Abstract
636
Objectives We have recently reported the causal relationship of KRAS/BRAF mutations and 18F-fluorodeoxyglucose (FDG) accumulation (Clin Cancer Res 2012;18:1696), demonstrating FDG accumulation was higher in colorectal cancer with KRAS/BRAF mutations. The purpose of this study was to investigate whether FDG-PET/CT findings can predict the differentiation between KRAS-mutated tumors and wild type tumors.
Methods A total of 71 patients (M:F=41:30) with colorectal cancer who underwent FDG-PET/CT scan for staging before treatment were included. We took region of interest in a primary tumor, normal liver, and mediastinal blood pool (MBP), calculating seven parameters as follows: maximal standardized uptake value (SUV) in a primary tumor (SUVmax), SUVmax-to-normal liver uptake ratio (TLR), metabolic tumor volume defined as volume of tumor uptake greater than MBP + 2 standard deviations (MTV), mean SUV of a whole tumor (SUVmean), standard deviation of a whole tumor (SD), total glycolytic volume defined as MTV multiplied by SUVmean (TLG), and retention index (RI). RI was calculated only in 37 patients with additional delayed scan at 2h postinjection. The diagnostic performance of each parameter was assessed using the ROC analysis.
Results All parameters except RI were higher in KRAS-mutated tumors than in wild type tumors with statistical significance (p values ranging from 0.0001 to 0.025), although no significant difference of RI was observed between the two groups. The highest sensitivity (71%), specificity (84%), and accuracy (79%) were obtained by applying 2.8 as a cut-off value for SD, followed by SUVmean (89%, 67%, 76%, 4.8, and 68%, 81%, 76%, 5.1, respectively), and SUVmax (61%, 84%, 75%, 14.6). The areas under the curve of SD, SUVmean, and SUVmax were 0.815, 0.772, and 0.740, respectively.
Conclusions Our data suggest that quantitative parameters acquired in FDG-PET/CT scan could help to differentiate between KRAS-mutated tumors and wild type tumors in patients with colorectal cancer.