The aim of the study was to determine the most accurate quantitative morphological parameters on computed tomography (CT) that correlate with fluorodeoxyglucose (FDG)-avid para-aortic nodes (PANs) in patients with cervical cancer.
A single-institution retrospective evaluation was performed of women with cervical cancer who underwent pretreatment positron emission tomography (PET)/CT and radiotherapy therapy planning CT between 2009 and 2020. A node-by-node correlation between pretreatment CT and PET/CT was performed for the reference standard of FDG avidity for short- and long-axis diameters, volume, and long-/short-axis ratio (L/S). The FDG-avid PANs were defined as PET-positive and non–FDG-avid PANs from patients without PET-determined PAN metastasis were defined as PET negative. Area under the receiver operator curve was calculated to access diagnostic accuracy of the different quantitative parameters.
A total of 94 women (mean age ± standard deviation, 52 ± 13 years) with cervical cancer were included. Forty-seven patients had PET-positive PANs (181 PET-positive PANs) and 47 patients had no PET-positive PANs (141 PET-negative PANs). The area under the receiver operator curve for volume (0.945) was greater (P < 0.001) than that of short axis (0.895), long axis (0.885), and L/S (0.583). At a specificity set point of 0.90 (127/141 PANs), the cutoff for volume was 0.443 cm3 or greater (0.85 sensitivity [154/181 PANs]; 95% confidence interval, 0.83–0.93) and for short-axis diameter was 5.9 mm or greater (0.75 sensitivity [135/181 PANs]; 95% confidence interval, 0.68–0.81).
Para-aortic lymph node volume demonstrated that improved node-by-node correlation between CT and PET/CT compared with short-axis diameter, long-axis diameter, and L/S and is an alternative to improve detection of PAN suspicious of metastatic diseases in locations without access to PET/CT.