Purpose: To determine the heterogeneity of glucose uptake applying fractal analysis on positron emission tomography/computed tomography (PET/CT) with 2-deoxy-2-[18F]fluoro-D-glucose ([18F]FDG) images in patients with non-small cell lung carcinoma (NSCLC) before surgery, and to assess whether this heterogeneity was associated with disease-free survival (DFS). Procedures: [18F]FDG PET/CT scans of 113 patients’ prior surgery were retrospectively revised. PET DICOM images were analyzed for fractal geometry using a ad hoc software to automatically determine the following indexes: (a) mean intensity value (MIV), (b) standard deviation (SD), (c) relative dispersion (RD), (d) three-dimensional (3D) histogram of the fractal dimension (3D HIST FR DIM), and (e) fractal dimension in 3D (3D-FD). All the fractal indexes were subsequently compared with metabolic parameters and disease-free survival (DFS). Results: We found a significant correlation between 3D-FD and SUVmax, SUVmean, MTV, and TLG. Additionally, positive correlations between MIV, SD, and all metabolic parameters were also detected. Patients with high 3D-FD tumor (≥ 1.62) showed significantly higher values of SUVmax, SUVmean, MTV, and TLG than those with lower 3D-FD. In univariate analysis, median 3D-FD and median TLG were significantly associated with DFS (p = 0.04 and p = 0.03, respectively). These findings were confirmed on log-rank test. On multivariate analysis, among age, stage disease, histotype, 3D-FD, and metabolic parameters, only 3D-FD was identified as independent prognostic factor for DFS (p = 0.032; HR 0.418, 95 % CI 0.189–0.926). 3D-FD was different between adenocarcinoma and squamous cell carcinoma (1.60 versus 1.88, p = 0.014), and 3D-FD value was found higher in advanced stage disease. Conclusions: Metabolic heterogeneity determined applying fractal principles on PET images can be considered as a novel imaging biomarker for survival in patients with NSCLC.
- Disease-free survival
- [F]FDG PET/CT
ASJC Scopus subject areas
- Radiology Nuclear Medicine and imaging
- Cancer Research