Patients

The study involved patients with histopathologically confirmed FIGO stages IB–II invasive cervical cancer determined by a conventional workup, including pelvic and rectovaginal examination, routine laboratory testing, chest radiography, cystoscopy, and sigmoidoscopy. All the patients underwent preoperative workup, including both MRI and PET/CT scans between January 2004–December 2008. Institutional review board approval was obtained in advance for this retrospective study. The informed consent requirement was waived because of the retrospective nature of the study.

Study procedures and determination of lesion status

Both MRI and PET/CT scans were performed within 1 week before radical surgery. If a distant metastasis was confirmed histologically before surgical treatment, surgery was to be abandoned, and concurrent chemoradiotherapy (CCRT) or palliative therapy was determined according to current clinical standard.

MRI

All pelvic MR examinations were performed using Magnetom 1.5 T (Siemens, Erlangen, Germany) or Signa 1.5 T (General Electric Medical Systems, Milwaukee, WI) scanners. All patients had spin-echo T1-weighted images and T2-weighted fast-spin-echo images. Gadolinium-enhanced T1-weighted fat saturation sequences were performed before and after intravenous injection of contrast media. Buscopan (20 mg) was injected intramuscularly before imaging to suppress bowel motion.

A pelvic phased-array coil was used in all cases. T2-weighted imaging was performed in the axial plane, with a repetition time of 4000–5000 milliseconds and an echo time of 80–85 milliseconds (4000–5000/80–85), 3 signals were acquired, an echo-train length of 8, and a 512 × 256 matrix. The section thickness was 5 mm, with a 1-mm intersection gap and a 22-cm field of view. T2-weighted images were obtained in the oblique sagittal (parasagittal) planes that were parallel to the longitudinal axis of the uterus. After injection of gadopentetate dimeglumine (Magnevist; Schering, Berlin-Wedding, Germany) at a dose of 0.1 mmol/kg bodyweight, T1-weighted sagittal images and fat-suppressed T1-weighted axial images were also obtained. These images were obtained parallel to the parasagittal T2-weighted images and were centered at the imaging plane where the endometrial and cervical epitheliums were observed. All sequences were performed with a superior spatial presaturation pulse, as well as with a saturation band placed anteriorly and posteriorly to eliminate the high-signal-intensity phase-shift artifact caused by subcutaneous fat.

PET/CT technique

The patients were studied by a dedicated PET/CT system (Gemini; Philips Medical Systems, Andover, MA). The patient was asked to fast at least for 4 hours before undergoing PET/CT, and 125 mL of a barium sulfate solution (EZCT [1.5% weight-volume barium sulfate suspension]; Taejoon Pharm, Seoul, Korea) was administered orally 1 hour before imaging to opacify the bowel for the CT portion of the study. In addition, 0.15 mCi/kg bodyweight of FDG was administered intravenously 1 hour before imaging. CT was performed befor PET; the resulting data were used to generate an attenuation correction map for PET, and the PET images were reconstructed.

Lymph node dissection and histologic evaluation

Bilateral pelvic lymph node dissection was performed from the deep circumflex iliac vessel to the midportion of the common iliac vessel. All lymph nodes around the external, internal, and common iliac vessels and in the obturator fossa were removed. All lymph nodes around the aorta and inferior vena cava area between bilateral ureters were removed.

Each primary tumor and lymph nodes were sliced and stained with hematoxylin and eosin and examined microscopically by a pathologist. The numbers of lymph nodes retrieved in each area and the presence or absence of metastases were recorded.

Image analysis

MRI images were interpreted by two experienced radiologists, and PET/CT images were interpreted by 1 nuclear physician. The result of a lesion observed by MRI was considered positive when a lymph node had a short-axis dimension ≥1 cm. MRI images were examined in the picture archiving and communication system (PACS). Any lymph node with 1 cm or a little less in length, giving an overall equivocal impression, was considered negative in the study. PET/CT images were interpreted as negative when no areas of abnormal F-FDG uptake were seen. Every focus of increased F-FDG uptake was recorded and classified as malignant, equivocal, or benign on the basis of the shape, size, and intensity of the uptake. When focal F-FDG uptake, with intensity higher than that of surrounding tissues, was seen in areas unrelated to physiologic or benign processes, it was defined as malignant. Lesion with equivocal impression in this study was considered negative, as nuclear physicians gave more weight on benign or inflammatory processes rather than malignancy.

The readers were informed of patient history in that they knew all patients had undergone systemic lymph node dissection for cervical cancer; however, the readers were blinded to the presence of metastatic lymph nodes for all patients.

Statistical analysis

Calculation of sensitivity, specificity, accuracy, and positive and negative predictive values for metastasis detection was performed, in relation to results at histologic analysis after surgery. For the purposes of statistical analysis, a true-positive lesion was a lesion seen on images and found to be positive for tumor tissue at histologic analysis. A false-positive lesion was a lesion seen on images but found to be negative for tumor tissue at histologic analysis. A true-negative lesion was indicated when no lesion was seen on images and results of the histologic analysis were negative for tumor tissue. A false-negative lesion was a lesion that was missed at image analysis but was found to be positive for tumor tissue at histologic analysis. We calculated sensitivity, specificity, accuracy, and positive and negative predictive values of MRI and PET/CT on the basis of patient-based analysis for comparison.

Differences in sensitivity, specificity, and accuracy between MRI and PET/CT were estimated using the McNemar exact test. P values (from 2-sided tests) less than .05 were considered statistically significant. Data were analyzed using SPSS software for Windows (version 12.0; SPSS, Inc, Chicago, IL).