions. Average operative blood loss, for six studies, ranged from 200 ml to more than a half liter of blood loss (508 ml). Two studies reported decreases in hemoglobin ranging from 1.8 g/dl to 3.1 g/dl. The study that evaluated mesna to assist resection reported the lowest blood loss in the mesna arm (0.9 g/dl) and a more conventional decrease, 1.7 g/dl, in the placebo control group. Finally, one study reported that 31 percent of patients had a blood loss greater than 500 ml. Most of these studies reported mean estimated blood loss without commenting on the handling of extremes of minimal or excessive blood loss. Few authors commented on other measures of centrality such as the median or any skew in the data.
Transfusion risk was reported in eight studies. The percentage of women requiring transfusions ranged from < 1 percent to 21 percent. In four studies, the numbers of transfusions ranged from 1 to 7. Use of intravenous oxytocin (compared with placebo) for reducing blood loss did not provide evidence of advantage when comparing mean blood loss; the study was underpowered to evaluate influence of risk of transfusion.148 The publications that reported on minilaparotomy and laparoscopically assisted minilaparotomy provided too little detail to determine if these approaches were associated with reports of higher or lower blood loss.
Emergency hysterectomy at the time of abdominal myomectomy is most often a response to excessive bleeding. The two studies that best reflect general practice (including a large number of surgeons at community sites) are one in the United Kingdom and a U.S. insurance database. These studies reported that 4 percent and 3.7 percent (respectively) of women presenting for abdominal myomectomy had their procedure converted to a hysterectomy.133, 140
Fever. Clinicians believe that febrile morbidity is a common occurrence after myomectomy. Definitions of febrile morbidity in this literature ranged from a single temperature recorded at 38° degrees Centigrade (C) or higher, to requiring repeated measures of fever over a number of hours. The three studies that reported low febrile morbidity (2.9 percent,136 3.3 percent,121) and 12 percent (aggregate of two small study arms with 3 and 20 percent per arm)146 based their information on undefined “fever” from chart review136, 146 or a requirement for temperature of 38° C on two occasions at least 6 hours apart, excluding the first day after surgery.121 The remainder of studies reporting on febrile morbidity all reported that temperature elevation occurred in 15 percent or more of the study population (15.9, 17, 23.5, 26, 26.2, and 38 percent).
The clinical relevance of a high proportion of postoperative patients having fevers is related to the degree to which clinical examinations and diagnostic testing are done to evaluate the patient and rule out other sources of infection including urinary tract infection, pelvic operative site, and wound infection. Regardless of cost and effort required to evaluate the febrile patient, the occurrence of fever also influences length of stay; virtually all authors reported that a common clinical criterion for discharge is that the patient be afebrile.
Other Complications. Frank infectious complications and wound healing abnormalities are known outcomes of all surgical procedures. Women having myomectomies are generally young and healthy and rates of such complications are low. Endometritis and wound infections were reported at rates below 1 in 100 women. Wound healing complications, which can be difficult to distinguish from wound infection, were more common, affecting between 2 percent and 6 percent of participants in studies of abdominal procedures. Only one minilaparotomy study reported on wound healing complications, which occurred in 2 percent of their participants.122 Because wound healing complications such as seromas and hematomas generally require opening the incision and either allowing it to heal by secondary intention with daily wound care and dressing changes or reclosing the incisions with suture or staples after debridement, they present significant morbidity for the patient.
Other complications (data not shown in Table 9) such as intraoperative bowel and bladder injuries were rare. Readmissions were rare as were postoperative bowel and bladder complications such as ileus and urinary retention. No perioperative deaths were reported in these studies.
Utilization Measures. Table 9 also contains information for abdominal myomectomy or its variants on operative times and length of stay. Operative times among the seven studies reporting them all exceeded 1 hour (range from 62 minutes to 180 minutes). Length of stay varied from 13.6 hours to 6.9 days in ten studies.
Abdominal Myomectomy: Longer-Term Outcomes. Nine studies, six of poor quality and three of fair quality, followed up participants months to years after abdominal myomectomy (Table 10).70, 90, 122, 123, 128, 133, 135, 137, 140
The longest followup included women who were contacted an average of 49 months from the initial abdominal myomectomy; shortest follow-up periods were generally 24 months (except for one study that had a range including some women followed for as short a time as 2 months).
Improved Symptoms. After abdominal myomectomy, more than half of women studied had improvements in the symptoms for which they sought care. Outcomes evaluated, most often by survey or telephone interview, included the following: “improved symptoms,” 68 percent; no recurrence of heavy bleeding over 5 years, 50 percent; and “completely” or “significantly” improved menorrhagia in 64 percent of women, pain in 54 percent, and mass effects in 91 percent. One study with 30 participants who had abdominal myomectomies specifically addressed satisfaction with treatment outcomes. At an average follow-up time of 49 months, 10 percent of women had had no improvement or worsening of symptoms and 21 percent were very dissatisfied with the therapy, indicating that 69 percent had satisfactory results.70 The investigators for the studies reported here did not carry out formal health-related quality of life, functional status, or detailed satisfaction surveys.
Subsequent Interventions. Incidence of fibroids rises through the late reproductive years. For that reason, recurrence of fibroids after myomectomy is expected, either through growth of small fibroids that could not be identified or removed at the time of first surgery or through appearance of new fibroids.30 In some proportion of such cases, further surgery or other interventions may be advised and carried out.
Two studies of fair quality assessed all participants for recurrence through uniform use of imaging at regular intervals (both were RCTs comparing abdominal myomectomy with other surgeries); they reported that 18 percent of women at 32 months137 and 23 percent of women at 40 months135 had newly identified fibroids. Hanafi, using data linked to clinical records of ultrasounds done after the index surgery, found that 62 percent of women (followed for an average of 38 months) had fibroids on subsequent ultrasound.123
In studies of longer-term operative outcomes, recurrence is presumed to be the underlying cause for subsequent surgeries; this association, however, is generally not proven by documenting recurrence to the reports of the proportions who have subsequent procedures. In the six studies that sought self-report, medical record evidence, or prospective follow-up data about subsequent intervention, between 1.4 percent and 17 percent of women had another surgery, but we found only limited information to describe what proportion of these procedures were hysterectomy compared to myomectomy.
This literature is limited by the dominance of retrospective case series and cohorts that do not have sufficient opportunity to operationalize outcome definitions and unify measurement for research purposes. As throughout this review, we emphasize that (with the exception of two community-based sources of data) these outcomes reflect the experience of women receiving care in academic centers and specialty clinics with an explicit interest in fibroid care. The community studies suggest higher rates of complications than those observed in academic centers. Outcomes cannot be predictably generalized to all abdominal myomectomies performed in all care settings.
With this concern noted, we can summarize that transfusion and febrile morbidity are expected to be common. Consent for abdominal myomectomy should specifically address the real possibility of transfusion. Exploring autologous blood banking and use of cell-saver and other technologies may be advisable to reduce risks from heterologous transfusion. However, autologous and cell-saver technologies are not without risk themselves. Thus, in general, strategies for minimizing blood loss are preferable to increased use of tools to accommodate high blood loss.
Laparoscopic Myomectomy: Overview. In all, 16 studies (17 articles) dealt with laparoscopic myomectomy alone. Thirteen studies, nine of fair quality and four of poor quality, (14 publications) provided information about perioperative outcomes of laparoscopic myomectomy (Table 11).115–117, 119, 121, 125, 127, 128, 134, 137–140, 145 Five publications provided some information about longer-term outcomes that include resolution of symptoms and subsequent surgeries—four already noted128, 135, 137, 140 and one additional study.118 Three of the longer-term outcome publications were retrospective analyses; two were RCTs comparing abdominal with laparoscopic myomectomy.135, 137 One study reported short-term operative outcomes stratified by whether the participant subsequently achieved a pregnancy; the means and ranges are not provided in aggregate for all participants. The data by pregnancy status are not presented here because the analysis was conducted to examine what characteristics at the time of surgery predicted improved reproductive outcomes. This study is addressed in more detail in KQ 3.147
In short, the overlap in this literature is small. This lack of continuity is important because it means that the findings of follow-up studies do not reflect the outcomes of the populations studied in the perioperative studies. Overwhelmingly, data were not prospectively gathered to capture details about how surgical events influence long-term outcomes.
Laparoscopic Myomectomy: Perioperative Outcomes. The 13 studies with perioperative outcomes (Table 11) covered essentially the same outcomes as reported for abdominal myomectomy. Complications, however, are different insofar as they can include conversion of this particular operative procedure to one or another form of abdominal myomectomy. These studies involved study populations of small to moderate size with a total of 2,887 participants and an average of 222 participants per study. The range of size was 18 participants to 635 in a European specialty clinic case series.
Fibroids Removed. All but two studies provided some information about fibroids removed. In seven studies, the number of fibroids removed was, on average, fewer than 2; in four others, the number ranged from just over 2 to almost 6; and in one study, the fibroid weight removed was 151 grams.
Blood Loss or Transfusions. Nine studies reported data on average operative blood loss or on postoperative change in hematocrit or hemoglobin. Among those studies that reported estimated blood loss, the mean reported was 235 ml, with a range from 102 ml to 378 ml (in one arm of a trial). When direct comparisons are made within a single study population, laparoscopic myomectomy is statistically associated with lower operative blood loss (data not shown) and decreased length of stay,113, 128, 137, 138 though not in each case statistically significant.121
Transfusion was rare—less than 1 percent across studies. Seven studies reported no transfusions; of the remainder, the number ranged from one to ten.
Fever. Febrile morbidity was variably defined by authors; typically, they did not document operational criteria (such as interval of temperature measurement and duration of elevation). Ten studies had data on febrile morbidity. In terms of numbers of subjects with any fever, the values ranged from 1 (of 28) to 12 (of 368); using percentages as the metric, the values for any febrile morbidity ranged from 1.1 percent to 16 percent.
Complications. The primary adverse outcome was conversion from laparoscopic procedure to abdominal myomectomy, attributed commonly to difficulty with controlling bleeding, accommodating challenging anatomy laparoscopically, or closing the defect in the uterine wall.
Three Italian studies, each with highly specialized laparoscopic surgeons, reported no conversions among a total of 400 participants.115, 121, 145 Another large Italian series, also with highly specialized surgeons, reported a conversion rate below 1 percent among 635 procedures.117 Including these studies, the risk of conversion to an open incision, averaged across studies, was 6.1 percent. Excluding these reports, approximately 9 percent of women had conversion to abdominal myomectomy with a range from less than 1 percent to 29 percent.
Conversion in the study based on a large insurance database was 13.3 percent (to abdominal procedures), with an additional 2.8 percent conversion to hysterectomy.140 In a U.S. retrospective cohort, conversion was 12 percent;138 and in a group of 11 Italian university and community hospitals, it was 29 percent.128 This spectrum from highly specialized to more generalized practice suggests that, in conventional clinical practice, women and their care providers should anticipate a conversion rate of 10 percent or higher when discussing likely outcomes of laparoscopic myomectomy and planning for postoperative recovery.
Utilization Measures. In the 12 studies reporting on average operative times, all studies reported average times longer than 1 hour (range 62 minutes to 223 minutes) except for one study reporting its own range of 30 minutes to 140 minutes.
Across 11 studies, the length of postoperative admission (i.e., length of stay) generally averaged fewer than 3 days. One study reported a median of 2 days. Most studies apparently discharged their laparoscopic myomectomy patients by the middle of the second postoperative day. European studies tended to report somewhat longer lengths of stay than those done in the United States. This is the case across types of surgery and likely reflects underlying differences in practice styles rather than real differences in the trajectory of postoperative recovery.
Laparoscopic Myomectory: Longer-Term Outcomes. Resolution of symptoms and satisfaction with surgical outcomes were not investigated in the studies that we identified for this review. Five studies did report on recurrence (Table 12). Of these, three of fair quality used regularly repeated ultrasounds for all participants during followup over (on average) 31 months to 47 months;118, 135, 137 they documented recurrence rates of 12.7 percent at 1 year (or 16.7 percent by 5 years) to 22 percent or 27 percent (between 10 and 30 months). Contrasted with the estimated 2.5 percent recurrence of fibroids in a poor-quality study based on retrospective documentation of clinical findings and symptoms,128 these higher rates document the value of prospective surveillance for presence of uterine fibroids as a research tool.
Three studies, two of fair and one of poor quality, with a total of 637 participants followed for, on average, 24 months to 47 months, sought to document subsequent surgeries.118, 137, 140 One Italian group reported that no subsequent procedures were performed over a mean of 31 months; a French study reported that 4.6 percent of women had further surgery (predominantly myomectomies) over an average of 47 months; and the U.S. insurance database study showed that 12.3 percent of women had a subsequent myomectomy or hysterectomy within 2 years.
Hysteroscopic Surgery: Overview. Eight studies, with 10 publications, provided information about perioperative outcomes of hysteroscopic myomectomy. Seven of these also provided some information about longer-term outcomes including subsequent surgeries (Table 13).108–111, 124, 130, 132, 140, 153, 154
Studies ranged in size from a small comparison of endometrial ablation techniques with 42 participants to a case series of 948 participants. Two studies reported on combining hysteroscopic myomectomy with endometrial ablation during the same hysteroscopic procedure;111, 154 a single study reported primarily on use of endometrial ablation as a method of controlling bleeding for women with uterine enlargement from fibroids.153
Hysteroscopic Myomectomy: Perioperative Outcomes. The five studies of hysteroscopic resection (myomectomy) without associated endometrial ablation included 2,061 women (top panel of Table 13).102, 108–110, 130, 132, 140 Generally, the authors provided relatively little information about operative complexity and perioperative complications.
Fibroids Removed, Blood Loss, and Transfusion. Few authors documented the number of fibroids removed at the time of hysteroscopy. Authors often reported the number of fibroids present on ultrasound (details are recorded on Evidence Table 7 in Appendix C *), but the identification of fibroids does not necessarily equate to the number that were able to be resected at the time of surgery. Authors did not routinely report blood loss or transfusion risk; the latter appears to be low but is poorly documented.
Fluid Absorption. Three of five groups reporting on hysteroscopic resection with no other procedure described fluid absorption. This is an important measure because volume imbalances can lead to volume overload and/or hyponatremia, which can be life-threatening. Fluid absorption happens when the fluid used to distend the uterine cavity enters the blood stream via the rich network of blood vessels that serve the endometrium (lining) of the uterus and that can be exposed during hysteroscopic resection. The reported range of fluid absorption is wide, from a mean of 281 ml to a mean of more than 1 liter; one group reported only that they observed no cases of volume overload. Good overall operative technique and low average fluid volumes do not prevent adverse events; for example, the publication reporting a mean of 281 ml fluid absorption also reported one case of volume overload.
Perforations. Perforations of the uterus at the time of the procedure occurred in approximately 1 percent of women who had hysteroscopic myomectomy (10 of 1,009 women for whom data were available in the Agostini and colleagues study and the Munoz et al., study).
Utilization Measures. Length of surgery was not routinely reported by authors.
Other Complications. Seven women (0.7 percent) experienced hemorrhage. Infection was rare, affecting 0.55 percent of women in the two studies that tracked infection risk (data not shown). Incomplete procedures and conversion to other types of procedures were the most common undesired outcomes. In one study, 7.4 percent of cases were converted to abdominal myomectomy and 1.5 percent to hysterectomy. Other studies reported 13.1 percent to 18.3 percent incomplete resections.
This variation in conversion and incomplete procedures likely reflects practice patterns and routines for obtaining preoperative consent of patients. Surgeons who prefer to conduct a second procedure to attempt to complete the hysteroscopic myomectomy are less likely to obtain consent for same-day conversion to abdominal or laparoscopic myomectomy or hysterectomy (unless it is an emergency procedure). In the study that reported conversions, the proportion of these that followed from advanced contingency plans to continue to more definitive surgery in order to have a high level of certainty that symptoms would be resolved is not clear; some may have been responses to operative complications such as hemorrhage or perforations.
In summary, general details are poorly reported in these studies. Serious complications are inconsistently reported for hysteroscopic myomectomy, but they likely occur in fewer than 1 percent of procedures. However, incomplete procedures or immediate conversion to another surgery may occur at rates higher than 5 in 100 women.
Hysteroscopic Myomectomy With Other Procedures: Perioperative Outcomes. Two publications reported on hysteroscopic myomectomy with concurrent endometrial ablation (middle panel of Table 13).102, 111 They were both relatively small studies (73 and 149 participants). One reported an average operative time of 43 minutes. Fluid absorption averages were again wide, from 292 to 1,031 ml, with the same study that reported higher fluid absorption for hysteroscopic myomectomy reporting averages over a liter for the combined procedure as well. One study reported a 10 percent perforation rate.111
Endometrial Ablation: Perioperative Outcomes. A single small study that included 42 women with uterine size greater than 12 weeks compared two methods of endometrial destruction: using a roller ball versus a resection approach.153 It reported mean fluid absorption of 645 ml and an operative time of 29 minutes. Five percent of participants were hospitalized for observation but the reasons were not clearly specified. Fibroids removed, blood loss, perforation, hemorrhage, and other serious complications were not reported.
Hysteroscopic Myomectomy: Long-Term Outcomes. Seven research groups followed up participants at time periods of a year or longer;102, 111, 124, 130, 132, 140, 153 the average length of followup was around 2 years, and the longest followup included women who were tracked for 10 years after the initial procedure (with the minimum followup in that cohort being 4 years).
Women who had hysteroscopic myomectomy alone were followed up for satisfaction and symptom control at a minimum of 12 months in one study and at 36 months in another.102, 111, 124, 130, 132, 140, 153 Outcomes were poorly operationalized in these studies; the authors gave no definition of how they collected these data. One study reported that 80.8 percent of women achieved “control of bleeding”; the other reported that 81 percent reported “good control” of bleeding with 6 percent reporting return of frank menorrhagia after one or two procedures. In this cohort, 13.1 percent had a second hysteroscopic resection of fibroids.
Across the four studies of hysteroscopic myomectomy reporting such information, between 11 percent and 22 percent of women elected to have subsequent surgical intervention related to fibroids and fibroid symptoms. With the exception of the study in which repeat procedures were common (13.1 percent), myomectomy and hysterectomy were the most common procedures, with hysterectomy being selected by 2 percent to 22 percent of women as definitive management.
Results for women with both hysteroscopic myomectomy and endometrial ablation suggest potential for better control of symptoms. The smaller study group (73 women) was followed up at a minimum of 12 months after their procedure; 95.9 percent reported “control of bleeding.” This study included a comparison group of participants (n = 104) who had hysteroscopic resection only, with 81 percent achieving “control of bleeding.” This difference as well as the rates of hysterectomy by group (22 versus 18 percent) within this cohort favor performance of endometrial ablation at the time of hysteroscopic myomectomy.102 Istre and Langebrekke studied the largest group (N = 188) and reported that 5 percent of women experienced recurrent fibroids, 4 percent had recurrent bleeding, and 6 percent had recurrent pain (not mutually exclusive) within a minimum follow-up period of 4 years. Eighteen percent of their participants had repeat hysteroscopic resection of the endometrium. Of those who had repeat procedures 36 percent eventually had hysterectomies.
In the single study of endometrial ablation alone, Eskander and colleagues collected more detailed outcomes than other authors reporting on resection and ablation at the time of hysteroscopic myomectomy but had only 42 patients. They reported 67 to 77 percent of women achieved complete absence of menses, 13 percent to 15 percent had light bleeding, and 93 percent to 96 percent were “very satisfied” with their treatment outcomes during 2 years of followup.153
Across studies of hysteroscopy with ablation, the rate of eventual selection of hysterectomy for fibroid management is similar to the rate in hysteroscopy alone: 2 percent to 18 percent. None of the studies can clearly delineate whether subsequent surgeries were indicated by the appearance of new fibroids. Several of these studies used survival analysis techniques or other approaches to define the trajectory of time to subsequent procedure. The majority of women who failed treatment in these studies with an average of more than 3 years of followup, did so early, seeking subsequent surgical intervention within 1 to 2 years of the initial procedures. This may reflect the fact that treatment failure is fairly immediately apparent and women choose to act quickly. An additional consideration is that, as women age, some proportion exit the window of chaotic bleeding patterns that can occur in the perimenopause and become frankly menopausal, markedly reducing the need for further fibroid-related treatment.
This literature is limited by a general lack of direct comparisons of intervention methods and by lack of comparison of hysteroscopic approaches to other surgical and medical management methods for outcomes, costs, and risks of harms. With that caveat, the identified case series and cohorts do document that serious complications are rare in the context of hysteroscopic intervention. Expertise and the number of procedures done by a surgeon have been shown to be related to decreased complications. Physicians and clinical care settings that have sufficient participant volume to publish results of case series and cohorts are likely to be more experienced and specialized than some community care settings. They also are likely to accumulate patients, and therefore study participants, who are referred with different expectations for symptom resolution and persistence of intervention to address symptoms than may be the case in general practice. The degree and direction of bias from lack of comparability of surgical skills and patient populations cannot be quantified.
Nonetheless, although repeat procedures and subsequent surgery are not uncommon, more than 80 percent of women followed across hysteroscopy studies for an average of more than 3 years do not have subsequent surgical interventions. Because hysteroscopic interventions are generally outpatient procedures and associated with rapid return to usual activities, these data suggest that the majority of women who have fibroids amenable to hysteroscopic intervention (which is not the case for all) can achieve good outcomes without resorting to more complex and costly procedures that also have a longer recovery time.
Hysterectomy: Overview and Nomenclature
This section presents the results of our literature search and findings about outcomes of hysterectomy, which is surgical removal of the uterus. Hysterectomy does not require removal of the ovaries, which is termed oophorectomy, however both procedures are often done concurrently. Surgery that removes the entire uterus and cervix as well as the ovaries is properly called total hysterectomy with bilateral salpingo-oophorectomy. Surgery that leaves the uterine cervix is called “supracervical” or “subtotal” rather than “total” hysterectomy. Hysterectomy is not a surgical option for women who wish to have future pregnancies or who wish to retain their uterus.
The content of the literature spans the range of surgical approaches currently available in routine clinical practice. These surgical approaches are described below. We did not identify any publication that met inclusion criteria and described outcomes of robotic surgery, which is becoming available at a limited number of highly specialized sites.
For convenience and consistency we have used uniform terminology and abbreviations to describe and discuss hysterectomy. The list that follows is approximately in the order of “invasiveness” as reflected by size and location of the surgical incision to be healed and the degree of disruption of nearby tissue and, therefore, the amount of healing required after the procedure.
Abdominal Hysterectomy. Abdominal hysterectomy consists of removal of uterus (with or without the associated surgery of removing ovaries and fallopian tubes) through an incision in the skin of the abdomen; this is also called a laparotomy incision. This includes midline incisions made along the imaginary line between the umbilicus and the pubic symphysis or “pelvic bone,” as well as incisions made lower on the abdomen at a right angle to that line. The surgeon operates with his or her hands and instruments in direct contact with the abdominal and pelvic organs.
Laparoscopically Assisted Hysterectomy. Laparoscopically assisted hysterectomy is the removal of the uterus assisted by use of a laparoscope and other instruments inserted through small incisions in the abdominal wall. Generally each incision is less than 1.0 to 1.5 centimeters size. The laparoscope is attached to a video camera and the surgeons conduct a portion of the procedure while watching the surgery progress on a display screen. In the majority of the cases described in this literature as laparoscopically assisted, the laparoscope was used to complete the portion of the surgery required to identify and transect the major blood supply to the uterus (and ovaries if they are to be removed), and the procedure, including closing the vaginal incision, was completed through a vaginal approach using conventional vaginal surgical techniques.
Laparoscopic Hysterectomy. Laparoscopic hysterectomy is the removal of the uterus (with or without the ovaries and fallopian tubes) using a laparoscope and instruments inserted through “ports” in the abdominal and pelvic wall to accomplish the entire surgery. The surgeon's hands are not directly in contact with the uterus or pelvic organs during the surgery. The surgery is accomplished and the vaginal incision is closed entirely through the laparoscope. During laparoscopic hysterectomy, the uterus and fibroids may be morcellated (i.e., cut into smaller pieces), to remove them from the abdomen through small openings. This can be accomplished with laparoscopic instruments like scissors or various forms of scalpels or with a specialized device termed a “morcellator.” For the purposes of this review we have indicated when laparoscopic hysterectomy was supracervical or total.
Vaginal Hysterectomy. Vaginal hysterectomy is the removal of the uterus (with or without the ovaries and fallopian tubes) via an exclusively vaginal approach. The operative incisions are made through the upper vagina to allow access to the uterus and pelvis, and the uterus is removed by operating through the vagina.
The approach to hysterectomy is in some part determined by a match between the size of the uterus, the patient's anatomy, the plan to perform or not perform oophorectomy, concerns about potential adhesions (which is scarring) from prior surgery like cesarean, and the surgeon's skill sets via the available approach. Vaginal hysterectomy is more challenging as the size and number of fibroids increases; the very large uterus is generally not compatible with vaginal removal, even when the surgeon uses techniques to divide the uterus or morcellate the segments. Abdominal approaches have traditionally been clinically taught to be appropriate for very large fibroids, i.e., those at and above the umbilicus. However, surgeons continue to compare open and laparoscopic approaches and to examine what size of uterus and fibroids can be safely removed. The influence of pretreatments with medical (pharmaceutical) interventions such as GnRH agonists, to diminish the size of fibroids prior to surgery, was discussed earlier.
The prior review on the management of uterine fibroids found that in prospective studies, hysterectomy resulted in improvement in symptoms and quality of life up to 2 years after the procedure in most women with sufficiently severe symptoms. Type of hysterectomy or short-term outcomes such as complications did not appear to influence longer-term outcomes.30
Studies, Designs, Populations, and Outcomes Measured. Eighteen articles from 17 distinct study populations address hysterectomy (Appendix C *, Evidence Table 8).47, 75, 76, 94, 144, 155–167 Five of these studies are retrospective case series or cohorts;144, 159, 161, 166, 167 the remainder are either RCTs47, 76, 155–158, 160, 163, 165 or nonrandomized prospective cohorts.75, 94, 162, 164
Five studies were conducted in Italy,155, 158, 160, 164, 165 four in the United States,94, 161, 166, 167 and the remainder in the United Kingdom,162, 163 France,156, 157 Sweden,159 Netherlands,76 Greece,47, 144 and Canada.75 Three were multicenter trials.76, 94, 162 One study was based on an inpatient registry,159 and the others were hospital-based studies.75, 144, 156, 157, 160, 161, 163–167
With the exception of three studies,75, 94, 159 no study examined outcomes beyond the immediate perioperative window. Most studies reported on the length of the procedure, intraoperative and postoperative complications, and length of hospital stay.
Two studies reported outcomes of hysterectomy from large case-series data.159, 162 Three studies compared UAE with hysterectomy.75, 76, 94 One study compared abdominal myomectomy with abdominal hysterectomy.144 All other studies compared different types of hysterectomy or modifiers of hysterectomy outcome. Six studies compared different types of hysterectomy: three studies compared vaginal hysterectomy with abdominal hysterectomy,94, 155, 161 two studies compared laparoscopically assisted vaginal hysterectomy (LAVH) with abdominal hysterectomy,160, 165 and one study compared LAVH with vaginal hysterectomy.156, 157 Eight studies addressed modifiers of hysterectomy outcomes.47, 76, 94, 158, 160, 163, 164, 167
Our findings are presented in Appendix C *, Evidence Table 8 and summary tables below. Two studies reported outcomes of hysterectomy from large case-series data (Table 14).159, 162 Of these, one poor-quality study drew upon data from the National Health Service and private hospitals from England, Wales, and Northern Ireland to report a severe operative complication rate of 4.4 percent and a severe postoperative complication rate of 1.2 percent in the 6-week period following surgery from 1994 and 1995.162 The other study, of fair quality, reported myocardial infarction rates from a national registry of patients from Sweden over an average of 8.9 years of followup.159 The relative risk of myocardial infarction for women with only fibroids rather than other indications for hysterectomy was not statistically significant (relative risk [RR], 1.1; 95% CI, 0.7–1.7). However, the relative risk of myocardial infarction for naturally menopausal women with fibroids compared with that for all other women was statistically significant but imprecise (RR, 6.2; 95% CI, 1.9–20).
Comparative Studies. Uterine Artery Embolization (UAE) Versus Hysterectomy. Three fair-quality studies compared the outcomes of UAE and hysterectomy (Table 15).75, 76, 94 Two were multicenter studies of UAE versus a mixed group of hysterectomies (abdominal, laparoscopic vaginal, LAVH) and focused on symptoms and clinical outcomes.76, 94
These studies consistently demonstrated shorter procedure and hospital times for the UAE group than for the hysterectomy group, but they were not consistent in the rate or direction of complications.76, 94 Hehenkamp and colleagues reported a significantly higher rate of minor complications at 6 weeks postprocedure in the UAE group than in the hysterectomy group;76 the Spies et al., study reported a significantly lower rate of minor complications and overall morbidity in the UAE group than in the hysterectomy group.94 Hehenkamp documented a higher rate of readmissions among UAE patients76 whereas Spies et al., did not find any significant differences in rates of readmission.94 Spies et al., reported significant differences in days before return to work favoring UAE (UAE: 10.7 days, hysterectomy: 32.5 days; P < 0.001), and significant differences in the proportion reporting improved pelvic pain at 12 months (UAE: 84 percent, hysterectomy: 98 percent; P = 0.021), favoring hysterectomy. They found no differences in other symptoms, quality of life, satisfaction, or overall health assessment.94
The third, hospital-based study compared UAE with laparoscopic hysterectomy to assess the risk of damage to ovarian function. The authors reported no differences between the groups in ovarian function between baseline and 6 months following the procedure.75
Abdominal Myomectomy Versus Abdominal Hysterectomy. One poor-quality study compared abdominal myomectomy with abdominal hysterectomy, seeking to provide evidence on whether abdominal myomectomy was associated with febrile morbidity.144 The study's retrospective analysis of 204 patients suggested no difference in the incidence or length of febrile morbidity. The study presented no other outcomes.
Vaginal Versus Abdominal Hysterectomy. Three studies compared vaginal hysterectomy with abdominal hysterectomy (Table 16); two were of fair quality155, 161 and one of poor quality.166 Two of the three were retrospective cohorts161, 166 and one was an RCT.155 All three studies focused on perioperative outcomes.
All three studies reported higher operative times for abdominal hysterectomy, although the difference was statistically significant in only one study.155 They reported no difference in blood transfusion or intraoperative complications. With regard to postoperative decrease in hemoglobin, the studies yielded inconsistent effects: one study reported no differences in hemoglobin,155 another reported a higher but nonsignificant decrease in postoperative hemoglobin,166 and a third reported a significantly lower postoperative decrease in hemoglobin with abdominal hysterectomy than with vaginal hysterectomy.161 All three studies reported higher but nonsignificant rates of blood transfusion among abdominal hysterectomy patients. All three studies reported either no differences in postoperative complication rates or higher rates of postoperative complications among the abdominal hysterectomy patients. These differences are significant for the risk of ileus in one study161 and for postoperative complications in another study.166 All three studies are consistent in reporting significantly longer hospital stays among abdominal hysterectomy patients.
Laparoscopically Assisted Vaginal Hysterectomy (LAVH) Versus Abdominal Hysterectomy. Two RCTS conducted in Italy reported on comparisons of LAVH and abdominal hysterectomy, one of fair quality165 and one of poor quality160 (Table 17). Both trials demonstrated significantly longer hospital stays for the abdominal route. Additionally, one study reported significantly shorter convalescence for the LAVH group (LAVH, 22.0 ± 11.3 days; abdominal hysterectomy, 36.0 ± 12.1 days; P < 0.001),165 and the other reported significantly reduced use of analgesia for the LAVH group (LAVH, 3 percent of 7 patients; abdominal hysterectomy, 77 percent of 24 patients; P < 0.001).160 Neither reported significant differences in the rates of blood transfusion or postoperative decrease in hemoglobin.
LAVH Versus Vaginal Hysterectomy. A single fair-quality study (two publications) compared outcomes following LAVH or vaginal hysterectomy (Table 18).156, 157 This RCT reported significantly longer operating times, higher rates of total perioperative complications, and longer hospital stays in the LAVH group. The study did not find significant differences in the rates for individual complications, use of paracetamol, use of nonsteroidal anti-inflammatory drugs, use of opioid drugs during hospitalization, or time of passing gas and stool.
Modifiers of Hysterectomy Outcomes. Eight studies reported on a variety of modifiers of outcomes of hysterectomy (Table 19): five of fair quality47, 76, 94, 164, 167 and three of poor quality.158, 160, 163
Few studies examined the variety of modifiers identified for KQ 5, such as age, race, or ethnicity, parity, breastfeeding, contraceptive choices, body habitus, insulin resistance, concurrent medical conditions such diabetes, or hormone replacement status. Two studies that compared UAE with hysterectomy found that factors such as uterine volume, previous therapies, age, and race94 or radiologists' experience, hospital experience, and type of hysterectomy76 did not predict perioperative complication rates.
Another study based on a prospective case series of vaginal hysterectomy found that generally considered contraindications to vaginal hysterectomy, such as large uterus, adnexal pathology, nulliparity, previous pelvic surgery, or more than one contraindication, were not significant predictors of complications.164
Two studies examined uterine weight as a modifier of outcomes of an RCT of LAVH and total abdominal hysterectomy160 or retrospective study of abdominal hysterectomy.167 One study found that uterine weight was a significant predictor of at least one complication (estimated blood loss > 500 mL, perioperative blood transfusion, major organ injury, postoperative antibiotic therapy, readmission);167 the other study reported that uterine weight was a significant predictor of conversion to laparotomy among LAVH patients.160
Three RCTs addressed clinical modifiers designed to reduce blood loss; these included use of bipolar electrocautery scissors vs. conventional scissors,158 vasopressin vs. placebo,163 and recombinant human erythropoietin (rHuEPO) plus iron supplementation vs. iron supplementation alone.47 Dessole et al., demonstrated lower operating time and number of ligations for the electrocautery group than for the conventional scissors group; they did not find differences in hemoglobin or hematocrit until day 5 following the procedure, when the electrocautery group did better than the conventional scissors group.158 Okin et al., reported lower estimated blood losses for the vasopressin group than for the placebo group, but they did not demonstrate significant differences in postoperative hemoglobin, change in hemoglobin, intraoperative transfusion, total operating room time, hysterectomy time, or hospital stays of 4 or more days.163 Doussias et al., reported improved hemoglobin levels at days 3, 7, and 14 postoperatively in the rHuEPO plus iron group than in the iron-only group. The study also found significantly higher rates of blood transfusion in the iron-only group but not differences in blood loss or length of hospital stay.47
Complementary and Alternative Medicine
The prior review on the management of uterine fibroids found a single study on Chinese herbal medicine.30 Similarly, we found a single poor-quality study that met our inclusion criteria for complementary and alternative medicine involving traditional Chinese medicine (Appendix C *, Evidence Table 9);48 it is also discussed in the section on expectant management. This nonrandomized cohort study compared a group of women who received weekly acupuncture, Chinese herbs, and nutritional therapy (N = 37) to a comparison group (N = 37); patients in the traditional Chinese medicine group also received pelvic bodywork, guided imagery, and meditation. Study investigators selected herbs and nutritional therapies for each patient but standardized them in accordance with traditional Chinese medicine tenets. Patients in the comparison group received progestational agents to stop excessive uterine bleeding, oral contraceptives to control menstrual bleeding, and NSAIDs for pain. Patients in the treatment group had significantly smaller fibroids after 6 month of treatment than in the comparison group (-0.8 cm vs. +1.9 cm; P < 0.01). A greater proportion were improved (that is, cured, reduced in size, stopped growth, or reduced rate of growth) than in the comparison group (22 [60 percent] versus 3 [8 percent]; P < 0.001). The traditional Chinese medicine treatment group was also more likely to be very satisfied with their treatment than the comparison group (14 [38 percent] versus 8 [22 percent]; P < 0.05). The author noted potential biases from the differences in degree of motivation between the two groups: the treatment group was selected from the author's practice or by word-of-mouth referral from current patients; the comparison group, although selected to match the treatment group in age, fibroid size, presenting symptoms, and health insurance status, was entirely drawn from a sample of women who used the emergency room.
KQ 3: Treatment for Goals Other than Symptom Relief
KQ 3 asks about treatment for goals other than symptom relief; specifically, the focus is on enhancing fertility, reducing adverse pregnancy outcomes, preventing further growth, or ruling out uterine malignancy. We found 10 studies relating to reproductive outcomes. We did not find publications about preventing growth of existing fibroids that compared treatment with either no treatment or alternative treatments. Information about fibroid recurrence after myomectomy was presented above for KQ 2. We did, however, find five studies examining uterine fibroid outcomes in postmenopausal women undergoing hormone replacement therapy for postmenopausal symptoms. No studies, in the time frame reviewed, reported on the probability of identifying a uterine malignancy when surgery or biopsy was done for treating or evaluating presumed uterine fibroids. These questions are new to our review, and had not been directly addressed by the prior review on the management of uterine fibroids.30
Pregnancy Outcomes: Overview
We identified 10 studies (10 articles) providing information about objectives for fibroid management other than symptom relief or treating the health consequences of the fibroids (e.g., anemia) (Appendix C *, Evidence Table 10). All concerned reproductive outcomes among women after treatment for their fibroids. All studies that contained fertility and pregnancy data and that met the review inclusion criteria of more than 40 women in a trial or cohort or more than 100 women in a case series who desired or achieved a pregnancy were related to outcomes after myomectomy.
Studies, Designs, and Populations. Table 20 provides information on these 10 studies in four blocks with information about specific subgroups; for that reason some studies appear more than once. For example, when authors reported data for both laparoscopic and hysteroscopic myomectomy, these cohort data will be recorded for each intervention group separately; we did not duplicate any study data in this table.
Seven publications focused on laparoscopic myomectomy; two were of poor quality112, 151 and five were of fair quality.116, 117, 127, 139, 147 Three publications of fair quality included more than one type of myomectomy;137, 141, 152 of these, one included a cohort of women who had either laparoscopic or hysteroscopic myomectomy and combined the outcome data;152 one assessed use of hysteroscopic or abdominal myomectomy as indicated by fibroid type before ART and presented outcomes separately;141 and the third was a randomized trial that examined conception and pregnancy outcomes after laparoscopic or abdominal myomectomy.137
This literature is exclusively from large academic, tertiary care centers and internationally recognized fibroid surgery centers. Except for one study conducted in the United States141 and one in Japan,147 the remainder were performed in Europe, mostly in Italy or France.
Fertility Status. The fertility status of the populations varied widely. One prospective cohort compared women with existing fibroids with those who had had myomectomy before in vitro fertilization and embryo transfer.112 One retrospective cohort made similar comparisons among six groups of women: those who had hysteroscopic myomectomy with and without donor oocytes, those who had laparoscopic myomectomy with or without donor oocytes, and a comparison group of women without fibroids with and without donor oocytes.141 The other retrospective cohort included only women with infertility as an indication for surgery; it compared laparoscopic myomectomy outcomes to those of the small group of women whose procedure was converted to an abdominal myomectomy. This study reported modest subsequent use of ART, which indicated potentially less severe fertility impairment.139 One trial compared myomectomy for unexplained infertility with expectant management among women who did not have ART;152 another randomized trial investigated different myomectomy methods among women with infertility and did not report use of ART;137 the remaining four are case series with varied rates of use of ART in their study populations.
Outcomes Measured. The majority of this literature relies on clinical followup, at times with individual contact when records were insufficient. One group conducted annual questionnaires,151 and several specified prospective followup but did not report how this was accomplished. Overall, loss to followup is minimal (< 5 percent) to modest (5 percent to 10 percent), although completeness of data and details about timing of attempted conception is limited by the nature of clinical records.
Ideally, data about ability to conceive would be reported as cycle- or even day-specific probability of conception, or fecundability, and the investigators would do analyses based on comparison of time-to-event across groups or by characteristics. By definition, rates, such as pregnancy rates, require documentation of a time period in which the event occurred among a known population. Overall, the poor quality of outcome assessment is a central challenge of interpreting this literature. Other than the ART studies, which reported outcomes for a group average of embryo transfer cycles, no authors reported per-cycle fecundability. Only one conducted a time-to-event analysis, estimating that 60 percent of women would conceive within 2 years.147
Several publications reported average time-to-pregnancy as a descriptor and not a focus of the data analysis. These data, however, cannot be equated to fecundability data because we cannot know whether all the elapsed time between the surgical intervention and the pregnancy was associated with cycles in which the women could have conceived or attempted to conceive. Two studies that calculated proportions of women who achieved a pregnancy did not note duration of followup; several reported broad ranges, such as from 12 months to 60 months; and yet others indicated that all participants had a minimum of some fixed time of followup such as 1 year. None adjusted for time attempting conception during or before followup.
Pregnancy Outcomes: Results
Among the three studies, all of fair quality, that included participants or identified a subgroup within the study with predominantly (> 93 percent) spontaneous conceptions (Table 20, first panel), two reported outcomes of laparoscopic myomectomy and one included some proportion of hysteroscopic procedures. In those three studies, the proportions of women attempting to conceive who had a subsequent pregnancy averaged 37 percent (range, 25 percent to 43 percent).117, 127, 152 Among spontaneous conceptions, the risk of spontaneous abortion (i.e., miscarriage) was 11 percent, 15 percent, and 37 percent of recognized pregnancies. The proportions of women who achieved a pregnancy and had a live birth in this group of predominantly spontaneous conceptions were 80 percent and 88 percent (not reported in one study). Overall, in this group of studies predominantly reflecting spontaneous conceptions, 20 percent to 36 percent of all women who desired a pregnancy had a live birth.
One study in this group had randomized women with intramural and/or submucous fibroids to receive myomectomy or forego surgery.152 The investigators reported an increase of more than 15 percent in the proportion of women who achieved a pregnancy among those who had surgery for fibroids with any submucosal component, which is a meaningful, statistically significant improvement. The trend also favored higher numbers of women achieving a pregnancy for intramural fibroids; however, the number of participants was small and the comparison across groups was not significant. This was also the case for comparing miscarriage rates; in each case the miscarriage risk was higher among women without surgery, but the authors did not comment on statistical significance, which was likely not reached given the limited power of this trial.
Among the three studies of fair quality of women who had and who had not had infertility treatment (Table 20, second panel), the proportion of women who achieved pregnancies was 37 percent to 56 percent.116, 139, 147 In this subset of studies with a small proportion of women receiving infertility treatment, 40 percent to 82 percent of women who achieved a pregnancy gave birth. Overall, 22 percent to 39 percent of women who desired to conceive after myomectomy were able to conceive and have a live birth. Outcomes were similar to these in the studies that did not specify the proportion of participants who had infertility treatment (Table 20, second and third panels).
Among these studies of fair quality, two compared outcomes by type of myomectomy. Soriano and colleagues compared women who had laparoscopic myomectomy (n = 88) with those who had complications at the time of laparoscopic myomectomy and whose procedure was converted to abdominal myomectomy (n = 18). Noting the small number of conversions, they did not find a statistically significant difference in the proportion of women who became pregnant (50 percent and 56 percent), although time to becoming pregnant was longer by approximately 7 months among those who had a conversion to open procedure (P < 0.001).139
Seracchiolli and colleagues randomly assigned participants to either abdominal (n = 65) or laparoscopic (n = 66) myomectomy. They reported similar numbers of pregnancy, miscarriage, preterm births, and cesarean birth across study arms; this finding suggests that the choice of method of myomectomy may exert little influence on outcomes.137
In the two studies that included exclusively ART patients, one of poor quality112 and one of fair quality141 (Table 20, bottom panel), the proportions who achieved a clinical pregnancy were 24 percent to 33 percent, with an overall higher miscarriage risk (29 percent, 39 percent, and 48 percent) than other studies had reported. This finding may relate to the very close surveillance of these embryo transfer pregnancies.
Surrey and colleagues retrospectively compared hysteroscopic and laparoscopic myomectomy in a population of women receiving ART.141 The method of myomectomy did not have a statistically meaningful influence on outcomes. Moreover, women who had myomectomy had neither better nor worse outcomes than a comparison group of women with no history of fibroids undergoing similar ART procedures.
Maternal age is a strong predictor of reproductive performance, especially in ART research. In this case, the authors did in effect adjust for some components of maternal age and oocyte quality by comparing groups with similar treatments who did and did not have oocyte donation, which would be from young, healthy donors. The findings were comparable for both those using donor eggs and their own.141 Births are not well reported in these studies, which are oriented toward immediate infertility care outcomes.
Across the other studies with data about route of birth, seven reported on cesarean births; among the women who had had myomectomy, 24 percent of births to 83 percent of births were accomplished by cesarean delivery.116, 117, 127, 137, 139, 147, 151The data are insufficient to understand what proportion of these births were planned as cesarean deliveries or resulted from difficulties during labor. Among the 314 births were three documented cases of uterine rupture; two were at the site of a prior cesarean scar and not in the location of the myomectomy scar. Thus, one rupture is properly attributed as related to the myomectomy.
In summary, the literature about pregnancy outcomes after care for fibroids is quite restricted in scope and of overall fair quality; we did not identify any good-quality studies. The majority of research is descriptive, conducted in clinical settings outside the United States, and is especially limited with respect to representativeness of the population, study size, and statistical analysis. The sole clinical trial with evidence comparing surgical intervention to none, without additional ART care, supports a benefit from removing fibroids that have a submucosal component (i.e., those in which the fibroid is immediately adjacent to or distorts the uterine cavity). The benefit reported in that study is substantial (> 15 percent absolute increase in proportion of women becoming pregnant) but limited, by small study size, to reflecting on only ability to conceive.152 Other outcomes were promising but not significant. Given how common and concerning fibroids are to women and their care providers, this literature will require expansion beyond infertility care with careful attention to design of large-scale prospective cohorts and intervention trials that shed light on the risks and benefits of intervening with fibroids only for the sake of modifying reproductive outcomes.
Preventing Further Growth: Overview
Despite the widespread use and effectiveness of hormone replacement therapy to reduce symptoms of menopause, clinicians are often hesitant to prescribe hormone therapy to postmenopausal women with fibroids because of the risk of fibroid growth.58
Studies, Designs, and Populations. We found five studies (one of good quality,168 two of fair quality58, 169 and two of poor quality170, 171) that evaluated the outcomes associated with menopausal hormone therapy (Table 21).58, 169–171
Preventing Further Growth: Results
Our findings are reported in Appendix C *, Evidence Table 11. Four of five studies included only postmenopausal women.58, 168–170 One study evaluated the risk of a first diagnosis of fibroids in peri- and postmenopausal women associated with prior use of estrogen and progestogen therapy.171 This study reported no statistically significant effects for all women; a subanalysis of women stratified by BMI status, however, demonstrated an increased risk of development of fibroids with prior combined estrogen-progestin therapy among women with a BMI less than 24 (ever-use: OR, 2.3; 95% CI, 1.2–4.3) and hormone therapy use for 5 or more years (OR, 4.0; 95% CI, 1.6–10.3). The remaining four studies reported on size changes.58, 168–170 One study compared an oral cyclic association of oestradiol valerate and cyproterone acetate with a sequential combination of transdermal E2 and orally administered medroxyprogesterone acetate on 240 postmenopausal women with and without uterine myomas. The study demonstrated a higher risk of uterine growth with the percutaneous-oral schedule of hormone replacement therapy than a single oral combination of oestradiol valerate and cyproterone acetate.170 The three remaining studies did not report significant increases in uterine volume with hormone therapy.
KQ 4: Costs of Fibroid Treatment
The prior review used multiple sources (2000 “Red Book” of wholesale drug prices, published literature on hospital costs for surgical management of uterine fibroids, primary data from the Nationwide Inpatient Sample, and primary data from Duke University Medical Center) but nevertheless concluded that “most administrative data sources do not provide sufficient clinical detail to allow comparison between procedures.”30 (p98) We, too, found only very limited evidence on the cost of treating uterine fibroids. We identified three studies on this topic (Table 22), all of poor quality; two examined costs from a hospital perspective172, 173 and one used an insurance claims database evaluation.140 Detailed information for these studies appears in Evidence Table 12 in Appendix C *.
One study compared 23 UAE patients with 17 myomectomy patients from Georgetown University Hospital.172 The UAE sample was significantly older on average (42.65 years) than the myomectomy sample (35.5 years) (P < 0.001). On average, the hospital, professional, and imaging costs were $6,708 for UAE and $7,630 for myomectomy. The authors attributed differences in costs to higher hospital care and operating room costs for myomectomy even though UAE had much higher professional costs, $2,220 for UAE and $1,611 for myomectomy (P = 0.002). Overall, the authors found a trend for UAE to be the least expensive option, but the difference was not significant (P = 0.086).
The second hospital-based study was a retrospective comparison of UAE with hysterectomy.173 Women who were treated with UAE were significantly younger (43.1 vs. 47.0 years; P < 0.001) and less likely to be white (69.6 percent vs. 77.0 percent; P = 0.01), had bigger fibroids (8.0 cm vs. 6.3 cm in diameter; P = 0.001), and had more fibroids (2.8 vs. 2.0; P < 0.001). The mean total hospital costs were significantly different for the two modalities—$8,223 for UAE and $6,406 for hysterectomy (P < 0.0001)—even though UAE had a significantly shorter length of stay than hysterectomy (0.95 days vs. 2.6 days; P < 0.0001).
The third study performed a retrospective database analysis of the costs involved in different types of myomectomies.140 The study measured facility and professional costs of inpatient and outpatient procedures. The authors found that outpatient hysteroscopic myomectomy ($4,291) was less than half the cost of inpatient abdominal myomectomy ($8,860). They also found that, because of repeated procedures (at the rate of about 16.5 percent over 2 years), the mean overall cost rose from $6,737 for the initial procedure to a mean of $8,001 at 2 years for the repeat procedure.
Chapter 4 discusses the findings for each of the KQs presented in Chapter 3. We also provide a further analysis of these findings responding to KQ 5 on modifiers, KQ 6 on comparisons, and KQ 7 on variations in treatment.
Appendixes cited in this report are provided electronically at http://ahrq
Agency for Healthcare Research and Quality (US), Rockville (MD)
Viswanathan M, Hartmann K, McKoy N, et al. Management of Uterine Fibroids: An Update of the Evidence. Rockville (MD): Agency for Healthcare Research and Quality (US); 2007 Jul. (Evidence Reports/Technology Assessments, No. 154.) 3, Results.