Prognostic Value of 16α-¹⁸F-Fluoro-17β-Estradiol PET as a Predictor of Disease Outcome in Endometrial Cancer: A Prospective Study

Patients

This prospective study included 67 patients with untreated EC (International Federation of Gynecology and Obstetrics [FIGO] stage IA–IVB) who were referred for pretreatment assessment to the University of Fukui Hospital between December 2004 and December 2015. The inclusion criteria were histologically confirmed primary EC diagnosed by endometrial biopsy, and surgical treatment, irrespective of age or menstrual status. Patients who had received hormone therapy, wished to preserve fertility, had participated in a clinical trial that was not a standard treatment, or had life-threatening complications were excluded.

All patients underwent ¹⁸F-FES and ¹⁸F-FDG PET/CT before initial treatment. All patients underwent surgery, and the surgical specimens from each patient were examined histologically. The World Health Organization classification was used for histopathologic diagnosis. Patients were followed up for at least 24 mo after the date of their first visit or until death. Of the present participants, 19 of 52 patients with G1–3 endometrioid carcinoma and 2 of 8 patients with carcinosarcoma were included in our previous studies (10,11).

The study was approved by the Institutional Review Board of the University of Fukui Hospital (approval 20108007), and all subjects signed an informed consent form before undergoing PET.

PET Procedure

¹⁸F-FES was synthesized as previously reported (8). ¹⁸F-FES PET was performed with a dedicated full-ring PET scanner (Advance; GE Healthcare) used for medical research, and ¹⁸F-FDG PET was performed with a combined PET/CT scanner (Discovery LS; GE Healthcare) used mainly for clinical purposes. The patients underwent both ¹⁸F-FES and ¹⁸F-FDG PET scanning, as described in our previous studies (10,11). The 2 PET scans were performed on 2 separate days within 1 wk of each other, in random order. We previously reported that the ¹⁸F-FES SUV of normal endometrium was significantly higher in the proliferative phase than in the secretory phase (6.03 ± 1.05 vs. 3.97 ± 1.29) (12). Therefore, to minimize the effects of normal endometrial uptake, the ¹⁸F-FES PET scans of premenopausal patients were obtained in the luteal phase. Approximately 185 MBq of tracer were administered via the antecubital vein for each ¹⁸F-FES or ¹⁸F-FDG PET study. The patients fasted for at least 4 h before each study. Fifty minutes after tracer injection, each patient was placed supine on the PET or PET/CT scanner bed. For PET, a 16-min emission scan was obtained, with 3-min scans of the pelvic region (2 bed positions) and 2-min scans of each remaining region (5 bed positions) to provide total coverage from the head to the inguinal area. After the emission scans, postinjection transmission scans lasting 2 min for the pelvis and 1 min for other areas were acquired using a ⁶⁸Ge/⁶⁸Ga rod source for attenuation correction. For PET/CT, the following CT scanning parameters were used for attenuation correction: automatic amperage (upper limit, 40 mA; noise index, 20), 140 kV, section thickness of 5 mm, table feed of 15 mm, and pitch of 4. After the CT transmission scan, a whole-body emission scan was performed from the head to the inguinal region at 2 min per bed position (7–8 bed positions). The iterative method was used to reconstruct PET data by selecting 14 subsets and 2 iterations. The reconstructed images were then converted to SUVs.

Image Analysis

MRI was performed before the 2 PET examinations to obtain a diagnosis and anatomic information on the pelvic organs. T1- and T2-weighted images of the pelvis were acquired in the axial, sagittal, and coronal planes using a 1.5- or 3.0-T superconducting MRI system (Signa; GE Healthcare). After injection of gadolinium diethylenetriamine pentaacetic acid (0.1 mmol/kg), contrast-enhanced MRI was performed with and without fat saturation in the axial and sagittal planes. The method of image analysis has been reported previously (9,13). To obtain the regional SUV_mean, multiple circular regions of interest with a fixed diameter of 8 mm were drawn on primary tumor lesions irrespective of the presence of metastatic lesions. The SUV at the center of the lesion was obtained on 2 or 3 sagittal or transaxial slices 4 mm thick. For small lesions, a single section at the center of the lesion was used to avoid substantial partial-volume effects on the SUV_mean. To avoid the effect of uptake by normal uterine tissue, individual MR images were referenced for placement of regions of interest in the appropriate region after coregistration of the PET and MR images (Body Guide; Advance Biologic Co.). Because all 3 images have the same spatial coordinates, regions of interest were applied to resliced ¹⁸F-FES and ¹⁸F-FDG PET images in the same location. The SUVs for each patient were averaged for all regions of interest to obtain the SUV_mean of the tumor for ¹⁸F-FES and for ¹⁸F-FDG. The ¹⁸F-FDG/FES SUV ratio for each lesion was also calculated.

Endpoints

The primary endpoints were progression-free survival (PFS) and overall survival (OS). The date on which endometrial biopsy was performed was used as the starting point for PFS and OS. Tumor progression was confirmed by either imaging or tissue biopsy showing evidence of progressive disease according to the World Health Organization RECIST guidelines. The secondary endpoint was the prediction of postoperative recurrence risk factors (surgical stage, lymph node metastasis, LVSI, myometrial invasion, and histology) by preoperative PET parameters.

Statistical Analysis

Calculation of sample size was based on previous results for ER expression as a predictive marker for PFS (14). All data were collected in a structured database and analyzed using SPSS statistics, version 25 (IBM). The Mann–Whitney U test was used to analyze relationships between clinical characteristics and PET parameters. Receiver-operating-characteristic curve analysis was used to identify optimal cutoffs for each PET parameter. The Kaplan–Meier method was used to estimate PFS and OS, and these were compared using the log-rank test. Cox proportional hazards regression modeling was used for univariate and multivariate analyses. Significance was defined as a P level of less than 0.05 (2-sided testing).