Patients
This is a prospective cohort study of 11 consecutive women with histologically proven advanced stage primary cervical cancer (FIGO stages ≥IB2-IVB), which were referred to Maastricht University Medical Centre between July 2014 and October 2015. All patients were discussed in a multi-disciplinary team consisting of gynecologic oncologist, medical oncologist, radiation oncologist, pathologist, radiologist, and a nuclear physician. Patients allocated to radiotherapy (FIGO stages ≥IB2, according to the clinical guidelines) were included.
Ethical approval was given and informed consent for the use of (coded) images was waived by the Maastricht University Medical Centre ethical committee, as the data was analyzed anonymously in accordance with the Institutional Review Board guidelines (IRB no. 13-02-2015).
Radiotherapy was performed according to the current GEC-ESTRO guidelines; this consisted of external beam radiation therapy (EBRT) followed by MRI-guided brachytherapy (BCT). The first BCT was applied during the fifth week of radiotherapy, and the overall treatment time for radiotherapy was 6–7 weeks.
Patients were treated with either concurrent chemotherapy or hyperthermia (if concurrent chemotherapy was contraindicated or in case of FIGO 4B disease after neoadjuvant chemotherapy) according to local clinical guidelines.
PET/MRI imaging protocol
A whole-body [18F]-FDG-PET/MRI was performed on a 3.0-Tesla Biograph mMR PET/MRI scanner (Siemens) with a 4.4-mm PET resolution (full width at half maximum).
Patients were fasting for at least 6 h before the examination; a blood glucose level (all < 10 mmol/L) was obtained in all patients. Body weight adapted intravenous administration of 2 Mbq/kg of 18 fluorodeoxyglucose was performed 45 min before the PET-MRI. The PET images were acquired in 5-min bed positions. Attenuation correction was performed with a dual-echo VIBE DIXON that separates water and fat with TE1/TE2 = 1.23 ms/2.46 ms, TR = 3.6 ms, left-right FOV = 500 mm, and anterior-posterior FOV = 300 mm. The PET acquisition was reconstructed using the Siemens HD reconstruction algorithm (3 iterations, 21 subsets, 4 mm Gauss, matrix size 300) and corrected for attenuation, scatter, randoms, dead-time, and radioactive decay. PET/MRI images were fused and analyzed using the dedicated DICOM software (Osirix MD, Geneva).
Quantification of FDG uptake was performed by assessing the standardized uptake value (SUV; measured activity concentration [Bq/ml] × body weight [g]/injected activity [Bq]).
A total body (from skull-base to groin) 2D T2-weighted fast spin-echo image in two planes (sagittal and axial) was performed (TR 2150/ TE 138 ms, 33 ETL, 1NSA, (0.98 × 0.98) × 6.50 mm3).
Second, diagnostic pelvic MRI images were performed in three planes (sagittal, axial, and coronal), with the axial and coronal planes angled perpendicular and parallel to the cervical axis, respectively (2D T2-weighted (T2 W) fast spin-echo images (TR 4000/TE 102 ms, 25 ETL, 3 NSA, (0.98 × 0.98)× 3.00 mm3 voxel at 3.0 T). Furthermore, the protocol consisted of a T1 in the coronal plane (TR 700/TE 11 ms, 57 ETL, 1 NSA, (0.98 × 0.98) × 1.1 mm3 voxel at 3.0 T) and diffusion weighted images (TR 9000/TE 80 ms, 1 ETL, 3 NSA, (0.98 × 0.98) × 5.0 mm3 voxel at 3.0 T) with B-values (50, 400, and 800).
Patients did not receive bowel preparation, bladder catheterization, or anti-spasmodic agents.
Additional scans
The pre-treatment MRI was performed according to the same protocol on the 3.0 Tesla Bioraph mMR PET-MR (Siemens Magnetom Avanto, Erlangen, Germany) or on a 1.5-Tesla MRI unit (Intera Achieva); Philips Healthcare, Best, The Netherlands (in 5 patients). The protocol used in the latter was a standard protocol for diagnostic pelvic MRI images comparable with the 3.0 Tesla PET-MR protocol consisting of T2W, T1W, and diffusion weighted images.
PET pre-treatment was performed according to the same protocol on the 3.0 Tesla PET-MR or in 5 patients on a PET-CT scanner Gemini TF TOF 64 (Philips Healthcare, Best, The Netherlands) according to a similar protocol as described for the PET-MR.
Image evaluation
In The Netherlands, radiologic assessment is done by a radiologist and assessment of nuclear imaging by a nuclear physician. If combined imaging, like PET and CT, is used, the radiologist and nuclear physician evaluate their findings together and give their combined result to the clinician. The MRI images were retrospectively independently analyzed by a radiologist (FB) and the PET images by a nuclear physician (SdV) with experience in oncologic imaging blinded to all patient information and patient outcome. After the initial analysis, both readers evaluated the PET-MRI images together, they discussed their separate findings and came to consensus. ROC curves to determine area under the curve were performed.
The readers were asked to assess the presence of residual tumor and/or metastasis based on a “subjective” visual assessment of the images using the following confidence level scores: 0, definitely no residual tumor/metastasis; 1, probably no residual tumor/metastasis; 2, unclear; 3, probably residual tumor/metastasis; 4, definitely residual tumor/metastasis. The readers were not given any instructions or asked to search for certain criteria and were free to interpret the scans based on prior experience.
Second, at the consensus meeting, they scored diagnostic confidence. This was scored on a similar 5-point Likert scale: 0, certainly no increase in confidence; 1, probably no increase; 2, unclear; 3, probably increase; 4, definitely increase. Change in diagnostic confidence was calculated with a predefined cut-off (score 2); only scores 3 and 4 were seen as a change of diagnostic confidence.
All measurements were performed on a dedicated DICOM system (Osirix MD, Geneva).
Standard of reference
Local residual tumor was defined as tumor at the original tumor site (cervix, vagina, parametria, bladder, or rectum).
The presence or absence of a local residual tumor was determined by:
-
(1)
Histopathology of the surgical resection specimen.
-
(2)
The results of a post-treatment gynecologic examination (under anesthesia, with or without biopsy) performed 3 months after completion of the entire radiation treatment and at least 12 months of documented follow-up. Follow-up consisted of three monthly check-up by alternating a gynecologic and radiation oncologist and if indicated with additional imaging and/or tumor markers conform clinical guidelines.
Regional residual tumor was defined as residual and/or recurrent lymphatic tumor within the irradiated volume (i.e., pelvic or lower para-aortic lymph nodes). Distant metastasis was defined as metastasis outside of the irradiated volume. Regional tumor and distant metastasis were proven either by histopathology or the detection of a growing tumor mass during subsequent imaging analysis.
After radiation therapy in case of isolated local residual disease without metastasis, patients were subjected to salvation surgery. If distant metastasis was present, palliative therapy was proposed. In case of no evidence of disease, follow-up was performed according to the clinical guidelines.
Statistical analysis
Statistical analyses were performed using SPSS Statistics v20.0 (SPSS Inc., Chicago, Illinois) and Stata v11.0 (StataCorp LP, Texas).
Receiver operating characteristic (ROC) curves were constructed to evaluate diagnostic performance for (a) MRI, (b) PET, and (c) combined PET-MRI analysis. Corresponding areas under the ROC curve (AUC) were compared according to the method described by De Long et al. [13]. P-values less than 0.05 were considered statistically significant.