Abstract P3-16-01: Using the new pan-cancer clinical data resource (TCGA-CDR) to identify breast cancer genomic correlates associating with different survival outcome endpoints

Abstract

Abstract Introduction The Cancer Genome Atlas (TCGA) generated abundant high quality molecular data, however its relatively short-term patient follow-up limited its immediate clinical utility. We led a PanCanAtlas effort to systematically collate, integrate, and quality check the large body of acquired clinicopathologic data, generated 4 primary clinical outcome endpoints for each case, and created a new Pan-Cancer Clinical Data Resource (TCGA-CDR) for public use. We report here on the utility and validity of this TCGA-CDR in relating breast cancer (BC) genomic information to survival endpoints. Methods Clinicopathologic data from all data files were integrated and processed. Overall survival (OS), disease-specific survival (DSS, an approximation), progression-free interval (PFI), and disease-free interval (DFI)were derived.Tests of the adequacy of the follow-up intervals for each endpoint were performed, and quality evaluation of these endpoints was established by their comparison with different clinical features. As a case study we compared each survival endpoint for significant association (FDR <0.2) with chromosomal aneuploidy. Results The 4 endpoints were derived for 1097 TCGA BC cases having a median follow-up time of 27.7 months. Median times to events/censorship for OS, DSS, PFI, and DFI were 41.8/25.0, 32.6/26.0, 26.0/25.0, and 25.4/25.0 months respectively. PFI and DFI passed tests for adequate follow-up times; OS and DSS partially passed the same tests signaling some caution with their use in genomic associations. Using the endpoints, outcomes of patients with ER+ and ER- tumors were compared, along with those of patients with low (I&II) and high (III&IV) stage breast tumors. Univariate analyses suggested patients with ER+ tumors had significantly better survival than patients with ER- tumors when using PFI (p=0.005), DFI (p=0.001), and DSS (p=0.009), with OS not reaching significance (p=0.09). Patients with low stage tumors showed significantly better outcomes than patients with high stage tumors for each endpoint (p<0.001). The 4 endpoints were also evaluated for their significant associations with chromosomal arm aneuploidy. Adjusted for patient age and AJCC stage, tumors with a loss of 8q and 8p (p=0.019, FDR=0.37) had worse PFI; and those with loss of 8q, 20q, and 8p had worse DFI. Tumors with gain of 11q or loss of 14, 7q, 12q, 18q, 20q, 3p, 7p, 8p, 18p, and 20p had worse OS. In contrast, tumors with loss of 16q had better DSS, while those with loss of 3q, 12q, 17q, 18q, 19q, 20q, 3p, 8p, 12p, 18p, 19p, and 20p had worse DSS. The finding that 8p loss associated with worse survival for all 4 endpoints, while 18p loss associated with worse OS and DSS, agrees with literature reports. Conclusion These findings confirm that PFI and DFI, as extracted from the TCGA-CDR, are valid and appropriate BC survival endpoints, while OS and DSS may be recommended with some caution when employing TCGA data to evaluate new relationships between breast cancer genomic abnormalities and clinical outcomes. The views expressed in this article are those of the author and do not reflect the official policy of the Department of Army/Navy/Air Force, the Department of Defense, or U.S. Government. Citation Format: Liu J, Lichtenberg T, Hoadley KA, Cherniack A, Poisson L, Kovatich AJ, Benz C, Thorsson V, TCGA PanCanAtlas Research Network, Shriver CD, Hu H. Using the new pan-cancer clinical data resource (TCGA-CDR) to identify breast cancer genomic correlates associating with different survival outcome endpoints [abstract]. In: Proceedings of the 2017 San Antonio Breast Cancer Symposium; 2017 Dec 5-9; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2018;78(4 Suppl):Abstract nr P3-16-01.