Patients
Data of patients with pathologically proven endometrial carcinoma, who underwent PET/CT and PET/MRI from December 2017 to January 2021 were retrospectively analyzed. All procedures were approved by the ethics committee of the hospital. Written informed consent was obtained from all patients.
We screened the patients based on the following inclusion criteria: (1) patients who underwent PET/CT and PET/MRI before treatment, (2) patients who received surgical treatment that confirmed endometrial carcinoma by postoperative pathology, (3) patients with complete postoperative pathological data, and (4) patients who had < 2 weeks between imaging and surgery. The exclusion criteria were as follows: (1) patients who received other treatments (e.g., radiotherapy and chemotherapy) before imaging, (2) patients who had other types of malignant tumor history before endometrial carcinoma diagnosis, (3) patients in whom the measurement of parameters was affected by artifacts in PET/MRI images, and (4) patients who belonged to the FIGO stage IIIC2 or IV. Finally, 57 patients were enrolled in this study (Fig. 1). All 57 patients underwent total abdominal or laparoscopic hysterectomy with bilateral salpingo-oophorectomy, and 41 patients underwent pelvic lymphadenectomy.
PET/CT scanning and image acquisition
Before non-contrast-enhanced 18F-fluorodeoxyglucose (18F-FDG) PET/CT scanning, all patients fasted for ≥ 6 h and had blood glucose levels of ≤ 7 mmol/L. They were injected with 3.70–5.55 MBq/kg 18F-FDG in the resting state. After 60 min, 18F-FDG PET/CT scanning was performed with a Discovery PET/CT 690 scanner (GE Healthcare, Waukesha, WI, USA) while the patients were lying on the scan bed. The scan ranged from the calvarium to the middle thigh (120 s/bed). The slice thickness, tube voltage, and tube current for CT scans were 3.75 mm, 120–140 keV, and 80 mA, respectively.
PET/MRI scanning and image acquisition
Pelvic 18F-FDG PET/MRI scanning was performed 33 ± 12 min after PET/CT scanning. All images were acquired from the scans that were performed using Signa PET/MRI (GE Healthcare, Waukesha, WI, USA), integrating time-of-flight–PET, and 3.0 T MRI (GE Signa 750w) scanners. PET images and MRI images were acquired simultaneously. A 32-channel coil (upper anterior array) served as the cavitary scanning coil. The pelvic axial scan ranged from the vaginal level to the superior iliac boundary. For pelvic MRI scanning, T2-weighted images (T2WI) were acquired in the axial, sagittal, and coronal planes using the following T2WI parameters: repetition time (TR), 2600–3400 ms; echo time (TE), 60–90 ms; section thickness, 5 mm; interval, 1 mm; matrix, 384 × 384; field of view (FOV), 240 × 240 mm. Axial T1-weighted images parameters were as follows: TR, 500 ms; TE, 8 ms; section thickness, 5 mm; interval, 1 mm; matrix, 384 × 384; FOV, 240 × 240 mm. Axial diffusion-weighted imaging (DWI) with b-values of 0 and 800 s/mm2 were obtained. For PET scanning, the correction for γ-ray attenuation was performed with the Dixon MRI sequence. PET scanning was conducted under list mode. Images were reconstructed by iterative ordered subset expectation maximization. The pelvic scan time for PET/MRI was approximately 24 min.
Image analysis
All 18F-FDG PET/CT and PET/MRI images were uploaded to a GE AW4.6 workstation (GE Healthcare, Waukesha, WI, USA) and then post-processed using PET volume computer-assisted reading (VCAR) software. Three radiologists/nuclear medicine physicians, each with double board certifications and with > 5 years of medical imaging diagnosis experience, evaluated the PET/CT and PET/MRI images independently and sequentially at two time points. PET/MRI image evaluation was performed 4–5 weeks after PET/CT image evaluation, thereby eliminating subjective image bias. The readers were unaware of the other readers’ evaluation results. To achieve a final summary of the results, diagnosis was determined through negotiation and consensus when their opinions were inconsistent.
For maximum standardized uptake value (SUVmax) measurements of PET/MRI, by applying the iterative adaptive algorithm, PET VCAR enables automatic segmentation. SUVmax was defined as the maximum value of SUV.
IMAgenGINE MRToolbox (Vision Tech, Hefei, Anhui, China) software was used for apparent diffusion coefficient (ADC) measurements. Regions of interest (ROIs) were drawn on the ADC maps with T1WI and T2WI sequences as references. ROIs were manually delineated on the slices containing the largest diameter of the tumors as much as possible, avoiding necrotic and hemorrhage tissue. Necrotic areas were defined as areas with relatively low signal intensity on DWI and T1WI, and high signal intensity on T2WI compared with solid tumors. Hemorrhage areas were defined as areas with relatively high signal intensity on DWI and T1WI compared with solid tumors. ADCmean and ADCmin were defined as the average and the lowest ADC values in ROI, respectively. Three readers independently measured all parameters, and the average was calculated.
Diagnostic criteria
PET/CT and PET/MRI diagnostic criteria for staging endometrial carcinoma were based on the FIGO staging system. Postoperative pathological staging was applied as the gold standard.
To diagnose myometrial invasion, the proportion of the tumor thickness that invaded the myometrial layer was used to calculate the myometrial invasion. The superficial myometrial invasion ratio was < 50%. The deep myometrial invasion ratio was ≥ 50%. The diagnosis of cervical invasion was based on high uptake tumor invasion at the cervix in the PET/CT image, low signal disappearance of the cervical interstitial layer under the T2WI sequence, and high signal in the DWI image at the cervix. On PET/MRI images, tumor invasion of adjacent structures was determined primarily on the basis of MRI findings, with reference to PET findings. On PET/CT images, tumor invasion of adjacent structures was determined primarily on the basis of PET findings, with reference to CT findings.
On PET/CT and PET/MRI images, lymph node metastasis diagnosis was based on the abnormally high uptake of FDG in the pelvic lymph nodes (i.e., exceeding normal muscle or exceeding normal lymph nodes at the same level in the contralateral pelvis that corresponded to the lymph node chains), regardless of whether their short-axis diameter was higher than 1 cm. In addition, abnormally high signal intensity on DWI was also considered as a positive sign of lymph node metastasis.
Histological examination
Tumor staging was performed according to the 2018 FIGO criteria. All surgical specimens were examined and reported by a gynecologic pathologist. A comprehensive histological evaluation was performed for each lesion, including histological type, tumor grade, myometrial and cervical stromal invasion, and lymph node status. For patients who did not undergo pelvic lymph node dissection, lymph node metastasis was confirmed at follow-up (at least 12 months). A decrease in lymph node size after treatment was considered a sign of malignancy.
Statistical analysis
The McNemar test was used to determine the statistical significance of differences in the accuracy of staging as determined by PET/MRI and PET/CT. The inter-reader agreement in PET/CT and PET/MR evaluation from the three readers was assessed using Kendall’s concordance coefficient (W). The associations between PET/MRI parameters and clinicopathological characteristics were analyzed using independent sample t-test (two sets of variables) and one-way analysis of variance, followed by Bonferroni posthoc test (three sets of variables). The correlations between PET/MRI parameters were assessed using Pearson’s correlation coefficient test. Receiver operating characteristic (ROC) curve was used to evaluate the value of PET/MRI quantitative parameters to differentiate FIGO stage I and FIGO stage II + III endometrioid carcinoma. The Youden index was used to obtain the cut-off value. The SPSS 22.0 software (IBM Corp., Armonk, NY, USA) was used for all statistical analyses. Statistical significance was set at p < 0.05.