Abstract P2-11-03: A tumor microenvironment of metastasis (TMEM) biomarker in early breast cancer provides prognostic information that is complementary to IHC4: A prospective validation study

Abstract

Abstract Background: Multiparameter gene expression assays (eg, Oncotype DX®) provide prognostic and predictive information in estrogen receptor (ER)+ Her2- negative disease, but are driven largely by proliferation and ER signaling captured by immunohistochemistry (IHC) assays such as IHC4 (Cuzick et al. J Clin Oncol. 2011 10;29:4273-8). However, these assays do not reflect the capacity of individual tumor cells to metastasize, nor interaction with their microenvironment. Using a bench-to-bedside approach, we have identified in animal models and human breast cancer by IHC methods a microanatomic landmark (TMEM) precisely where Mena-overexpressing invasive tumor cells intravasate and hematogenously metastasize by directly interacting with endothelial cells and macrophages (Robinson et al. Clin Cancer Res 2009;15:2433-41). In the present study, we performed a prospective validation study to establish analytic and clinical validity of TMEM in comparison with IHC4 in accordance with recommended guidelines (Simon et al. JNCI 2009;101:1446-52). Methods: A case-control study was performed nested within a cohort of 3760 invasive ductal breast carcinomas, in which cases (subsequent distant recurrence) and controls (no recurrence) were selected using incidence density sampling matched on age at and calendar year of primary diagnosis, resulting in 481 subjects (259 case-control pairs). TMEM and IHC4 were performed as previously reported, and stained slides were read independently by 5 pathologists blinded to outcome. Odds ratios (OR)(95% CI) for the association with TMEM were estimated using logistic regression, with adjustment for clinical variables, including age, tumor size, grade, number of positive lymph nodes, ER, PR, and Her2 expression, lymphovascular invasion (LVI), and use of chemotherapy or hormonal therapy. The ORs for TMEM was compared with those for IHC4. Results: in a multivariate model adjusted for other covariates, TMEM was associated with increased risk of distant metastasis in 62% of subjects with ER+/Her2- tumors (OR [95%CI]high vs. low tertile = 2.70 [1.39-5.26], Ptrend = 0.004), whereas IHC4 had a borderline positive association (OR10 unit increase = 1.06 [1.00-1.13]); the association for TMEM persisted after adjustment for IHC4 (OR [95%CI]high vs. low tertile = 2.68 [1.31-5.49], Ptrend = 0.004). TMEM scores ranged up to a nearly 200-fold from lowest to highest, and did not correlate with tumor size, number of positive lymph nodes, nor IHC4 score (Spearman correlation 0.12 for IHC4). The AUC for a TMEM composite score (0.77), derived from a multivariate logistic regression model with TMEM and clinical variables, was significantly higher than those for TMEM alone (0.65) and for IHC4 (0.62). Neither TMEM nor IHC4 was independently associated with metastatic risk in the triple negative (20%) or HER2+ (18%) tumors. Conclusion: TMEM score, a mechanism-based assay, is positively associated with risk of distant metastasis in ER+/Her2- breast cancer and provides prognostic information that is complementary to IHC4 and other clinicopathologic risk factors. Additional validation studies designed to establish clinical validity are currently being planned. Citation Information: Cancer Res 2013;73(24 Suppl): Abstract nr P2-11-03.