Transvaginal ultrasound for detecting parametrial involvement in suspected deep pelvic endometriosis: updated meta-analysis

This updated systematic review and meta‐analysis was performed according to the Synthesizing Evidence from Diagnostic Accuracy TEsts (SEDATE) guideline and the Preferred Reporting Items for Systematic Reviews and Meta‐Analyses (PRISMA) statement 13 , 14 . The protocol was registered in the PROSPERO database (registration number: CRD42024581246). Institutional review board approval was waived because of the nature and design of the study. We used data from our previous meta‐analysis 11 and conducted an additional search in three databases (Web of Science, PubMed and Scopus) for studies published between January 2021 and May 2024 evaluating TVS for detecting parametrial involvement in women with suspected deep pelvic endometriosis. The search terms used in all databases were as follows: ‘endometriosis’, ‘parametrium’, ‘parametrial’, ‘ultrasound’ and ‘sonography’. No language restriction was applied. The search strategy is shown in Appendix  S1 . Three authors ( J.L.A., C.M., C.Can.) screened the titles and abstracts of reports identified by the literature search to exclude articles not strictly related to the topic of the review, narrative reviews, Letters to the Editor and case reports. Full‐text versions of the remaining articles were obtained to identify potentially eligible studies according to the following inclusion criteria: (1) participants were premenopausal women with a clinical suspicion of deep pelvic endometriosis; (2) the index test was TVS; (3) the reference standard was laparoscopy with or without histological confirmation; and (4) sufficient data were reported to construct a 2 × 2 table for parametrial involvement. We reviewed the reference lists of the selected studies to identify further potentially eligible articles. In the case of missing relevant data, we sought to contact the authors to solicit this information. For each study included in the meta‐analysis, the following data were retrieved: year of publication, country, study design, type of series (consecutive or not), number of patients, mean age of patients, inclusion and exclusion criteria, TVS scanning technique, sonographic definition of parametrial involvement, number of patients with parametrial involvement, laterality of parametrial involvement, numbers of true positives, true negatives, false positives and false negatives, number of participating centers, number of examiners, experience level of examiners, blinding of examiners, definition of reference standard and time elapsed between ultrasound and surgery. We focused specifically on the lateral parametrium. Risk‐of‐bias assessment of the included studies was conducted using the Quality Assessment of Diagnostic Accuracy Studies‐2 (QUADAS‐2) tool 15 , which includes four domains: (1) patient selection; (2) index test; (3) reference standard; and (4) flow and timing. Three authors (C.M., C.Can., C.Cat.) independently assessed the methodological quality using a standard form with quality assessment criteria. Disagreements were resolved by discussion with a fourth author (J.L.A.) to reach a consensus. The authors determined the risk of selection bias based on the description of the inclusion and exclusion criteria of the studies. The descriptions of the TVS scanning technique and criteria used to establish the presence of disease were used to assess the index test domain. To evaluate the reference standard domain, the method used to determine the presence of endometriosis in the parametrium (laparoscopy alone or with histological confirmation) was assessed. For the flow and timing domain, the time elapsed between the index test assessment and the reference standard result was evaluated. Concerns regarding applicability to the review question were evaluated for the patient selection, index test and reference standard domains. The primary outcomes of the quantitative synthesis were the pooled sensitivity, specificity, positive likelihood ratio, negative likelihood ratio and diagnostic odds ratio of TVS in the detection of parametrial endometriosis. Post‐test probabilities were calculated and plotted on Fagan nomograms. Forest plots were created, and the I 2 index was used to assess for the presence of heterogeneity. Meta‐regression was conducted if heterogeneity was found ( I 2  > 60%). The variables assessed for meta‐regression were year of publication, number of patients in the series, prevalence of parametrial involvement and whether or not laterality assessment was performed. A summary receiver‐operating‐characteristics (sROC) curve was plotted to illustrate the relationship between sensitivity and specificity. A subgroup analysis was performed to determine the diagnostic performance of TVS according to the laterality of the parametrial involvement (left vs right). We decided to assess publication bias only if 10 or more studies were found. Statistical analysis was performed using Meta‐analytical Integration of Diagnostic Accuracy Studies (MIDAS) and METANDI commands in STATA version 12 for Windows (Stata Corp., College Station, TX, USA). Forest plots were developed using RevMan 5.4.1 software (Cochrane Collaboration, 2020). P  < 0.05 was considered statistically significant.

In this updated meta‐analysis, the literature search yielded 115 records (Figure  1 ). After 36 duplicates had been removed, the titles and abstracts of the remaining 79 records were screened. Sixty‐three records were excluded (58 were not related to the topic of the review, four were narrative reviews and one was a case report). The full texts of the remaining 16 articles were reviewed. Twelve articles were further excluded (two narrative reviews, two studies not related to the topic of the review, four studies that did not report data on the lateral parametrium, one Letter to the Editor and three studies that did not report sufficient data to construct a 2 × 2 table), and four studies were selected for inclusion 16 , 17 , 18 , 19 . Flowchart summarizing inclusion of studies in meta‐analysis. In our previous meta‐analysis, we had included four studies 20 , 21 , 22 , 23 . Therefore, eight studies were ultimately included in this updated meta‐analysis 16 , 17 , 18 , 19 , 20 , 21 , 22 , 23 . Excluded studies are listed in Appendix  S2 . The eight included studies reported data from 6728 women (13 456 parametria). The number of parametria with endometriotic involvement was 4225 (mean prevalence, 21.3% (range, 2.5–39.3%)) (Table  1 ). Main characteristics of studies included in meta‐analysis on diagnostic accuracy of transvaginal ultrasound (TVS) for detecting parametrial involvement in women with suspected deep pelvic endometriosis Only first author is given for each study. Study did not specify whether diagnosis was confirmed by histology (H). Median (range). LP, left parametrium; LPS, laparoscopic surgical findings; NA, not applicable; NS, not stated; Prosp, prospective; Retro, retrospective; RP, right parametrium; TAS, transabdominal ultrasound; US, ultrasound. Five studies had a prospective design 17 , 18 , 19 , 20 , 23 and three were retrospective analyses 16 , 21 , 22 . All studies had clearly defined inclusion criteria, but two did not define exclusion criteria 16 , 19 . All studies used TVS as the index test, but one added transabdominal ultrasound 17 . Four studies were performed in a single center 16 , 17 , 21 , 22 , three studies were multicentric 19 , 20 , 23 (although in one study, the same examiner performed all ultrasound scans 20 ) and in one study this information was not available 18 . Information regarding which parametrium was involved (left and/or right) was reported in five studies 16 , 17 , 18 , 19 , 20 . The right parametrium was involved in 1886 women (mean prevalence, 23.9% (range, 8.4–39.0%)) and the left parametrium was involved in 2274 women (mean prevalence, 27.8% (range, 7.7–41.0%)). All studies reported data on the lateral parametrium and three also reported data on the ventral and dorsal parametria 16 , 17 , 19 . We performed the quantitative synthesis only for the lateral parametrium. Evaluation of study quality using the QUADAS‐2 tool is summarized in Table  2 . Four studies were considered as high risk for selection bias because of the study design, inadequate patient exclusion or lack of defined exclusion criteria 16 , 18 , 19 , 22 . Six studies were considered as low risk for the index test domain because they described clearly the TVS scanning technique and criteria for parametrial involvement 16 , 17 , 18 , 19 , 20 , 21 . One study was considered to have an unclear risk of bias in this domain because these characteristics were not described adequately 23 . The remaining study was considered high risk in this domain because the parametrium was specified as the broad ligament without describing the identification of the structure 22 . Quality assessment of studies included in meta‐analysis using Quality Assessment of Diagnostic Accuracy Studies‐2 (QUADAS‐2) tool Only first author is given for each study. Six studies were considered to have low risk of bias for the reference standard 16 , 17 , 18 , 19 , 20 , 21 . One study was considered to have an unclear risk, as it did not specify whether the diagnosis was confirmed by histology 22 . In the remaining study, confirmation was obtained only by surgery, and it was therefore considered high risk 23 . Regarding flow and timing, two studies did not provide relevant information and so were considered to have unclear risk of bias in this domain 18 , 23 . One study was considered as high risk because it included patients for whom surgery was performed more than 180 days after ultrasound evaluation 21 . The remaining five studies were considered as low risk 16 , 17 , 19 , 20 , 22 . Regarding applicability, all studies were deemed to include patients who were relevant to the review question. For the index test domain, six studies were considered to have low concern for applicability, as the index test was described sufficiently well for study replication. One study was considered to have high concern because of the anatomical area assessed (broad ligament) 22 and one study was considered as unclear 23 . All studies had low concern regarding applicability for the reference standard domain. The pooled sensitivity, specificity, positive likelihood ratio, negative likelihood ratio and diagnostic odds ratio of TVS for detecting endometriotic infiltration of the parametrium were 63% (95% CI, 31–86%), 98% (95% CI, 96–99%), 35.0 (95% CI, 13.2–96.5), 0.40 (95% CI, 0.17–0.87) and 94.9 (95% CI, 21.3–423.3), respectively (Figure  2 ). Heterogeneity was high for both sensitivity ( I 2  = 99.2%) and specificity ( I 2  = 97.7%). Meta‐regression identified laterality assessment and prevalence of parametrial involvement as explanatory variables for heterogeneity in sensitivity and specificity, respectively. Forest plots of sensitivity and specificity of transvaginal ultrasound for detection of parametrial involvement in women with suspected deep pelvic endometriosis. Only first author is given for each study. The sROC curve for parametrial involvement is shown in Figure  3 . The area under the curve was 0.97 (95% CI, 0.96–0.98). The Fagan nomogram showed that a positive result on TVS increased significantly the probability of identifying parametrial endometriotic involvement on laparoscopy, from 21% pretest to 90% post‐test, while a negative TVS result decreased the probability from 21% pretest to 10% post‐test (Figure  4 ). Publication bias was not assessed owing to the number of included studies being less than 10. Subgroup analysis was conducted including only the five studies that reported whether the left and/or right parametria were involved 16 , 17 , 18 , 19 , 20 . The pooled sensitivity, specificity, positive likelihood ratio, negative likelihood ratio and diagnostic odds ratio of TVS for detecting endometriotic infiltration of the left parametrium were 85% (95% CI, 64–95%), 98% (95% CI, 92–99%), 35.5 (95% CI, 9.2–137.1), 0.15 (95% CI, 0.06–0.42) and 230 (95% CI, 31–1727), respectively (Figure  S1 ). Heterogeneity was high for both sensitivity ( I 2  = 98.5%) and specificity ( I 2  = 97.7%). The area under the sROC curve for the left parametrium was 0.98 (95% CI, 0.96–0.99) (Figure  S2 ). The Fagan nomogram showed that a positive result on TVS increased significantly the probability of left parametrial endometriotic involvement, from 28% pretest to 93% post‐test, while a negative TVS result decreased the probability from 28% pretest to 6% post‐test (Figure  S3 ). The pooled sensitivity, specificity, positive likelihood ratio, negative likelihood ratio and diagnostic odds ratio of TVS for detecting endometriotic infiltration of the right parametrium were 84% (95% CI, 61–94%), 97% (95% CI, 92–99%), 30.0 (95% CI, 9.1–97.1), 0.17 (95% CI, 0.06–0.46) and 177 (95% CI, 26–1288), respectively (Figure  S4 ). Heterogeneity was high for both sensitivity ( I 2  = 98.3%) and specificity ( I 2  = 96.9%). The area under the sROC curve for the right parametrium was 0.98 (95% CI, 0.96–0.99) (Figure  S5 ). The Fagan nomogram showed that a positive result on TVS increased significantly the probability of right parametrial endometriotic involvement, from 24% pretest to 90% post‐test, while a negative TVS result decreased the probability from 24% pretest to 5% post‐test (Figure  S6 ). Despite finding high heterogeneity, meta‐regression was not performed because of the limited number of studies.

In this updated meta‐analysis, the pooled sensitivity of TVS for detecting endometriotic involvement in the lateral parametrium in women with suspected deep pelvic endometriosis was significantly higher than that reported in our previous meta‐analysis (63% vs 31%). Furthermore, subgroup analysis including studies that assessed parametrial laterality reported higher pooled sensitivity (85% for left parametrium and 84% for right parametrium). This can be explained by the fact that the studies included in the previous meta‐analysis did not follow a standardized approach for TVS scanning of the parametrium, whereas the studies reported more recently employed a defined scanning protocol 16 , 17 , 18 , 19 . This protocol was not identical in all the studies but shared significant similarities. It is apparent that, when the examiner follows a standardized TVS scanning approach, the chance of identifying lesions is much higher. In our opinion, this is the main difference between the ‘old’ and ‘new’ studies included in this meta‐analysis. Summary receiver‐operating‐characteristics curve ( ) for transvaginal ultrasound in detecting parametrial involvement in women with suspected deep pelvic endometriosis. , 95% confidence contour; , 95% prediction contour; , observed data; , summary operating point (sensitivity, 0.63 (95% CI, 0.31–0.86); specificity, 0.98 (95% CI, 0.96–0.99)). Fagan nomogram for transvaginal ultrasound in detecting parametrial involvement in women with suspected deep pelvic endometriosis. LR, likelihood ratio; prob, probability. The pooled number of patients included in the updated meta‐analysis was significantly higher than that in the previous review (6728 vs 560 women). Improved knowledge of pelvic sonographic anatomy, particularly that of the parametrium, might have contributed to the better diagnostic performance observed herein, as sonographers may now be more skilled at identifying this structure 24 . These results reinforce the value of the IDEA group's addendum regarding the ultrasound evaluation of the parametrium in women with suspected pelvic endometriosis 12 . A major limitation of our review is that we conducted our literature search in only three databases, which means that some studies addressing the review question could have been missed. Indeed, despite representing an improvement over the previous meta‐analysis, the number of included studies is still low. Moreover, we observed that the quality of the included studies is limited, particularly with regard to patient selection, with two failing to define exclusion criteria 16 , 22 . We also consider as a limitation the fact that all included studies published since 2022 came from the same country and from expert groups of examiners, which could call into question the generalizability of our findings. Furthermore, no study assessed the intra‐ or interobserver reproducibility of TVS assessment. In addition, we observed high heterogeneity in sensitivity and specificity between the included studies. Despite performing meta‐regression and sensitivity analyses, heterogeneity remained high. We identified laterality assessment and prevalence of parametrial involvement as sources of heterogeneity in sensitivity and specificity, respectively. As such, our results should be interpreted with some caution. Finally, most of the studies did not adhere to standardized IDEA terminology, but instead used their own methodology. Despite these limitations, our findings are encouraging and demonstrate that parametrial involvement can be assessed using TVS with a high degree of accuracy. This is particularly relevant for surgeons, as knowledge of the extent of parametrial endometriotic involvement may help to tailor surgical management of these patients. We believe that our data support the adoption of a standardized scanning protocol for assessing parametrial involvement in all women suspected to have deep pelvic endometriosis, as proposed recently by the IDEA group 12 . There has been recent debate regarding the adoption of simplified scanning protocols for endometrial assessment 25 , 26 , 27 . However, our data reveal that parametrial involvement is not uncommon (it is found in one in five women suffering from deep endometriosis), and the clinical implications might be important (for example, ureteral involvement with silent loss of renal function). We believe that specific training programs for sonographers for assessing the parametrium in women with suspected deep pelvic endometriosis should be developed. Questions remain regarding TVS assessment of the parametrium in women with suspected deep pelvic endometriosis, such as operator training and the learning curve required to obtain competence. Future research should focus on assessing interobserver reproducibility, delineating the learning curve and comparing the diagnostic performance of TVS with that of other imaging techniques, such as MRI. In conclusion, this updated meta‐analysis found that standardized assessment of the parametrium by TVS has high accuracy for detecting endometriotic lesions. Our results call for the adoption of standardized protocols for ultrasound scanning of the parametrium in clinical practice and the performance of further validation studies.

Endometriosis is a relatively common benign chronic disease affecting mainly women of reproductive age 1 . Transvaginal ultrasound (TVS) is considered the first‐line imaging technique for diagnosing this disease 2 . Several meta‐analyses have shown that TVS has good diagnostic performance for endometriosis, similar to that of magnetic resonance imaging (MRI) 3 , 4 , 5 , 6 . In 2016, the International Deep Endometriosis Analysis (IDEA) group published a consensus opinion proposing a standardized approach for the sonographic evaluation and reporting of findings in women with suspected pelvic endometriosis 7 . Many studies published since then have employed this approach. Among the various pelvic organs and anatomical structures that may be affected by endometriosis, the parametrium is of particular importance. Parametrial involvement is clinically relevant because it has been associated with more severe dysmenorrhea, more frequent voiding problems and constipation 8 . More importantly, parametrial involvement may cause progressive extrinsic or intrinsic ureteral obstruction, leading to silent functional loss of the ipsilateral kidney 9 . However, managing women with parametrial endometriotic involvement is challenging. This is particularly true when surgery is considered because of the advanced surgical skill and multidisciplinary team required 10 . Therefore, accurate diagnosis of parametrial involvement is essential for adequate management. Assessment of the parametrium by TVS in women with pelvic endometriosis has been chronically underutilized. In 2021, our group published a systematic review and meta‐analysis addressing this topic 11 . We reported that very few studies had been published to that date and that, while pooled specificity was high, pooled sensitivity was low. In that study, we did not assess which parametrial region (anterior, lateral or posterior) was involved. Since then, the IDEA group has published an addendum to their original consensus statement in order to stress the importance of evaluating the parametrium and provide instruction on how to perform this assessment 12 . In recent years, several studies focusing on TVS assessment of parametrial involvement in endometriosis have been published. Meta‐analysis of these new studies could offer fresh insight on this topic, specifically whether standardized scanning techniques and sonographer expertise in regard to parametrial anatomy could improve the detection of parametrial involvement in deep endometriosis. The aims of this study were to update our previous meta‐analysis with data from more recently published studies and to assess whether their inclusion modifies our pooled findings on the diagnostic performance of TVS for detecting parametrial involvement in women with suspected deep pelvic endometriosis.

Appendix S1 Search strategy. Appendix S2 Studies excluded from meta‐analysis. Figures S1–S3 Forest plots (Figure S1), summary receiver‐operating‐characteristics curve (Figure S2) and Fagan nomogram (Figure S3) of transvaginal ultrasound for assessment of left parametrial involvement in women with suspected deep pelvic endometriosis. Figures S4–S6 Forest plots (Figure S4), summary receiver‐operating‐characteristics curve (Figure S5) and Fagan nomogram (Figure S6) of transvaginal ultrasound for assessment of right parametrial involvement in women with suspected deep pelvic endometriosis.