Abstract:
The purpose of this study was to assess various volume-based PET quantification metrics, including metabolic tumor volume and total lesion glycolysis (TLG) with different thresholds, as well as background activity-based PET metrics (background-subtracted lesion activity [BSL] and background-subtracted volume) as prognostic markers for progression-free and overall survival (PFS and OS, respectively) in early-stage I and II non-small cell lung cancer (NSCLC) after resection. Methods: Patients (n = 133) underwent an adequate 18F-FDG PET/CT scan before surgery between January 2003 and December 2010. All PET activity metrics showed a skewed distribution and were log-transformed before calculation of the Pearson correlation coefficients. Survival tree analysis was used to discriminate between high- and low-risk patients and to select the most important prognostic markers. The Akaike information criterion was used to compare 2 univariate models. Results: Within the study time, 36 patients died from NSCLC and 26 patients from other causes. At the end of follow-up, 70 patients were alive, with 67 patients being free of disease. All log-transformed PET metrics showed a strong linear association, with a Pearson correlation coefficient between 0.703 and 0.962. After multiple testing corrections, only 1 prognostic marker contributed a significant split point in the survival tree analysis. Of 10 potential predictors including 7 PET metrics, a BSL greater than 6,852 (P = 0.017) was chosen as split point, assigning 13 patients into a high-risk group. If BSL was removed from the set of predictors, a 42% TLG (TLG42%) of greater than 4,204 (P = 0.023) was chosen as split point. When a dichotomized BSL or TLG42% variable was used for a univariate Cox model, the Akaike information criterion difference of both models was smaller than 2; therefore, the data do not provide evidence that 1 of the 2 prognostic factors is superior. Conclusion: Volume-based PET metrics correlate with PFS and OS and could be used for risk assessment in stage I-II NSCLC. The different PET metrics assessed in this study showed a high correlation; therefore, it is not surprising that there was no significant difference to predict PFS or OS within this study. Overall, patients with large and metabolically active tumors should be considered high risk and might need further treatment after resection. Because all analysis steps were done with the same data, these results should be validated on new patient data.