What is the best T1-weighted Dixon protocol for diagnosis of endometriosis?
What is the best T1-weighted Dixon protocol for diagnosis of endometriosis?
Ambra Santonocito, Barbara J. Fueger, Nina Poetsch, Thomas H. Helbich, Pascal A.T. Baltzer, Paola Clauser
Endometriosis is a common and multifaceted condition, which affects approximately 10% of women of reproductive age. Magnetic resonance imaging (MRI) has gained a central role in the diagnosis and treatment planning, particularly for women affected by deep pelvic endometriosis (DPE). The current guidelines for MRI of pelvic endometriosis recommend the use of 2D non fat-suppressed T2-weighted sequences (axial, sagittal and oblique) together with fat-suppressed T1-weighted imaging. In the last decade, the Dixon technique has been increasingly used as fat-suppressed T1-weighted sequence. The Dixon technique is a sequence based on chemical shift that allows a homogenous fat suppression and the possibility to reconstruct images with and without fat suppression in a single acquisition. T1-weighted Dixon sequences have been suggested in MRI endometriosis protocols, but the evidence on their performance is limited. 3D T1-weighted Dixon sequences also allow the reconstruction of the acquired images in orthogonal planes, thus theoretically allowing the evaluation of the pathology in different planes without acquiring multiple imaging sequences. The aim of this study was to evaluate the detection of endometriosis with MRI using a standard protocol including 3D T1-weighted Dixon acquired in the axial or sagittal plane.
This retrospective analysis was approved by the Institutional Review Board (IRB), and the need for signed informed consent was waived. Images of consecutive patients undergoing MRI with the clinical and/or sonographic suspicion of DPE were evaluated. Exclusion criteria were incomplete MRI examinations, the absence of a reference standard or impossibility to correlate the MRI finding with the available histology and the absence of T1-weighted Dixon sequences in the MRI protocol or an incomplete protocol. The reference standard was the histology obtained by surgery. MRI protocol included a T2W-TSE sequence of the pelvis in three plane, axial diffusion weighted imaging and a T1W-Dixon acquired in the axial and sagittal plane. Images were evaluated by two readers in consensus, blinded to patient clinical data and to histopathological results. After the readings, one of the readers correlated the imaging findings with the location defined at histopathology. Three reading settings were used, including the three plane T2W-TSE together with: 1) T1W-Dixon axial (T1-AX); 2) T1W-Dixon axial and reconstructed sagittal (T1-REC); T1W-Dixon axial and acquired sagittal (T1-AS). The washout period between readings was at least two weeks to reduce bias. Three typical endometriosis sites were analyzed in each reading: ovaries, rectal-sigmoid colon and DPE sites (fallopian tubes, sacro-uterine ligaments, vesicouterine space). Detection of endometriosis site was calculated and compared using the different acquisition protocols using the McNemar test. The false positive rate was calculated.
A total of 120 histologically verified lesions from 44 patients (mean age 31 years; range 21-46) were included. Results are summarized in Figure 1. Detection for endometriomas was 92% with T1-AX, 89% with T1-REC and 92% with T1-AS without statistically significant differences (p>0.05.). The FRP was 0.4 with T1-AX and 0.3 with both T1-REC and T1-AS. Detection for DPE was 53% with T1-AX, 72% with T1-REC and 77% with T1-AS, with a significant difference between T1-AX versus T1-REC (p=0.03) and versus T1-AS (p=0.004) but not between T1-AS versus T1-REC (p>0.05). The FPR was 0.53 for both T1-AX and T1-REC, and 0.45 for T1-AS. Detection in the recto-sigmoidal colon was 64% with T1-AX, 68% with T1-REC and 95% with T1-AS. There was a significant difference between T1-AX versus T1-AS (p=0.02) and T1-REC versus T1-AS (0.03), but not T1-AX versus T1-REC (p>0.05). The FRP was 0.5 with T1-AX and 0 with both T1-REC and T1-AS. The overall detection of endometriosis site was 68% with T1-AX, 77% with T1 REC and 85% with T1-AS. The false positive rate (FPR) was 0.48, 0.42 and 0.35 respectively. Significant differences were found between T1-AX versus T1-AS (p<0.001), but not between T1-AX versus T1-REC (p=0.07) and T1-AS versus T1-REC (p=0.08). An example is given in Figure 2.
Figure 1. Lesion detection and false positive rate (FPR) for all the investigated location overall and separately. 95% confidence intervals are given in brackets.
Figure 2: 35 years old patient with endometriomas and deep pelvic endometriosis (DPE). A large endometrioma was identified in the left ovary with an associated DPE in the retrouterine space (axial and sagittal T2 weighted images in A and D). Based on the axial T1 images (B), no involvement of the sigma was suspected. When the T1 axial and sagittal images were evaluated (C acquired, E reconstructed), the involvement of the serosal surface of the sigma was suspected. This was confirmed at surgery.
We obtained a good detection of endometriosis sites using a standard protocol including 3D T1W-Dixon sequence. The acquisition of the sequence in both the axial and sagittal plane allows for the highest lesion detection with lowest FPR. The acquisition of sagittal T1W images should not be substitute with the use of reconstructed images.
