Comparison of imaging-based gross tumor volume and pathological volume determined by whole-mount serial sections in primary cervical cancer

OBJECTIVE To investigate the accuracy of imaging-based gross tumor volume (GTV) compared with pathological volume in cervical cancer. METHODS Ten patients with International Federation of Gynecology and Obstetrics stage I-II cervical cancer were eligible for investigation and underwent surgery in this study. Magnetic resonance imaging (MRI) and fluorine-18 fluorodeoxyglucose positron emission tomography ((18)F-FDG PET)/computed tomography (CT) scans were taken the day before surgery. The GTVs under MRI and (18)F-FDG PET/CT (GTV-MRI, GTV-PET, GTV-CT) were calculated automatically by Eclipse treatment-planning systems. Specimens of excised uterine cervix and cervical cancer were consecutively sliced and divided into whole-mount serial sections. The tumor border of hematoxylin and eosin-stained sections was outlined under a microscope by an experienced pathologist. GTV through pathological image (GTV-path) was calculated with Adobe Photoshop. RESULTS The GTVs (average ± standard deviation) delineated and calculated under CT, MRI, PET, and histopathological sections were 19.41 ± 11.96 cm(3), 12.66 ± 10.53 cm(3), 11.07 ± 9.44 cm(3), and 10.79 ± 8.71 cm(3), respectively. The volume of GTV-CT or GTV-MR was bigger than GTV-path, and the difference was statistically significant (P < 0.05). No significant difference was observed between GTV-PET and GTV-path (P > 0.05). Spearman correlation analysis showed that GTV-CT, GTV-MRI, and GTV-PET were significantly correlated with GTV-path (P < 0.01). There was no significant difference in the lesion coverage factor among the three modalities. CONCLUSION The present study showed that GTV defined under 40% of maximum standardized uptake value in PET images was very similar to the pathological volume of cervical cancer. This result should be replicated in a larger number of patients with cervical cancer in a future study of ours.