Systematical searches were performed on PubMed, Embase, Web of Science, and Cochrane Library for studies published online before October 2022 on postoperative recurrence of ovarian endometriosis. The search terms included “endometriosis” (e.g., “endometrioses”, “endometrioma”, and “endometriomas”), “recurrence” (e.g., “recurrences”, “relapse”, “relapses”, and “recrudescence”) and “factors” (e.g., “reasons” and “element”), as shown in Supplementary Table 1.

With the search terms and inclusion and exclusion criteria, relevant studies were retrieved from the above databases. The relevant literature was initially screened by reading the titles and abstracts, and those that failed to meet the requirements were excluded. Then, the full texts were read, and those that failed to meet the requirements were further excluded. Disagreements on literature inclusion were resolved through discussion or judgment by a third researcher. Studies with similar contents were carefully checked in terms of the title, author, and affiliation to determine whether they were duplications. Only one of the confirmed duplications was included.

After the literature selection, two researchers extracted the data from the included literature using the prepared table. Any discrepancies were settled by a third researcher. The extracted data included the first author, year of publication, age at surgery, age of menarche, dysmenorrhea, postoperative pregnancy, infertility, endometriosis operation history, combined myoma, combined adenomyosis, postoperative medication, family history, preoperative carbohydrate antigen (CA-125), ovarian cyst, body mass index (BMI), laterality of endometriosis (EMs), r-American Fertility Society (r-AFS) stage, and extent of surgery. The included studies were evaluated using the Newcastle-Ottawa Scale (NOS). Studies with a NOS score $\ge$6 were considered high quality. The NOS scores of each study are included in Tables 1,2 (Ref. [7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24]). Supplementary Table 2 shows the findings on the certainty of evidence.

Review Manager 5.4 (the International Cochrane Collaboration, Nordic Cochrane Centre, Oxford, United Kingdom) was employed for the analysis. Efficacy analysis statistics were expressed by odds ratio (OR) for dichotomous variables and by mean difference (MD) and 95% confidence interval (CI) for continuous variables to evaluate the impact of related factors on recurrence. The test level was p 0.05, and the I${}^2$ statistic was used to assess heterogeneity. The fixed effects model was adopted since homogeneity was indicated by p $>$ 0.1 and I${}^2$ $\le$50%. If p $\le$ 0.1 and I${}^2$ $>$50%, the random effects model was used considering the large heterogeneity among studies. Sensitivity analysis was performed to assess the stability and reliability of the pooled results. Publication bias was evaluated using funnel plots.

The literature retrieval process yielded 6768 potentially eligible studies. In addition, 3 studies were identified from the reference lists of the included studies. A total of 4615 studies were screened out by two reviewers independently after excluding duplicate studies. After reading the titles and abstracts, 4542 studies not associated with recurrent risk factors for ovarian endometriosis were further excluded. The remaining 73 studies were then fully reviewed. Among them, 5 were excluded due to insufficient data; 3 studies were excluded due to inconclusive postoperative pathologic findings; 25 studies were disqualified due to the lack of a control group or a non-recurrent group as the control group; 19 studies were excluded as cyst recurrence was not selected as an outcome; and 14 studies were disregarded due to insufficient postoperative follow-up. Finally, this meta-analysis included 18 studies satisfying the inclusion criteria. The flow chart of the literature selection process is shown in Fig. 1.

The included literature was retrospective studies published between 2006 and 2022, which covered 6388 patients with endometriosis, with 1138 and 5250 in the relapsing and non-relapsing groups. The basic information of the included literature is presented in Table 3 (Ref. [7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24)]. The comprehensive analysis of relapsing-related factors is listed in Table 4. The analyzed potential risk factors included age of menarche, age (years), infertility, postoperative pregnancy, BMI (kg/m${}^2$), preoperative CA-125 (U/mL), family history, dysmenorrhea, r-AFS stage, presence of adenomyosis, presence of myoma, cyst size, laterality of EMs, cystectomy of the affected site, postoperative medication, and previous endometriosis-related surgery. With 18 studies having NOS scores between 6 and 8, the included studies were of moderate to high quality (Tables 1,2).

Age at surgery: In total, 12 studies [8, 10, 11, 12, 13, 14, 16, 20, 21, 22, 23, 24] with 4611 patients reported the connection between age at surgery and postoperative endometriosis recurrence. Since the heterogeneity was significant (I${}^2$ = 52%, p = 0.02), the random-effects model was applied. The findings revealed that the recurrence rate of endometrioma decreased significantly with age (MD = –0.69, 95% CI = –1.33~–0.05, Z = 2.11, p = 0.04; Fig. 2A).

Age of menarche: A total of 3 studies [20, 23, 24] reported the connection between the age of menarche and postoperative recurrence (I${}^2$ = 0%, p = 0.72). With little heterogeneity among the 3 studies, the fixed-effects model was used, and the results revealed no significant difference in the age of menarche between the recurrence group and the non-recurrence group (MD = –0.18, 95% CI = –0.44~0.08, Z = 1.35, p = 0.18; Fig. 2B).

Dysmenorrhea: The effect of dysmenorrhea on recurrence was reported in 6 studies [9, 11, 16, 17, 21, 22]. With significant heterogeneity between the studies (I${}^2$ = 57%, p = 0.04), the random-effects model was applied, and the findings revealed that dysmenorrhea had no effect on recurrence (OR = 1.95, 95% CI = 0.95~4.02, Z = 1.82, p = 0.07; Fig. 2C).

Postoperative pregnancy: There were 7 studies [8, 12, 14, 17, 19, 20, 21] investigating the effect of postoperative pregnancy on recurrence. As the heterogeneity among the studies was significant (I${}^2$ = 83%, p 0.00001), the random-effects model was employed. The results revealed that patients impregnated after surgery had a lower risk of recurrence (OR = 0.40, 95% CI = 0.19~0.82, Z = 2.52, p = 0.01; Fig. 2D).

Infertility: Exactly 8 studies [8, 9, 11, 12, 14, 16, 19, 21] covered the relationship between infertility and postoperative endometriosis recurrence. The fixed-effects model was used due to the absence of significant heterogeneity among the 8 studies (I${}^2$ = 41%, p = 0.10), which demonstrated no significant difference in infertility between the recurrence group and the non-recurrence group (OR = 0.95, 95% CI = 0.71~1.26, Z = 0.37, p = 0.71; Fig. 2E).

Endometriosis operation history: There were 4 studies [8, 10, 14, 16] investigating the effect of previous relevant surgery history on postoperative endometriosis recurrence. Due to the little heterogeneity among the 4 studies (I${}^2$ = 45%, p = 0.14), the fixed-effects model was applied. The results revealed that previous relevant surgery history significantly affected the recurrence, and the recurrence rate was high in patients with previous related surgeries (OR = 1.90, 95% CI = 1.45~2.51, Z = 4.58, p 0.00001; Fig. 2F).

Combined myoma: The effect of combined myoma on recurrence was reported in 8 studies [8, 11, 12, 16, 18, 19, 20 23], with no significant heterogeneity (I${}^2$ = 39%, p = 0.12). The findings revealed that the recurrence rate increased with combined myoma (OR = 1.44, 95% CI = 1.07~1.94, Z = 2.42, p = 0.02; Fig. 2G).

Combined adenomyosis: In total, 7 studies [8, 11, 12, 16, 18, 19, 23] with 1287 patients reported the connection between combined adenomyosis and the postoperative recurrence rate of ovarian endometriosis, with minimal heterogeneity (I${}^2$ = 4%, p = 0.40). The findings based on the fixed-effects model revealed that combined adenomyosis enhanced the recurrence rate (OR = 1.53, 95% CI = 1.11~2.11, Z = 2.60, p = 0.009; Fig. 2H).

Postoperative medication: There were 8 studies [8, 12, 14, 17, 20, 21, 22, 24] reporting the effect of postoperative medication on postoperative recurrence, with significant heterogeneity (I${}^2$ = 71%, p = 0.001). Analysis of the results based on the random-effects model revealed that patients treated with medication had a lower risk of recurrence (OR = 1.64, 95% CI = 1.02~2.62, Z = 2.06, p = 0.04; Fig. 2I).

Family history: The effect of family history on recurrence was reported in 2 studies [10, 21], which had no significant heterogeneity between them (I${}^2$ = 47%, p = 0.17). The findings based on the fixed-effects model revealed that family history affected the recurrence (OR = 2.18, 95% CI = 1.10~4.29, Z = 2.24, p = 0.02; Fig. 2J).

Preoperative CA-125: In total, 5 studies [10, 12, 14, 22, 23] with 3548 patients reported the connection between preoperative CA-125 and postoperative recurrence rate of ovarian endometriosis. Since the heterogeneity was significant (I${}^2$ = 78%, p = 0.001), the random-effects model was applied. The findings showed a statistically significant correlation between CA-125 and the endometriosis recurrence rate (MD = 24.08, 95% CI = –7.55~40.61, Z = 2.85, p = 0.004; Fig. 2K).

Size of ovarian cyst: There were 10 studies [8, 10, 11, 14, 16, 20, 21, 22, 23, 24] concerning the relationship between cyst size and postoperative recurrence. With significant heterogeneity among the 10 studies (I${}^2$ = 85%, p 0.00001), the random-effects model was used, which showed no significant difference in cyst size between the recurrence group and the non-recurrence group (MD = 0.48, 95% CI = –0.21~1.17, Z = 1.37, p = 0.17; Fig. 2L).

BMI: The effect of BMI on recurrence was reported in 9 studies [8, 10, 12, 14, 16, 20, 21, 22, 23], which had no significant heterogeneity (I${}^2$ = 0%, p = 0.58). Using the fixed-effects model, the results revealed that the difference between the recurrence group and the non-recurrence group was not statistically significant (OR = –0.21, 95% CI = –0.46~0.03, Z = 1.73, p = 0.08; Fig. 2M).

Laterality of EMs: In total, 11 studies [8, 9, 10, 12, 14, 17, 18, 19, 21, 22, 23] with 4586 patients reported the effect of laterality of EMs on the recurrence rate. Since the heterogeneity was significant (I${}^2$ = 48%, p = 0.04), the fixed-effects model was applied. The findings revealed that the recurrence rate of patients with bilaterality of cyst was significantly higher than that of patients with unilaterality (OR = 1.19, 95% CI = 1.00~1.40, Z = 2.02, p = 0.04; Fig. 2N).

r-AFS stage: There were 6 studies [7, 10, 12, 13, 19, 22] that reported on the difference between stages I and II versus stages III and IV on endometriosis relapse, which had little heterogeneity (I${}^2$ = 43%, p = 0.12). The fixed-effect model was used to reveal the effect of the r-AFS stage on postoperative recurrence. The recurrence rate of stage I and II patients was 0.30 compared to stage III and IV patients (OR = 0.30, 95% CI = 0.19~0.46, Z = 5.46, p 0.00001; Fig. 2O).

There were also 13 studies [7, 8, 9, 10, 12, 13, 14, 15, 16, 17, 18, 19, 22] reporting the difference between stage III and stage IV on endometriosis relapse, with little heterogeneity (I${}^2$ = 0%, p = 0.46). The fixed-effect model was used to reveal the effect of the r-AFS stage on postoperative endometriosis recurrence. The recurrence rate of stage III patients was 0.57 compared to stage IV patients (OR = 0.57, 95% CI = 0.48~0.66, Z = 7.13, p 0.00001; Fig. 2P).

Cystectomy of the affected site: The effect of cystectomy of the affected site on postoperative recurrence was reported in 3 studies [12, 14, 23], which had significant heterogeneity (I${}^2$ = 98%, p 0.00001). Using the random-effect model, the results revealed no effect on recurrence by removing only the affected part (OR = 3.94, 95% CI = 0.07~216.50, Z = 0.67, p = 0.50; Fig. 2Q).

Sensitivity analysis was performed on postoperative pregnancy, infertility, combined myoma, combined adenomyosis, endometriosis operation history, ovarian cyst size, BMI, and r-AFS stage. After removing the data one by one, no significant change was found in the analysis results, indicating the reliability of the summary results. Sensitivity analysis was also performed on dysmenorrhea. Following the exclusion of the study by Wacharachawana et al. [16], the heterogeneity was statistically changed (I${}^2$ = 44%, p = 0.13).

The results revealed that dysmenorrhea had an effect on endometriosis recurrence, and the recurrence rate of patients with dysmenorrhea was high (OR = 2.44, 95% CI = 1.26–4.73, Z = 2.65, p = 0.008).

r-AFS stage was used for the funnel plot analysis to assess the publication bias. As shown in Fig. 3, all the included studies are in inverted funnel plots with good symmetry. Therefore, the publication bias of the included studies has relatively little impact on the results.

Age at surgery: Studies have revealed that the endometriosis recurrence rate decreases with age, while the age at the time of surgery is a risk factor for recurrence. Higher estrogen levels in young women induce persistent growth of residual lesions. The study by Yang et al. [32] also demonstrated that younger age at surgery was a significant risk for postoperative recurrence of ovarian endometriosis. Therefore, more attention should be attached to young female patients, and the number of follow-up visits should be increased.

Family history: This study suggests that family history has some impacts on postoperative endometriosis recurrence and confirms the existence of genetic trait changes in the pathogenesis of endometriosis. The genetic differences lead to molecular biological differences between the eutopic endometrium of patients and normal subjects.

Preoperative CA-125: Carbohydrate antigen CA-125 is a glycoprotein derived from coelomic epithelial cells and expressed in normal tissues, which could offer some insights into the diagnosis of endometriosis [33]. The above studies concluded that postoperative CA-125 was a risk factor for postoperative endometriosis recurrence. However, CA-125 is a sensitive but poorly specific indicator, and many other diseases also cause marked increases in CA-125. Therefore, CA-125 can be used as a follow-up index but not a diagnostic index of postoperative recurrence. In addition, CA-125 levels can vary with the menstrual cycle. An increased sample size is needed to improve the accuracy of this study.

Laterality of EMs: This study revealed that the recurrence rate of patients with bilaterality of cysts was significantly higher than that of patients with unilaterality. It is speculated that bilateral cysts may increase the possibility of residual lesions. This conclusion is consistent with the current clinical consensus but contrary to the conclusion of Han et al. [34] and Koga et al. [35], suggesting no significant association between bilateral cysts and postoperative recurrence. Therefore, it is necessary to expand the sample for further study.

Presence of adenomyosis: The results revealed that adenomyosis was also a risk factor for postoperative endometriosis recurrence as the pathogenesis of endometriosis and adenomyosis overlap [36, 37]. Therefore, the correct identification of endometriosis combined with adenomyosis and the formulation of personalized long-term postoperative management plans play a crucial part in recurrence prevention.

Presence of myoma: Fibroids in the uterus were regarded as estrogen-dependent tumors. An animal model of fibroid xenotransplantation revealed that steroids, including estradiol and progesterone, were necessary for tumor growth [37]. Therefore, the high estrogen in patients with uterine fibroids affects the endometriosis recurrence rate. Clinically, long-term management plans and regular follow-ups should be made for postoperative patients with uterine fibroids to prevent recurrence.

Previous endometriosis-related surgery: The results suggested that patients with previous endometriosis surgeries had a higher postoperative recurrence rate, possibly because of the more aggressive endometriosis type in patients with previous related surgeries. The mainstream view is that whether the lesion is completely removed during the operation determines the postoperative recurrence. Therefore, surgeons should remove the lesion as much as possible during the operation rather than relying on postoperative adjuvant drugs. Radical surgery is recommended for patients at high risk of recurrence and without fertility requirements.

r-AFS stage: r-AFS stage is determined based on endometriosis lesion size, location, degree of adhesion, etc. [1, 38]. This study concluded that the postoperative recurrence rate was higher with higher surgical scores and more severe stages. In short, extensive lesion adhesions during surgery tend to cause residual lesions and increase the risk of recurrence. Therefore, through clinical analysis, the r-AFS stage is of certain reference significance to evaluate the prognosis and disease severity. When selecting treatment plans, the patient’s r-AFS stage should be considered.

Postoperative pregnancy: The combined data analysis confirmed the benefit of postoperative pregnancy in preventing endometriosis recurrence, possibly because pregnancy and postpartum lactation can lead to ovarian non-ovulation and endometrium decidual-like changes, causing the necrosis of endometriosis lesions and reducing the postoperative recurrence rate [39]. Therefore, Vercellini et al. [40] also found that after endometriosis surgery, patients with fertility requirements were preferably impregnated sooner.

Postoperative medication: The integrated data revealed that patients treated with medication had a lower risk of endometriosis recurrence. Gonadotropin-releasing hormone agonist (GnRHa) can effectively inhibit the secretion of endogenous gonadotrophins and lead to a hypoestrogenic state in vivo, causing deciduosis of endometrium and necrosis of endometriosis lesions [41]. The study by Adachi et al. [42] showed that the endometriosis recurrence rate was significantly reduced after 24 months of dienogest (DNG) administration. In recent years, as an important means of long-term maintenance management after surgery, drug therapy has become a common clinical treatment. In the meantime, Chiu et al. [43] also showed that postoperative adjuvant plus maintenance therapy might be the most effective intervention to prevent endometriosis recurrence. However, the variety of drugs and the complex action mechanisms require more accurate and extensive studies to clarify the efficacy and safety of drugs [44].

Age of menarche: Ponomarenko et al. [45] concluded that some single nucleotide polymorphisms (SNPs) for the age of menarche were associated with endometriosis. Meanwhile, Vercellini et al. [40] found in their study on endometriosis that the early age of menarche in modern women was closely related to the high incidence of endometriosis. However, this meta-analysis suggested that the age of menarche had no significant effect on postoperative endometriosis recurrence. The increased incidence of contemporary endometriosis may be attributed to the improved detection rate. The understanding of the disease is gradually deepened, and the disease becomes treatable, thus increasing the detection rate of endometriosis.

Infertility: The mechanisms by which endometriosis induces infertility are multifactorial, involving mechanical, molecular, genetic, and environmental causes. New research has found that changes in gene expression and gene defects are also responsible [46]. However, infertility could be caused by a variety of factors, the most common one being female pelvic inflammatory diseases, which cause the fallopian tubes to adhere to prevent the ovum and sperm from bonding and have no significant association with endometriosis incidence. Therefore, the fallopian factor accounting for a large proportion of infertility may affect the accuracy of the study results.

Cystectomy of the affected site: It is generally accepted in clinical practice that lesion resection alone leads to an increased likelihood of endometriosis recurrence. However, the combined evidence here suggested that postoperative recurrence rates were not significantly increased in patients with lesion resection alone.

Cyst size: Jiang et al. [47] revealed that greater cyst diameters led to higher possibilities of endometriosis recurrence. On the one hand, the intraoperative residual lesions of large cysts are more likely to conceal the potential factors for recurrence. On the other hand, large cysts often represent greater ectopic cell vitality, thus increasing the recurrence rate. Nevertheless, the findings revealed no association between cyst size and recurrence. We concluded from the overall review that the possible cause was the lack of an objective standard for the intraoperative cyst size evaluation, which was mainly based on the subjective judgment of the surgeon, thus bringing errors to cyst size calculation. Therefore, improving the objective evaluation criteria for cyst size is suggested to increase the accuracy of results.

BMI: As the endometrial tissue grows outside the uterine cavity, the progesterone and estrogen signals are interfered with, leading to progesterone resistance and estrogen dominance in patients with endometriosis [27, 48]. Hormone disorders may lead to obesity in the patients and thus increased BMI. It can be inferred that BMI may be related to the recurrence of endometriosis. The study of the relationship between ovarian endometriosis and obesity here revealed no significant association between BMI and endometriosis recurrence.

Dysmenorrhea: Previous studies have shown that the direct and indirect effects of local bleeding in endometriosis tissue, the actions of celiac inflammatory cytokines, and stimulation or direct infiltration of pelvic floor nerve may be the causes of dysmenorrhea [49]. Meanwhile, patients with dysmenorrhea generally have more extensive lesions, which are difficult to completely remove by surgery, thus explaining the increased recurrence rates in patients with dysmenorrhea. However, this analysis suggested that dysmenorrhea was not a risk factor for recurrent endometriosis, contrary to previous findings by Chon et al. [22]. After removing the study by Wacharachawana et al. [16] with heterogeneous source, we found that patients with dysmenorrhea had a higher recurrence rate after surgery (OR = 2.44, 95% CI = 1.26~4.73, Z = 2.65, p = 0.008), which was consistent with the research findings of Chon et al. [22]. After carefully studying the research by Wacharachawana et al. [16], no cause for heterogeneity was found. Therefore, an increased sample is crucial for further exploration to enhance the reliability of the results.