Validation of the AJCC 8th edition prognostic stage in breast cancer.

Abstract

e23186 Background: The 8th edition of the AJCC staging manual introduced a prognostic stage incorporating biologic factors. The goal this study was to validate the prognostic stage in a single institution cohort and large population database. Methods: Patients were identified in a prospective institutional database and the California Cancer Registry (CCR). Factors included were T, N, m stage, grade, estrogen receptor, progesterone receptor and HER2 status. Survival was calculated by Kaplan Meier method. Harrell’s concordance index (C index) was used to quantify models’ predictive performance, and Akaike Information Criterion (AIC) to compare model fits. Results: A total of 3,327 Stage I-IIIC breast cancer patients treated 2007-2013 at a single institution with complete clinicopathologic data were identified. Compared to the anatomic stage, the prognostic stage upstaged 29.5% of patients and downstaged 28.1%. Median follow up was 5 yrs. The prognostic staging system’s C index (0.84) and AIC (816.8) reflect a more accurate model predictive of disease specific survival (DSS) (Table) than the anatomic staging (C index 0.74, P < .0001; AIC 1039.8). A total of 49,767 Stage I-IIIC patients diagnosed 2005-2009 with complete data were identified in the CCR. 31.7% of patients were upstaged and 21.1% downstaged. Median follow up was 7.15 years. Prognostic staging performed better (C index 0.81, AIC 65924.1) than anatomic staging (C index 0.77, P < .0001; AIC 66847.7). Conclusions: The newly released 8th AJCC prognostic staging system, validated in both a single institution and a large population database, provides more accurate prognostic information than the anatomic staging system. [Table: see text]