Risk factors for postoperative recurrence of deep infiltrating endometriosis during a 6- to 12-year follow-up

This study included 320 patients with a mean age of 34.75±5.70 years (range, 18–44 yrs) and a mean BMI of 20.58±3.66 kg/m 2 . The characteristics of the study patients are provided in Table  1 . The median gravidity and parity were 1 (0–8) and 1 (0–4), respectively. Before surgery, the patients with dysmenorrhea accounted for 92.2% (295/320) of the sample. The median preoperative visual analog scale (VAS) scores and dysmenorrhea time were 0 (0–10) and 3 (0–20 years), respectively. Other pain symptoms, such as chronic pelvic pain (CPP), dyspareunia and tenesmus, were reported by 7.5% (24/320), 17.5% (56/320) and 40.3% (129/320) of patients, respectively. The infertility rate before surgery was 27.5% (88/320), including primary infertility in 18.4% (59/320) and secondary infertility in 9.1% (29/320) of patients. A total of 30.9% (99/320) of patients had a history of surgery for endometriosis, of which 25.3% (81/320) had one surgery and 5.6% (18/320) had two surgeries. A total of 70.9% (227/320) of cases were associated with ovarian endometrioma, 36.5% (117/320) of cases were associated with bilateral ovarian endometrioma, and 34.3% (117/320) of cases were associated with unilateral ovarian endometrioma (15.9%, 51 left, 18.4%, 59 right). Adenomyosis was present in 35.6% (114/320) of cases. Table 1 Characteristics of all 320 patients included in study. Variable N (%) Mean age [range], years 34.75 ± 5.70 [18–44] BMI (kg/m 2 ) 20.58 ± 3.66 Gravidity 1 (0–8) Parity 1 (0–4) Dysmenorrhea 295 (92.2%) Preoperative VAS 8 (0–10) Duration of dysmenorrhea, y 3 (0–20) CPP 24 (7.5%) Dyspareunia 56 (17.5%) Tenesmus 129 (40.3%) Infertility 88 (27.5%) Primary 59 (18.4%) Secondary 29 (9.1%) History of EMs surgery 99 (30.9%) one 81 (25.3%) Two 18 (5.6%) With EMA 227 (70.9%) Left 51 (15.9%) Right 59 (18.4%) Bilateral 117 (36.6%) Adenomyosis 114 (35.6%) rAFS stage  Stage III 38 (11.9%)  Stage IV 282 (88.1%) Operation  Number of lesions 4.14 ± 1.60 (1–8)  Hysterectomy 52 (16.2%)  Time (min) 216.8 ± 93.9  Blood loss (ml) 186.4 ± 136.2  GnRH-a 246 (76.9%)  0 month 74 (23.1%)  3 months 45 (14.1%)  6 months 201 (62.8%) Postoperative dysmenorrhea relief  Postoperative VAS 0 (0–5)  Postoperative pregnancy 83 (25.9%)  Recurrence 45 (14.1%) BMI body mass index, VAS visual analog scores, Preoperative VAS were assessed at admission, Postoperative VAS were assessed at 6 months after surgery, CPP chronic pelvic pain, EMA ovarian endometrioma, GnRH-a gonadotrophin releasing hormone agonist. Characteristics of all 320 patients included in study. BMI body mass index, VAS visual analog scores, Preoperative VAS were assessed at admission, Postoperative VAS were assessed at 6 months after surgery, CPP chronic pelvic pain, EMA ovarian endometrioma, GnRH-a gonadotrophin releasing hormone agonist. All study patients underwent complete excision of endometrioma and DIE lesions. The mean number of DIE lesions per patient was 4.14 ± 1.60 (range, 1–8). DIE lesions were widely distributed; there were 30 anatomical sites recorded in the present study (Supplementary Table S1 ). At laparoscopy, total hysterectomy was performed in 52 (16.2%) cases. The mean operation time and bleeding volume were 216.8±93.9 min and 186.4±136.2 ml, respectively. Based on individual characteristics: disease severity at surgery, surgical completeness and patient tolerance, postoperative treatment with GnRH-a for 3–6 months. A total of 246 (76.9%) patients were treated with GnRH-a after the operation, of whom 45 (14.1%) were treated for 3 months and 201 (62.8%) were treated for 6 months. After more than 6 years of follow-up (median 8, 6–12 years), the rates of pain recurrence, cyst recurrence, and DIE lesion recurrence were 1.25% (4/320), 11.6% (37/320) and 1.25% (4/320), respectively. The cumulative rate of endometriosis recurrence at 6, 7, 8, 9, 10, 11 and 12 years after laparoscopic surgery for DIE was 7.1%, 8.7%, 11.6%, 12.8%, 12.8%, 14.1% and 14.1%, respectively. Forty-five cases were in the recurrence group, and 275 cases were in the nonrecurrence group. Univariate analysis and Cox multivariate proportional hazard analyses were performed. Differences were found between the two groups in terms of age at surgery (HR: 0.853, 95% CI: 0.807–0.901, p  < 0.001), gravidity (HR 0.407, 95% CI: 0.196–0.846, p  = 0.016; reference: 0 pregnancies), parity (HR: 0.411, 95% CI: 0.212–0.796, p  = 0.008; reference: 0 deliveries), presence of left side EMA (HR: 4.882, 95% CI: 2.340-10.184, p  < 0.001; reference: absence) and use of GnRH-a for 6 months (HR: 0.258, 95% CI: 0.142–0.467, p  < 0.001; reference: <6 months). Multivariate Cox regression identified four independent predictors of endometriosis recurrence (Table  2 ): Left-sided EMA presence demonstrated the strongest association with increased recurrence risk (HR: 6.413, 95% CI: 2.961–13.889, p  < 0.001), while advancing age at surgery (HR: 0.836/year, 95% CI: 0.788–0.887, p  < 0.001), extended GnRH-a therapy (6-month vs. none: HR: 0.324, 95% CI: 0.178–0.589, p  < 0.001), and postoperative pregnancy (HR: 0.365, 95% CI: 0.179–0.748, p  = 0.006) emerged as protective factors. Notably, gravidity (HR: 1.096, p  = 0.903) and parity (HR: 0.910, p  = 0.898) showed non-significant trends toward modifying recurrence risk. Table 2 Univariate and multivariate analysis of risk factors in the endometriosis and/or endometriosis-related pain recurrence and non-recurrence groups. Univariate Multivariate HR 95% CI P value HR 95% CI P value Mean age, y 0.853 0.807–0.901 < 0.001 0.836 0.788–0.887 < 0.001 BMI (kg/m 2 ) 1.050 0.993–1.110 0.086 Gravidity  0 1.000  1 0.838 0.428–1.639 0.606  2 (and above) 0.407 0.196–0.846 0.016 1.069 0.368–3.104 0.903 Parity  0 1.000  1 0.411 0.212–0.796 0.008 0.910 0.214–3.865 0.898  2 (and above) 0.281 0.039–2.039 0.209 Dysmenorrhea  VAS 0–4 1.019 0.516–2.013 0.956  VAS 5–10 0.981 0.497–1.938 0.956  Time of dysmenorrhea, y 1.030 0.961–1.103 0.406  CPP (%) 0.468 0.113–1.946 0.296  Dyspareunia (%) 0.636 0.250–1.614 0.341  Tenesmus (%) 0.797 0.428–1.483 0.474 Infertility (%)  Primary 1.267 0.626–2.562 0.510  Secondary 0.639 0.198–2.064 0.454 History of EMs surgery  Once 1.231 0.643–2.390 0.539  Twice 0.899 0.217–3.729 0.883 With EMA  Left 4.882 2.340-10.184 < 0.001 6.413 2.961–13.889 < 0.001  Right 0.793 0.369–1.707 0.554  Bilateral 1.746 0.969–3.145 0.063  Adenomyosis 0.943 0.512–1.739 0.852 DIE lesions site  USL 2.036 0.860–4.821 0.106  Recto-vaginal Septum 1.014 0.362–2.838 0.979  Vagina 0.706 0.329–1.517 0.372  Intestine 0.488 0.227–1.047 0.066  Ureter 1.551 0.656–3.669 0.318  Fallopian tube 0.486 0.118–2.006 0.318 Operation  Hysterectomy 0.327 0.079–1.356 0.123 GnRH-a  0 month 2.730 1.467–5.081 0.002  3 months 2.736 1.371–5.460 0.004  6 months 0.258 0.142-0467. < 0.001 0.324 0.178–0.589 < 0.001  Postoperative pregnancy 0.935 0.473–1.850 0.847 0.365 0.179–0.748 0.006 BMI body mass index, VAS visual analog scores, CPP chronic pelvic pain, EMA ovarian endometrioma, GnRH-a gonadotrophin releasing hormone agonist. Univariate and multivariate analysis of risk factors in the endometriosis and/or endometriosis-related pain recurrence and non-recurrence groups. BMI body mass index, VAS visual analog scores, CPP chronic pelvic pain, EMA ovarian endometrioma, GnRH-a gonadotrophin releasing hormone agonist. ROC analysis established 33.5 years as the optimal age threshold for recurrence prediction (AUC = 0.782, 95% CI: 0.720–0.843, Fig.  2 a). Given that age was recorded as whole numbers in clinical practice, we pragmatically stratified patients using 34 years as the clinically applicable cut-off, patients < 34 years exhibited 5-fold higher recurrence rates than older counterparts (26.5% vs. 5.32%, p  < 0.001, Fig.  2 b). Fig. 2 ( a ) ROC analysis of age at surgery in recurrent patients. Area under curve: 0.782, cutoff value: 33.5 years. ( b ) Kaplan‒Meier curves presenting the cumulative rate of recurrence according to age at surgery (< 34 years. or ≥ 34 years). There were significant differences between the two groups according to the log-rank test analysis (χ 2  = 27.305, p  < 0.001). ( a ) ROC analysis of age at surgery in recurrent patients. Area under curve: 0.782, cutoff value: 33.5 years. ( b ) Kaplan‒Meier curves presenting the cumulative rate of recurrence according to age at surgery (< 34 years. or ≥ 34 years). There were significant differences between the two groups according to the log-rank test analysis (χ 2  = 27.305, p  < 0.001). Extended GnRH-a therapy duration significantly impacted outcomes (log-rank χ²=23.249, p  < 0.001, Fig.  3 ). The 6-month regimen achieved lowest recurrence rates (8.23% vs. 27.5% in 3-month and 30.6% in untreated groups), demonstrating duration-dependent protective effects. Fig. 3 Kaplan‒Meier curves presenting the cumulative rate of recurrence based on the duration of GnRH-a usage (0 months, 3 months or 6 months). There were significant differences between the three groups according to the log-rank test analysis (χ 2  = 23.249, p  < 0.001). Kaplan‒Meier curves presenting the cumulative rate of recurrence based on the duration of GnRH-a usage (0 months, 3 months or 6 months). There were significant differences between the three groups according to the log-rank test analysis (χ 2  = 23.249, p  < 0.001). Stratification by EMA localization revealed striking risk disparities (log-rank χ²=30.77, p  < 0.001, Fig.  4 ). Left-sided EMA conferred worst prognosis with cumulative recurrence rate of 47.4% versus 4.80%-13.6% in other groups. Fig. 4 Kaplan‒Meier curves presenting the cumulative rate of recurrence based on the side of EMA. There were significant differences between the four groups according to the log-rank test analysis (χ 2  = 30.77, p  < 0.001). Kaplan‒Meier curves presenting the cumulative rate of recurrence based on the side of EMA. There were significant differences between the four groups according to the log-rank test analysis (χ 2  = 30.77, p  < 0.001).

This study was approved by the Ethics Committee (institutional review board of the First Affiliated Hospital of Sun Yat-sen University) and exempted from ethics application. Written informed consent consent was obtained from all patients at their follow-up interviews. All methods were performed in accordance with the relevant guidelines and regulations. We identified 320 patients who underwent surgery at the First Affiliated Hospital of Sun Yat-sen University and were diagnosed with deep infiltrating endometriosis between January 2010 and June 2016. The inclusion criteria were as follows: (1) the operation was performed laparoscopically; (2) the diagnosis was confirmed by pathologists; (3) complete clinical records and follow-up information were available; and (4) the duration of follow-up was at least 6 years. The exclusion criteria were as follows: (1) age  45 years; (2) laparoscopic surgery converted to open surgery; (3) previous history of hysterectomy or oophorectomy; (4) pelvic malignant tumors; and (5) incomplete medical records or loss to follow-up. The patient enrollment and study design flowchart in Fig.  1 . Fig. 1 Patient enrollment and study design flowchart. Patient enrollment and study design flowchart. For each patient, the following medical information was recorded: age, body mass index (BMI), parity, location of the endometrioma (right, left or bilateral), history of surgical treatment for endometriosis, pain symptoms (including dysmenorrhea, dyspareunia, chronic pelvic pain, and bowel pain), operative time and intraoperative blood loss. All operations were performed laparoscopically under general anesthesia. The size of the uterus, the obliteration of the pouch of Douglas, and the size, location and adhesion of ovarian endometriosis cysts and DIE lesion were recorded. Based on the intraoperative lesions of the patient and the experience of the surgeon, radical resection of DIE lesions was performed. The aim of the surgical procedures was to remove all visible implants of endometriosis, complete the opening of the pouch in the operation and perform lysis of adhesion, thus achieving anatomical restoration. All laparoscopic procedures were performed by the same experienced surgeon. All women were followed up according to an internal protocol. A standard gynecological examination and a transvaginal ultrasound were conducted at 3, 6, and 12 months after surgery and then yearly after surgery. During follow-up, patients were asked whether they experienced dysmenorrhea, pelvic pain, or dyspareunia. Pain was rated on the basis of a 10-cm visual analog scale (VAS), and the intensity was divided into none (0), mild (1–4), or severe (5–10). Recurrence of endometriosis in this study was defined as the occurrence of 1 or more of the following: Recurrence of pelvic pain after at least 3 months of remission after surgery. Clinical findings suggesting a recurrence (pelvic masses, pelvic tenderness, or nodulations at pelvic examination). Ultrasound or MRI diagnosis of recurrence. Recurrence of pelvic pain after at least 3 months of remission after surgery. Clinical findings suggesting a recurrence (pelvic masses, pelvic tenderness, or nodulations at pelvic examination). Ultrasound or MRI diagnosis of recurrence. All patients were followed up in outpatient clinics or contacted by telephone interview, and follow-up ended in December 2022. Postoperative information was collected during the long-term follow-up, including postoperative medications, pain, pregnancy, related symptoms, imaging results, and recurrence time. IBM SPSS 27.0 software was used for statistical analysis. Survival curves were plotted using the Kaplan‒Meier method and compared using the log-rank test. Univariate and multivariate Cox proportional hazards models were used to analyze the relationship between clinical features and lesion recurrence. Variables with p  < 0.2 in univariate analyses were subsequently entered into the multivariate Cox proportional hazards regression model using forward elimination. The estimated hazard ratio (HR) and 95% confidence interval (CI) were calculated. Significance was defined as p  < 0.05.

Endometriosis, a prevalent benign condition within the field of gynecology, is characterized by its invasive nature and tendency for recurrence 5 . The rates of reoperation following endometriosis surgery in which the ovaries are preserved have been reported to range from 27% to 58% 6 , 7 . Notably, previous studies have reported the recurrence rate and the factors associated with recurrence of endometriosis within 5 years of initial surgery, but there is little information about the recurrence of DIE beyond 5 years postoperation 4 . In our investigation, we observed an increase in the cumulative recurrence rate over time; however, the recurrence rate observed in patients with DIE was found to be lower than those reported in prior studies. It is widely recognized that the regeneration of residual hidden lesions post-surgery is a significant contributor to recurrence 8 . All patients involved in this study underwent surgery performed by the same experienced physician, who aimed to excise as many visible endometriotic lesions as feasible to minimize the likelihood of recurrence. Additionally, 30.9% of the participants in this study had undergone one or two prior surgical procedures for endometriosis, which may have adversely affected ovarian function. This factor could partially explain the comparatively low cumulative recurrence rate observed among DIE patients in our analysis. Some studies have indicated that the age at surgery is an important determinant for DIE recurrence 4 , 9 . Viqnali et al. 10 studied 115 patients with DIE, 43 of whom experienced relapse after surgery. Cox multivariate analysis revealed that young age was an independent risk factor for postoperative recurrence. Busacca et al. reported that women ≥ 34 years have a decrease in recurrence compared with women ≤ 33 years. Similar results were reported in a retrospective study by Nirgianakis et al., who found that age < 31 years was an independent predictor of DIE recurrence 11 . In our study, multivariate analysis suggested that age was a protective factor against recurrence in DIE patients. This means that the risk of DIE recurrence decreases with age. The results of this study show that 33.5 years is a cutoff age in the ROC age analysis (Fig. 2 a). Patients who are younger than 34 years may have a higher risk for recurrence (Fig. 2 b). Endometriosis is a hormone-dependent disease, and the higher circulating estrogen levels in younger women may produce a more aggressive form of endometriosis; therefore, the younger women are more likely to experience recurrence. In particular, more attention should be given to long-term postoperative management among patients younger than 34 years of age. Mounting evidence suggests postoperative recurrence of endometriosis predominantly originates from residual microscopic lesions inadvertently retained during primary surgery 10 , 12 . This biological rationale underscores the therapeutic value of adjuvant medical therapy aimed at suppressing estrogen-dependent pathways. In this context, our study evaluated the efficacy of gonadotropin-releasing hormone agonists (GnRH-a), which induce a clinically significant hypoestrogenic state. We found that patients treated with GnRH-a for 6 months after surgery experienced a significantly lower recurrence rate (16.4%) compared to those not treated (28.9%), supported by multivariate analysis showing GnRH-a use as a protective factor (HR: 0.324, 95% CI: 0.178–0.589, p < 0.001). This aligns with previous research, including a meta-analysis indicating that a 6-month regimen can effectively reduce recurrence, whereas shorter durations showed minimal benefit 13 . However, GnRH-a therapy is no longer considered the standard of care for long-term maintenance due to well-documented concerns, such as bone mineral density loss that may occur within 6 months 14 , 15 . To address the shift in clinical practice, current guidelines 16 now favor progestin-based therapies as first-line maintenance options; these alternatives offer comparable efficacy in suppressing lesion growth but with fewer long-term risks, such as reduced bone health impacts 17 . Our results, while demonstrating GnRH-a’s short-term effectiveness, should be interpreted within the historical context of the study cohort (patients treated from 2010 to 2016), a period before progestin-based strategies were widely adopted as routine maintenance. Consequently, this study focused solely on GnRH-a’s role and did not explore other agents, reflecting the era-specific treatment paradigm. Busacca et al. 18 demonstrated that there was no significant association between pregnancy and a decreased likelihood of recurrence among patients with deep infiltrating endometriosis. They postulate that this observation might stem from the divergent effects of hormonal stimulation on deep endometrial lesions. Nevertheless, these conclusions are derived from a relatively small sample size. Many studies have found that the prevalence of recurrent endometriosis varies according to whether there is a successful pregnancy after surgical treatment of endometriosis 19 – 21 . Donnez et al. 22 , in a prospective study of 500 patients with conservative surgical treatment of deep rectovaginal endometriotic nodules, noted that the recurrence rate was significantly lower in women who became pregnant than in those who did not. In a recent large retrospective study involving 1092 patients undergoing surgery for endometriosis, Roman et al. pointed out that postoperative pregnancy was a statistically significant independent predictor for a reduction in the probability of having a repeated surgery 23 . In our study, postoperative pregnancy was also a protective factor against postoperative recurrence in DIE patients (HR: 0.365, 95% CI: 0.179–0.748, P = 0.006). In addition, the overall pregnancy rate in women who tried to become pregnant was approximately 25.9% (Table 1 ). Although this proportion may seem low, it is consistent with the results of the few randomized clinical trials on the efficacy of surgery in infertile women 18 . We encourage DIE patients who plan to become pregnant to actively complete pregnancy after surgical treatment, thereby reducing the risk of disease recurrence and enabling patients to have a better prognosis. A relation between the location of the primary endometriotic lesions and anatomic differences of the pelvic organs was previously described. Vercellini et al. 24 found that the likelihood of involvement of the left adnexa was significantly higher than the expected proportion of 50%. Ghezzi et al. 25 reported that the recurrence rate was higher when the left ovary was involved than when it was not (29% vs. 7.3%; p < 0.05). In our cohort, women with left-sided endometrioma (EMA) exhibited a higher incidence of endometriosis recurrence, as illustrated in Fig. 4 . Studies have shown that left ovarian cysts are frequently associated with more dense adhesions, during the surgical dissection of which ectopic endometrial tissue is prone to persist in the peritoneum. Notably, patients with incomplete removal of peritoneal lesions in adhesion areas exhibit a 5.9% recurrence rate, with 81.3% of relapses topographically corresponding to the original adhesion regions 26 , 27 . In addition, the sigmoid colon and filmy adhesions that frequently cover the left adnexa create an area isolated from the remaining peritoneal cavity and, therefore, may be less exposed to the peritoneal current 28 . Cells regurgitated through the right tube are more exposed to the clockwise peritoneal current and may be removed by the macrophage disposal system 25 . In our previous study of distribution characteristics in 304 DIE patients, we also found that the distribution of DIE lesions was greater on the left side than on the right side 29 . The clinical importance of our finding is that counseling among patients with DIE should account for not only age, duration of GnRH-a use, and postoperative delivery but also the anatomic location of the lesion. Moreover, because the location of EMA is determined before surgery in most patients and the rate of recurrence is almost 6.5-fold higher in the case of left ovarian involvement, the performance of a more extended operation (i.e., monolateral or bilateral salpingo-oophorectomy) should be discussed with patients who do not wish to become pregnant. Additionally, for DIE, patients with left EMA should have more rigorous long-term management after surgery, and those who wish to become pregnant should be encouraged to try to conceive immediately after surgery. Beyond recurrence, it should be acknowledged that large epidemiological studies suggest endometriosis—particularly with ovarian endometrioma—confers a small but clinically meaningful increased risk of specific epithelial ovarian cancers (notably clear cell and endometrioid subtypes) 30 , 31 . Although the absolute risk remains low and malignant transformation is rare, this association underscores the need for long-term vigilance in surgically managed DIE patients, especially those with recurrent disease or persistent endometriomas 30 . Our finding that left ovarian endometrioma independently predicted recurrence further highlights a subgroup warranting continued monitoring. Future biomarker studies should explore potential links between recurrence biology and oncogenic pathways. In summary, we performed a long-term follow-up investigation extending beyond six years, during which the cumulative recurrence rate exhibited an upward trend over time. Our research revealed that left side EMA presence was independent risk factors of DIE patients for the recurrence after surgery during the long-time follow-up. Age, use of GnRH-a for 6 months and postoperative pregnancy were protective factors against recurrence. For younger DIE patients or those who have left-side EMA, we should pay more attention to postoperative medication management, and it is recommended that GnRH-a be used continuously for 6 months after surgery. We should also encourage patients with pregnancy plans to actively conceive after surgical treatment.

Endometriosis is defined as the presence of active endometrial tissue outside the uterine cavity 1 . According to histopathological classification, endometriosis can be categorized into three principal types: superficial endometriosis, ovarian endometrioma (EMA), and deep infiltrating endometriosis (DIE) 2 . DIE is specifically defined as the infiltration of endometrial glands and stroma beyond 5 mm into the subperitoneal tissue, predominantly affecting the rectovaginal septum, uterosacral ligaments, pararectal space, and vesicouterine pouch 3 . DIE involves a wide range of lesions and deep infiltration of lesions, the surgery is difficult and seriously affects the patient’s quality of life. Surgery is the definite and effective method for both the diagnosis and management of DIE. Post-excisional recurrence rates for DIE range from 2% to 43%, influenced by the duration of follow-up 4 . The reasons underlying this variability remain ambiguous, potentially attributable to differences in recurrence definitions, sample sizes, and the demographic characteristics of study populations. Postoperative follow-up studies on DIE are limited, and most of these studies had a small sample size or a follow-up duration of less than 5 years. Consequently, this study aims to assess the recurrence rates and identify the risk factors associated with recurrence following surgical intervention for DIE. By elucidating the factors that may contribute to recurrence, gynecologists can better identify patients at heightened risk, optimize the timing of surgical procedures, and effectively strategize both preoperative and postoperative management.

Below is the link to the electronic supplementary material. Supplementary Material 1 Supplementary Material 1