S. Zhang2, T. Boyle1,2, C. Williams1,2, S. Antonia1,2, A. Chiappori1,2, J. Gray1,2, T. Tanvetyanon1,2, B. Creelan1,2, E. Haura1,2, M. Shafique1,2, J. Fontaine1,2, J. Cox1,2, F. Kaszuba1,2, R. Keenan1,2, V. Nair1,2, E. Toloza1,2 1Moffitt Cancer Center And Research Institute,Tampa, FL, USA 2University Of South Florida College Of Medicine,Tampa, FL, USA
Introduction: When tumor biopsies are not feasible, liquid biopsy of peripheral blood circulating tumor DNA (ctDNA) and protein has been shown to capture genetic and proteomic data that represent the entire tumor burden of each cancer patient. We sought to investigate whether liquid biopsy can correlate histopathologic factors, treatment, or outcomes with peripheral blood ctDNA mutations and proteomic signatures.
Methods: We retrospectively analyzed data from all non-small cell lung cancer (NSCLC) patients who underwent liquid biopsy analysis of ctDNA and proteins on peripheral blood samples from August 2016 to June 2018. This ctDNA analysis detected presence of targetable mutations, and proteomic analysis grouped patients into either Good or Poor status. Patients with targetable mutations were excluded. Liquid biopsy results were then correlated with histopathologic factors, such as tumor histology, grade of differentiation, tumor (T) status, nodal (N) status, metastasis (M) status, pathologic stage (pStage), and treatment. Student’s t-test, Kruskal-Wallis test, or Chi-square test were used to compare these factors between groups, and Kaplan-Meier curves were used to compare survival. Statistical differences were significant at p≤0.05.
Results: Of 522 patients analyzed by liquid biopsy, 92 (17.6%) mutation-positive patients were excluded. Of 430 (82.4%) mutation-negative patients, 376 (87.4%) had proteomic Good status, and 54 (12.6%) had proteomic Poor status. Mean age did not differ between Good and Poor groups (68.4 yr vs. 65.1 yr; p=0.07). Mean primary tumor size did not differ between Good and Poor groups (p=0.342). Histology (i.e. adenocarcinoma, squamous cell carcinoma, neuroendocrine carcinoma, etc.) did not differ between Good and Poor groups (p=0.11). However, tumor grade of differentiation, N status, M status, and pStage differed between Good and Poor groups, with the Poor group having more patients with poorly-differentiated (G3) tumors (p<0.01), with N2 or N3 status (p<0.01), with M1 status (p<0.01), and with pStage III and IV cancers (p<0.01). Similarly, treatment differed between Good and Poor groups, with the Good group more likely to have surgery and the Poor group more likely to receive systemic therapy (p<0.01). In Kaplan-Meier survival analysis, the Good group had 1-year overall survival (1-yr OS) of 88.5% compared to a 1-yr OS of 48.4% for the Poor group (p<0.01).
Conclusion: Using a commercially-available peripheral blood liquid biopsy kit, mutation-negative NSCLC patients were identified by ctDNA analysis and as Good or Poor status by proteomic analysis. While age and tumor size did not correlate with Good versus Poor status, the Poor group had significantly more poorly-differentiated tumors, more mediastinal LN (N2 and N3) involvement, more distant metastases, and higher pStaged cancers, required systemic therapy more often, and had significantly worse 1-yr OS than proteomic-Good patients.