Abstract 3197: Augmenting 18F-Fluorodeoxyglucose PET with a gene classifier for outcome prediction in patients with resected, non-small cell lung cancer: A pilot study

Abstract

Abstract Introduction: Biomarkers that identify cancer patients at increased risk of death from their disease are particularly attractive as an aid in management for those who may require toxic therapies in addition to traditional surgical treatment. While investigators have retrospectively established tumor gene expression profiling and 18F-Fluorodeoxyglucose (FDG) PET uptake as biomarkers that identify patients at high risk of death from non-small cell lung cancer (NSCLC) despite resection, their additive utility has not been investigated to date. Methods: We retrospectively analyzed 40 patients at UCSF with resected NSCLC, bronchoalveolar cell carcinoma (BAC) subtype, for which FDG was shown previously to predict non-survival as a threshold value. We compared a previously validated gene expression score that was developed using tumor specimens from the UCSF Thoracic Oncology Tissue Bank and tumor FDG uptake, expressed as a maximum tumor standard uptake value (SUV), to patient outcome using Cox-proportional hazards regression modeling and Kaplan-Meier product limit analysis. We assessed test performance for each metric by using logistic regression to develop ROC curves. Results: Twenty two patients were included for this study after excluding those with neoadjuvant therapy and no gene expression data. FDG uptake was dichotomized at a threshold value of 2.5 and gene expression data was dichotomized at the median composite score of the log-normalized transcript levels for Wnt3a, ERBB, LCK and RND3. Of the 22 patients with both FDG uptake and gene score data, mean age was 70 (±6) years, 86% had stage I disease and one patient received adjuvant chemotherapy. Median tumor SUV was 2.9 (Interquartile Range [IQR] 1.7-6.0) and median follow-up time was 45 months (IQR 25-59). At five years from surgery, 73% of patients with a low gene score were alive compared to 88% of patients with low tumor FDG uptake. Tumor size was the most significant predictor of non-survival (1.3, confidence interval [CI] 1.0-1.8), while increasing FDG uptake and gene score as continuous variables showed a trend only towards increased mortality. In a multivariate model including tumor size, FDG uptake and gene score, tumor size attenuated the effect of tumor FDG uptake while both gene score and tumor size remained significant predictors. The ROC (C-statistic) for test performance was modest and slightly better for binary FDG uptake (0.74) than for binary gene score (0.68). Combining the two variables resulted in improved test performance (0.85) Conclusion: This pilot study suggests that FDG uptake may add additional information to a robust gene assay in patient prognosis. Particular attention to patient selection and biomarker validation will be crucial for the success of these types of models, and further studies in larger cohorts of patients are warranted and underway. Citation Format: {Authors}. {Abstract title} [abstract]. In: Proceedings of the 102nd Annual Meeting of the American Association for Cancer Research; 2011 Apr 2-6; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2011;71(8 Suppl):Abstract nr 3197. doi:10.1158/1538-7445.AM2011-3197