Results

Katherine E. Hartmann; Christopher Fonnesbeck; Tanya Surawicz; Shanthi Krishnaswami; Jeffrey C. Andrews; Jo Ellen Wilson; Digna Velez-Edwards; Shannon Kugley; Nila A. Sathe

NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.

Hartmann KE, Fonnesbeck C, Surawicz T, et al. Management of Uterine Fibroids [Internet]. Rockville (MD): Agency for Healthcare Research and Quality (US); 2017 Dec. (Comparative Effectiveness Review, No. 195.)

Management of Uterine Fibroids [Internet].

Show details

Contents

< Prev Next >

Results

This chapter presents the evidence to address our four Key Questions (KQs): KQ1, effectiveness of interventions; KQ2, factors that modify effectiveness; KQ3, risk of leiomyosarcoma at the time of fibroid surgery; and KQ4, influence of patient or fibroid characteristics or morcellation approach on survival after leiomyosarcoma. When a final outcome is not discussed it means that the literature did not provide evidence to help understand whether the intervention had any effect on that outcome.

For KQ1 and KQ2, we screened 11,169 records and excluded 7,945 at the time of abstract review. We retrieved the full text of 1,313 publications; 1,192 were excluded for one or more reasons. We identified 121 publications representing 97 unique studies (Figure 2). We included 160 studies for KQ3 (Figure 3) and 28 unique studies (24 of which contributed data to the survival analysis) for KQ4 (Figure 4). In all, we retained 311 publications, representing 285 unique studies, to address one or more KQs in this review. We did not receive any scientific information submissions through the portal on the Effective Health Care Web site.

Appendix D includes a list of excluded publications and reason for exclusion.

Figure 2

Literature flow diagram for Key Question 1 and Key Question 2.

Figure 3

Literature flow diagram for Key Question 3.

Figure 4

Literature flow diagram for Key Question 4.

Content of the Literature About Effectiveness

We included 97 unique randomized controlled trials (reported in 121 publications).¹³^-²³^,³⁶^,³⁸^,⁵³^-¹⁶⁰ These studies included 9,179 women with the majority of studies conducted in Europe (Table 5). Forty three studies included a medical intervention (including intrauterine devices [IUDs]);⁵⁴^,⁵⁹^,⁶⁴^-⁶⁷^,⁷⁵^,⁷⁹^,⁸⁴^,⁸⁷^,⁸⁸^,⁹⁴^,⁹⁸^,¹⁰²^,¹⁰³^,¹⁰⁶^,¹⁰⁸^,¹¹⁵^,¹¹⁶^,¹¹⁸^,¹¹⁹^,¹²²^-¹²⁴^,¹²⁸^,¹²⁹^,¹³⁶^-¹⁴⁰^,¹⁴²^-¹⁵³ 28 assessed a procedural intervention;¹³^-¹⁶^,¹⁸^,²⁰^,²²^,²³^,³⁸^,⁵³^,⁵⁷^,⁵⁸^,⁶⁰^,⁶²^,⁶⁸^,⁷⁸^,⁸¹^,⁹²^,¹⁰⁴^,¹⁰⁵^,¹⁰⁸^,¹¹³^,¹¹⁴^,¹¹⁷^,¹²⁷^,¹⁵⁵^,¹⁵⁶^,¹⁵⁸ and 37 assessed surgical treatments.¹⁴^,¹⁶^,²⁰^,²²^,⁵⁶^,⁵⁷^,⁶³^,⁶⁸^,⁷²^,⁷⁶^,⁷⁷^,⁸⁰^,⁸⁵^,⁸⁶^,⁸⁹^,⁹⁰^,⁹⁵^,¹⁰⁰^,¹⁰¹^,¹⁰⁷^,¹¹⁰^,¹¹²^-¹¹⁴^,¹²⁰^,¹²¹^,¹²⁵^,¹²⁶^,¹³⁰^-¹³²^,¹³⁴^,¹³⁵^,¹³⁷^,¹⁴¹^,¹⁴⁸^,¹⁶⁰ We included two studies for their expectant management arms only, because the intervention arms used tibolone ¹³⁹ or asoprisnil ¹⁰³, neither of which are approved for use in the United States). Eleven studies compared interventions from more than one category (e.g., procedure vs. surgery).¹⁴^,¹⁶^,²⁰^,²²^,⁵⁷^,⁶⁸^,¹⁰⁴^,¹¹³^,¹¹⁴^,¹³⁷^,¹⁴⁸

Table 5

Characteristics of studies included for Key Question 1.

We assessed risk of bias for all studies included for KQ1. We considered 18 (18.6%) of these studies to have low risk of bias (good quality), 27 (27.8%) to have moderate risk of bias (fair quality), and 52 (53.6%) to have high risk of bias (poor quality). The most common shortcoming was failure to blind assessors or participants to treatment status. For most studies we assessed risk of bias from the published report only; we also identified the study protocol for 19 of the included studies (Appendix F). We enumerate the risk of bias assessments and source of bias for all studies in Appendix E.

Key Question 1. Effectiveness of Treatment for Uterine Fibroids

Key Points

This summary reflects synthesis of outcomes across arms of studies that used the intervention. If a study included different types of interventions, each arm is included in the related synthesis and discussion. Assessment of harms was limited to those reported by randomized trials and these are reported in the text.

Expectant Management

The number of women followed without intervention is small (n=514) and data is insufficient to project the outcomes of expectant management (insufficient strength of evidence).

Medications

Gonadotropin-releasing hormone (GnRH) agonists reduced the size of fibroids and the overall size of the uterus (moderate strength of evidence).
GnRH agonists, with or without add-back therapy, improved bleeding symptoms, (moderate strength of evidence) and improved quality of life (low strength of evidence)
Mifepristone reduced fibroid size and improved bleeding outcomes and improved quality of life (moderate strength of evidence).
Ulipristal reduced fibroid size, improved bleeding outcomes and improved quality of life measures (moderate strength of evidence).
Raloxifene did not change fibroid size and did not cause changes in hemoglobin in postmenopausal women (low strength of evidence).

Uterine Artery Embolization (UAE) or Occlusion

UAE reduces the size of fibroids (high strength of evidence) with data from two long term studies demonstrating effectiveness up to 5 years.
Bleeding outcomes and fibroid-related quality of life measures also improved following UAE (moderate and moderate strength of evidence, respectively).
Data are inadequate to comment on other methods of occlusion (insufficient strength of evidence).

High Intensity Focused Ultrasound (HIFU) for Fibroid Ablation

HIFU reduces fibroid and uterine size (low strength of evidence).
The effects on bleeding, quality of life outcomes, or other patient reported outcomes are unknown (insufficient strength of evidence).

Myomectomy

Fibroid-related quality of life improved following myomectomy (low strength of evidence).

Hysterectomy

Fibroid-related quality of life improved following hysterectomy (low strength of evidence).

Direct comparative effectiveness

Because of the small number of women studied there was insufficient strength of evidence for outcomes of ulipristal versus GnRH, and UAE versus Myomectomy

Estimation of Subsequent Treatment for Uterine Fibroids

Modelling estimates that for women in their 30s, the probability of subsequent intervention for fibroids over 2 years varied from 6-7 percent after medical treatment or myomectomy to 44 percent after UAE. For women in their 40s and 50s, modelled 2-year reintervention rates were 9-12 percent following medical treatment or UAE, and 0 percent after myomectomy.

Expectant Management: Overview

We did not identify any studies intentionally designed to determine outcomes of no intervention, also called expectant management or watchful waiting. However, 16 randomized controlled trials (RCTs)(reported in 17 publications) included 514 women in expectant management arms (defined as no treatment, placebo or sham treatment, or minimal intervention such as multivitamin use) compared to active treatment.³⁶^,⁵⁹^,⁶¹^,⁶⁵^,⁶⁷^,⁷⁴^,⁷⁹^,¹⁰³^,¹⁰⁶^,¹⁰⁷^,¹¹⁸^,¹²³^,¹²⁸^,¹²⁹^,¹³⁹^,¹⁴⁷^,¹⁵⁰^,¹⁵⁶ Two of these trials were of good quality, six fair, and eight poor. Expectant management arms assessed changes in fibroid or uterine size (13 studies),³⁶^,⁶⁵^,⁷⁴^,⁷⁹^,¹⁰³^,¹⁰⁶^,¹¹⁸^,¹²³^,¹²⁸^,¹³⁹^,¹⁴⁷^,¹⁵⁰^,¹⁵⁶ bleeding patterns (4 studies),⁶⁵^,⁷⁴^,⁷⁹^,¹⁴⁷ pain, pressure, or symptom severity (8 studies),³⁶^,⁶⁵^,⁷⁴^,¹⁰³^,¹⁰⁶^,¹²⁹^,¹⁴⁷^,¹⁵⁶ sexual function (3 studies),⁶⁵^,¹⁰³^,¹⁴⁷ and pregnancy outcome (1 study).¹⁰⁷

Expectant Management: Results

The majority of women evaluated for expectant management came from 14 drug trials.³⁶^,⁶⁵^,⁶⁷^,⁷⁴^,⁷⁹^,¹⁰³^,¹⁰⁶^,¹¹⁸^,¹²³^,¹²⁸^,¹²⁹^,¹³⁹^,¹⁴⁷^,¹⁵⁰ Other no-intervention groups include one study arm from a myomectomy trial to evaluate pregnancy outcomes¹⁰⁷ and one from a placebo-controlled pilot study of MRI-guided focused ultrasound (MRgFUS).¹⁵⁶ The studies were small and almost half (7 of 16) did not report masking those conducting or interpreting the imaging measurements to group status. The placebo-controlled trials did describe credible placebos, which diminishes concern that imaging measures would be modified by participant report of their intervention status. Unless knowledge of study arm was directly available to those interpreting measures from imaging, the effect of bias may not be substantial.

Expectant Management and Fibroid Characteristics

The overall evidence suggests the size of fibroids does not meaningfully change over short timespans, based on an average followup time of 5 months (range, 3 to 12 months) (Table 6). Neither of the two studies¹²⁸^,¹³⁹ with women who were postmenopausal and followed for a full year detected an increase in total volume of fibroids.

Table 6

Change in fibroid and uterine size with expectant management by study arm.

Expectant Management and Bleeding

Bleeding characteristics, measured by bleeding (days and severity) or hemoglobin, did not change meaningfully during followup for those with expectant management (Table 7). Some studies reported no statistically significant change in other bleeding outcomes such as heaviness of periods,¹³⁹ monthly hemoglobin measures within normal range,¹⁶² and number and severity of heavy bleeding episodes over 12 months.¹²⁸

Some groups with expectant management experienced modest improvements; 6.3 percent of one placebo group had resolution of intermenstrual bleeding over three months.⁶⁵ Symptom severity score improved slightly in the placebo group after 12 weeks (4.2 ± 6.5).⁷⁹ In a double-blind study of women who presented with heavy menstrual bleeding, 26 of 37 (70%) in the expectant management group reported resolution or improvement at final visit (24 weeks after enrollment).¹⁴⁷

The proportion of the 514 women enrolled in these trials who presented specifically with problem bleeding, as opposed to other fibroid-related symptoms, is not known. However, the data suggest that women with fibroids should not expect that bleeding patterns will worsen over the near term.

Table 7

Change in bleeding characteristics with expectant management by study arm.

These findings of minimal change over followup periods of a year or less are compatible with a prior review that included observational cohorts.¹² The number of women in the literature followed without intervention is small (n=514) and these participants may be fundamentally different from other women with fibroids since they were willing to risk randomization to no intervention. Because none of these studies were designed to evaluate expectant management, the overall quality of the research is poor to inform choice of expectant management over other options and strength of the evidence is insufficient.

Medical Management: Overview

We sought studies that addressed whether medications can reduce symptoms or delay the need for other management. Our intended scope was wide, including common clinical interventions such as continuous oral contraceptives to avoid menstrual periods, nonsteroidal anti-inflammatory agents to improve bleeding or dysmenorrhea, and agents such as stool softeners to prevent constipation from bulky fibroids. However, we identified RCTs for five types of clinically less common medications: GnRH agonists, progesterone receptor agents, estrogen receptor agents, hormone replacement therapy, and antifibrinolytic treatment. We did not review trials in which medications were used as adjuncts and in which all participants were scheduled for surgery.¹⁶³ We excluded medications that cannot be prescribed in the United States.

Medical Management: Results

We identified 43 studies (48 publications) assessing effectiveness of medical treatment for uterine fibroids⁵⁴^,⁵⁵^,⁵⁹^,⁶¹^,⁶⁴^-⁶⁷^,⁷⁵^,⁷⁹^,⁸⁴^,⁸⁷^,⁸⁸^,⁹⁴^,⁹⁸^,¹⁰²^,¹⁰³^,¹⁰⁶^,¹⁰⁸^,¹¹⁵^,¹¹⁶^,¹¹⁸^,¹¹⁹^,¹²²^-¹²⁴^,¹²⁸^,¹²⁹^,¹³⁶^-¹⁴⁰^,¹⁴²^-¹⁵³^,¹⁵⁷^,¹⁶⁴ We rated four studies as good quality (low risk of bias), 12 as fair (moderate risk of bias), and 27 as poor quality (high risk of bias). Common reasons for classification as poor quality included: no description or unclear description of randomization method (4 studies),⁹⁸^,¹¹⁸^,¹³⁸^,¹⁴⁵ no report of assessment of medication adherence,¹²⁸ and failure to blind outcome assessors.⁹⁴^,⁹⁸

Eleven studies included a placebo or no treatment comparison group³⁶^,⁶⁵^,⁶⁷^,⁷⁴^,¹⁰⁶^,¹¹⁸^,¹²³^,¹²⁸^,¹²⁹^,¹⁴⁷^,¹⁵⁰. Eight studies compared two or more medications⁷³^,⁸⁴^,⁸⁷^,¹⁰²^,¹²⁴^,¹³³^,¹³⁶^,¹⁴⁹^,¹⁵³ and 10 compared doses of the same drug.⁵⁴^,⁶⁴^,⁶⁶^,⁷⁵^,⁷⁹^,⁸⁸^,¹¹⁹^,¹²³^,¹⁴⁰^,¹⁴⁶ Several studies evaluated dose schedules or regimens that change over time. Another eight studies⁹⁴^,⁹⁸^,¹²²^,¹³⁸^,¹⁴²^,¹⁴⁴^,¹⁴⁵^,¹⁵¹ examined the role of an additional drug (i.e., add-back) given to decrease the side effects of the primary GnRH treatment. Six studies of medical interventions are discussed in other sections of this report: the placebo arm of two studies are discussed in the section on expectant management, because the active agents (tibolone, asoprisnil) are not approved for use in the United States; ¹³⁹ ¹⁰³; two studies are discussed in the section on comparison of effectiveness across intervention categories⁸⁴^,¹⁰²; and two studies that compared a GnRH agonist to surgery¹³⁷^,¹⁴⁸ are summarized in the surgical intervention section.

Approximately 35 percent of the medication studies (15/43) were industry sponsored.⁵⁴^,⁵⁹^,⁶⁴^-⁶⁷^,⁷³^-⁷⁵^,⁹⁸^,¹⁰³^,¹³⁷^,¹⁴⁰^,¹⁴²^,¹⁴⁷ ¹³⁷ The duration of followup ranged from no additional followup after the end of treatment with the medication to 5 years after conclusion of the medication. Women included in the studies were predominantly premenopausal (39 studies). Four studies¹²⁴^,¹²⁸^,¹³⁶^,¹³⁹ enrolled postmenopausal women.

We have organized this section to first present the evidence about effectiveness for each category of drug when an important outcome has been measured by multiple studies. We also note if several specific doses or routes of delivery of the drug (e.g., injection vs. oral) have been investigated. We reserve discussion of direct comparisons between categories of medications to the end of the section.

To summarize outcomes we move from changes in the fibroids, to changes in symptoms, including bleeding characteristics, pain, and sexual function. When reported we also summarize fertility status and pregnancy outcomes as well as satisfaction with treatment and subsequent treatments over time. Only hemoglobin/hematocrit laboratory values, severity of uterine bleeding, and standardized quality of life and functional status measures were reported using validated approaches.

GnRH Agonists

Eighteen studies reported in 20 publications evaluated GnRH agonists.⁷³^,⁸⁷^,⁹⁴^,⁹⁸^,¹¹⁶^,¹²²^,¹³³^,¹³⁸^,¹⁴⁰^,¹⁴²^-¹⁴⁷^,¹⁴⁹^-¹⁵¹^,¹⁵³ Thirteen studies (2 good¹⁰¹^,¹⁰⁸, 2 fair¹²⁴^,¹³⁶, and 9 poor quality⁷³^,⁸⁷^,⁹⁴^,⁹⁸^,¹¹⁶^,¹²²^,¹³³^,¹³⁸^,¹⁴⁰^,¹⁴²^-¹⁴⁷^,¹⁴⁹^-¹⁵¹^,¹⁵³) evaluated leuprolide (seven with an add-back agent), three poor quality studies⁹⁸^,¹³⁷^,¹⁴⁹ included goserelin, and two poor quality studies used triptorelin as the GnRH agonist. ⁶⁸^,¹²⁶^,

A total of 912 participants were randomized to GnRH treatment; study size ranged from 16 to 101 women. This small study size limits power for discerning differences across treatment groups and virtually prohibits meaningful evaluation of factors that may influence outcomes within groups. In general, study size was selected to detect differences in fibroid size and bleeding characteristics that are measured as continuous variables. The clinical significance of small to modest changes in fibroid size is unknown. Few studies were specifically designed to assess if treatment improved patient reported outcomes such as quality of life, sexual function, or satisfaction with treatment.

As in much of the fibroid literature, lack of followup over time is a limitation. Most studies completed their followup of participants when treatment ended. Only seven studies⁹⁸^,¹⁴³^,¹⁴⁶^,¹⁴⁷^,¹⁴⁹^,¹⁵⁰^,¹⁵³ followed women from 3 to 9 months after end of treatment, limiting the information about how durable the effects may be. Only one study re-contacted participants 3 years after treatment and found 23/59 (39%) had undergone hysterectomy.¹³⁷

Effects of GnRH Agonist Treatment on Fibroid Characteristics

GnRH agonists reduce the size of fibroids, with reductions in volume of fibroids documented between 64 and 175 cm³ and reductions in the total volume of the uterus between 131 and 610 cm³ (Table 8). As a point of reference, the volume of a golf ball is 40 cm³. It may be that change in size is related to initial size, in other words bigger fibroids have more capacity to shrink, but these studies are not able to assess that hypothesis. The duration of treatment was not directly related to reduction in volume in this literature. Two studies that measured fibroids more than once across the course of treatment found the change in the first round of imaging to be the greatest,¹⁴⁶^,¹⁴⁷ but another small study reported the largest volume reduction two months after treatment ended.¹⁵³

Table 8

Change in fibroid and uterine size with GnRH agonists by study arm.

Seven studies provided information on durability of treatment effects on fibroid size from 3 to 9 months after the end of treatment.⁷³^,¹⁴³^,¹⁴⁶^,¹⁴⁷^,¹⁴⁹^,¹⁵⁰^,¹⁵³ All of these reported increases or regrowth often back to pre-treatment levels.⁷³^,¹⁴³^,¹⁴⁶^,¹⁴⁷^,¹⁴⁹^,¹⁵⁰^,¹⁵³ A single study reported that two doses (1.88 mg and 3.75 mg) of GnRH were equally effective in reducing uterine volume.¹⁴⁶ Use of add-back raloxifene to GnRH agonists -resulted in greater reduction in fibroid size but no change in fibroid related symptoms.¹²² Add-back therapy of tibolone was protective for bone mineral density without interfering with fibroid size reduction.⁹⁸ Only one trial found that effects can be maintained over 2 years.¹⁴³ Two thirds of 51 women continued GnRH treatment while 17 (33%) discontinued treatment, three for other medication options, five had myomectomies, and nine had hysterectomy.

Effects of GnRH Agonist Treatment on Bleeding

GnRH agonists produce hypogonadism and many women completely stop bleeding (Table 9). Of the 20 published reports, five reported absence of bleeding, three noting statistical significance for clinically important reduction from baseline. One study reported reduction in days of bleeding⁹⁸ without a statistical test, and four reported improvement in hemoglobin levels with three of the four reporting significance. No study reported an increase in bleeding or worsening in measures such as hemoglobin or hematocrit within a treatment group. In several studies, some women discontinued treatment because bleeding became more irregular or did not decrease.

The symptoms and side effects of hypogonadism caused by GnRH are often treated with add-back hormonal therapy. GnRH plus add-back therapy was evaluated in eight studies and the effects on bleeding were mixed. Women who received medroxyprogesterone (MPA) as add-back therapy had improved hemoglobin levels reported in two small trials.¹⁴²^,¹⁵¹ A single trial that evaluated raloxifene as add-back therapy in conjunction with leuprolide acetate noted no difference in the proportion of women who continued to bleed after six cycles of raloxifene (6.3%) vs. placebo (8.3%).¹²² Another small study that compared estrogen-progestin to progestin only add-back with leuprolide acetate depot reported improved hemoglobin levels in both groups.¹⁴⁴. Women receiving goserelin plus tibolone had significantly higher mean number of days of bleeding (6.3 days) compared with the goserelin and placebo group (2.9 days).⁹⁸ In another 6-month study of leuprolide acetate with or without tibolone, both groups had reductions in the number of women reporting bleeding, but a small number of women continued to bleed with the add-back of tibolone compared with none in the placebo arm.¹³⁸ Bleeding outcomes were not assessed in the third study.⁹⁴

Table 9

Change in bleeding characteristics with GnRH agonists by study arm.

Effect of GnRH Agonist Treatment on Fibroid-Related Pain

Compared to baseline, GnRH treatment significantly improved pain symptoms including pelvic pressure,⁹⁴^,⁹⁸^,¹²²^,¹⁴³ pelvic and abdominal pain,⁹⁴^,⁹⁸^,¹²²^,¹⁴³ and dysmenorrhea.⁹⁸ Other studies reported similar improvements but without statistical comparisons of baseline to followup.¹⁴⁴^,¹⁴⁷

Effects of GnRH Agonist Treatment on Grouped Symptoms

Palomba and colleagues have conducted multiple studies treating women for 6 months with leuprolide and placebo or a comparator arm not available in the United States.⁹⁴^,¹²²^,¹³⁸ Within the leuprolide arms of these studies, women experienced a significant improvement in fibroid-related symptoms that were scored on a 1 to 10 point validated scale that includes menorrhagia, pelvic pressure, pelvic pain, urinary frequency, and constipation. Total scores and each individual scale item were improved, in each study bleeding and constipation completely resolved and other scores improved by 3 to 5 points, a substantial and likely clinically significant change. Mood and quality of life were also improved by treatment.⁹⁴ In studies with raloxifene add-back, similar improvements were documented in both the raloxifene and placebo add-back groups.¹²²

Using a similar, but not identical, five-item scale, Friedman and colleagues also demonstrated improvements in menorrhagia, bulk symptoms, pelvic pressure, urinary frequency, and pelvic pain that were sustained over 1 and 2 years of treatment with leuprolide and either estrogen and progestin add-back or just progestin add-back; with an overall advantage for the combined estrogen and progestin add-back group.¹⁴³^,¹⁴⁴

Harms Reported in Studies of GnRH Agonist Treatment

Because of estrogen suppression, GnRH is associated with onset of menopausal symptoms,⁹⁴^,⁹⁸^,¹³⁸^,¹⁴⁷ unfavorable changes in lipid profile,¹³⁸^,¹⁴⁴ and bone loss (ranging from 2.6% to 5.5%).⁹⁸^,¹³⁸^,¹⁴³ These effects increase motivation for investigating add-back therapy. Estrogen and progesterone together normalized adverse lipid effects, while progesterone only did not.¹⁴⁴ Addition of raloxifene protects bone¹³⁸ and estrogen-progestin or progestin add-back stabilized bone loss when initiated after a 12-week period of GnRH only.¹⁴⁴

Six months of treatment with leuprolide was associated with declines in cognitive function and memory as measured by the Mini-Mental Status Exam and the Wechsler Memory Scale..⁹⁴

GnRH Agonists Summary

GnRH agonists reduce the size of fibroids and the overall size of the uterus. Both with and without add-back therapy, bleeding symptoms, hemoglobin, fibroid-related pain and other symptoms improve over baseline. Add-back medication relieves associated menopausal symptoms and can ameliorate bone loss and lipid changes. Only one trial examined outcomes of treatment after more than 24 months. Extended followup of women after they discontinue GnRH agonists is not available, thus information about potential harms to guide care is limited.

Progesterone Receptor Agents: Anti-Progestins, Selective Receptor Modulators, and Levonorgestrel IUD

This section includes 14 studies designed to test the effectiveness of medications that work through progesterone pathways.³⁶^,⁵⁴^,⁵⁵^,⁵⁹^,⁶¹^,⁶⁴^-⁶⁶^,⁷³^-⁷⁵^,⁷⁹^,⁸⁸^,¹⁰⁶^,¹¹⁵^,¹¹⁹^,¹⁵⁷^,¹⁶⁵ They include seven studies of mifepristone,⁶⁴^-⁶⁶^,⁷⁵^,⁸⁸^,¹⁰⁶^,¹¹⁹; six of ulipristal,³⁶^,⁵⁴^,⁵⁹^,⁷³^,⁷⁴^,⁷⁹^,¹⁵⁷; and a single study of a progesterone-containing IUD.¹⁵²

Mifepristone

Seven studies (eight publications) provide data about outcomes of mifepristone treatment.⁶⁴^-⁶⁶^,⁷⁵^,⁸⁸^,¹⁰⁶^,¹¹⁹ This literature is dominated by two teams: a group led by Carbonell and colleagues who conducted five of the included studies in Cuba; and by Eisinger and Fiscella at University of Rochester School of Medicine who conducted two included studies.¹⁰⁶^,¹¹⁵^,¹¹⁹ Overall, 5 good quality and 2 poor quality trials provided information on 690 women. Two studies compared a 5 mg dose to placebo,⁶⁵^,¹⁰⁶ one study compared 2.5 mg and 5 mg doses,⁶⁴ the remainder compared 5 mg and 10 mg doses. Four groups included followup after treatment with mifepristone had ended. Average length of time for off-medication followup was 11 months with the longest untreated followup being 18 months.⁶⁶^,¹⁶⁶

Effects of Mifepristone on Fibroid Characteristics

All studies observed a decrease in the size of fibroids at the completion of active treatment. The decrease in size of the largest fibroid ranged from 37 cm³ to 95 cm³, with an average of 71 cm³ among the 575 women studied (Table 10).⁶⁴^-⁶⁶^,⁷⁵^,⁸⁸ Likewise, total uterine volume decreased in all groups receiving mifepristone.⁶⁴^-⁶⁶^,⁷⁵^,⁸⁸^,¹⁰⁶^,¹¹⁵^,¹¹⁹ This was consistent across doses from 2.5 mg to 10 mg each day, with statistically significant reductions at 5 mg and 10 mg doses documented in three trials.⁸⁸^,¹⁰⁶^,¹¹⁹ Because most trials were designed to compare doses, authors often did not provide statistical comparisons within groups from baseline to followup.

In the studies designed to determine if changes in fibroid size were durable, all four trials reported no statistically significant change in the size of the largest fibroid or uterine volume after completion of treatment.⁶⁴^,⁶⁶^,⁷⁵^,¹¹⁵ However, while volume of the largest fibroid remained smaller than baseline at nine months of followup off medication, the total uterine volume was slightly increased over baseline. With 12 and 18 months of followup, fibroid and uterine volume tended to increase, often above baseline,⁶⁶^,⁷⁵ suggesting that treatment suspends fibroid growth but does not have lasting effects to forestall future growth of fibroids.

It is also important to note in these studies that the number of women available at followup was often fewer than at enrollment. This loss to followup includes those who did not continue medication or who did not improve and had subsequent treatments including surgery. Since intention-to-treat analyses with last uterine volume carried forward were not done, this means as the women are lost to follow up, the measures may under-represent changes in fibroids if we speculate that those who were lost could be more likely to have increase in size over time.

Table 10

Change in fibroid and uterine size with mifepristone by study arm.

Effects of Mifepristone on Bleeding

All studies that assessed bleeding reported heaviness of bleeding was reduced by treatment (Table 11). Those that made comparison found the active drug superior to placebo.⁶⁵^,¹⁰⁶ Women were as likely to have decreased bleeding or absent menses on the lower doses compared with the higher doses.⁶⁶^,⁸⁸^,¹¹⁵ When bleeding occurred it was often described as spotting or staining.⁶⁴^,⁶⁵^,⁷⁵

Table 11

Change in fibroid volume with uterine artery embolization by study arm.

Effects of Mifepristone on Fibroid-Related Pain

Each of six publications that evaluated pelvic pain before treatment and at conclusion of treatment noted substantial improvements.⁶⁴^-⁶⁶^,⁷⁵^,⁸⁸^,¹⁰⁶ At baseline, 68 to 100 percent of women in these trials reported pelvic pain. By three months of treatment, this was reduced to a range of 9 to 28 percent with those in the lower dose groups having lower or equivalent prevalence of pelvic pain.⁶⁴^,⁶⁵^,⁸⁸ Similar findings persisted at conclusion of 6 and 9 months of treatment. Once off treatment, prevalence of pelvic pain remained meaningfully lower, with 6.3- 37.0 percent of women affected at 9 months,⁶⁴^,⁶⁶ 16.2 to 18.6 percent at 12 months,⁷⁵ and 10 to 11 percent at 18 months.⁶⁶ The Rochester group reported change in pain using the McGill Pain Questionnaire and documented a steady decline from a high score of approximately 20 to about 6 points during the 6 months of treatment.¹⁰⁶ Strength of evidence was not graded for pain.

Effects of Mifepristone Treatment on Quality of life

Four studies (374 women) reported quality of life metrics: for the three that used the UFSQOL, composite scores improved,⁶⁴^-⁶⁶ as much as 50 of a possible 100 points (with placebo controls improving by 17 points). Dose was not convincingly related to quality of life scores.⁶⁴^,⁶⁶ The study that used the Short Form 36 subscales documented improvements in energy and fatigue, health status, and pain domains.¹⁰⁶ Statistically significant improvements from baseline were noted in these aspects across studies at one or more doses:

Symptoms⁶⁴^,⁶⁵^,¹⁰⁶
Concern⁶⁴^,⁶⁵^,¹⁰⁶
Activity⁶⁴^,⁶⁵
Inhibition/Self-consciousness⁶⁴^,¹⁰⁶
Control⁶⁴^,⁶⁵
Sexual function⁶⁴^,¹⁰⁶
Energy and mood⁶⁴^,⁶⁵

Other Effects of Mifepristone Treatment

In the Cuban studies, improvements were observed for other symptoms including pelvic pressure, urinary symptoms, lumbar pain, rectal pain, and dyspareunia. In each case, the proportion with the symptom dropped by one to two-thirds or more. Improvement was sustained with less than 6 percent increase in prevalence over the additional 9 to 18 months of followup. ⁶⁴^-⁶⁶^,⁷⁵^,⁸⁸ The Rochester group also reported that improvements in pelvic pressure, urinary frequency, low back pain, rectal pain, and pain with intercourse improved across treatment with active drug compared with placebo, with significant benefits for reducing pain with intercourse.¹⁰⁶

Harms Reported in Studies of Mifepristone Treatment

No unanticipated adverse drug effects were identified in these trials. All trials conducted surveillance for the most serious known risk of harm, which is development of endometrial hyperplasia. All seven trials (eight publications) conducted an endometrial biopsy at an interim point or at the completion of treatment, unless the subject declined.⁶⁴^-⁶⁶^,⁷⁵^,⁸⁸^,¹⁰⁶^,¹¹⁵^,¹¹⁹ In four of these studies, biopsies were conducted at more than one followup time point.⁶⁴^,⁶⁶^,⁷⁵^,¹¹⁵^,¹¹⁹ The proportion of participants with at least one post-treatment biopsy result ranged from 73 percent to 99 percent. The total number of biopsies subsequent to treatment in all studies combined was 782.

Two placebo controlled studies with 5 mg mifepristone as the active intervention reported no hyperplasia in either group after 3 or 6 months of treatment.⁶⁵^,¹⁰⁶ One study compared 5 mg to 2.5 mg mifepristone and reported no hyperplasia in either group after 3 months of treatment.⁶⁴ Four studies (reported in five publications) compared 10 mg to 5 mg mifepristone with treatment durations of 3 months,⁸⁸ 6 months,⁷⁵^,¹¹⁹ 9 months,⁶⁶ and 12 months.¹¹⁵ One study¹¹⁹ reported an additional 6-month continuation, with a revised report of pathology at the 6-month timepoint.¹¹⁵ Table 12 summarizes the number of women across the four studies who had biopsies with the indicated findings at specific timepoints.⁶⁶^,⁷⁵^,⁸⁸^,¹¹⁵^,¹¹⁹ There were no reported cases of atypical hyperplasia. The counts of simple hyperplasia at 3 months of treatment were 2/92 (2%) for 5 mg dosage and 2/118 (2%) for 10 mg dosage. The counts of simple hyperplasia at 6 months of treatment were 2/85 (2%) for 5 mg dosage and 6/100 (6%) for 10 mg dosage. Data for 9 months and 12 months are sparse. In aggregate, there were 13 cases of simple hyperplasia detected among 446 biopsies.

Table 12

Number of mifepristone-treated women with indicated endometnal status upon biopsy.

All seven trials also monitored liver function enzymes as elevations have been reported but not overt or sustained liver damage.. The percentages of women with elevated transaminases following mifepristone treatment ranged from 5.0 to 12.7 percent in the Cuban studies. ⁶⁴^-⁶⁶^,⁷⁵^,⁸⁸. The maximum values when reported did not exceed 100 IU. Increases in liver enzymes were also noted in zero and 8 percent of the women in two U.S. studies ¹¹⁹.¹⁰⁶ Abnormal liver enzymes were similar between the group treated with mifepristone (49/652, 7.5%) and the group who received placebo (5/67, 7.5%). There were no reports of any liver damage.

Mifepristone Summary

Moderate evidence from 5 fair and 2 poor quality studies supports that mifepristone reduces size of fibroids and overall uterine volume and improves quality of life. Heaviness of bleeding is reduced during treatment and measures of anemia improve. Information is unavailable to contribute to dose selection between higher and lower doses.

Some evidence suggests fibroids do resume growth after treatment; however, the majority of women achieve symptomatic relief for a year or more after cessation of active treatment. Few participants in these trials pursued subsequent treatment during medical management or in the time after concluding active treatment suggesting that treatment with mifepristone can provide sufficient management of fibroid-related symptoms.

Ulipristal Acetate

Six trials, reported in nine publications, investigated use of ulipristal acetate as a treatment for fibroids, enrolling 691 women in the ulipristal arms. Two small, poor quality trials were conducted at National Institutes of Health³⁶^,⁷⁹ and four in research networks in Europe (the PEARL trials), of which two were good and two fair quality.⁵⁴^,⁵⁵^,⁵⁹^,⁶¹^,⁷³^,⁷⁴^,¹⁵⁷ The two earliest studies³⁶^,⁷⁹ followed participants to the end of 12 weeks of treatment with active drug. Two trials reported on drug efficacy in women seeking surgery for management of heavy bleeding and fibroids.⁷³^,⁷⁴ PEARL I randomized women to 5 or 10 mg doses or placebo for three months⁷⁴ and PEARL II randomized women to 5 or 10 mg doses or leuprolide acetate.⁷³ The PEARL III study assessed up to four courses of 10 mg dose for 12 weeks⁵⁹ followed with a daily progestin (norethisterone acetate) or placebo. PEARL IV compared 5 and 10 mg doses in women for four repeated 12 week cycles.¹⁵⁷ There was a drug-free interval following each treatment cycle, with the next treatment round resuming at the start of the second menstrual cycle. Final followup was conducted three months after the completion of medication.⁵⁴^,¹⁵⁷

Effects of Ulipristal on Fibroid Characteristics

All six studies found ulipristal effective for reducing the size of individual fibroids and the overall fibroid burden as measured by total fibroid and uterine volume (Table 13). A single course of 5 or 10 mg reduced fibroids size by 17 to 38 percent;³⁶^,⁵⁴^,⁵⁹^,⁷⁹ a repeated course of treatment reduced volume of the three largest fibroids by 54 to 58 percent from baseline to completion of both cycles.⁵⁹

A study that monitored women in ulipristal trial arms who did not go on to have surgery found minimal resumption of fibroid growth (8.1%) that was not statistically significant at six months after completion of treatment regardless of ulipristal dose.⁷³ A related trial with similar design found that the 10 mg dose sustained lack of fibroid growth beyond six months while there was increase in size on the 5 mg dose that was similar to placebo.⁷⁴ Durability of effects remains unknown. Extended followup after treatment cessation is needed because fibroid growth patterns may differ for women in these trials who did proceed to surgery.

Table 13

Change in fibroid and uterine size with ulipristal acetate by study arm.

Effects of Ulipristal on Bleeding

Ulipristal, as intended, resulted in absent menses for the majority of women during treatment (range 62 to 100%) and the large majority reported improved bleeding (Table 14).³⁶^,⁵⁴^,⁷³^,⁷⁴^,⁷⁹ This was also documented by improved or stable hematocrit or hemoglobin levels.³⁶^,⁵⁴^,⁷⁹ Cessation of bleeding at onset of treatment was prompt, ranging from a mean of 4 to 7 days.⁵⁴^,⁵⁹^,⁷³ Absence of bleeding was achieved more consistently with higher doses.⁵⁴^,⁷³^,⁷⁹

Other Effects of Ulipristal

Compared with placebo, all ulipristal doses resulted in improved overall fibroid-related quality of life or subscale scores as measured by the UFS-QOL scale³⁶^,⁷⁹ though some time points lack statistical testing. Similar improvements were seen in fibroid-related quality of life as measured by the Short Form 36 (SF-36),⁷⁴ and two trials also documented improvement in pain.⁵⁴^,⁷⁴

Harms Reported in Studies of Ulipristal Treatment

Among 978 biopsies at completion of treatment, six cases of confirmed hyperplasia (one with atypia) were reported (0.6%). The two smallest studies reported modest elevations of liver function enzymes during treatment; one larger trial documented change in liver function enzymes was comparable to those taking placebo. About 2 to 10 percent of women taking ulipristal experienced hot flashes.

Ulipristal Summary

Moderate evidence supports the effectiveness of ulipristal for reducing the size of fibroids and improving bleeding and quality of life. Bleeding is reduced with most women reporting absence of menses and measures of anemia stabilized or improved during treatment. Evidence is insufficient to contribute to dose selection between higher and lower doses. Data on extended followup are lacking to gauge whether fibroids resume growth after treatment. Use of a progestin for 10 days to prompt onset of menses shortened the time between treatment cycles in a single study.

Table 14

Change in median PBAC score with ulipristal acetate by study arm.

Levonorgestrel (LNG) Intrauterine Device

One small, poor quality study compared a daily oral progestin (norethindrone acetate) with LNG-IUD for improving bleeding patterns among 60 premenopausal women with uterine fibroids. ¹⁵² No placebo was used and women were not blind to intervention group.

Effects of LNG-IUD on Fibroid Characteristics

The study did not report changes in fibroid volume.

Effects of LNG-IUD on Bleeding

Participants used the Visual Blood Scoring system,¹⁶⁷ a standardized pictorial method for reporting blood loss in a diary over the course of treatment. Visual blood loss scores improved by 6 months in both groups, with greater improvement in the LNG-IUD group (p=0.03). Improvement in hemoglobin likewise occurred in both groups with a statistically greater improvement among those with an IUD.

Other Effects of LNG-IUD

Women with the LNG-IUD were more satisfied and more likely to continue treatment than women taking norethindrone acetate. This study did not report harms of LNG-IUD.

LNG-IUD Summary

This trial suggests the LNG- IUD can improve bleeding even among women whose fibroid symptoms were considered appropriate for surgical intervention. However, the quality of the study was poor and thus evidence to guide care is inadequate. Based on the criteria for this review, evidence is insufficient for effects of LNG-IUD on bleeding, fibroid size, and quality of life.

Estrogen Receptor Agents and Combined Hormonal Therapy

Six studies included agents that act at the estrogen receptor.¹¹⁸^,¹²³^,¹²⁴^,¹²⁸^,¹²⁹^,¹³⁶ Three studies (two fair and one poor quality) investigated raloxifene in 73 women compared to placebo.¹¹⁸^,¹²³^,¹²⁸ Two were conducted in Italy by Palomba and colleagues with a total of 160 participants,¹²³^,¹²⁸ and the third, a smaller study with 25 women in Austria.¹¹⁸ Two of these studies focused on premenopausal women; one enrolled only post-menopausal women.¹²⁸ A single poor quality pilot study (n=20) evaluated tamoxifen. ¹²⁹ Two small poor quality trials conducted in Italy (n=38) and Turkey (n=46) had arms that provided estrogen plus progestin replacement to postmenopausal women with fibroids.¹²⁴^,¹³⁶

Effects of Estrogen Receptor Agents on Fibroid Characteristics

Fibroid size decreased by 4.4 cm³ to 34.2 cm³ in two studies of raloxifene and did not change size in another raloxifene study (to put this in perspective, 40 cm³ is the volume of a golf ball). Change in fibroid characteristics was not reported in the trial of tamoxifen (Table 15).

Table 15

Change in fibroid and uterine size with estrogen receptor agents by study arnr.

Effects of Estrogen Receptor Agents on Bleeding

In studies of raloxifene with premenopausal women, neither bleeding pattern¹¹⁸^,¹²³^,¹²⁸ nor hemoglobin levels¹²³ improved compared with placebo, and a lower versus higher dose had similar results for days and severity of bleeding.¹²³ Among postmenopausal women, most women remained amenorrheic (83% in the raloxifene group and 86% in the placebo group at 9 months); the number of episodes of spotting and severity of bleeding were similar among women in the treated and control group. In the pilot study, tamoxifen use in premenopausal women also did not influence length or severity of bleeding compared with placebo.¹²⁹

Effects of Estrogen Receptor Agents on Other Symptoms

Only the small pilot tamoxifen comparison to placebo assessed pain; 70 percent of participants had pain at enrollment. The treatment group reported significantly less pain after four months of treatment but not earlier.¹²⁹ The study did not report on improvement in other symptoms or quality of life.

Risk of Harms With Estrogen Receptor Agents

These studies reported no drug-related adverse events, and withdrawal from treatment for perceived side effects or adherence was rare and equal to placebo groups.¹¹⁸^,¹²³^,¹²⁸ Simple ovarian cysts occurred in raloxifene treated women which resolved once off medication.¹¹⁸ Endometrial thickening occurred with tamoxifen, and biopsies in this very small study were normal.¹²⁹ However, these randomized controlled trails were not designed or powered to detect all harms related to this therapy.

Combined Hormone Replacement Therapy

Two poor quality RCTs had a total of 42 women receiving transdermal estrogen plus cyclic oral medroxyprogesterone acetate after menopause. ¹²⁴^,¹³⁶ They compared hormone therapy to tibolone (not available in United States) for menopausal symptom management with attention to whether treatment increased size of fibroids. Combined estrogen plus progestin therapy resulted in timed cyclic bleeding as intended among postmenopausal women.¹³⁶ In the HRT arm fibroid size increased by approximately 10 cm³ after 6 months.¹²⁴ In the longer study there was no further growth between 6 and 12 months.¹³⁶

Summary of Estrogen Receptor Agents

These agents were related to no or small decreases in fibroid size without improvement in bleeding among those who were premenopausal.

These studies provide a low strength of evidence that raloxifene is unlikely to prompt significant fibroid growth or to exacerbate bleeding if they are needed to treat women with fibroids for other conditions such as breast cancer prophylaxis. Two small studies of combined hormone replacement therapy did not find any changes in fibroid size after 6 or 12 months of therapy. This is insufficient to gauge if estrogen plus progestin therapy for menopausal symptoms does or does not promote fibroid growth.

Tranexamic Acid

We include one study, a pooled analysis of data from two independent trials of tranexamic acid treatment versus placebo for heavy uterine bleeding.⁶⁷ We did not include the primary studies that contributed data to the pooled analysis because those studies did not meet our review inclusion criteria (outcomes were not reported by fibroid status). The pooled analysis included a subset of women with uterine fibroids from each study. Women with fibroids who received tranexamic acid reported statistically significant (p<0.001) reductions in menstrual blood loss at treatment cycle three compared with placebo. We did not rate the quality or risk of bias for the pooled analysis or the primary studies from which the data was drawn.

Procedures for Fibroid Intervention: Overview

In this section we include studies of procedures to treat uterine fibroids including uterine artery embolization (UAE) and occlusion, high intensity focused ultrasound (HIFU) for fibroid ablation, and radiofrequency fibroid ablation. 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 to surgery are discussed in the section about direct comparisons.

Procedures: Results

We include 28 studies addressing UAE, HIFU including MRgFUS, and radiofrequency fibroid ablation.¹³^-²³^,³⁸^,⁵³^,⁵⁷^,⁵⁸^,⁶⁰^,⁶²^,⁶⁸^,⁷⁰^,⁷¹^,⁷⁸^,⁸¹^,⁸²^,⁹¹^-⁹³^,⁹⁶^,⁹⁷^,⁹⁹^,¹⁰⁴^,¹⁰⁵^,¹⁰⁸^,¹⁰⁹^,¹¹¹^,¹¹³^,¹¹⁴^,¹¹⁷^,¹²⁷^,¹⁵⁵^,¹⁵⁶^,¹⁵⁸ Studies included 2,149 women and were conducted in 15 different countries, most frequently in the United States (7 studies) or China (6 studies).) Many (15) of these¹⁴^,¹⁵^,¹⁸^,²²^,²³^,⁵³^,⁵⁸^,⁶²^,⁸¹^,⁹²^,¹⁰⁵^,¹⁰⁸^,¹¹⁷^,¹²⁷^,¹⁵⁵ did not report source of funding. Three studies, two comparing embolic agents¹³^,⁷⁸ and one that evaluated radiofrequency fibroid ablation,⁵⁷ were industry supported. Nine studies compared a procedural intervention to hysterectomy or myomectomy.¹⁴^,¹⁶^,²⁰^,²²^,⁵⁷^,⁶⁸^,¹⁰⁴^,¹¹³^,¹¹⁴ The longest duration of followup after the end of treatment was 5 years in two studies.

Uterine Artery Embolization or Occlusion

We identified 21 studies (reported in 34 publications) that randomized women to UAE or uterine artery occlusion.¹³^-²³^,³⁸^,⁶²^,⁷⁰^,⁷¹^,⁷⁸^,⁸²^,⁹¹^-⁹³^,⁹⁶^,⁹⁷^,⁹⁹^,¹⁰⁴^,¹⁰⁵^,¹⁰⁸^,¹⁰⁹^,¹¹¹^,¹¹³^,¹¹⁴^,¹¹⁷^,¹²⁷^,¹⁵⁵^,¹⁵⁸. Of 18 studies of UAE (30 publications), ten studies compared an embolization agent to a different agent or different size,¹³^,¹⁵^,¹⁸^,²³^,⁶²^,⁷⁸^,⁹²^,¹¹⁷^,¹⁵⁵^,¹⁵⁸ seven studies compared UAE to surgery,¹⁴^,¹⁶^,²⁰^,²²^,¹⁰⁴^,¹¹³^,¹¹⁴ and one compared UAE to a GnRH agonist (goserelin).¹⁰⁸ We assessed six studies to be good quality, six were of fair quality and six were poor quality.

We identified three studies (one fair quality and 2 poor quality) reported in four publications that assessed uterine artery occlusion. Two studies (reported in three publications) compared transcatheter³⁸ or laparoscopic⁸³^,¹⁰⁵ uterine artery occlusion to UAE; one compared laparoscopic bipolar coagulation alone to laparoscopic bipolar coagulation plus laparoscopic ligation of uterine nerves to determine if the addition of laparoscopic uterine nerve ablation would improve postoperative pain and dysmenorrhea.¹²⁷

Much of the information on safety and long-term outcomes of UAE is from two large trials (EMMY¹¹⁴ and REST¹⁰⁴). The duration of followup ranged from 2 months to 60 months after treatment, with an average of 15 months.

Effects of UAE on Fibroid Characteristics

Fibroid and uterine volume decreased consistently and significantly following UAE (up to 12 months postprocedure) regardless of the embolization agent or size of particles used to occlude the fibroid arteries (Table 16). Additional longer-term followup reports from the EMMY trial confirm that fibroid and uterine volume reductions persist up to 5 years after UAE; however in four studies with two to five years of followup, the proportion of women seeking subsequent treatment following UAE ranged from 19 percent to 38 percent Appendix G includes additional tables with data on changes in uterine volume.

Table 16

Change in fibroid volume with uterine artery embolization by study arm.

Effects of UAE on Bleeding

Changes in bleeding were reported as incidence of amenorrhea, change in bleeding score using a scale from −5 to +5, and self-reported dysmenorrhea or menorrhagia (Table 17).

Table 17

Change in bleeding outcomes with uterine artery embolization by study arm.

Effects of Uterine Artery Embolization or Occlusion on Fibroid-Related Pain

Most women who underwent uterine artery occlusion via laparoscopic bipolar coagulation reported improvement in dysmenorrhea symptoms at 6 months after procedure (76.2% of women who were treated by coagulation of uterine vessels alone). ¹²⁷ At baseline, 73 women (90.1%) in the UAE arm of the EMMY trial complained of lower abdominal pain. At 24 months of followup, 84.9 % of women reported moderate improvement of pain.⁹⁹ In another study, nine of the 27 women (33%) in the UAE arm reported dysmenorrhea at baseline while only four (15%) complained of dysmenorrhea at the 2 year followup, thus showing a reduction of 56 % from baseline.²⁰

Other Treatment Effects of Uterine Artery Embolization or Occlusion

Quality of Life

Overall improvement in symptoms and physical well-being following UAE were reported using UFS-QOL, SF-36, and European Quality of Life 5D (EQ-5D™) scores. Quality of life was not reported following laparoscopic bipolar coagulation or transcatheter uterine artery occlusion³⁸^,⁸³^,¹⁰⁵ ¹²⁷

Quality of Life: UFS-QOL

After UAE, significant improvement in symptoms was reported by the UFS-QOL in a small study (n=36) ²³ The 2014 study by Shlansky-Goldberg and colleagues¹³ reported changes in total quality of life or symptom and subscores on the UFS-QOL in figures only. Changes in the total quality of life score were reported in figures only in another small study (n=44) comparing UAE with trisacryl gelatin microspheres to UAE with polyvinyl alcohol particles.⁷⁸ Total quality of life scores improved at 3 months after uterine artery embolism in both groups (UAE with trisacryl gelatin microspheres: 36.0 ± 25.5 and UAE with polyvinyl alcohol particles: 23.1 ± 23.4, p-values not reported).¹¹⁷ A 2012 study was powered to detect a 10-point difference in quality of life outcomes among premenopausal women with symptomatic fibroids following abdominal myomectomy or UAE.¹⁶ Authors reported significant improvements from baseline in overall quality of life and severity scores after UAE (p=NR).¹⁶

Quality of Life: SF-36

Four trials reported SF-36 quality of life outcome measures following UAE. Two assessed at 6 months,¹⁴^,¹⁵⁵ and two, the REST¹⁰⁴ and EMMY⁹³ trials, up to 60 months (Table 18). At 6 months, both trials identified significant improvements in quality of life scores from baseline (p=NR and p<0.05).¹⁴^,¹⁵⁵ The REST trial¹⁷ reported a gain in quality of life after UAE with five year measures of SF-36 scores being comparable to normative data. The EMMY trial⁹³ also found those in the UAE group had improved general health-related quality of life at 6 months and later when compared with baseline values (p<0.05). Using the Body Image Scale, the EMMY trial⁹⁶ reported that body image also improved significantly from baseline (p<0.05) in the UAE group at 6 (−1.34), 18 (−1.24) and 24 (−1.06) months with lower scores representing favorable body image and negative numbers indicating improvement.

Table 18

Change in quality of life (SF-36) with uterine artery embolization by study arm.

Quality of Life: EQ-5D

The REST trial did not report the change in EQ-5D or symptom status score from baseline at 12 months or at 5 years, though the absolute scores showed improvement.¹⁷^,¹⁰⁴ Significant improvements (p<0.05) from baseline in EQ-5D scores were observed at 6 months and afterwards in the EMMY trial.⁹³

Satisfaction

Out of the seven studies comparing UAE with surgeries, satisfaction rates were reported in all but one study (FUME trial¹⁶). Satisfaction with outcome was measured by asking women if they would undergo the same treatment again,²² if they obtained symptom relief,²¹^,¹¹³ if they were satisfied with the treatment,¹⁹^,²⁰ and if they would recommend treatment to a friend.¹⁴^,¹⁹^,¹⁰⁴ One trial also reported satisfaction without providing details of the criteria.¹⁴ Satisfaction rates (Table 19) ranged from 78 to 89 percent at 6 months²¹^,²² to 82 to 88 percent at 1 year¹⁴^,¹⁰⁴ to about 90 percent at 2 years.¹⁹^,²⁰ Satisfaction rates remained high (84% to 90%) at 5-year followup.¹⁷^,¹⁹

Table 19

Patient satisfaction with uterine artery embolization.

Recurrence and Subsequent Treatment

Two studies²¹^,⁷¹ reported fibroid recurrence. In one trial, women were followed for a mean period of 26 months after UAE.²¹ By 2 years, there were six women (10.3%) with regrowth or recurrence of fibroids.²¹ The REST trial reported fibroid recurrence in five out of 68 women (7%) 5 years after UAE treatment.⁷¹

Subsequent treatment was reported in 11 trials with length of followup ranging from 6 to 60 months (Table 20). Hysterectomy was the most frequent intervention (8.9%) followed by repeat embolization (4.2%), myomectomy (3.6%), medication or IUD (1.1%) and endometrial ablation (0.1%).

Table 20

Subsequent fibroid treatment following uterine artery embolization by study arm.

Effects of UAE on Ovarian Reserve and Pregnancy Outcomes

Ovarian reserve and pregnancy outcomes were not uniformly evaluated following UAE.⁸³^,¹⁰⁵^,¹²⁷ (Table 21) Ovarian failure, measured by follicle stimulating hormone (FSH) >40 IU/L and anti-Mullerian hormone (AMH), was reported in two trials.⁸²^,⁹⁷ In the EMMY trial 88 women were assigned to UAE. Their average age at baseline was 45. In this group FSH increased significantly by 12.1 IU/L compared with baseline (p=0.001) by 24 months after treatment with UAE. Ovarian failure (FSH >40 IU/L) was reported in 12 percent and 18 percent at 12 and 24 months, respectively.⁹⁷ Levels of AMH were significantly lower, indicating ovarian aging at each followup up to 24 months after UAE (p<0.05).⁹⁷ These changes in FSH and AMH were comparable to those randomized to hysterectomy (p=0.37). The only predictor of becoming menopausal in each group was being older than 45 at randomization. A similar proportion of 73 women (11%) were observed to have menopausal levels of FSH at 12 months after UAE in the REST study. This was also comparable to levels in the surgical arm of their trial (p=0.47). Participants in REST also had an average age in their mid-forties at the time of randomization.⁸²

The trial by Mara and colleagues included 58 women randomized to UAE. The average age of participants in their study was more than a decade younger than the other trials.²¹ In this younger study population the risk of elevated FSH >10 IU/L after intervention was higher among those with UAE (13.8%) than myomectomy (3.2%; p<0.05), though no participants became frankly menopausal. This study, though under-powered, was the only study that prespecified pregnancy and live birth as outcomes of interest. By two years, there were 13 pregnancies (50%) and five live births (19.2%) reported out of 26 women wanting to conceive. The REST trial¹⁰⁴ did not prespecify pregnancy outcomes, but did report seven pregnancies after UAE at 12 months which included four miscarriages, two livebirths, and one intrauterine fetal death at 33 weeks with no known cause. The EMMY trial⁹⁹ reported one unplanned pregnancy after UAE at 24 months in a 39-year-old multipara, who delivered a healthy child after secondary cesarean section for fetal distress.

Table 21

Pregnancy and fertility status following uterine artery embolization.

Harms in Studies of UAE

Transfusion

Three studies of UAE compared with myomectomy or hysterectomy reported incidence of transfusion.²¹^,²²^,¹¹⁴ None of the 186 patients required transfusion after UAE.

Other Major Complications

The proportion of major complications, including unplanned hysterectomy, rehospitalization, ovarian failure, and pulmonary embolism during and following UAE ranged from 1.2 to 6.9 percent within a month of the procedure, up to 3 percent by 1 year,¹⁶ and about 5 percent at 2 years.⁹¹ The rates of major complications were highest in two studies that reported long-term followup (21% at 5 years in the REST trial¹⁷ and 16.8% at 32 months in a second study¹⁰⁴.

Complication rates were generally low in the studies comparing embolic agents¹⁵^,¹⁸^,⁶²^,⁹²^,¹¹⁷ (Appendix G), however the duration of followup among these studies was 9 months or less.

Summary of Uterine Artery Embolization or Occlusion

There was high strength of evidence that UAE is effective for reducing fibroid volume. The strength of evidence supporting improvements in bleeding and quality of life is moderate for UAE. Five-year followup data were available from two large good quality trials in which well over half the women who received an embolization did not need a subsequent intervention (including hysterectomy). The effect of UAE on reproductive outcomes is not well studied and evidence is insufficient to guide care or determine safety.

Because of small numbers and heterogeneity of methods, there is insufficient evidence to make any conclusions about uterine artery occlusion.

High Intensity Focused Ultrasound for Fibroid Ablation

The prior review¹² included findings from a prospective case series that was conducted to support an application for FDA approval of the MRgFUS.¹⁶⁸^,¹⁶⁹ Since then, we identified six studies reported in seven publications,⁵³^,⁵⁸^,⁶⁰^,⁶⁸^,⁶⁹^,⁸¹^,¹⁵⁶ assessing HIFU as treatment for uterine fibroids. Interventions included MRgFUS in one study,¹⁵⁶ HIFU in two studies with three publications,⁶⁸^,⁶⁹^,⁸¹ HIFU plus contrast enhanced ultrasound in two studies,⁵³^,⁶⁰ and HIFU plus ethanol injection into the fibroid in one study.⁵⁸ These trials were published between 2010 and 2016. Four studies were conducted in China and one each in Italy and the United States. Study size ranged from 20 to 100 women and included a total of 316 participants. The MRgFUS pilot study was fair quality and the other studies were poor quality.

Effects of HIFU for Fibroid Ablation on Fibroid Characteristics

Four studies reported fibroid volume following HIFU (Table 22).⁵³^,⁵⁸^,⁶⁰^,¹⁵⁶ The magnitude of fibroid volume reduction was greater at 12 months⁵³ after HIFU than at 1 month post-treatment.⁵⁸

Table 22

Change in uterine fibroid volume following HIFU by study arnr.

Effects of HIFU for Fibroid Ablation on Bleeding and Fibroid-Related Pain

There was no change in hemoglobin levels after twelve weeks in one small study.¹⁵⁶ Patient-reported bleeding symptoms were not reported in these trials.

Other Treatment Effects of HIFU for Fibroid Ablation

One month after HIFU, the mean change in UFS-QOL score was increased by 16 or more points among 37 patients (baseline not reported); however, four patients experienced persistent symptoms and underwent a second HIFU procedure.⁵³ A study that compared myomectomy to HIFU reported treatment effects on sexual function at 6 months postprocedure among 100 women. The total sexual function score using the brief index of sexual function for women (BISF-W) improved from 24.6 ± 6.6 at baseline to 26.7 ± 5.2 at 6 months in the HIFU group (n=48; p<0.05).⁶⁸

Harms Reported in Studies of HIFU for Fibroid Ablation

No major harms were observed postprocedure in the 316 patients who received HIFU for fibroid treatment.⁵³^,⁵⁸^,⁶⁰^,⁶⁸^,⁸¹^,¹⁵⁶ One study reported on transfusion; none of the 48 women who received HIFU required a transfusion (Appendix G).⁶⁸

HIFU for Fibroid Ablation Summary

HIFU reduced fibroid and uterine size, but strength of evidence is low because of short followup and poor quality of overall study design. Studies of HIFU for fibroid ablation reported intra- and postprocedural outcomes, specifically technical success and safety of the technique.. Publications did not assess symptoms or long-term outcomes, including pregnancy. Limited data suggests that QOL and sexual function do not worsen post-procedure. Evidence related to patient reported outcomes is insufficient.

Radiofrequency Fibroid Ablation

We included two studies (4 publications) that assessed outcomes of radiofrequency fibroid ablation. These studies were conducted in China and Germany and included 75 patients. Both were assessed as poor quality. The Chinese study published in 2010 randomized women to radiofrequency ablation (n=50) or HIFU (n=50).⁸¹ The smaller German study compared a different radiofrequency ablation device (n=25) with myomectomy (n=25).⁵⁷^,¹⁵⁴^,¹⁵⁹ One and two year followup data included quality of life, bleeding outcomes, reinterventions, and pregnancy outcomes.

Effects of Radiofrequency Fibroid Ablation on Fibroid Characteristics

The effect of the procedure on fibroid volume was only reported as technical success during the procedure. In the German study, authors reported that radiofrequency ablation successfully excised 71 of 72 fibroids (98.6 percent) in 25 patients.⁵⁷ The Chinese study reported ablation success of 86 percent (by volume) and 90 percent (by diameter) following radiofrequency ablation among 50 women.⁸¹

Effects of Radiofrequency Fibroid Ablation on Bleeding and Pain

At one year post-ablation, 94.4 percent of women (n=21) reported improved or no change in menstrual blood loss as measured by the Menstrual Impact Questionnaire.¹⁵⁹ The German study assessed pain and quality of life outcomes at 2 years. Uterine pain did not improve: 3 of 25 women (12%) reported uterine pain at baseline and 5/21 (24%) reported pain at 2 years after ablation.¹⁵⁴ The authors did not take into account recurrent fibroids. Mean health-related quality of life scores improved from 77.1 at baseline to 89.4 at 2 years (p=0.08).

Effects of Radiofrequency Fibroid Ablation on Pregnancy Outcomes

Three pregnancies, culminating in three live births were noted fin the German study after two years of followup for 21 women.¹⁵⁴

Harms Reported in Studies of Radiofrequency Fibroid Ablation

With the exception of one case of rehospitalization for unexplained vertigo after treatment,⁵⁷ authors reported no complications following radiofrequency ablation treatment.⁸¹

Radiofrequency Fibroid Ablation Summary

We identified two small studies, both poor quality, that evaluated radiofrequency ablation techniques for fibroid symptoms. Two-year followup noted improvements in symptoms and quality of life. This study plans for a total of five years of followup to obtain long-term pregnancy and satisfaction outcomes.⁵⁷^,¹⁵⁶^,¹⁵⁹

Surgical Intervention: Overview

Surgical Interventions: Results

We identified 37 studies¹⁴^,¹⁶^,²⁰^,²²^,⁵⁶^,⁵⁷^,⁶³^,⁶⁸^,⁷²^,⁷⁶^,⁷⁷^,⁸⁰^,⁸⁵^,⁸⁶^,⁸⁹^,⁹⁰^,⁹⁵^,¹⁰⁰^,¹⁰⁴^,¹⁰⁷^,¹¹⁰^,¹¹²^-¹¹⁴^,¹²⁰^,¹²¹^,¹²⁵^,¹²⁶^,¹³⁰^-¹³²^,¹³⁴^,¹³⁵^,¹³⁷^,¹⁴¹^,¹⁴⁸^,¹⁶⁰ evaluating a surgical intervention to treat women with uterine fibroids. Studies were conducted in seven countries (Brazil, China, France, Germany, Italy, Korea, and Taiwan) and randomized 3,172 women with uterine fibroids to endometrial ablation, hysterectomy, or myomectomy. We assessed study quality as good in ten studies,⁵⁶^,⁷⁶^,⁸⁶^,⁸⁹^,⁹⁰^,⁹⁵^,¹¹⁴^,¹²⁰^,¹²¹ fair in ten studies,¹⁶^,²²^,¹⁰⁰^,¹⁰⁴^,¹¹³^,¹³⁰^-¹³²^,¹³⁴^,¹³⁵^,¹⁴¹ and poor in 17 studies.¹⁴^,²⁰^,⁵⁷^,⁶³^,⁶⁸^,⁷²^,⁷⁷^,⁸⁰^,⁸⁵^,¹⁰⁷^,¹¹⁰^,¹¹²^,¹²⁵^,¹²⁶^,¹³⁷^,¹⁴⁸^,¹⁶⁰

Endometrial Ablation

We identified one study of endometrial ablation.¹³² This fair quality study reported amenorrhea, bleeding, hemoglobin, patient satisfaction and the incidence of harms in patients treated by roller-ball (n=54) vs. thermal balloon endometrial ablation (n=42).

Effects of Endometrial Ablation on Bleeding

Menorrhagia, reported by pictorial blood loss chart, decreased significantly (p<0.0001) in both groups from baseline to 12 month after procedure. This patient-reported outcome was confirmed by a clinically significant increase in mean hemoglobin in both groups (p<0.001).¹³² Table 23 reports changes in hemoglobin.

Table 23

Change in hemoglobin with endometnal ablation by method.

Other Treatment Effects of Endometrial Ablation

The rate of reintervention was similar following rollerball (8.3%) and thermal balloon ablation (8.9%). Rates of dissatisfaction were high in both the rollerball (33%) and thermal balloon (39%) groups.¹³²

Harms Reported in Studies of Endometrial Ablation

The endometrial ablation procedure using rollerball necessitates general anesthesia whereas the thermal balloon ablation procedure can be conducted under local anesthesia. The number of intraoperative complications was correspondingly higher in the group who underwent the more invasive procedure (five complications including one case of cervical injury). There were no complications reported during procedures in the thermal ablation group (Appendix G). Overall rates of patient satisfaction were equally poor as noted above.¹³²

Endometrial Ablation Summary

Evidence is insufficient to assess the effectiveness of endometrial abalation to improve fibroid symptoms. Limited data suggest that bleeding outcomes remained improved one year following ablation.

Myomectomy

We included 20 studies (reported in 24 publications) that assessed myomectomy for treatment of uterine fibroids.¹⁶^,⁵⁷^,⁶³^,⁶⁸^,⁷²^,⁷⁶^,⁷⁷^,⁸⁰^,⁸⁵^,⁸⁶^,⁹⁰^,⁹⁵^,¹⁰⁰^,¹⁰⁷^,¹¹⁰^,¹¹³^,¹³⁰^,¹³⁴^,¹⁴¹^,¹⁶⁰ Of these, 15 studies reported final health outcomes following myomectomy for fibroids. Five studies reported harms only.⁷²^,⁷⁶^,⁸⁶^,⁹⁰^,⁹⁵ We considered four RCTs to be good quality, six to be fair quality and ten to be poor quality. Myomectomy techniques include laparoscopic, laparotomy, minilaparotomy, laparoscopically-assisted minilaparotomy, and hysteroscopic approaches.

Effects of Myomectomy on Fibroid Characteristics

Because fibroids are removed at the time of myomectomy, reduction in fibroid and uterine volume are not often reported as an outcome. Fibroid recurrence was reported in seven studies with duration of followup ranging from 6 to 40 months.⁷⁷^,⁸⁵^,¹¹⁰^,¹¹³^,¹³⁰^,¹³⁴^,¹⁴¹ No recurrences were reported in two studies that evaluated 114 women by ultrasonography 6 months following myomectomy.⁸⁵^,¹¹⁰ In five studies, recurrence rates ranged from 2.5 to 24.7 percent (56 recurrences reported in 456 women). Recurrence rates did not differ by type of incision in four studies that compared different surgical approaches.¹¹⁰^,¹¹³^,¹³⁰^,¹³⁴^,¹⁴¹

Effects of Myomectomy on Bleeding

One trial reported absence of heavy menstrual bleeding at 12 months after the procedure for 12 of 15 (87%) women who had reported it as a problem at baseline.¹⁵⁹ Another noted persistent abnormal menstruation in three (1.9%) women who had myomectomy plus uterine artery occlusion at 2-year followup.⁷⁷ No other studies reported changes in symptomatic fibroid-related bleeding, such as heaviness of menses or total days of bleeding. Lack of this outcome means evidence is insufficient for determining if myomectomy improves an extremely common concern among women who seek intervention for fibroids.

Other Treatment Effects of Myomectomy

Return to Usual Activity

Recovery time ranged from 15 days up an average of 30 days as reported in three studies.¹¹⁰^,¹¹³^,¹⁴¹ Type of incision was a major determinant in recovery time in two studies with more women reporting feeling fully recovered by day 15 following laparoscopy compared with minilaparotomy¹¹⁰ or abdominal myomectomy.¹⁴¹ Recovery time from myomectomy averaged 30 days in another small study.¹¹³

Quality of Life and Symptom Status

Improvement in quality of life or symptom status was reported in five studies (7 publications).¹⁶^,²¹^,⁵⁷^,⁷⁷^,⁸⁰^,¹⁵⁴^,¹⁵⁹ Quality of life significantly improved in three studies that assessed it. The FUME trial comparing myomectomy to UAE reported improved quality of life for both groups after one year as measured by the UFS-QOL.¹⁶ A large Chinese study reported improvements in all four domains (physical, psychological, environment, and social relationship) after two years following laparoscopic uterine artery occlusion plus myomectomy using the abbreviated World Health Organization Quality of Life (WHOQOL-BREF) measure. A small study of 25 women documented significant improvements in health related quality of life scores after two years as measured by UFS-QOL (p=0.04) and EQ-5D (p=NR).¹⁵⁴

Measures of symptom status were reported in two studies. Symptom relief from six symptoms including heavy menstrual bleeding, dysmenorrhea, dyspareunia, pelvic pain, dysuria, and pressure improved for 88 percent (51/58) of women after six months assessed by a questionnaire Symptom improvement including constipation, urinary frequency and menorrhagia improved for 85 percent of women postoperatively in a study evaluating loop ligation for women with larger fibroids.⁸⁰ Patient satisfaction with scarring measured using a visual analog scale was comparable in a single study that compared single-port laparoscopically- assisted transumbilical ultraminilaparotomic with single-port laparoscopic myomectomy. ¹⁶⁰

Effects of Myomectomy on Pregnancy Outcomes

Fertility and Pregnancy

Reproductive outcomes were reported in five studies.²¹^,¹⁰⁰^,¹⁰⁷^,¹³⁴^,¹⁵⁴ One study described no significant difference in pregnancy outcomes following laparoscopic myomectomy (54%) compared with myomectomy by laparotomy (56%).¹³⁴ An RCT (n=136) compared myomectomy approaches (laparoscopic vs. minilaparotomic) and assessed 12-month reproductive outcomes for a total of 556 and 669 cycles, respectively.¹⁰⁰ There was no significant difference in the cumulative pregnancy rate, or the cumulative live-birth rate. However, the time to first pregnancy (5 months vs. 6 months) and live birth (14 months vs.15 months) was lower after laparoscopic than minilaparotomic myomectomy(p<0.01). A post-hoc subgroup analysis stratified fibroid patients according to indication: symptoms vs. unexplained infertility. In patients without infertility, the cumulative pregnancy rate, pregnancy rate per cycle (11.1% vs. 5.4%) and live-birth rate per cycle (9.9% vs. 4.8%; p<0.05) were significantly higher following laparoscopic than minilaparotomy myomectomy. In patients without infertility, the times to first pregnancy and live birth were significantly lower after laparoscopic than minilaparotomic approach. In patients with unexplained infertility, no difference in any reproductive outcomes assessed was observed between the two myomectomy approaches.¹⁰⁰

One RCT enrolled 181 women with fibroids and infertility (trying to conceive for at least 1 year without success) then subdivided the women according to the location of the fibroid (submucous, intramural, subserosal) and randomized them to myomectomy or no surgery.¹⁰⁷ For women with subserosal or intramural fibroids, there was no significant difference in the pregnancy rate, comparing myomectomy with no treatment. For women with submucous fibroids, the group who underwent myomectomy had a greater pregnancy rate (40.4%) than those who did not undergo surgery (21.4 percent).¹⁰⁴ A subset of women from a RCT comparing myomectomy to UAE among women with reproductive plans noted that among women who attempted to conceive following myomectomy, 31 of 40 were pregnant at 13 months after fibroid removal and the delivery rate was 47.5 percent (19/40).²¹ In a small study comparing ablation (n=26) to myomectomy (n=25), there were six pregnancies culminating in four live births, one induced abortion, and one ongoing pregnancy among twenty-two women after 2 years of followup after myomectomy.¹⁵⁴

Harms Reported in Studies of Myomectomy

Transfusion

Following treatment for fibroids with myomectomy, transfusion rates were most often zero (1,040 participants in 10 studies).⁶³^,⁷²^,⁷⁶^,⁸⁰^,⁸⁵^,⁹⁵^,¹⁰⁰^,¹⁰¹^,¹¹⁰^,¹³⁰^,¹³⁴ Six studies⁶⁸^,⁷⁶^,⁸⁰^,¹¹³^,¹³⁴^,¹⁶⁰ reported 25 transfusions among 502 participants treated by myomectomy (Appendix G). Seven studies did not report transfusion following myomectomy.¹⁶^,⁵⁷^,⁷⁷^,⁸⁶^,⁹⁰^,¹⁰⁷^,¹⁴¹ One study reported no significant difference in transfusion outcomes following laparoscopic myomectomy compared with laparotomic myomectomy.¹³⁴ One study (n=166) reported no significant difference in transfusion outcomes following laparoscopic myomectomy plus uterine artery occlusion compared with laparoscopic myomectomy alone.⁷² One study (n=384) reported fewer transfusions following gasless laparoscopic myomectomy compared with conventional laparoscopic myomectomy.⁷⁶ Postoperative transfusion risk was found to be comparable in one study comparing a single-port laparoscopically-assisted transumbilical minilaparotomy approach (n=3, 6.5%) with single-port laparoscopic myomectomy (n=4, 8.7%).¹⁶⁰

Other Surgical Complications

One study (n=80) comparing isobaric laparoscopically-assisted myomectomy with myomectomy via minilaparotomy reported no organ perforations in either group, but was underpowered to assert comparability.⁸⁶ Another study (n=148) reported one small bowel injury and one conversion to laparotomy due to hemostasis difficulties in the laparoscopic myomectomy group while no complications were noted in the minilaparotomy group.¹¹⁰

Readmission

One study (n=80) comparing isobaric laparoscopically-assisted myomectomy with myomectomy via minilaparotomy reported no readmissions or reoperations, but was underpowered to assert comparability.⁸⁶ Another study (n=148) reported that one woman had acute peritonitis and underwent abdominal surgery 10 days after laparoscopic myomectomy, with no complications following minilaparotomic myomectomy.¹¹⁰ The Mara trial²¹ (n=121) reported just one readmission (1.6%) within 30 days after myomectomy.

Reintervention and Recurrence

One study (n=136) reported technical failure (conversion to laparotomy) was more common after myomectomy by minilaparotomy (9%) compared with laparoscopic myomectomy (0%) (p=NR).¹⁰⁰ The reintervention rate was 3.2 % due to fibroid recurrence in another study including 121 women. ²¹ One woman required hysterectomy 7 months after myomectomy in a second study including 163 women.¹⁶

Total Complications

In one study (n=136) that reported total complications, these were higher among mini-laparotomy (16%) compared to those who had laparoscopic myomectomy (3%).¹⁰⁰ The FUME trial¹⁶ (n=163) reported six major complications (8%), all occurring during the hospital stay after myomectomy while another trial (n=121) reported “unexpected intrauterine penetration” (not defined) after myomectomy in three cases (5%).²¹

Myomectomy Summary

There is low strength of evidence that women had improved quality of life following myomectomy. Evidence is insufficient to determine if myomectomy improves bleeding. This procedure is an option for women desiring future fertility, but evidence is insufficient to define the potential benefit. There is some risk of fibroid recurrence (reported ranges from 0 to 25%) that does not vary by type of surgical technique.

Hysterectomy

We identified 14 studies reported in 23 publications ¹⁹^,²⁰^,²²^,⁵⁶^,⁷⁰^,⁷⁷^,⁸⁹^,⁹¹^,⁹³^,⁹⁶^,⁹⁷^,⁹⁹^,¹⁰⁹^,¹¹¹^,¹¹²^,¹¹⁴^,¹²⁰^,¹²¹^,¹²⁵^,¹²⁶^,¹³¹^,¹³⁵^,¹³⁷ assessing hysterectomy in women with uterine fibroids. Seven reported harms only (i.e., did not report final health outcomes for effectiveness).⁵⁶^,⁸⁹^,¹²⁰^,¹²⁵^,¹²⁶^,¹³¹^,¹³⁵ One study¹³⁷ did not report results for the women who were randomized to immediate hysterectomy; the results from the comparator arm (women treated with a GnRH) is reported in the medication section above and in the comparative effectiveness section below.

Studies addressed the following interventions: intrafascial supracervical hysterectomy,⁷⁷ laparoscopic assisted vaginal hysterectomy,⁵⁶^,⁸⁹^,¹²⁰^,¹²⁶^,¹³¹^,¹³⁵ total laparoscopic hysterectomy,⁵⁶^,¹²⁵, vaginal hysterectomy,⁵⁶^,⁸⁹^,¹¹²^,¹²⁰^,¹²¹^,¹³¹^,¹³⁵ and total abdominal hysterectomy.²²^,¹¹²^,¹²⁰^,¹²¹^,¹²⁵ Three studies used more than one hysterectomy approach (i.e., type and route not standardized) or did not describe the type of hysterectomy.²⁰^,¹¹⁴^,¹³⁷ Assessment duration (where clearly reported) ranged from 15 days to 24 months. We assessed five studies as good quality,⁵⁶^,⁸⁹^,¹¹⁴^,¹²⁰^,¹²¹ three as fair quality,²²^,¹³¹^,¹³⁵ and six as poor quality for effectiveness outcomes.²⁰^,⁷⁷^,¹¹²^,¹²⁵^,¹²⁶^,¹³⁷

Effects of Hysterectomy on Bleeding

Two studies reported increases in hemoglobin levels at 24 months after surgery.²⁰^,¹¹⁴ The mean increase reported in the EMMY trial was statistically significant (2.03 g/dL; 95% CI, 1.44 to 2.61, p<0.0001).

Effects of Hysterectomy on Pain

Two studies reported pain or pressure outcomes. At 24 months after hysterectomy, 28 women (93%) reported total or substantial improvements in pressure symptoms. Lower abdominal pain decreased from 16 participants (53%) at baseline to three (10%) after 2 years.²⁰ A majority (78%) of the women who had lower abdominal pain at baseline reported improvement at 24 months after hysterectomy.⁹⁹ Of those without pain at baseline (n=14), one woman reported worsening of symptoms after hysterectomy.⁹⁹

Other Treatment Effects of Hysterectomy

Return to Usual Activity

Mean time to return to work or usual activity ranged from 22 to 41 days as reported in four studies.²⁰^,²²^,¹²⁰^,¹²⁵ Women who received a vaginal or laparoscopic vaginal hysterectomy had a significantly faster recovery (mean 22 to 30 days) compared with total abdominal hysterectomy (mean 36 to 41 days) as reported in two studies¹²⁰^,¹²⁵ Mean recovery time for abdominal hysterectomy averaged 37 days in one study²² and 35 days in another study that used a variety of surgical procedures.²⁰ We did not assess the strength of evidence for return to usual activities

Quality of Life and Symptom Status

Improvements in quality of life or symptom status were reported in four trials. Two studies reported short terms results (1-month after surgery) ¹¹²^,¹²¹ and two studies reported results after 24 months.⁷⁷^,⁹³ Short-term results compared vaginal to abdominal hysterectomy in 179 women. Quality of life assessed by the SF-36 was better after vaginal hysterectomy than abdominal hysterectomy.¹¹² Patient satisfaction with treatment assessed using an unvalidated questionnaire was higher for women after vaginal than abdominal hysterectomy.¹²¹ After two years, quality of life was significantly improved after hysterectomy in the EMMY trial⁹³ assessed by HRQOL and SF-36. A Chinese study reported improved quality of life two years after hysterectomy as measured by the WHOQOL-BREF questionnaire.⁷⁷

Harms Reported in Studies of Hysterectomy

Transfusion

The proportion of women requiring transfusion following hysterectomy ranged from zero to 20 percent in 890 women from 11 studies (Appendix G, Table G-9). Six studies reported no significant difference in transfusion outcomes following different hysterectomy approaches for fibroids: laparoscopic-assisted vaginal versus total abdominal,¹³⁵ laparoscopic-assisted vaginal versus vaginal ⁵⁶^,¹³¹ laparoscopic-assisted vaginal versus total laparoscopic,⁵⁶ vaginal versus total abdominal, ¹¹² vaginal versus total laparoscopic, ⁵⁶ and total laparoscopic versus total abdominal,¹²⁵ Authors reported 5 percent (3/61) of patients randomized to laparoscopic hysterectomy with bipolar coagulation of uterine vessels or laparoscopic hysterectomy alone required transfusion.¹²⁶

Thromboembolism

Two thromboembolic events (one pulmonary embolism and one deep vein thrombosis) were reported following hysterectomy in 227 patients from four studies that reported thromboembolism after hysterectomy (Appendix G, Table G-7).²²^,¹¹⁴^,¹²¹

Perforation of Organs

Two studies (n=179) reported no organ perforation following different hysterectomy interventions for fibroids.¹¹²^,¹²¹ One RCT (n=122) reported that one woman randomized to total laparoscopic hysterectomy was converted to total abdominal hysterectomy because of intraoperative bowel injury.¹²⁵

Hysterectomy Summary

There is low strength of evidence that women had improved quality of life following hysterectomy, regardless of surgical approach. Overall, patient satisfaction and recovery time following hysterectomy was better for women who received a vaginal hysterectomy compared with total abdominal hysterectomy, but the strength of this evidence was not graded. Harms, including the need for blood transfusion and organ perforation, were similar for all types of hysterectomy.

Hysterectomy or Myomectomy

We found three studies reported in six publications (one good quality¹⁰⁴ and two poor quality¹⁴^,¹⁴⁸) that randomized women to an arm that allowed either myomectomy or hysterectomy.

Quality of life was the primary outcome for two studies comparing UAE to surgery (myomectomy or hysterectomy). ¹⁴^,¹⁰⁴ In the REST trial, scores on the SF-36 and the EQ-5D for the surgery groups were significantly improved after 12 months.¹⁷^,⁷¹^,⁸²^,¹⁰⁴ In the trial by Jun and colleagues, there was limited improvement for the surgery group after 6 months as assessed by the SF-36 questionnaire. Scores were improved for the mental health and vitality measures only, while physical health, social function and emotional role were almost unchanged.¹⁴

One study compared a GnRH agonist to myomectomy or hysterectomy.¹⁴⁸ Symptom improvement and fertility outcomes were reported.¹⁴⁸ Three of five women who underwent myomectomy for infertility had term pregnancies.

Surgical Intervention Summary

The strength of evidence was low for improvement in quality of life after myomectomy and hysterectomy. We found insufficient evidence that myomectomy improves bleeding. Evidence is insufficient to assess the effectiveness of endometrial abalation in improving fibroid symptoms. With notable exceptions,¹⁶^,⁸²^,⁹⁷ the majority of these studies did not follow patients beyond the postoperative period. Therefore, many studies did not report important patient outcomes such as change in fibroid-related pain or bleeding. Many of the studies with surgical or procedural interventions reported intermediate outcomes such as technical success, hospital length of stay, or estimates of blood loss related to the invasive procedure (e.g., postoperative hemoglobin, intra- or postoperative transfusion rate). While these are important measures, they do not provide evidence about patient-centered outcomes of surgery for fibroids such as relief from symptoms, improvement in quality of life, and sexual function.

Comparative Effectiveness Studies

Content of Literature for Comparing Effectiveness of Interventions

Direct comparisons of alternative treatment strategies are crucial to support informed decisions by women and their care providers. Direct comparative effectiveness studies provide unbiased comparisons by randomly assigning similar groups of women to different interventions and assessing the same outcomes with the same yardstick. When different approaches within the same category of intervention were compared (e.g., uterine artery occlusion vs. UAE, or myomectomy via laparoscopy vs. minilaparotomy) those findings are reviewed in the previous sections for each category.

In this section, we review comparisons across categories of interventions. In total we found 17 studies which compared the effectiveness of different categories of interventions.¹⁴^,¹⁶^,²⁰^,²²^,⁵⁷^,⁶⁸^,⁷³^,⁷⁷^,⁸¹^,⁸⁴^,¹⁰²^,¹⁰⁴^,¹¹³^,¹¹⁴^,¹²⁷^,¹³⁷^,¹⁶⁵ Available comparisons are summarized in Table 24.

Table 24

Randomized trial comparisons across categories of interventions.

Comparisons of Medical Management

We identified four studies designed to compare outcomes across different categories of drugs.⁷³^,⁸⁴^,¹⁰²^,¹⁵² One good quality study compared ulipristal acetate to leuprolide.⁷³ Two compared a GnRH agonist to cabergoline, which is a dopamine receptor agonist.⁸⁴^,¹⁰² The fourth compared the levonorgestrel IUD to daily oral progesterone.¹⁵² A single study randomized women interested in hysterectomy to immediate hysterectomy or goserelin to determine if medical management allowed some women to avoid surgery.¹³⁷ Overall, these studies were small and inadequately powered to provide definitive evidence; they are briefly described below.

Comparison of GnRH Agonist to Ulipristal Acetate

A single good quality, multisite European trial compared two doses of daily oral ulipristal to monthly injection of 3.75 mg leuprolide.⁷³ They recruited 307 women whose bleeding was severe enough that they were considering surgery. The three groups had similar outcomes for change in size of the three largest fibroids and improvements in pain and quality of life measures. The overall reduction in uterine size was superior for leuprolide. Across three arms, 5 mg and 10 mg of ulipristal achieved results at least as good, or better, than leuprolide for control of bleeding. In 90 percent, 98 percent, and 89 percent, respectively bleeding improved to normal as assessed by the standardized and validated pictorial blood loss assessment chart. The average time to absence of bleeding in women with complete cessation was shorter for ulipristal than leuprolide (5-7 days vs. 21 days) (p<0.001). Hot flashes were substantially more common among those receiving leuprolide (40% vs. 10%-11% in ulipristal arms). At 13 weeks the proportion of adverse events and individuals discontinuing medication were equivalent, though women in the leuprolide group had evidence of greater bone resorption as measured by biomarkers. Endometrial samples were benign in both groups. Forty-nine percent of these participants who had initially sought evaluation for surgery had not had surgery by the end of 6 months of followup. In a subgroup of these women, growth of fibroids at one month had resumed in some women who received leuprolide while at 6 months fibroid size remained reduced in the majority of women who received ulipristal. In summary, evidence was insufficient to choose between ulipristal and leuprolide, based on a single study with small numbers.

Comparison of GnRH Agonist to Cabergoline

The GnRH agonist used in these two small, poor quality studies (Diphereline) is not available in the United States, however similar drugs are. ⁸⁴^,¹⁰² Cabergoline is available in the United States for off-label use. In both trials, GnRH was given for one month and cabergoline for 6 weeks of treatment. Fibroid volume decreased in both groups and was not statistically different.⁸⁴^,¹⁰² ⁸⁴^,¹⁰² ⁸⁴^,¹⁰² ⁸⁴^,¹⁰² ⁸³^,¹⁰¹⁷⁹^,⁹⁷ Those on GnRH agonists were more likely to stop bleeding and to have hot flashes. In both arms bleeding days decreased compared with baseline, with the fewest days in the GnRH agonist arm.⁸⁴^,¹⁰² Pain was similarly reduced in both groups.⁸⁴^,¹⁰² Side effects of GnRH matched those reviewed above; cabergoline was associated with headaches and nausea in the first week of use.⁸⁴^,¹⁰² Carbergoline side effects caused one woman to discontinue treatment, compared with none in the GnRH group.⁸⁴^,¹⁰²

In summary, evidence was insufficient to choose between the GnRH agonist and cabergoline but them.

Comparison of Oral Versus IUD Delivered Progesterone

One poor quality Turkish study randomly assigned 30 women in each arm to daily norethindrone acetate or levonorgestrel IUD (LNG-IUD) among premenopausal women with fibroids who had declined surgery.¹⁵² Visual blood loss scores improved by 6 months in both groups, with significantly greater improvement in the LNG-IUD group (p=0.028). Improvement in hemoglobin likewise occurred in both groups with a significantly greater improvement among those with an LNG-IUD (p<0.01). Women with the LNG-IUD were more satisfied and more likely to continue treatment than the oral treatment group.

This trial suggests that both norithindrone acetate and the LNG-IUD might improve bleeding even among women whose fibroid symptoms were considered appropriate for surgical intervention. However, evidence to choose between them is insufficient.

Comparison of Goserelin With Immediate Hysterectomy

Seventy-two premenopausal women older than 45, with at least one fibroid larger than 10 cm in diameter, heavy menses for three months or longer, and who were eligible for hysterectomy were randomized to immediate surgery (n=13) or treatment with a GnRH agonist (n=59) (goserelin 3.6 mg depo injection each month for three months) ¹³⁷ Participants were followed for 3 years and allowed up to two more rounds of treatment if their bone mineral density was not at risk. Those on medication could opt for hysterectomy at any time. Twenty three of the 59 women assigned to medication, 39 percent (95% CI, 26 to 62), had continued to hysterectomy by the end of the third year.

In summary, some women may be able to use GnRH agonist therapy to delay hysterectomy. The quality of this study was judged to be poor in part because of lack of researcher and participant masking to assigned intervention; if participants were inclined to want surgery this bias would tend to increase the proportion who opted for hysterectomy rather than forestall that choice. The evidence is intriguing but insufficient to encourage selection of medication if hysterectomy is planned.

Comparisons of Procedures

Two studies compared HIFU to alternate interventions: one to radiofrequency ablation⁸¹ and another to myomectomy.⁶⁸ The MRI guided HIFU system that is approved in the United States was not used in any direct comparison studies. Other direct comparisons of procedures included two comparisons of UAE to myomectomy,¹⁶^,¹¹³ three comparisons of UAE to hysterectomy,²⁰^,²²^,¹¹⁴ and two comparisons of UAE to surgery (myomectomy or hysterectomy).¹⁴^,¹⁰⁴

Comparisons of HIFU for Fibroid Ablation to Radiofrequency Ablation

One RCT (n=100) of premenopausal women compared HIFU to radiofrequency ablation. However, the study reported only technical success (58% for HIFU vs 90% with radiofrequency ablation) Followup of fibroids and symptom resolution were not studied so evidence is insufficient to guide choice among these options.⁸¹

Comparisons of HIFU for Fibroid Ablation to Myomectomy

One RCT randomized women to HIFU vs. myomectomy. Fifty-five women were randomized to each arm; only results for women who completed followup (n=48 HIFU and n=52 myomectomy) are reported. Sexual function was comparable in both groups after 6 months of followup. Women having HIFU had shorter hospitalizations, earlier ambulation, and faster recovery; they had no blood loss and no surgery-related symptoms.⁶⁸^,⁶⁹ Fibroid outcomes and symptom resolution were not studied, therefore, evidence is insufficient to help women choose between options.

Comparisons of UAE to Myomectomy

Two European RCTs reported in three publications¹⁶^,²¹^,¹¹³ compared UAE to myomectomy. The studies included 284 women ages 31 to 50 years; 96 percent of participants had symptoms, some had only fertility concerns. In total there were 121 embolizations and 122 myomectomies with followup. Technical success was similar.²¹ as were improvements in quality of life (measured by UFS-QOL) one year after treatment, ¹⁶ and symptom relief (88% at an average of 17 months after intervention.)¹¹³

Length of stay is significantly shorter for UAE (averaging 2 days) compared with myomectomy (4 to 6 days). Blood loss was associated only with myomectomy.¹⁶^,²¹ Time away from work was at least 10 days shorter for those who had UAE compared with myomectomy.²¹

Around 2 years of followup, women in the UAE group had experienced more subsequent interventions (re-embolizations and myomectomies) compared with the myomectomy group (32.8% vs. 3.2%; p<0.0001).²¹ However, subsequent intervention in the UAE group could have been driven by a protocol to recommend myomectomy if there was not reduction in fibroid size by 6 months or if or if a fibroid greater than 5 cm in size persisted. Risk of fibroid recurrence, satisfaction and quality of life are similar across groups. Thus, the likely cause of additional interventions among those with UAE was linked to the clinical protocol in this trial.

Reproductive outcomes were reported for 66 women (26 after UAE and 40 after myomectomy) in one study.²¹^,¹¹³ After 2 years pregnancy was significantly more common after myomectomy (78% vs. 50%; p<0.05) while miscarriage risk was higher after UAE (64% vs. 23%; p<0.05). Nineteen percent of women who had UAE had live births compared with 48 percent after myomectomy (p<0.05).²¹ All participants in this study received care in a fertility clinic, for other underlying causes of infertility and treatment options included in vitro fertilization. As a result findings may not represent the general population of women with fibroids who plan to conceive.

In summary, two small studies of fair quality suggest UAE achieved similar results to myomectomy with a shorter recovery time. Because of low power to detect differences in outcomes, the evidence is insufficient to determine if outcomes are better after myomectomy compared to UAE. Women who have UAE may be more likely to have future interventions though this may have been driven by clinical protocol in these trials. Evidence is insufficient to guide choice between these options.

Comparisons of UAE to Hysterectomy

Three RCTs comparing UAE with hysterectomy are published in 12 reports. Study participants were all from Europe. The EMMY trial¹⁹^,⁷⁰^,⁹¹^,⁹³^,⁹⁶^,⁹⁷^,⁹⁹^,¹⁰⁹^,¹¹¹^,¹¹⁴ is a multicenter trial (28 hospitals) conducted in the Netherlands. A Spanish trial by Pinto and colleagues²² and a trial from Finland by Ruuskanen and colleagues²⁰ were single-center trials. Combined, these studies included 291 women aged 33 to 57 years with symptomatic uterine fibroids, who were candidates for hysterectomy. The EMMY trial excluded women with submucosal leiomyoma while the Spanish trial excluded women with leiomyomas larger than 10 cm in diameter. The third trial ⁶² did not exclude women based on the size or location of fibroids. Across studies, 153 women were allocated to UAE and 138 to hysterectomy. Followup duration of these three trials varied from 6 months²² to 2 years²⁰^,²² and 5 years.¹⁹

Early Procedure Outcomes and Recovery

Blood loss from the procedure was negligible for UAE and higher for hysterectomy.¹¹⁴ In the surgical group of the Finnish trial,²⁰ 19 percent of the vaginal or laparoscopic hysterectomies were converted to abdominal hysterectomy due to technical difficulties; and the EMMY trial reported four conversions of laparoscopic or vaginal hysterectomy to laparotomy (5.3%)¹¹⁴ Length of hospital stay was shorter (p<0.001) for UAE (1.3 to 1.7 days) compared with hysterectomy (3.5 to 5.9 days).²⁰^,²²^,¹¹⁴ Re-admission rates were the same (5%) in both arms of the Spanish study²² while the EMMY trial reported significantly higher near term re-admission rates after UAE (11% vs. 0%, p<0.003) compared with hysterectomy.¹⁰⁹^,¹¹⁴ Despite this, time to return to usual activities was significantly (p<0.001) shorter in the UAE group (fewer than 20 days) compared with hysterectomy group (more than 30 days).²⁰^,²²^,¹¹⁴

Fibroid Characteristics

By definition, women who have hysterectomy have 100% reduction in fibroids. For the UAE groups, effects on fibroid characteristics are included in Table 16.

Bleeding

By definition, women who have hysterectomy stop bleeding. Improvement in bleeding in the UAE arms is included in Table 17. Increase in hemoglobin levels from baseline at two years was significantly higher for the hysterectomy group compared with the UAE group (+2.03 vs. +1.37 g/dL; p=0.037) in EMMY⁹⁹ but not in the Finnish trial (p=0.16),²⁰ which may reflect inadequate power to detect modest differences.

Symptoms

Overall relief of symptoms was good across groups (82% UAE vs. 93% hysterectomy; p=0.173) in the Finnish study at 2 years.²⁰ For both groups, there was a significant decrease in pain from baseline at all time points (p≤0.03) and by 24 months women reported similar improvement of pain (84.9% after UAE vs. 78% after hysterectomy; p=0.30).⁹⁹ Greater improvement in pressure symptoms was reported after UAE compared with hysterectomy (95% vs. 69%; p=0.03).²⁰ More urinary bladder symptoms were reported after hysterectomy than after UAE (30% vs. 7%; p=0.03).

Quality of Life

In the EMMY trial, overall health related quality of life improved in both trial arms and was comparable across groups by completion of 5 years of followup.¹⁹ One secondary outcome (change in bowel movements) was worse in the hysterectomy group than UAE at 5 years from baseline (p=0.01) while the UAE group had improvement from 6 month onwards.¹⁹ At 6 months, body image scores improved significantly more in UAE group than in hysterectomy group (p=0.02), but there was no group difference in body image or sexual function scores at 24 months or later.¹⁹

Subsequent Treatment

Three studies assessed subsequent treatment at 6 months to 5 years. Subsequent treatment was higher in the UAE group than in the hysterectomy group at each time point in followup.¹⁶^,²⁰^,¹¹⁴ At 6 months, 1-year, 2-year, and 5-year followup, the EMMY trial reported significantly higher rates of any subsequent gynecologic intervention (9.8%, 17.3%, 28.3%, and 34.6%) after UAE versus 1.3 percent, 5.3 percent, 8.0 percent, and 10.7 percent after hysterectomy.¹⁹ The risk of additional intervention was reported at 2-year followup in another study (19.2% with UAE and 10.3% after hysterectomy; p=0.24).²⁰ Some early reinterventions are repeat UAE, myomectomy, or hysterectomy among the small percentage of women (1.0% to 4.9%) for whom effective occlusion of the uterine vessels could not be achieved at the time of the initial procedure.¹⁶^,¹¹⁴ These data are included in the meta-analysis of subsequent intervention section.

Compared to women with UAE, women who received hysterectomy reported higher satisfaction rates after 1 (p=0.001) and 2 years (p=0.02) but satisfaction levels for both groups were comparable by 5 years (p=0.13) (Table 25). Forty six percent of women after UAE and 56 percent of women after hysterectomy were very satisfied with the outcome by 5 years (overall satisfaction, 84% vs. 88%; p=0.37).

Table 25

Satisfaction in studies of uterine artery embolization versus hysterectomy.

Reproductive Status

Ovarian reserve, pregnancy, and sexual function related outcomes were reported only in the EMMY trial.⁹⁷ Reduction in ovarian reserve was based on FSH and AMH hormone levels. Levels of FSH, suggesting diminished reserves, increased from baseline (p=0.001) in both groups at 24 months followup but there was no group difference (p=0.32). The number of women with FSH >40 IU/L, confirming menopause, at 24 months was 14/80 (18%) in the UAE group compared with 17 women (21%) in the hysterectomy group, with no differences over continued followup (p=0.37). The change scores of AMH levels from baseline were significantly different between the groups only at 6 weeks of treatment (p=0.005) and the AMH levels remained decreased only in the UAE group during the followup.

Harms

Harms are reviewed for each category of intervention within the sections for UAE and specific surgeries above and in Appendix G.

Summary

Compared to hysterectomy, UAE provides similar relief of pressure and pain symptoms, similar improvements in quality of life, and fewer side effects. Though subsequent intervention is more common after UAE than hysterectomy, the majority (66%) of women randomly assigned to have UAE avoided hysterectomy for the duration of followup, which included up to 5 years of surveillance in the largest study. However, at present evidence is insufficient to shape clinical decisions.

Comparisons of UAE to Patient Choice of Myomectomy or Hysterectomy

A larger trial of good quality conducted in the United Kingdom¹⁰⁴ and a poor quality study conducted in China¹⁴ made this comparison. Combined, these studies randomly assigned 169 women to UAE and 115 women to conventional surgeries for uterine fibroids.¹⁴^,¹⁰⁴ Both found UAE to have shorter hospital stay and recovery time. UAE was associated with fewer complications at the time of the procedure but a greater proportion of women had a subsequent intervention in followup. Similar to the studies detailed above, the study that included 1-year followup reported that among women assigned to UAE, 12.6 percent had subsequent interventions for inadequate control of symptoms. In the UK trial, quality of life (measured by the SF-36) at 1 month the UAE group had significant improvements in multiple areas of function, however by one year improved similarly in both groups.

In summary, these studies contribute to a growing body of evidence that UAE has a shorter recovery and is as effective as surgery in major domains of patient outcomes. The primary caveat is that this comes with risk of subsequent intervention, which is also addressed in our meta-analysis.

Comparisons of Surgeries

The majority of surgical studies that made comparisons did so within a category of intervention, for instance comparing myomectomy conducted by laparoscopy to myomectomy through a laparotomy incision. These are reviewed within categories of interventions above and when clinically and statistically significant advantages have been demonstrated they are noted. Comparisons across types of surgical interventions were meager, and the three related studies are noted below.

Comparison of Laparoscopic Bipolar Coagulation With and Without Uterine Nerve Ablation

In this small (n=85), poor quality study of women with fibroids and menstrual pain, 41 women were randomly assigned to also have laparoscopic uterine nerve ablation at the time of laparoscopic occlusion (bipolar coagulation) of the uterine vessels.¹²⁷ Both groups had equal reduction in size of fibroids and in reduction of bulk symptoms. Women in the nerve ablation group had less postoperative pain at 1 month as measured by a non-standardized five-point scale (p<0.05) and by 6 months painful menses were improved in 92.1 percent of the nerve ablation group compared with 73.8 percent of the occlusion only group (p<0.05).¹²⁷ The results are insufficient to inform a decision to include uterine nerve ablation at the time of coagulation of the uterine vessels.

Comparisons of Radiofrequency Ablation With Laparoscopic Myomectomy

A small study, reported in three publications, compared laparoscopic use of radiofrequency (n=26) to laparoscopic myomectomy (n=25).⁵⁷^,¹⁵⁴^,¹⁵⁹ Ultrasound was performed during the procedures to definitively identify fibroids and document immediate intervention outcomes. Intermediate outcomes (technical success, blood loss and length of stay) were better in the radiofrequency group compared with the myomectomy group ⁵⁷ At 1 and 2 years, both groups experienced improvements in health related quality of life and symptom severity as assessed by UFS-QOL and EQ-5D scores. The majority of women were moderately or very satisfied with their treatment at 12 months (86% in both groups).¹⁵⁹ Three pregnancies, culminating in three live births were noted for the radiofrequency ablation group and six pregnancies (four live births, one induced abortion, and one ongoing) were reported for the myomectomy group.¹⁵⁴

Radiofrequency appears to offer some advantages to myomectomy, but there is insufficient evidence to determine clinical decisions.

Comparisons of Myomectomy Plus Uterine Artery Occlusion with Hysterectomy

A single moderate size study of poor quality conducted in China compared laparoscopic uterine artery occlusion (n=158) plus myomectomy to supracervical hysterectomy (n=174).⁷⁷ Supracervical hysterectomy is rarely performed in the United States. This RCT assessed quality of life using the WHOQOL-BREF. At two months after surgery, physical and social domains were superior in the myomectomy plus occlusion compared with the hysterectomy groups; no other areas were meaningfully different. By two years, myomectomy plus occlusion was superior to hysterectomy in all domains except environment (p<0.01) and both study groups had meaningful improvements compared with their baseline. Women treated with myomectomy plus uterine artery occlusion had a 2.5 percent fibroid recurrence rate.

In summary this single study provides insufficient strength of evidence that myomectomy plus uterine artery occlusion differs from supracervical hysterectomy for improving quality of life in physical, psychological, and relationship domains.

Comparison of Transfusion Requirements

Transfusion was reported in 40 arms across 23 studies. Transfusions by intervention category are summarized in Table 26.

Table 26

Transfusion by intervention category.

Analysis of Subsequent Treatment Following Initial Treatment for Uterine Fibroids

We estimated the probabilities of subsequently receiving additional treatment for fibroids after randomization to a given initial treatment of medication, UAE, or myomectomy for uterine fibroids from data reported in in 46 studies (Table 27).¹³^-¹⁶^,²⁰^,²²^,²³^,⁵⁴^,⁵⁷^,⁶⁴^,⁶⁶^,⁷⁹^,⁸⁰^,⁸⁴^,⁸⁵^,⁸⁷^,⁹²^,⁹⁸^,¹⁰⁰^,¹⁰²^,¹⁰⁴^,¹⁰⁵^,¹⁰⁷^,¹¹⁰^,¹¹³^,¹¹⁴^,¹¹⁸^,¹²²^,¹²³^,¹²⁷^-¹²⁹^,¹³⁴^,¹³⁶^,¹⁴⁰^,¹⁴⁸^,¹⁵⁰^,¹⁵¹ Subsequent treatments were grouped into these categories: 1) no intervention; 2) LNG-IUD; 3) UAE; 4) HIFU for fibroid ablation; 5) myomectomy, and 6) hysterectomy (Appendix H). Rates of subsequent intervention ranged from zero up to 44 percent. For women in their 30s, the model predicted that the probability of subsequent intervention for fibroids over 2 years varied from 6 to 7 percent after medical treatment or myomectomy to 44 percent after UAE. For women in their 40s and 50s, modelled 2-year reintervention rates were 9 to 19 percent following medical treatment or UAE, and 0 percent after myomectomy. Overall, fewer than half of women had another intervention within 24 months. Rates of subsequent intervention were lowest for initial medical management and higher following myomectomy or UAE. UAE was most often followed by myomectomy among those in their 30s and by hysterectomy among those in their 50s. Younger women who initially had myomectomy were most likely to have repeat myomectomies over the 2 years of followup. After medical treatment, few (6-11 percent) women in any age group had subsequent treatment within 2 years. Appendix G includes data for subsequent treatment at up to 6 and 12 months of followup.

Table 27

Estimated probability of subsequent treatment by age at up to 24 months of followup.

Key Question 2. Influence of Patient/Fibroid Characteristics on Effectiveness

Key Points

Among 97 RCTs of interventions, none were explicitly designed to address whether intervention effectiveness varied by patient or fibroid characteristics.
Six studies provided some information about influence of characteristics on outcomes within or across arms.

Description of Studies

For this KQ we systematically reviewed all included trials (n=97) for each category of intervention. We sought: (1) indication that the study aimed to determine the influence of patient or fibroid characteristics on effectiveness and/or (2) described statistical analyses that allowed determination of whether patient of fibroid characteristics modified outcomes.

At its core, this question is about effect modification, also called interaction, which can be used to inform clinical decisions. For instance if women with five or more fibroids are found to have less improvement of their hematocrit 6 months after an intervention than those with fewer than five initial fibroids, and the p-value for that comparison is significant, we would say there is modification of the effectiveness of the intervention by fibroid number, such that women with fewer fibroids may experience superior improvement in hematocrit. No studies were explicitly powered to investigate effect modification.

We identified only six trials that contributed related analyses: two trials of medications (in three publications),¹¹⁵^,¹¹⁹^,¹²³ two studies that compared UAE to surgery, each with multiple publications (REST⁷¹^,⁸²^,¹⁰⁴ and EMMY¹⁹^,¹⁰⁹^,¹¹⁴), and one surgical trial.¹⁰⁷ One additional study addressed whether baseline characteristics influenced likelihood of success across procedure arms.⁸¹ We assessed two studies as good quality,¹⁹^,⁷¹^,⁸²^,¹⁰⁴^,¹⁰⁹^,¹¹⁴ two as fair quality,¹¹⁵^,¹¹⁹^,¹²³ and two as poor quality.⁸¹^,¹⁰⁷

Detailed Synthesis of Effect Modifiers

Medical Intervention

In a dose comparisons trial of oral mifepristone (5 mg vs. 10 mg), the authors reported no difference in uterine fibroid volume size when analyses were adjusted for baseline volume, dose, and treatment duration (6 or 12 months). There was no significant difference in uterine volume reduction between the two dose groups (p=0.94). However in pooled analyses, for every 10 cm³ larger increment in baseline fibroid volume, fibroid volume reduction increased, on average, by 3.8 cm³ after adjusting for dose and time (95% CI, 2.7 to 4.9; p<0.001).¹¹⁵

Within a raloxifene arm (60 mg for six cycles) authors found the drug demonstrated selective action on leiomyoma by menopausal status, such that postmenopausal women were more likely to achieve decreased uterine and fibroid size. Thirteen of 31 postmenopausal women had decreased size of fibroids¹²⁸ compared with one of 29 premenopausal women.¹²³

Procedures

Uterine Artery Embolization

In one followup publication from the REST trial comparing UAE to surgery (hysterectomy or myomectomy), the authors found that 5-year reintervention rate after UAE did not differ (p=0.123) by the degree of infarction achieved during the initial procedure, as measured by MRI.⁷¹ Study authors reported no effect of age (p=0.77), uterine volume (p=0.50) or fibroid diameter (p=0.57) on the degree of infarction as measured by MRI at 6 months. Similarly, age (p=0.13), uterine volume (p=0.81) and fibroid diameter (p=0.81) did not modify the need for reintervention.⁷¹ The presence of a single fibroid (OR=6.21; 95% CI, 1.65 to 23.41) and small uterine volume (less than 500 cm³) (OR=10.8; 95% CI, 1.25 to 93.36) were associated with higher risk of procedural failure.

The finding of no significant group difference in the rate of ovarian failure as measured by FSH >40 IU/L at 1 year after treatment (UAE vs. any surgery) was not modified by age less than 45 years or age of 45 years and older.⁸² The risk for major complications (OR=5.68; 95% CI, 2.05 to 15.75) and high pain scores (higher than score of 5) (OR=1.97; 95% CI, 1.08 to 3.58) increased with each extra vial of polyvinyl alcohol used, though there was no significant association (p=NS) between high pain scores and uterine size, fibroid size, or total number of fibroids at 6 weeks after the procedure.¹⁰⁹

A multivariate analysis indicated that compared with baseline, after 24 months of treatment, higher number of fibroids was associated with lower risk of poor sexual function (OR=0.69; 95% CI, 0.51 to 0.94); while the presence of a comorbid condition was associated with an increased risk of a worse sexual function (OR=3.2; 95% CI, 1.38 to 7.41).⁹⁶

HIFU for Fibroid Ablation

A study comparing radiofrequency fibroid ablation to HIFU for fibroid ablation examined intermediate outcomes across groups by fibroid characteristic (fibroid blood supply and size).⁸¹ The rate of complete ablation was significantly different between groups with different grades of blood supply. The radiofrequency ablation technical success rate was 89.3 % versus 54.2 % among individuals with Grade II blood supply (direct vessels to fibroid readily visualized). No difference was seen within groups with no clear blood supply or widespread “halo” like blood supply; however the size of this trial was too small for meaningful assessment of true effect modification (total n =100). Completeness of ablation was similar among the subgroups of patients with fibroid diameter between 2 cm and 4 cm, whereas the technical success for radiofrequency ablation was superior to HIFU in patients with fibroid diameters between 4 and 6 cm and between 6 and 8 cm (p<0.05).⁸¹

Surgical Interventions

An RCT with 181 women with fibroids who had been trying to conceive for at least 1 year without success, subdivided the women according to the location of the fibroid (i.e., submucous, intramural, subserosal) and randomized to myomectomy (laparoscopic myomectomy or hysteroscopic myomectomy) or no surgery.¹⁰⁷ For women with subserosal or intramural fibroids, there was no significant difference in the pregnancy rate, comparing myomectomy with no treatment. For women with submucous fibroids, the group who underwent myomectomy had a greater pregnancy rate (40.4%) than those who did not undergo surgery (21.4%) (p<0.05).¹⁰⁷

Summary of Effect Modification

Overall, there is insufficient evidence for women to choose one intervention over another based on individual characteristics or the characteristics of their fibroids. Too few studies have been adequately powered to determine within arms if one subgroup or another has superior outcomes within a treatment. Such information is required as a first step towards using individual characteristics to inform treatment choice.

Key Question 3. Risk of Leiomyosarcoma When Mass Thought To Be a Fibroid

Key Points

The overall risk of discovering a leiomyosarcoma during surgery for presumed fibroids is 0.02 percent (range: 0% to 0.09%) in prospective studies and 0.08 percent (range: 0.05% to 0.13%) in retrospective studies. In other words, using data from 160 studies, an unexpected leiomyosarcoma will be identified in fewer than one and up to 13 of every 10,000 surgeries performed for symptomatic fibroids.

Overview

The risk of fragmenting and disseminating a leiomyosarcoma into the pelvic cavity is at the heart of the FDA and professional organizations concerns about the safety of power morcellation. The defining component of this risk is determining how likely it is that a surgeon who is operating for a fibroid encounters a leiomyosarcoma. To address KQ3 we pursued evidence in the literature to estimate the prevalence of uterine leiomyosarcoma among women having surgery for uterine fibroids.

Description of Studies

We sought literature from studies of myomectomy or hysterectomy for presumed benign disease that included histopathologic analysis of all excised fibroid specimens. In the course of our work, Elizabeth Pritts and her colleagues published such an estimate using a similar approach, with a stated aim to estimate the prevalence of occult leiomyosarcoma at time of treatment for presumed benign tumors (fibroids).¹⁰ We confirmed our search method included their articles and then updated their search using similar eligibility criteria to identify papers published since the end of their inclusion period in 2014. In addition to the 133 unique studies included in the prior analysis, we identified 27 additional studies, including twenty-four retrospective cohorts,¹⁷⁰^,¹⁷²^-¹⁹⁴ two prospective cohorts,¹⁹⁵^,¹⁹⁶ and one RCT.¹⁹⁷ The RCT was included with the prospective studies in the analysis.

Detailed Synthesis

The 2015 Pritts analysis extracted data from 133 publications including 30,193 women.¹⁰ The 27 new studies included an additional 106,002 women bringing the total to 160 studies and 136,195 women. Among prospective studies 56.7% focused on myomectomy findings; 35.8% hysterectomy, and 6.0% both types of surgery. Among retrospective studies 31.9% focused on myomectomy; 48.9% hysterectomy, and 19.2% both. Over 40,000 (29%) of the included women are from prospective studies. In prospective studies, subjects had an age range of 20 to 83 with a mean age of 38.5±6.0 years, while retrospective studies included women from age 18 to 96 with a mean age 43.4±5.6.

Following methods described in Pritts and colleagues¹⁸ we fit a Bayesian binomial random effects models to update the estimate of prevalence of leiomyosarcoma and achieved good model fit. The point estimates and credible intervals for prevalence are summarized below for the original Pritts study and for our five new models. (Table 28)

We identified some errors in counts from the data extraction in the Pritts analysis. ¹⁰ We ran analyses with the original (model 1) and corrected data (model 2). We added data from 27 new studies (model 3, most inclusive, largest data set). Then, we excluded prospective studies that included hysteroscopic fibroid resection because of concerns that hysteroscopy might yield incomplete tissue for pathology (model 4). Because we had noticed some discrepancies, we reviewed all publications and reclassified them based on our confidence that complete histopathologic evaluation was performed for all subjects. We attempted to contact authors to confirm when necessary. For model 5, we restricted our analysis to those publications for which we had high confidence of complete histopathologic evaluation for every subject, as the most refined estimate (model 5).

We present estimates for prospective and retrospective studies separately since statistical models suggest meaningful heterogeneity is introduced by study design. Regardless of model assumptions, all estimates from retrospective data produced higher estimates (5.1 to 8.5 cases per 10,000 surgeries) than prospective studies (0.5 to 2.9 cases per 10,000 surgeries). An aggregate estimate combining both retrospective and prospective studies is 8.3 per 10,000 surgeries (95% CI: 1.17, 9.39) but this model is difficult to interpret because assessment of statistical heterogeneity suggests the estimates are distinctively different, thus a combined estimate is not appropriate. We have greater confidence in the ability of prospective studies using standardized protocols to evaluate histology to detect incident cases than retrospective studies that rely on clinical pathology reports and retrospective determination of inclusion. Although we planned to estimate the effect of age on leiomyosarcoma risk, the lack of granular data (especially for non-cases) prevented us from doing so.

In her report, Pritts repeated the FDA analysis using a Bayesian binomial model. ¹⁰^,²⁴ To keep comparisons on the same scale, we plot these two published estimates on the same Bayesian binomial scale as our conservative model 3 (Figure 5.) While the point estimates differ, confidence intervals overlap.

Table 28

Leiomyosarcoma prevalence estimates.

Figure 5

Risk of leiomyosarcoma at surgery for presumed fibroids. Notes: Point estimates (cases per 10,000 surgeries) and 95% credible interval for published estimates and current model. Grouped by prospective (solid box) vs. retrospective (dashed box) study design. (more...)

Summary

The literature investigating the prevalence of leiomyosarcoma in presumed fibroids has grown rapidly and has added more data but not greater precision to estimates of this rare event. From 160 prospective and retrospective studies, the estimate of leiomyosarcoma ranges from 0 to 13 cases out of 10,000 surgeries. Our estimate based on 68 prospective studies biased in favor of detection (model 3), estimates that two (range: fewer than one and up to 9) women in every 10,000 who have surgery for fibroids may be found to have a leiomyosarcoma.

Key Question 4. Influence of Morcellation and Patient/Fibroid Characteristics on Leiomyosarcoma Survival

Key Points

Uterine leiomyosarcoma has high mortality regardless of surgical approach.
For women with uterine leiomyosarcoma, survival time appears shorter for those where power morcellation was used compared to those where sharp morcellation with a scalpel was used or to those who had intact removal of the uterus (no morcellation), however, confidence intervals are wide and overlap.
We found insufficient data on the influence of patient or fibroid characteristics on leiomyosarcoma survival.

Description of Studies

Twenty-eight studies (29 publications) provided data about disease progression and vital status for women who had a leiomyosarcoma identified at the time of an initial surgery and for whom the method of removal of the surgical specimens was known and survival time data could be extracted.¹⁷⁰^-¹⁷²^,¹⁷⁷^,¹⁸⁴^-¹⁹²^,¹⁹⁴^,¹⁹⁸^-²¹² The research was conducted in 14 different countries, including nine from the United States. The largest studies were from Norway and the United States. The majority identified baseline surgical data and outcomes after the events had occurred or relied on prospective registries and were able to provide followup for participants present at baseline. These studies included 715 women with leiomyosarcoma and the time of their initial surgeries ranges from the 1980s through 2015. This overlaps well with the period of growth in minimally invasive surgery for fibroids and with the use of power morcellation.

We reviewed studies for information about whether individual characteristics of the women or presumed fibroid status helped to identify those most at risk of harm. Similar to KQ2, we sought evidence that investigated effect modification by factors such as age, menopausal status, and imaging characteristics which can be known before surgery and have potential to inform decisions about surgical approach. Twenty-four studies (384 women) contributed data to models to compare survival time based on use of power morcellation, scalpel morcellation, or no morcellation. For studies that did not explicitly provide individual survival data¹⁸⁶^,¹⁹⁴^,¹⁹⁹^,²⁰⁷ we were able to manually extract data from published survival curves using an online digitizing tool.²¹³ Another four studies (reported in five publications) provided information about survival, but we could not confidently extract data to include in the analysis because we either could not determine event times with confidence¹⁷⁰^,¹⁷¹^,¹⁹⁸^,²⁰⁴ or because the data in the tables and text did not align with those shown in the survival curves.²⁰⁸

Detailed Synthesis

Our purpose for this aim was to determine if leiomyosarcoma dissemination was influenced by method of morcellation and to compare this with no use of morcellation while also assessing characteristics of patients and fibroids that might be associated with risk of dissemination.

There is no clinical way to detect dissemination of leiomyosarcoma at the time of surgery. If the tumor is disrupted, both visible and microscopic particles may be spilled. However, even without visible tumor disruption, microscopic or hematogenous dissemination may occur. As a result, stage of disease, progression or recurrence of disease, and survival become surrogates for recognizing dissemination. Thus we hypothesize that stage and survival would be worse for those in whom leiomyosarcomas were removed by power morcellation compared with scalpel morcellation and that both of these would be inferior to no breach of the integrity of the tumor by removing the tumor intact.

To estimate survival for each surgical intervention, we fit parametric survival models using a Bayesian hierarchical approach, using the data extracted from publications that made it available. To account for heterogeneity among studies, we included a study-level random effect in the hierarchical baseline survival parameter. A simple exponential survival function was found to be a poor fit to the data, therefore we fit a Weibull survival function that resulted in an adequate fit. Figure 6 captures cumulative hazard estimated by these models, with the shaded areas indicating the 95% Bayesian credible intervals (BCI) for each intervention group: no morcellation, scalpel morcellation and power morcellation.

Figure 6

Estimated survival after surgical intervention for leiomyosarcoma by morcellation approach. Notes: Survival curves plot x-axis as follow up time in months with hazard ratio indicated by y-axis. Dotted line indicates baseline comparison which is no use (more...)

By 5 years of followup (60 months), a typical interval considered in cancer outcomes, 30% (95% BCI, 13% to 61%) of women for whom power morcellation was used were alive. This is compared to 59% (95% BCI, 33% to 84%) of women for whom scalpel morcellation was used and 60% (95% BCI, 24% to 98%) with no use of morcellation. While the point estimate of power morcellation survival was much lower than for either non-morcellated or scalpel morcellation patients, the uncertainty in these estimates was very large, particularly at longer followup times. However this literature is evolving rapidly and more than half of the cases and papers that contribute to our estimates have appeared in the literature in 2015 or 2016.

Five papers, as described below, did not contribute to our estimates. Two small studies, one with 18 cases of leiomyosarcoma,¹⁹⁸ the other with 56 cases spread over 21 years,²⁰⁸ suggested increase in disease recurrence and worse survival after morcellation. In contrast, a final paper that did not report individual level data that could be extracted identified 53 patients with leiomyosarcoma and found rates of pelvic recurrence did not differ by use of morcellation at three or six months of followup with comparable disease-free survival rates in both groups.²⁰⁴ Three studies did not present data in a way to allow inclusion in our aggregate survival estimates. They include the findings of a recent review and lifetable analysis by Pritts and colleagues²¹⁴ and an analysis of data from Norway.¹⁷⁰^,¹⁷¹ These studies do not find a statistically meaningful disadvantage to morcellation when aggregate data are used to calculate survival.

Individually, few authors had sufficient number of cases to address differences in risk of dissemination or survival by other characteristics of the women found to have leiomyosarcoma at the time of surgery for presumed fibroids. If we consider only those studies with more than 10 cases with scalpel or power morcellation, only five publications with total size of 15 to 56 participants have potential to contribute information.¹⁹⁹^,²⁰⁴^,²⁰⁷^,²⁰⁸^,²¹⁵ Two do not provide adjusted multivariable models or stratification by characteristics other than operative approach.²¹⁵^,²¹⁶ None report assessment of effect modification by any trait other than surgical approach to removal of the uterus or fibroids.

Characteristics reported not to confound the association between risk of dissemination and outcome in multivariable time-to-event models included: age,²⁰⁷^,²⁰⁸ menopausal status,²⁰⁷^,²⁰⁸ adjuvant treatment including radiotherapy,²⁰⁷^,²⁰⁸ and BMI.²⁰⁸ In the publications authored by Perri and Park, only surgical approach (grouped as total abdominal hysterectomy or other approaches with any morcellation or breech of the tumor capsule) compared to removal intact significantly influenced outcomes.²⁰⁷^,²⁰⁸ Lin and colleagues adjusted for age, tumor size, and mitotic count, but including these covariates in the model did not meaningfully change estimates.¹⁹⁹ Thus, this literature lacks information to identify those most likely to have a more aggressive course of disease beyond pathology features of tumor differentiation and stage. This is not unexpected since leiomyosarcoma is rare and power is limited. It is helpful, however, that larger studies do not find other characteristics act as confounders. This implies that our aggregate estimate and those of others are not likely to be seriously confounded by commonly measured clinical factors.

In summary, this literature provides data to indicate that method of morcellation is a potential determinant of outcomes, with power morcellation being associated with decreased 5 year survival. However, confidence intervals overlap, and even those who have leiomyosarcoma with removal of the uterus intact have substantial mortality risk.

Bookshelf ID: NBK537744

Contents

< Prev Next >

PubReader
Print View
Cite this Page
Hartmann KE, Fonnesbeck C, Surawicz T, et al. Management of Uterine Fibroids [Internet]. Rockville (MD): Agency for Healthcare Research and Quality (US); 2017 Dec. (Comparative Effectiveness Review, No. 195.) Results.
PDF version of this title (4.4M)

Results - Management of Uterine Fibroids
Results - Management of Uterine Fibroids
Mus musculus strain C57BL/6J unplaced genomic scaffold scaffold_1690, whole geno...
Mus musculus strain C57BL/6J unplaced genomic scaffold scaffold_1690, whole genome shotgun sequence
gi|317096526|gnl|WGS:AEKR01|scaffol 0|gb|GL590820.1|
Nucleotide

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on