Abstract
237
Purpose: The metabolic activity of tumor can be assessed by 18F-fluorodeoxyglucose positron emission tomography (FDG PET). High FDG tumor uptake is associated with worse prognosis in patients with non small cell lung cancer (NSCLC). On the other hand, high immune cell infiltration in primary tumor has been linked to better prognosis in NSCLC. Considering these contrary prognostic impacts, the prognostic value of tumor FDG uptake may be confounded by the infiltrating immune cells, because tumor FDG uptake is indistinguishable between cancer cells and infiltrating immune cells. Thus, we hypothesized that prognosis of the patients with lung adenocarcinoma can be further stratified by integration of FDG PET parameters and infiltrating immune cell scores assessed by genomic analysis.
Methods: We obtained transcriptomic data, FDG PET data, and clinical data from 96 lung adenocarcinoma patients from ‘NSCLC-radiogenomics’ data set in the cancer imaging archive. The patients were divided into three subtypes, terminal respiratory unit (TRU), proximal proliferative (PP), and proximal inflammatory (PI) using RNA sequencing data [1]. In addition, maximum standardized uptake value (SUVmax), metabolic tumor volume (MTV), total lesion glycolysis (TLG), and coefficient of variation (COV) were obtained through primary tumor segmentation. To normalize FDG uptake of tumor, tumor-to-liver uptake ratio with SUVmax (TLR) was calculated using FDG uptake of liver in each patients. For quantification of infiltrating immune cell, Bindea signatures for 24 immune cells, immune score and cytolytic score (CYT) were obtained from RNA sequencing data [2, 3]. Immune score was obtained using xCell and CYT was calculated as the geometric mean of PRF1 and GZMA gene expression. ANOVA test, Chi-square test and correlation analysis were conducted to determine the differences between the subtypes (Figure 1). Kaplan-meier survival analysis with log-rank test was done after the FDG PET and immune cell parameters were divided into low and high groups.
Results: Sex distribution, stage, immune signatures, TLR and COV were significantly different (P = .004, .002, .001, .038 and .001) according to the subtypes (Table S1). In particular, it was found that the TRU had lower FDG PET parameters, a larger portion of female, and lower stage than the other subtypes. Meanwhile, PI group showed significantly higher immune score and CYT (P = 0.001, 0.001). Although FDG PET and immune cell parameters showed no association, in a subgroup analysis, PI group showed weak positive correlation (Figure S1). This finding is probably due to higher immune cell infiltration in PI group than the other groups. In assessment of prognostic values of the FDG PET parameters, high TLR (median as a cut-off = 1.51, P = .01) and high COV (median as a cut-off = 0.25, P = .04) were associated with worse prognosis. Also, among immune cell related scores, low CYT (median as a cut-off = 2.722, P = .05), high CD8 T cell (median as a cut-off = 0.0472, P = .036), and high T follicular helper cell (TFH) (median as a cut-off = 0.12, P = .005) were found to be associated with better prognosis (Figure S2). Furthermore, TLR and TFH were predictive of prognosis even after adjusting for clinicopathologic features and the other (P = .024 and .047) (Table S2). Survival analysis for four groups divided by TLR and TFH using median as cut offs showed prognostic value (P = .002) (Figure S3).
Conclusions: High TLR was associated with worse prognosis while high scores of CD8 T cell and TFH predicted better prognosis in lung adenocarcinoma. Furthermore, TLR and TFH could predict prognosis independently and the combination of the two parameters could further stratify the prognosis of the patients with lung adenocarcinoma. This finding suggests that the integrative analysis of tumor FDG uptake and infiltrating immune cell based on genomic analysis would provide better prognostic biomarker than using single modality in cancer patients.