Risk of Recurrence and Reintervention After Uterine-Sparing Interventions for Symptomatic Adenomyosis: A Systematic Review and Meta-Analysis

We considered all studies, of any study design (randomized controlled studies, cohort studies, cross-sectional studies, and case–control studies), that described the risk of recurrence or reintervention after uterine-sparing interventions for symptomatic adenomyosis. Case reports were excluded. The articles were screened according to the eligibility criteria as follows. Inclusion criteria: 1) patients with confirmed symptomatic adenomyosis; 2) studies regarding uterine-sparing interventions such as adenomyomectomy, UAE, and image-guided thermal ablation (HIFU, RFA, or MWA) in the management of symptomatic adenomyosis; 3) in the studies that compared the aforementioned uterine-sparing interventions with other therapies, the data from the single uterine-sparing treatment group were included; 4) studies that reported at least one of the outcome measures of interest (specified below); 5) studies published in peer-reviewed journals; and 6) English-language articles. Exclusion criteria: 1) studies that recommended hysterectomy for symptomatic adenomyosis; 2) studies that did not report any outcome measure of interest; 3) studies with fewer than 10 patients; 4) overlapped or repeated studies; 5) review articles, commentaries, animal studies, abstracts, letters, case reports, secondary analyses, and reports from meetings; and 6) non–English-language studies. The outcome measures of interest were: 1) Recurrence rate: the reappearance of symptoms (painful menses or heavy menstrual bleeding) after significant or complete remission or the reappearance of adenomyotic lesions confirmed by ultrasonography or magnetic resonance imaging. Recurrence rate=number of patients with recurrence/number of total patients×100%. 2) Reintervention rate: the performance of an additional intervention owing to symptomatic recurrence of adenomyosis. The main additional intervention was hysterectomy. Other uterine-sparing techniques were also included, such as UAE and thermal ablation interventions. The need for only medical therapy was not considered to be a reintervention. Reintervention rate=number of patients receiving reintervention therapy/number of total patients×100%. Two independent researchers (L.L. and H.T.) performed study selection and evaluation according to the screening criteria using Excel. In case of discrepancy, a third researcher (D.L.) was consulted. First, the titles and abstracts of articles were screened for potentially relevant studies. Second, all relevant articles were obtained in full text and assessed for eligibility. In addition, we scrutinized the references of the included articles and relevant reviews. The assessment of the potential risk of bias and evidence quality of the included studies was carried out independently by two researchers (L.L. and D.L.). The risk of bias for observational nonrandomized studies (including prospective and retrospective cohort studies) was evaluated with the use of the Newcastle-Ottawa scale. Analysis of the risk of bias for randomized controlled trials (RCTs) was performed using the Cochrane Collaboration’s tool. Visual inspection of the funnel plot was used to assess for publication bias. Two researchers (L.L. and L.Z.) independently extracted clinical data from each article identified for inclusion using pretested data-collection forms. We recorded characteristics of the included articles, such as the first author, time of publication, type of intervention, the number of participants, duration of follow-up, the age of participants, the volume of uterus and adenomyosis, and outcome measures. We extracted the recurrence rate and reintervention rate after the treatment. The data were pooled for meta-analysis. We used Stata 14.0 for statistical analysis. Results for individual studies were pooled according to the type of therapy (adenomyomectomy, UAE, or image-guided thermal ablation). The outcome measures were presented as the frequency with percentage and pooled with 95% CI. The pooled proportion was estimated using Freeman–Tukey double arcsine transformation. Heterogeneity reflects the variance from each included study and comes from differences in the participants, study design, treatments, duration of follow-up, analysis methods, participants’ inclusion criteria, or other characteristics among studies. We estimated the heterogeneity among studies using the I 2 statistic and its 95% CI. A random-effect model was used if there was significant heterogeneity (I 2 >50%), and a fixed-effect model was used in the absence of statistically significant heterogeneity (I 2 ≤50%). If I 2 values showed significant heterogeneity, we considered performing the subgroup and sensitivity analyses. The statistical significance was further assessed by Egger’s linear regression asymmetry test.

A total of 387 articles were identified after the duplicates were removed. After screening the titles and abstracts, 114 articles were further assessed for eligibility. Full texts were reviewed. Articles that did not include an outcome measure of interest (n=39), repeated articles (n=9), reviews articles (n=12), and case reports (n=12) were excluded. Forty-two articles 7 – 48 were deemed eligible for inclusion in this meta-analysis. Included articles were composed of 14 studies 7 – 20 regarding adenomyomectomy, 16 studies 21 – 36 regarding UAE, and 12 studies 37 – 48 regarding image-guided thermal ablation (HIFU nine studies, 37 – 45 RFA three studies, 46 – 48 MWA 0 studies) in the management of symptomatic adenomyosis (Fig. 1 ). Liu. Recurrence Risk After Interventions for Adenomyosis. Obstet Gynecol 2023. A total of 5,877 patients were identified within the 42 studies that met eligibility criteria. Of these patients, 1,790 (30.5%) underwent adenomyomectomy, 816 (13.9%) underwent UAE, and 3,271 (55.6%) underwent image-guided thermal ablation. The included studies were observational nonrandomized studies, and most of the studies were single-arm clinical research without a control group. The included studies had varying lengths of follow up (ranging from 6 to 95 months) and the length of follow up may affect the recurrence and reintervention outcomes. Among the adenomyomectomy group, one study had length of follow-up less than 24 months and 13 studies had length of follow-up 24 months or longer. In the UAE group, six studies had length of follow-up less than 24 months and 10 studies had length of follow-up 24 months or longer. Among the image-guided thermal ablation group, six studies had length of follow-up less than 24 months and six studies had length of follow-up 24 months or longer. Tables 1 – 3 summarize the details on the first author, time of publication, treatments, the number and age of participants, follow-up duration, the volume of uterus and adenomyosis, and outcome measures of included studies. The average/median age ranged from 33.6 to 43.4 years in the adenomyomectomy group, from 36.7 to 50.0 years in the UAE group, and from 34.0 to 42.4 years in the image-guided thermal ablation group. The average/median uterus volume ranged between 164.5 cm 3 and 306.3 cm 3 in the adenomyomectomy group, between 216.0 cm 3 and 563.0 cm 3 in the UAE group, and between 150.7 cm 3 and 362.1 cm 3 in the image-guided thermal ablation group. The average/median diameter of adenomyosis ranged between 3.5 cm and 6.5 cm in the adenomyomectomy group. The average/median adenomyosis volume ranged between 75.0 cm 3 and 189.0 cm 3 in the UAE group, and between 54.2 cm 3 and 147.3 cm 3 in the image-guided thermal ablation group. Due to data deficiencies, it was not possible to calculate whether the age of participants, follow-up duration, uterus volume, and adenomyosis volume at baseline were statistically different among the different uterine-sparing interventions. Characteristics of the Studies Regarding Recurrence and Reintervention Risk After Adenomyomectomy (n=14) Characteristics of the Studies Regarding Recurrence and Reintervention Risk After Uterine Artery Embolization (n=16) Characteristics of the Studies Regarding Recurrence and Reintervention Risk After Image-Guided Thermal Ablation (n=12) An assessment of the risk of bias was performed using the Newcastle-Ottawa scale to evaluate selection, comparability, and outcome. All the studies were classified as poor-to-medium quality (66.7% poor quality and 33.3% medium quality): 1) comparability of case and control groups: most included studies (85.7%) were single-arm studies rather than comparative studies, and no RCTs were available; 2) all studies had significant risk of selection bias; 3) loss to follow-up in 73.8% included studies, and the average rate of loss to follow-up was 9.1%; 4) the different length of follow-up period among studies: a short follow-up period of less than 24 months in 30.9% studies limited our ability to analyze the outcome measures; and 5) quality of the statistical analyses. Table 4 presents the details of the synthesized results to assess the risk of recurrence and reintervention after these uterine-sparing interventions for the treatment of symptomatic adenomyosis. Figures 2 and 3 and Appendices 1–5 (Appendices 1–5 are available online at http://links.lww.com/AOG/D32 ) show the forest plots for the outcome measures, subgroup analysis, and sensitivity analysis. Results of Meta-Analysis, Subgroup Analysis, and Sensitivity Analysis Liu. Recurrence Risk After Interventions for Adenomyosis. Obstet Gynecol 2023. Liu. Recurrence Risk After Interventions for Adenomyosis. Obstet Gynecol 2023. The recurrence rates were 12.6% (95% CI 8.9–16.4%, P <.001; I 2 =85.0%; 12 studies) for adenomyomectomy, 29.5% (95% CI 17.4–41.5%, P <.001; I 2 =86.5%; eight studies) for UAE, and 10.0% (95% CI 5.6–14.4%, P <.001; I 2 =87.7%; eight studies) for image-guided thermal ablation (Table 4 , Fig. 2 ). The overall recurrence rate was 15.2% (95% CI 12.0–18.4%, P <.001, I 2 =90.7%; 28 studies) after all uterine-sparing treatments. Because there were no studies that reported the recurrence risk after MWA and RFA, the recurrence rate for image-guided thermal ablation was reported only in patients undergoing HIFU. However, all analyses demonstrated significant statistical heterogeneity (I 2 >50%) (Table 4 , Fig. 2 ). The reintervention rates were 2.6% (95% CI 0.9–4.3%, P =.060; I 2 =48.3%; eight studies) in the adenomyomectomy group, 12.8% (95% CI 7.2–18.4%, P <.001; I 2 =84.7%; 11 studies) in the UAE group, and 8.2% (95% CI 4.6–11.9%, P <.001; I 2 =91.2%; nine studies) in the image-guided thermal ablation group (HIFU or RFA) (Table 4 , Fig. 3 ). The overall reintervention rate after these therapies was 7.4% (95% CI 5.3–9.5%, P 50%) (Table 4 , Fig. 3 ). Because I 2 values in most analyses demonstrated significant heterogeneity, we performed the subgroup analysis based on different treatments. According to whether the participants received perioperative or postoperative adjuvant hormone therapy, such as gonadotropin-releasing hormone agonist and levonorgestrel-releasing intrauterine system, a subgroup analysis of recurrence rate after adenomyomectomy was performed. The result showed that the recurrence rates were 16.6% (95% CI 10.5–22.8%, P <.001; I 2 =89.8%; 11 studies) for adenomyomectomy without adjuvant hormone therapy, and 8.0% (95% CI 5.0–11.1%, P =.092; I 2 =44.9%; six studies) for adenomyomectomy with adjuvant hormone therapy (Table 4 ). Heterogeneity was reduced in the adenomyomectomy with adjuvant hormone therapy group significantly (I 2 ≤50%). Similarly, according to whether the participants received perioperative or postoperative adjuvant hormone therapy, a subgroup analysis of the reintervention rate after adenomyomectomy was performed. The reintervention rates were 3.1% (95% CI 0.8–5.4%, P =.025; I 2 =60.9%; five studies) for the participants who had undergone adenomyomectomy without adjuvant hormone therapy, and 1.3% (95% CI −0.9 to 3.6%, P =.646; I 2 =0%; two studies) for the participants who had undergone adenomyomectomy with adjuvant hormone therapy. Heterogeneity was significantly reduced in the adenomyomectomy with adjuvant hormone therapy group (I 2 ≤50%). However, only two studies were included in the adenomyomectomy with adjuvant hormone therapy group. In addition, we performed a subgroup analysis of the reintervention rate after image-guided thermal ablation according to the different thermal energy sources (HIFU or RFA). The subgroup analysis showed that the reintervention rates were 5.4% (95% CI 2.3–8.5%, P <.001; I 2 =86.3%; six studies) for HIFU, and 19.4% (95% CI −2.2 to 40.9%, P 50%) (Table 4 and Appendices 1–3; Appendices 1–3 are available online at http://links.lww.com/AOG/D32 ). Because I 2 values still demonstrated significant heterogeneity in most results after subgroup analysis, we performed sensitivity analysis for the studies with follow-up periods of 24 months or longer only by excluding studies with follow-up periods of less than 24 months. The result showed that the recurrence rates were 13.3% (95% CI 9.1–17.5%, P <.001; I 2 =86.3%; 11 studies) for adenomyomectomy, 36.1% (95% CI 23.5–48.6%, P <.001; I 2 =83.9%; six studies) for UAE, and 14.5% (95% CI 3.2–25.8%, P <.001; I 2 =90.6%; three studies) for image-guided thermal ablation (Table 4 ). The overall recurrence rate was 18.6% (95% CI 13.7–23.4%, P 50%). The reintervention rates were 3.7% (95% CI 1.1–6.2%, P =.086; I 2 =48.2%; 5 studies) in the adenomyomectomy group, 16.8% (95% CI 11.5–22.2%, P =.170; I 2 =33.8%; 7 studies) in the UAE group, and 11.0% (95% CI 3.2–18.8%, P <.001; I 2 =95.3%; four studies) in the image-guided thermal ablation group (Table 4 ). The overall reintervention rate after these therapies was 10.3% (95% CI 6.9–13.7%, P <.001, I 2 =87.4%; 16 studies). Heterogeneity was significantly reduced in the UAE group (I 2 ≤50%) (Appendices 4 and 5, http://links.lww.com/AOG/D32 ).

This systematic review and meta-analysis protocol was registered through the PROSPERO online database (ID: CRD42021261289) and performed conforming to the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines. Institutional review ethics board approval was not required because this study was a systematic review of the literature. According to PRISMA guidelines, we searched electronic databases such as Web of Science, MEDLINE, Cochrane Library, EMBASE, ClinicalTrials.gov , and Google Scholar from January 2000 to January 2022 for articles that reported uterine-sparing interventions for symptomatic adenomyosis. We used a combination of MeSH terms and keywords. The following search strategy was used on MEDLINE: (adenomyosis [Title/Abstract]) AND ((relapse [Title/Abstract]) OR (recrudesce [Title/Abstract]) OR (recurrence [Title/Abstract]) OR (recur [Title/Abstract]) OR (reappear [Title/Abstract]) OR (reintervention [Title/Abstract])). We also manually searched the reference lists of the included articles and relevant reviews for additional studies.

This systematic review demonstrated overall low rates of recurrence (less than 30% for all studies and less than 37% for studies with follow-up periods of 24 months or longer) and reintervention rates (less than 13% for all studies and less than 17% for studies with follow-up periods of 24 months or longer) for all uterine-sparing techniques. Because none of the included studies directly compared the recurrence or reintervention risk after adenomyomectomy, UAE, or image-guided thermal ablation and the results were heterogeneous, we cannot definitively conclude which treatment has a lower risk of recurrence or reintervention. However, the results are promising that uterine-sparing techniques are successful. Further improvement in recurrence and reintervention rates were seen in the subgroup analyses with adjuvant hormone therapy, indicating medical adjuvant treatment could provide longer durable symptom control than surgery alone. Using the current data in the literature, those undergoing UAE had a higher recurrence rate (29.5%) and higher reintervention rate (12.8%) than patients who underwent adenomyomectomy or image-guided thermal ablation. The efficacy of UAE for adenomyosis has been established in the short‐term, but long‐term durability is less well established and some studies demonstrated high rates of symptom recurrence. 22 , 26 However, we noticed that patients treated with UAE had much larger uteri and larger adenomyosis than the other two groups, indicating that selection bias into UAE may have a significant effect on the outcomes. The larger size of the uteri and adenomyosis usually indicates more severe disease and it is more likely to be diffuse adenomyosis, which is more likely to recur. This selection bias was a confounding factor that could, in part, explain the higher recurrence and reintervention rates in UAE group. Uterine artery embolization is a minimally invasive angiographic interventional technique in which small particles are injected into the uterine arteries, resulting in ischemic necrosis of lesions. 5 , 49 Therefore, the size and type of the embolization agent and the blood supply characteristics of the lesions are important factors that affect the risk of recurrence. 24 , 33 Incomplete necrosis due to the size and type of particle used for embolization is one of the factors associated with UAE failure. 50 Theoretically, the smaller the size of the embolization agent, the lower the risk of recurrence. For example, Pelage et al, 22 who used trisacryl microspheres with a diameter of 500–900 micrometers, report that midterm results after UAE were disappointing, with a 44.44% recurrence rate after 2 years, whereas Kim et al, 24 who used polyvinyl alcohol particles measuring 250–355 micrometers or 355–500 micrometers, suggest that UAE is effective and has an acceptable long-term success rate. Kim et al 24 also suggest that the arteries of adenomyosis are not as large as those of leiomyomas and that more vigorous embolization seems to be needed for adenomyosis than for leiomyomas (use of a smaller-size embolization agent for complete embolization). Furthermore, the vascularity of adenomyosis is also related to recurrence risk. Zhou et al 33 investigated the association between blood supply of lesion and treatment outcomes after UAE according to the degree of vascularity at the adenomyosis lesion site (hypervascular, isovascular, and hypovascular). The results indicated that better-vascularized lesions (hypervascular lesions) showed a better response to UAE. However, this study only reported the improvements in painful menses and heavy menstrual bleeding symptoms without the risk of recurrence. We referred to the research regarding UAE for uterine leiomyomas and found that Isonishi et al 50 report that hypervascular uterine leiomyomas were more vulnerable to embolization with a lower recurrence rate after UAE compared with hypovascular lesions. However, most of the included studies did not describe the size of the embolization agent and the blood supply characteristics of lesions. Adenomyomectomy can be divided into laparotomy and laparoscopic surgery. Usually, recognition of the border of the lesion and separating the myometrial tissue invaded by the lesion from normal myometrium is difficult, especially in diffuse adenomyosis. 46 During the adenomyomectomy and thermal ablation, most of the adenomyosis tissue was excised or ablated while leaving at least 3–5 mm myometrium from the endometrial layer and serosa respectively to maintain the integrity of endometrium and serosa. However, the incompletely eliminated lesion near the endometrium or serosa may continue to grow and the symptoms may recur. In addition, the effect of HIFU and RFA is limited to the location and size of the lesions. During HIFU, the focused ultrasound waves penetrate the tissue and transform sound absorption energy into heat energy to generate coagulative necrosis of the lesion without an ablative electrode. 51 Because the energy attenuation occurs gradually during transmission of the ultrasonic wave, the ultrasound wave cannot penetrate deep enough and the deep lesions (such as the adenomyosis in the posterior wall) may not be completely ablated. 52 Radiofrequency ablation is effective only in lesions with a relatively small size, because increasing impedance limits further deposition of electricity and limits the temperatures that can be achieved in ablation lesions. 53 Therefore, various types of additional reinterventions and medical therapy were required to relieve the recurrent symptoms. 47 In addition, the type of adenomyosis (focal or diffuse) and the length of follow-up duration were also confounding factors for evaluating the risk of recurrence and reintervention after these uterine-sparing interventions. Diffuse adenomyosis is more difficult to eliminate completely and is more likely to recur. The length of follow up may affect the recurrence and reintervention outcomes. The assessment of recurrence and reintervention risk is improved with longer duration of follow-up. We have performed sensitivity analyses for the studies with follow-up periods of 24 months or longer only by excluding studies with follow-up periods of less than 24 months. The results showed that the overall recurrence rate (18.6%) and reintervention rate (10.3%) were significantly higher than before (the overall recurrence rate 15.4% and reintervention rate 7.4%). We also noticed that the image-guided thermal ablation group with the highest proportion of studies with follow-up periods of less than 24 months (50.0%) had the lowest recurrence rate (10.0%). After sensitivity analysis for the studies with follow-up periods of 24 months or longer only, the recurrence rate in image-guided thermal ablation group increased to 14.5%, which was higher than that in adenomyomectomy group (13.3%). Because all of the included studies were observational nonrandomized studies rather than RCTs, the patients’ age, size of the uterus, extent of adenomyosis, treatment protocol (such as the surgical experience of different operators and whether the participants received adjuvant hormone therapy), and postoperative follow-up (such as the duration of follow-up and the rate of loss to follow-up) varied significantly among the included studies. The subgroup analysis according to whether the participants received adjuvant hormone therapy showed that heterogeneity was significantly reduced in some groups, indicating adjuvant hormone therapy could in part explain the heterogeneity; however, there was still heterogeneity when RFA and HIFU were analyzed separately. We have also performed sensitivity analysis for the studies with follow-up periods of 24 months or longer only. The result of reintervention rate showed that heterogeneity was significantly reduced in the UAE group (I 2 ≤50%), indicating that a short follow-up period of less than 24 months may be one of the sources of heterogeneity. We have also analyzed the studies with similar uterine size and the heterogeneity was still significant (I 2 >50%). However, we cannot explore all sources of heterogeneity because some studies did not report the information of the type and extent of adenomyosis. There were several limitations in this meta-analysis. Most included studies were single-arm studies rather than comparative studies and no RCTs were available, which might affect the accuracy of the results. The length of follow-up duration was different in the included studies, which was the confounding factor in evaluating the outcome measures. Most of the studies were classified as poor-to-medium quality. Substantial heterogeneity was observed in most of the outcome measures, and the heterogeneity was still significant in several results after subgroup analysis and sensitivity analysis. We excluded non–English-language studies. No direct comparison of these interventions for treating adenomyosis has been reported, so this meta-analysis’ synthesizing of the currently available literature will provide valuable evidence for clinical decisions. To overcome these limitations, standardized outcome measures and RCTs are needed. Studies that investigate treatments for adenomyosis have used many different outcome measures, making it challenging to perform a meta-analysis or systematic review and, thus, severely curtailing the usefulness of research to inform clinical practice and guidelines. For example, Tellum et al 54 report that 14 different outcome measures were used for painful menses and 17 for heavy menstrual bleeding after different uterine-sparing treatments for adenomyosis. Encouragingly, Tellum et al 55 have developed a standardized core set of outcomes that will help researchers when designing and reporting the results of future studies on the treatment of adenomyosis. There were no RCTs available and most included studies were single-arm studies. Similarly, UAE for leiomyomas underwent the same process over time. When UAE was first used, there were registries of a single intervention; over time, data on prospective studies were accumulated, and more RCTs of UAE compared with other interventions were available. There is a need for more prospective RCTs regarding treatments for adenomyosis; only then will we be able to reach any meaningful conclusions. This systematic review and meta-analysis revealed overall low recurrences rates of 15.2% (18.6% for studies with follow-up periods of 24 months or longer) and reintervention rates of 7.4% (10.3% for studies with follow-up periods of 24 months or longer) for uterine-sparing techniques. Current data are limited by heterogeneity and likely selection bias, because hysterectomy remains the mainstay treatment for adenomyosis. Despite the limitations, our study provided valuable information on the risk of recurrence and reintervention after these uterine-sparing treatments (including adenomyomectomy, UAE, and image-guided thermal ablation) for adenomyosis.