Progression of Deep Infiltrating Rectosigmoid Endometriotic Nodules

Antoine Netter; Perrine d'Avout-Fourdinier; Aubert Agostini; Isabella Chanavaz-Lacheray; Marta Lampika; Marilena Farella; Clotilde Hennetier; Horace Roman


Hum Reprod. 2019;34(11):2144-2152. 

In This Article

Materials and Methods

Study Design

We conducted a monocentric case series study. Data were retrospectively and anonymously collected. The study was approved by the local ethics committee (E2018–71).

Study Population

Women who were referred from September 2016 to March 2018 to two of the authors (H.R.—surgeon or I.C.-L.—fertility specialist) for advice on the management of deep endometriosis infiltrating the rectum or the sigmoid colon causing pain and/or infertility were enrolled. Women were included if they had undergone two consecutive pelvic MRIs prior to outpatient review, at an interval of >12 months, with one or more deep endometriotic nodules infiltrating the rectosigmoid muscularis layer demonstrated on at least one MRI. Exclusion criteria were as follows: any surgical management of rectosigmoid endometriosis between the two MRIs, the onset of menopause between the two MRIs, an age <20 years at the time of the first MRI, insufficient quality of one or both MRIs to allow accurate measurement of deeply infiltrating lesions (minimum requirement of three main sequences, images of high quality and without artefact hampering nodule assessment) and insufficient data in the patient medical file to assess the duration of amenorrhoea between MRIs.

MRI Measurements

Digital Imaging and Communications in Medicine (DICOM) files were recorded using Osirix MD 10.0 software (Pixmeo SARL, Geneva, Switzerland). MRIs were analysed by one senior radiologist (P.d'A.-F.), specializing in women's health imaging and with experience and expertise in endometriosis. The radiologist was blinded to patient symptoms and prior hormonal treatment. The radiologist performed an initial assessment of MRI quality. MRIs where the protocol did not include at least one T1-weighted fat saturation sequence (T1 FS) and two orthogonal T2-weighted sequences were considered inaccurate and were excluded from the analysis. MRIs of poor quality, which did not allow precise measurement of endometriotic nodules, were also excluded. Poor MRI quality was determined by the presence of excessive bowel peristalsis, patient movement, excessive presence of stools or metal artefact. For patients with more than two prior MRIs, the two with the longest interval and greatest quality were selected.

Rectosigmoid involvement on MRI was based on the disappearance of the hypointense signal of the anterior wall of the rectum and/or sigmoid colon on T2-weighted images and the presence of a tissue mass forming an obtuse angle with the wall of the rectosigmoid, extending to the anterior wall of the rectum and the inferior wall of the sigmoid colon (Bazot and Daraï, 2017). The length of bowel wall infiltration was measured in millimetres from the proximal to the distal edge of involvement, along the bowel segment, using the 'open polygon' function. The thickness of the lesion was also measured in millimetres as the largest anteroposterior diameter of the nodule orthogonal to the length (Figure 1). The height of the nodule was defined as the distance from the anal verge to the distal edge of the nodule along the rectum (millimetres), using the 'open polygon' function. Number, length, thickness and height of endometriotic rectosigmoid lesions were reviewed and recorded for each MRI. For patients who presented with more than one rectosigmoid lesion, we selected the lesion with the greatest length for review. Only deep endometriotic nodules infiltrating the rectosigmoid were included for review because their interpretation and measurement on MRI are reliably standardized.

Figure 1.

Measurement of the length (a) and the thickness (b) of deep endometriosis nodules infiltrating the rectosigmoid.

We defined 'progression' of any nodule as an increase of ≥20% in length or in thickness and 'regression' of any lesion as a corresponding decrease in length or thickness, between two MRIs. Any nodules where length and thickness variations were <20% were considered as 'stable'.

To confirm reproducibility of results, a second measurement of rectosigmoid nodules was conducted on 30 randomly selected MRIs by the same radiologist (P.d'A.-F.) and by another experienced radiologist (M.L.). This enabled the assessment of both intra- and inter-operator reliability.

Assessment of Amenorrhoea

We recorded the time (in months) between the two MRIs. To assess the duration of amenorrhoea between the two MRIs, data were extracted from the medical records of all patients included in the study. The total number of months where pregnancy or breastfeeding occurred, or hormonal treatment (GnRH agonists, macroprogestins or continuous combined oral contraception) was taken without interruption for at least 3 months, were recorded as periods of amenorrhoea.


The primary endpoint was the probability of progression of any rectosigmoid nodule between two MRIs.

The secondary endpoints were the total duration of amenorrhoea between the two MRIs and the proportion of time between the two MRIs where amenorrhoea occurred among the three groups of women with stable, progressive and regressive rectosigmoid nodules, respectively. The risk of progression of rectosigmoid nodules was also compared according to whether or not a pregnancy had occurred during the interval.

Statistical Analyses

Statistical analysis was performed using IBM SPSS Statistics 20.0 (IBM Inc., New York, USA). The study sample was described using mean ± SD for continuous variables and number (percentage) for categorical variables. ANOVA and χ 2 tests were used to compare, respectively, continuous and categorical variables among the three groups. Intraclass correlation coefficient (ICC) was estimated to assess the inter- and intra-observer concordance of measures for the length and the thickness of rectosigmoid nodules. All statistical analyses were two-tailed, and results were considered to be statistically significant when P < 0.05.