{"paper_id":"8a012aac-e486-4a8c-851d-9c8702c8e31c","body_text":"Article Discovery Medicine 2025; 37(203): 3064–3072\nhttps://doi.org/10.24976/Discov.Med.202537203.256\nCopyright: © 2025 The Author(s). Published by Discovery Medicine. This is an open access article under the CC BY 4.0 license .\nNote: Discovery Medicine stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.\nDevelopment of a 5-year Recurrence Risk Prediction\nModel After Conservative Surgery for Adenomyosis\nBased on Clinical and Imaging Features\nAnna Shen1,†, Xueqin Cao 2,†, Leilei He 1,*\n1Department of Gynaecology and Obstetrics, The Fourth Affiliated Hospital of Soochow University (Suzhou Dushu Lake Hospital), 215123 Suzhou,\nJiangsu, China\n2Department of Endocrinology, The Fourth Affiliated Hospital of Soochow University (Suzhou Dushu Lake Hospital), 215123 Suzhou, Jiangsu, China\n*Correspondence: helei0609@163.com (Leilei He)\n†These authors contributed equally.\nSubmitted: 23 September 2025 Revised: 30 October 2025 Accepted: 10 November 2025 Published: 20 December 2025\nBackground: Dysmenorrhea and menorrhagia are common consequences of adenomyosis. While conservative surgery can ef-\nfectively preserve fertility in women with adenomyosis, they are still vulnerable to postoperative recurrence, for which a reliable\nlong-term predictive tool is lacking. This study aimed to develop and validate a 5-year recurrence predictive model for adeno-\nmyosis patients after conservative surgery based on their clinical and imaging features.\nMethods: In this retrospective study, 150 women aged 18–50 years who underwent uterus-preserving surgery for adenomyosis\nwere analyzed. Clinical data, including imaging parameters, surgical characteristics, and postoperative management, were col-\nlected. Recurrence was defined as either a ≥3-point increase in Visual Analog Scale score for dysmenorrhea or a ≥50% increase\nin Pictorial Blood Assessment Chart score within five years. Multivariate Cox regression was used to construct a nomogram,\nwith its predictive performance evaluated using concordance index (C-index), time-dependent receiver operating characteristic\ncurves, calibration, and decision curve analysis (DCA).\nResults: Four independent predictors were identified: older age, larger uterine volume, shorter duration of postoperative hor-\nmonal therapy, and concomitant endometriosis. The nomogram demonstrated good discriminative ability (C-index 0.766; AUCs\n0.68, 0.73, 0.76 at 15, 24, and 48 months, respectively), along with reliable calibration and evident clinical net benefit. Kaplan–\nMeier analysis revealed that the nomogram effectively distinguished risk groups, with five-year recurrence-free survival rates of\n78% in the low-risk group and 17% in the high-risk group.\nConclusion: By integrating clinical and imaging variables, the nomogram developed in this study demonstrates strong clinical\napplicability, accurately predicting recurrence risk in adenomyosis patients after conservative surgery and guiding personalized\npostoperative management.\nKeywords: adenomyosis; conservative surgery; recurrence; nomogram; risk prediction\nIntroduction\nAdenomyosis is a common gynecological disorder\ncharacterized by the presence of ectopic endometrial glands\nand stroma within the myometrium, leading to progressive\nuterine enlargement, chronic pelvic pain, dysmenorrhea,\nand menorrhagia [ 1,2]. The condition affects up to 20–\n30% of women of reproductive age, imposing a substantial\nclinical and social burden due to impaired quality of life,\ninfertility, and the need for repeated interventions [ 3]. Al-\nthough hysterectomy remains the definitive treatment, most\npatients of childbearing age who wish to preserve fertility\nand maintain uterine integrity opt for conservative surgical\napproaches [4].\nConservative surgery is an important uterus-sparing\ntreatment option for adenomyosis, effectively relieving\npain and menorrhagia while preserving fertility in patients\nof reproductive age [ 4]. However, symptomatic recurrence\nremains a major clinical challenge in patients receiving sur-\ngical intervention, with approximately one-third of the pa-\ntients experiencing symptom relapse within five years, de-\npending on surgical technique, disease extent, and postop-\nerative management [ 5,6]. Identifying patients at higher\nrisk of recurrence is therefore essential to optimize post-\noperative hormonal therapy and follow-up strategies.\nPrevious studies have explored the predictive value of\nvarious clinical and surgical factors for adenomyosis recur-\nrence, such as age, parity, coexistence of endometriosis,\nextent of excision, and duration of postoperative hormonal\nsuppressive therapy [ 7,8]. In recent years, a growing body\nof studies have increasingly established imaging-derived\nparameters, including junctional zone (JZ) thickness, uter-\nine volume, and adenomyosis phenotype, as objective in-\n\n3065\ndicators of disease burden and myometrial involvement\n[9,10]. Nevertheless, most of these studies are limited by\nsmall sample sizes, heterogeneous inclusion criteria, rela-\ntively short follow-up durations, and inconsistent incorpo-\nration of imaging indicators, thereby diminishing the clini-\ncal applicability of the identified factors [ 11,12].\nCurrently, there is no long-term predictive model that\nincorporates imaging variables to assess postoperative re-\ncurrence in adenomyosis patients following conservative\nsurgery. Therefore, the present study aimed to develop and\ninternally validate a 5-year recurrence predictive model that\ncombines clinical and imaging features, including JZ thick-\nness, uterine volume, and adenomyosis phenotype, together\nwith key clinical parameters such as age, coexistence of en-\ndometriosis, and postoperative hormonal suppressive ther-\napy duration. This model is expected to provide a practi-\ncal tool for personalized postoperative risk assessment and\nmanagement.\nMethods\nStudy Design and Setting\nThis study was designed as a retrospective cohort anal-\nysis conducted at The Fourth Affiliated Hospital of Soo-\nchow University (Suzhou Dushu Lake Hospital), a tertiary\nreferral center specializing in gynecological surgery. A to-\ntal of 150 patients diagnosed with adenomyosis who un-\nderwent conservative surgical treatment between February\n2018 and February 2023 were screened for eligibility. The\nstudy period was defined to guarantee a minimum follow-\nup of 12 months for all patients by the database lock, with\na subset of patients followed for up to five years. All rel-\nevant clinical, surgical, imaging, and follow-up data were\nextracted from the institutional electronic medical record\nsystem and imaging archives.\nStudy Population\nParticipants were women aged 18–50 years who had a\nconfirmed diagnosis of adenomyosis. The diagnosis was\nestablished on the basis of either clinical manifestations\ncombined with imaging findings according to the Morpho-\nlogical Uterus Sonographic Assessment (MUSA) criteria\nor magnetic resonance imaging (MRI), or histopathologi-\ncal evidence when available [ 13,14]. Both diffuse and fo-\ncal types of adenomyosis were included. Eligible patients\nunderwent conservative uterus-preserving surgery without\nhysterectomy, including focal adenomyotic lesion excision\nor uterine reconstructive repair. Furthermore, only those\nwith standardized postoperative follow-up records from\noutpatient visits or structured telephone interviews, and\nwith sufficient documentation of symptom changes for at\nleast 12 months, were included in the analysis.\nPatients were excluded if they had a prior hysterec-\ntomy or underwent hysterectomy during index admission.\nAdditional exclusion criteria include: (1) diagnosis of ma-\nlignancy or severe systemic disease during follow-up; (2)\nmissing data exceeding 40% in any of the key variables,\nincluding age, body mass index (BMI), coexistence of en-\ndometriosis, JZ thickness, uterine volume, adenomyosis\nphenotype, and duration of postoperative hormonal sup-\npressive therapy; and (3) the need for re-intervention within\nthree months after conservative surgery for reasons other\nthan perioperative complications, such as suspected resid-\nual adenomyotic lesions confirmed by imaging, uncon-\ntrolled symptoms unresponsive to early medical therapy, or\nnewly developed severe uterine bleeding unrelated to sur-\ngical complications.\nData Collection and V ariables\nData were extracted from the institutional electronic\nmedical record system, operative notes, imaging archives,\nand follow-up records. V ariables were selected for analy-\nsis based on their clinical significance, previously reported\nassociations with adenomyosis recurrence, and availability\nwithin the dataset. Clinical variables (e.g., age, BMI, re-\nproductive history [pregnancy and parity], concomitant en-\ndometriosis, and postoperative hormonal therapy duration)\nwere chosen because they have been widely reported to in-\nfluence disease recurrence or response to treatment. Imag-\ning variables, including JZ thickness, uterine volume, and\nadenomyosis phenotype, were incorporated as objective in-\ndicators of disease burden and myometrial involvement ac-\ncording to published imaging studies [ 5,9]. All imaging\nassessments were independently performed by two expe-\nrienced radiologists, and discrepancies were resolved by a\nthird reviewer to ensure consistency.\nSurgical variables included the extent of lesion ex-\ncision (categorized as localized or extensive), the number\nof myometrial repair layers, and the method of hemosta-\nsis (categorized as suture, energy-based, or combined\ntechniques). Data on postoperative management encom-\npassed the type of hormonal suppressive therapy, such\nas gonadotropin-releasing hormone agonists, dienogest,\nlevonorgestrel-releasing intrauterine system, or combined\noral contraceptives, as well as the total duration of treat-\nment in months.\nOutcome Definition\nThe primary outcome of this study was symptom re-\ncurrence within five years after conservative surgery for\nadenomyosis. Recurrence was defined using standardized\nclinical criteria. Dysmenorrhea severity was assessed using\nthe Visual Analogue Scale (V AS), a validated and widely\nused scoring system for pain intensity [ 15]. Menstrual\nblood loss was quantified using the Pictorial Blood Assess-\nment Chart (PBAC), originally developed for objective es-\ntimation of menstrual blood volume [ 16]. Recurrence was\ndefined as an increase of ≥3 points in V AS score or ≥50%\nincrease in PBAC score from the patient’s lowest postoper-\native score during follow-up. Time to recurrence was cal-\n\n3066\nFig. 1. Nomogram for predicting postoperative recurrence after conservative surgery for adenomyosis. The nomogram includes\nsix variables: age, junctional zone (JZ) thickness, uterine volume (UterusV ol), adenomyosis phenotype (AM type; 0 = focal, 1 = diffuse,\n2 = mixed), duration of postoperative hormonal suppressive therapy (TxDuration), and concomitant endometriosis (0 = no, 1 = yes).\nculated from the date of surgery to the first documented oc-\ncurrence of symptom relapse, regardless of whether it was\ndysmenorrhea or menorrhagia. Patients without recurrence\nat the end of follow-up were censored at their last visit.\nStatistical Analysis\nAll statistical analyses were performed using IBM\nSPSS Statistics 26.0 (IBM Corp., Armonk, NY , USA)\nand R software version 4.3.2 (R Foundation for Statis-\ntical Computing, Vienna, Austria). Baseline character-\nistics were summarized descriptively. Continuous vari-\nables were tested for normality using the Shapiro–Wilk test.\nThe Mann–Whitney U test was employed to analyze inter-\ngroup data that do not follow a normal distribution, which\nare expressed as medians and interquartile ranges (IQRs),\nwhereas the Chi-square test was used to compare categor-\nical variables, which are presented as counts and percent-\nages.\nUnivariate Cox proportional hazards regression anal-\nyses were conducted to evaluate the association between\neach variable and the risk of recurrence. V ariables with\nclinical significance or p < 0.05 in the univariate analysis\nwere subsequently entered into multivariable Cox regres-\nsion to identify independent predictors. A nomogram that\nenables individualized prediction of recurrence risk at dif-\nferent postoperative time points was then developed based\non the final multivariable Cox model. V ariables of clinical\nsignificance, such as JZ thickness and adenomyosis pheno-\ntype, were retained in the final model regardless of their\nstatistical significance, in accordance with Transparent Re-\nporting of a Multivariable Prediction Model for Individual\nPrognosis or Diagnosis recommendations [ 17].\nModel performance was assessed in terms of its dis-\ncriminative ability, calibration, and clinical utility. Dis-\ncriminative performance of the model was evaluated by cal-\nculating the concordance index (C-index) and plotting time-\ndependent receiver operating characteristic (ROC) curves\nto determine the corresponding area under the curve (AUC).\nCalibration was examined using bootstrap resampling with\n1000 iterations, and plots were generated to compare pre-\ndicted and observed probabilities of recurrence. Decision\ncurve analysis (DCA) was applied to estimate the net clini-\ncal benefit of the predictive model across a range of thresh-\nold probabilities. Finally, patients were stratified into high-\nand low-risk groups according to their absolute predicted\nrecurrence risk at 24 months, and Kaplan–Meier survival\ncurves were used to compare recurrence-free survival be-\ntween groups, with significance determined by the log-rank\ntest.\n\n3067\nTable 1. Baseline characteristics.\nOverall\n(n = 150)\nNon-recurrence group\n(n = 97)\nRecurrence group\n(n = 53) Z/χ2 p\nAge (years) 34.00 [30.00, 37.00] 33.00 [29.00, 37.00] 34.00 [31.00, 39.00] –1.712 0.087\nBMI (kg/m2) 22.10 [19.52, 24.40] 22.60 [20.00, 24.80] 21.10 [19.30, 23.20] –1.980 0.048\nDuration of follow-up (months) 27.00 [18.00, 38.00] 30.00 [18.00, 42.00] 24.00 [18.00, 38.00] –0.538 0.591\nJZ thickness (mm) 14.85 [12.93, 16.80] 13.90 [12.00, 15.50] 16.30 [14.90, 18.00] –5.295 <0.001\nUterine volume (mL) 178.00 [141.00, 238.00] 161.00 [128.00, 201.00] 238.00 [193.00, 304.00] –5.772 <0.001\nTxDuration (months) 7.00 [2.00, 11.00] 10.00 [6.00, 12.00] 2.00 [0.00, 6.00] –5.958 <0.001\nPregnancy (%) 0.029 0.864\nNo 15 (10.0) 10 (10.3) 5 (9.4)\nY es 135 (90.0) 87 (89.7) 48 (90.6)\nCesarean section (%) 0.035 0.852\nNo 41 (27.3) 27 (27.8) 14 (26.4)\nY es 109 (72.7) 70 (72.2) 39 (73.6)\nEndometriosis (%) 45.730 <0.001\nNo 89 (59.3) 77 (79.4) 12 (22.6)\nY es 61 (40.7) 20 (20.6) 41 (77.4)\nAdenomyosis phenotype (%) 26.795 <0.001\nDiffuse 76 (50.7) 34 (35.1) 42 (79.2)\nFocal 42 (28.0) 36 (37.1) 6 (11.3)\nMixed 32 (21.3) 27 (27.8) 5 (9.4)\nExtent of lesion excision (%) 1.933 0.164\nLocalized 65 (43.3) 38 (39.2) 27 (50.9)\nExtensive 85 (56.7) 59 (60.8) 26 (49.1)\nRepair layers (%) 4.297 0.117\n1 13 (8.7) 5 (5.2) 8 (15.1)\n2 78 (52.0) 52 (53.6) 26 (49.1)\n≥3 59 (39.3) 40 (41.2) 19 (35.8)\nHemostasis method (%) 0.381 0.827\nSuture 52 (34.7) 32 (33.0) 20 (37.7)\nEnergy-based 31 (20.7) 21 (21.6) 10 (18.9)\nCombined 67 (44.7) 44 (45.4) 23 (43.4)\nPostopTx (%) 12.298 0.015\nGnRH agonist 26 (17.3) 18 (18.6) 8 (15.1)\nDienogest 33 (22.0) 24 (24.7) 9 (17.0)\nLNG-IUS 33 (22.0) 25 (25.8) 8 (15.1)\nCOCs 25 (16.7) 17 (17.5) 8 (15.1)\nNone 33 (22.0) 13 (13.4) 20 (37.7)\nNotes: V alues that do not conform to the normal distribution are presented as median [IQR].\nAbbreviations: BMI, body mass index; COCs, combined oral contraceptives; GnRH, gonadotropin-releasing hormone; JZ, junctional\nzone; LNG-IUS, levonorgestrel-releasing intrauterine system; TxDuration, duration of postoperative hormonal therapy; PostopTx, post-\noperative treatment.\nResults\nBaseline Characteristics\nA total of 150 patients were included, comprising 97\n(64.7%) in the non-recurrence group and 53 (35.3%) in\nthe recurrence group (Table 1). Compared with the non-\nrecurrence group, patients who experienced recurrence had\na significantly lower BMI (21.1 vs. 22.6 kg/m 2, p = 0.048),\ngreater JZ thickness (16.3 vs. 13.9 mm, p < 0.001), and\nlarger uterine volume (238 vs. 161 mL, p < 0.001). The dif-\nfuse type of adenomyosis was significantly more common\nin the recurrence group (79.2% vs. 35.1%), whereas focal\nand mixed types predominated in the non-recurrence cases\n(p < 0.001). In addition, patients with recurrence showed\na markedly higher prevalence of concomitant endometrio-\nsis (77.4% vs. 20.6%, p < 0.001) and a shorter duration of\npostoperative hormonal suppressive therapy (median 2 vs.\n10 months, p < 0.001). The proportion of patients without\npostoperative treatment was also higher in the recurrence\ngroup (37.7% vs. 13.4%, p = 0.015). Other baseline char-\n\n3068\nFig. 2. Model performance evaluation. (A–C) Time-dependent receiver operating characteristic (ROC) curves at 15, 24, and 48\nmonths with area under the curve (AUC) values of 0.68, 0.73, and 0.76, respectively. (D–F) Calibration plots for 15-, 24-, and 48-month\nrecurrence probabilities showing agreement between predicted and observed outcomes.\nacteristics, including age, parity, history of cesarean sec-\ntion, extent of lesion excision, number of repair layers, and\nhemostasis method, showed no significant differences be-\ntween the two groups (all p > 0.05).\nUnivariate and Multivariate Cox Regression Analysis\nUnivariate Cox regression showed that age (Hazard\nRatio [HR] = 1.072, 95% CI = 1.015–1.132, p = 0.013), JZ\nthickness (HR = 1.160, 95% CI = 1.063–1.266, p < 0.001),\nuterine volume (HR = 1.005, 95% CI = 1.003–1.008, p <\n0.001), adenomyosis phenotype (HR = 0.421, 95% CI =\n0.267–0.665, p < 0.001), concomitant endometriosis (HR\n= 4.850, 95% CI = 2.548–9.233, p < 0.001), postoperative\ntreatment (HR = 1.265, 95% CI = 1.035–1.547, p = 0.022),\nand treatment duration (HR = 0.857, 95% CI = 0.804–0.913,\np < 0.001) were significantly associated with recurrence.\nBased on the multivariate Cox regression results, older\nage (HR = 1.069, 95% CI = 1.004–1.138, p = 0.037), larger\nuterine volume (HR = 1.003, 95% CI = 1.001–1.006, p\n= 0.018), a significantly shorter duration of postoperative\ntherapy (HR = 0.905, 95% CI = 0.843–0.971, p = 0.006),\nand concomitant endometriosis (HR = 2.564, 95% CI =\n1.27–5.178, p = 0.009) remained significantly associated\nwith recurrence, making them the four independent predic-\ntors of recurrence. During the analysis, JZ thickness, ade-\nnomyosis phenotype, and postoperative treatment lost sta-\ntistical significance following adjustment (Table 2).\nNomogram Development and Model Performance\nV ariables with statistical significance (age, uterine\nvolume, duration of postoperative hormonal suppressive\ntherapy, and concomitant endometriosis) in the multivari-\nate Cox regression analysis were included in the nomogram\n(Fig. 1). In addition, JZ thickness and adenomyosis phe-\nnotype were incorporated because of their well-recognized\nclinical relevance as imaging indicators of disease burden\nand morphological subtype, despite not reaching statistical\nsignificance in multivariate analysis. The nomogram pro-\nvided an individualized prediction of recurrence risk at 15,\n24, and 48 months after surgery, with higher total scores\ncorresponding to higher recurrence probabilities.\nThe discriminative ability of the model was accept-\nable, with a C-index of 0.766 in the overall cohort. Boot-\nstrap validation with 1000 resamples confirmed robust in-\nternal validity, yielding a corrected C-index of 0.767. Time-\ndependent ROC analyses demonstrated AUC values of 0.68\n\n3069\nTable 2. Univariate and multivariate Cox regression analysis of risk factors for recurrence after conservative surgery for\nadenomyosis.\nV ariable Univariate analysis Multivariate analysis\nHR (95% CI) p HR (95% CI) p\nAge 1.072 (1.015–1.132) 0.013 1.069 (1.004–1.138) 0.037\nBMI 0.916 (0.833–1.008) 0.072\nPregnancy 1.063 (0.423–2.673) 0.896\nCesarean section 1.195 (0.648–2.203) 0.568\nEndometriosis 4.850 (2.548–9.233) <0.001 2.564 (1.27–5.178) 0.009\nDuration of follow-up 0.000 (0–Inf) 0.984\nJZ thickness 1.160 (1.063–1.266) <0.001 1.107 (0.98–1.251) 0.102\nUterine volume 1.005 (1.003–1.008) <0.001 1.003 (1.001–1.006) 0.018\nAdenomyosis phenotype 0.421 (0.267–0.665) <0.001 0.73 (0.427–1.249) 0.250\nExtent of lesion excision 0.737 (0.429–1.265) 0.268\nRepair layers 0.731 (0.48–1.113) 0.144\nHemostasis method 0.952 (0.702–1.29) 0.749\nPostopTx 1.265 (1.035–1.547) 0.022 0.957 (0.785–1.167) 0.665\nTxDuration 0.857 (0.804–0.913) <0.001 0.905 (0.843–0.971) 0.006\nAbbreviations: HR, hazard ratio; CI, confidence interval; Inf, infinity.\nFig. 3. Decision curve analysis (DCA) of the recurrence-prediction model. (A) DCA curves showing the net benefit of the Cox\nmodel, clinical rules, “all”, and “none” strategies across a range of threshold probabilities. (B) DCA at the fixed 24-month time point,\nillustrating the net benefit of the Cox model compared with clinical rules, “all”, and “none” strategies.\n(95% CI: 0.644–0.721), 0.73 (95% CI: 0.691–0.763), and\n0.76 (95% CI: 0.719–0.795) at 15, 24, and 48 months, re-\nspectively (Fig. 2A–C), indicating stable predictive perfor-\nmance over time. Calibration plots for 15-, 24-, and 48-\nmonth predictions (Fig. 2D–F) showed close agreement be-\ntween predicted and observed probabilities, suggesting no\nsignificant overfitting. The numbers of patients remaining\nat risk at these time points were approximately 140, 125,\nand 100, respectively, indicating adequate sample sizes to\nensure reliable calibration, particularly at the later time\npoint. Decision curve analysis further indicated that the\nnomogram consistently provided a greater net clinical ben-\nefit than the traditional clinical rules when the threshold\nprobability ranged between 0.15 and 0.45 (Fig. 3).\nRisk Stratification\nA 24-month predicted recurrence risk threshold of\n≥30% was applied, which was selected because it falls\n\n3070\nFig. 4. Kaplan–Meier curves of recurrence-free survival stratified by 24-month predicted recurrence risk. Patients with a predicted\nrisk of ≥30% at 24 months (high-risk group, n = 31) had significantly higher recurrence rates than those with <30% risk (low-risk group,\nn = 119).\nwithin the DCA-identified optimal threshold probability\nrange, where the model provides the greatest net clinical\nbenefit (Fig. 3B). Using this threshold, 31 patients (20.7%)\nwere classified as high-risk and 119 (79.3%) as low-risk.\nThe five-year recurrence-free survival was 78% in the low-\nrisk group and 17% in the high-risk group (Fig. 4), with a\nstatistically significant difference between the two groups\n(log-rank p < 0.0001).\nDiscussion\nIn this study, a nomogram integrating clinical and\nimaging parameters was developed and internally validated\nto predict 5-year postoperative recurrence after conserva-\ntive surgery for adenomyosis. The model showed good\ndiscriminative ability (C-index 0.766) and calibration, with\ntime-dependent AUCs increasing from 0.68 to 0.76 across\n15–48 months, reflecting stable and reliable performance.\nThese findings suggest that the selected variables—age,\nuterine volume, JZ thickness, adenomyosis phenotype, du-\nration of postoperative hormonal suppressive therapy, and\ncoexistence of endometriosis—may serve as valuable pre-\ndictors for long-term recurrence risk.\nThe recurrence rate of 35.3% observed in this study\naligns with previous research [5], which showed recurrence\nrates ranging from 30% to 40% following conservative\nsurgery. Several studies have identified age, uterine size,\nand incomplete lesion resection as major risk factors influ-\nencing postoperative outcomes [ 5,18]. Our findings con-\nfirm that larger uterine volume and shorter duration of post-\noperative hormonal suppressive therapy are independently\nassociated with higher recurrence risk, consistent with ear-\nlier literature emphasizing disease burden and insufficient\nhormonal suppression as key contributors.\nImaging indicators such as JZ thickness and adeno-\nmyosis phenotype have been recognized as objective mark-\ners of disease severity and morphological subtype [ 9]. Al-\nthough these parameters were not statistically significant in\nmultivariate analysis, they were retained in the final model\ndue to their strong biological plausibility and extensive\nprior validation as imaging indicators of myometrial inva-\nsion and treatment response [ 19]. Moreover, the model’s\nslight improvement in discriminative performance at later\nfollow-up time points (24–48 months) may reflect the de-\nlayed manifestation of true recurrences, as transient post-\noperative inflammatory or hormonal changes could obscure\nearly symptom differentiation [ 5,20].\nThe current nomogram offers a clinically practical ap-\nproach for individualized postoperative management. Pa-\ntients classified as high-risk may benefit from prolonged\npostoperative hormonal suppressive therapy, early imaging\nsurveillance, or fertility counseling. By integrating both\n\n3071\nclinical and imaging markers, the model connects tradi-\ntional risk factor analysis with individualized recurrence\nprediction, thereby facilitating evidence-based follow-up\nplanning.\nThe main strengths of this study include its relatively\nlarge, well-characterized surgical cohort, the use of stan-\ndardized imaging protocols, and long-term (5-year) follow-\nup data, which enhance the robustness of the findings.\nHowever, several limitations should be noted. As a single-\ncenter retrospective study, potential selection bias and in-\ncomplete control of confounding factors cannot be fully ex-\ncluded. The model was only internally validated, and exter-\nnal validation in larger, multicenter prospective cohorts is\nneeded to confirm its generalizability. In addition, hetero-\ngeneity in surgical techniques and postoperative treatment\nstrategies may have influenced the research outcomes, and\nimaging parameters could be affected by interobserver vari-\nability. Furthermore, molecular biomarkers and advanced\nimaging features such as radiomics, which have additional\npredictive value, were not included in our analysis. Future\nresearch should therefore focus on multicenter prospective\nvalidation, incorporation of molecular and radiomic signa-\ntures, and development of dynamic predictive tools that up-\ndate risk over time. Such efforts may refine individual-\nized risk assessment and ultimately improve long-term out-\ncomes for women undergoing conservative surgery for ade-\nnomyosis.\nConclusion\nThis study highlights the value of integrating clinical\nand imaging characteristics into a nomogram for predicting\nrecurrence risk after conservative surgery for adenomyosis.\nThe proposed model serves as a practical tool for individ-\nualized postoperative management, supporting more accu-\nrate risk stratification and clinical decision-making.\nAvailability of Data and Materials\nThe data used to support the findings of this study are\navailable from the corresponding author upon request.\nAuthor Contributions\nStudy concept and design: ANS and LLH; Anal-\nysis and interpretation of data: XQC; Drafting of the\nmanuscript: ANS; Critical revision of the manuscript for\nimportant intellectual content: ANS, XQC and LLH; Sta-\ntistical analysis: XQC; Study supervision: all authors. All\nauthors have read and approved the final manuscript and\nagreed to be accountable for all aspects of the work, ensur-\ning that questions related to the accuracy or integrity of any\npart of the work are appropriately investigated and resolved.\nEthics Approval and Consent to Participate\nThis study was approved by The Fourth Affiliated\nHospital of Soochow University (Suzhou Dushu Lake Hos-\npital) (2024-241023). Given the retrospective nature of the\nanalysis, the requirement to obtain patients’ informed con-\nsent was waived by the ethics committee. All data were\nanonymized to protect patient confidentiality in accordance\nwith the principles of the Declaration of Helsinki.\nAcknowledgment\nNot applicable.\nFunding\nThis research received no external funding.\nConflict of Interest\nThe authors declare no conflict of interest.\nReferences\n[1] Moawad G, Fruscalzo A, Y oussef Y , Kheil M, Tawil T, Nehme\nJ, et al . Adenomyosis: An Updated Review on Diagnosis and\nClassification. Journal of Clinical Medicine. 2023; 12: 4828.\nhttps://doi.org/10.3390/jcm12144828.\n[2] Cui L, Sang C, Li R, Zhao S. Inhibition of lysine-specific his-\ntone demethylase 1A suppresses adenomyosis through reduction\nin ectopic endometrial stromal cell proliferation, migration, and\ninvasion. CytoJournal. 2024; 21: 50. https://doi.org/10.25259/\nCytojournal_48_2024.\n[3] Martire FG, d’Abate C, Schettini G, Cimino G, Ginetti A,\nColombi I, et al . Adenomyosis and Adolescence: A Challeng-\ning Diagnosis and Complex Management. Diagnostics (Basel,\nSwitzerland). 2024; 14: 2344. https://doi.org/10.3390/diagnost\nics14212344.\n[4] Maheshwari A, Gurunath S, Fatima F, Bhattacharya S. Adeno-\nmyosis and subfertility: a systematic review of prevalence, di-\nagnosis, treatment and fertility outcomes. Human Reproduction\nUpdate. 2012; 18: 374–392. https://doi.org/10.1093/humupd/d\nms006.\n[5] Lu K, Zhong G, Lian B, Zhong X, Xie M, Wu Y . Recurrence\nrates and associated risk factors after conservative surgery for\nadenomyosis: a retrospective study. BMC Women’s Health.\n2024; 24: 619. https://doi.org/10.1186/s12905-024-03457-6 .\n[6] Ioannidou A, Louis K, Sioutis D, Panagopoulos P , Theofanakis\nC, Machairiotis N. Conservative Surgical Management of Ade-\nnomyosis: Implications for Infertility and Pregnancy Outcomes-\nA Perspective Review. Journal of Clinical Medicine. 2025; 14:\n6956. https://doi.org/10.3390/jcm14196956.\n[7] Fan LX, Zhang Y , Y ang LL, Ji XL, Wang Y , Huang YF, et\nal. Analysis of related factors influencing postoperative recur-\nrence of adenomyosis treated with HIFU. Archives of Gynecol-\nogy and Obstetrics. 2024; 309: 1765–1773. https://doi.org/10.\n1007/s00404-023-07340-x .\n[8] Holdsworth-Carson SJ, Chung J, Machalek DA, Li R, Jun BK,\nGriffiths MJ, et al. Predicting disease recurrence in patients with\nendometriosis: an observational study. BMC Medicine. 2024;\n22: 320. https://doi.org/10.1186/s12916-024-03508-7 .\n[9] Harmsen MJ, Trommelen LM, de Leeuw RA, Tellum T, Juf-\nfermans LJM, Griffioen AW, et al. Uterine junctional zone and\n\n3072\nadenomyosis: comparison of MRI, transvaginal ultrasound and\nhistology. Ultrasound in Obstetrics & Gynecology: the Official\nJournal of the International Society of Ultrasound in Obstetrics\nand Gynecology. 2023; 62: 42–60. https://doi.org/10.1002/uog.\n26117.\n[10] Rees CO, Nederend J, Mischi M, van Vliet HAAM, Schoot BC.\nObjective measures of adenomyosis on MRI and their diagnostic\naccuracy-a systematic review & meta-analysis. Acta Obstetricia\net Gynecologica Scandinavica. 2021; 100: 1377–1391. https:\n//doi.org/10.1111/aogs.14139.\n[11] Maghsoudlou P , Chatroux LR, Ajao MO, King LP , Groszmann\nYS. Diagnostic Accuracy of Ultrasound for Adenomyosis-A\nBlinded, Retrospective, Single-Expert V alidation Study of the\n2022 MUSA Guidelines. Journal of Minimally Invasive Gyne-\ncology. 2025; 32: 996–1001.e1. https://doi.org/10.1016/j.jmig\n.2025.07.024.\n[12] An XL, Zhang J, Y un XC, Lei XH, Zhao J, Y ang XH,et al. Per-\nspectives in transvaginal sonography for the diagnosis of ade-\nnomyosis. Frontiers in Medicine. 2025; 12: 1533648. https:\n//doi.org/10.3389/fmed.2025.1533648.\n[13] Kho KA, Chen JS, Halvorson LM. Diagnosis, Evaluation, and\nTreatment of Adenomyosis. JAMA. 2021; 326: 177–178. https:\n//doi.org/10.1001/jama.2020.26436.\n[14] Dason ES, Maxim M, Sanders A, Papillon-Smith J, Ng D,\nChan C, et al . Guideline No. 437: Diagnosis and Manage-\nment of Adenomyosis. Journal of Obstetrics and Gynaecol-\nogy Canada: JOGC = Journal D’obstetrique et Gynecologie du\nCanada: JOGC. 2023; 45: 417–429.e1. https://doi.org/10.1016/\nj.jogc.2023.04.008.\n[15] Huskisson EC. Measurement of pain. Lancet (London,\nEngland). 1974; 2: 1127–1131. https://doi.org/10.1016/\ns0140-6736(74)90884-8 .\n[16] Higham JM, O’Brien PM, Shaw RW. Assessment of menstrual\nblood loss using a pictorial chart. British Journal of Obstetrics\nand Gynaecology. 1990; 97: 734–739. https://doi.org/10.1111/\nj.1471-0528.1990.tb16249.x.\n[17] Moons KGM, Altman DG, Reitsma JB, Ioannidis JPA,\nMacaskill P , Steyerberg EW, et al . Transparent Reporting of a\nmultivariable prediction model for Individual Prognosis or Di-\nagnosis (TRIPOD): explanation and elaboration. Annals of In-\nternal Medicine. 2015; 162: W1–73. https://doi.org/10.7326/\nM14-0698.\n[18] He W, He M, Li J, Hou T, Y ang H, Lei N, et al. Improved sur-\ngical outcomes in adenomyosis: a novel insight into the superi-\nority of secretory phase adenomyomectomy. Archives of Gyne-\ncology and Obstetrics. 2025; 312: 919–925. https://doi.org/10.\n1007/s00404-025-08084-6 .\n[19] Turtóczki K, Cho H, Dastaran S, Kaposi PN, Tömösváry Z, Vár-\nbíró S, et al . Evaluation of junctional zone differential and ra-\ntio as possible markers of clinical efficacy in uterine artery em-\nbolisation of adenomyosis. CVIR Endovascular. 2024; 7: 55.\nhttps://doi.org/10.1186/s42155-024-00468-0 .\n[20] Zhao TT, Pang LL, Y ang LL, Li RN, Fan LX, Wen Y . Efficacy of\nhigh-intensity focused ultrasound combined with LNG-IUS for\nadenomyosis: a systematic review and meta-analysis. Archives\nof Gynecology and Obstetrics. 2023; 308: 351–362. https://doi.\norg/10.1007/s00404-022-06720-z .","source_license":"CC0","license_restricted":false}