Abstract P4-07-09: An approach to the identification of tumors driven by HER2 using the integrated activity of oncomiR miR-21 along with HER2 enriched genes

Abstract

Introduction: The initial identification of HER2 as a driver in a subset of breast cancers was at the level of DNA (amplification), and subsequently noted at the level of transcripts and protein as well. However, the clinical selection of patients for treatment with Trastuzumab, has been through either IHC (protein) or FISH (DNA amplification) and not through transcript abundance. Interestingly, in most studies that have estimated transcript abundance in primary tumors, the proportion of patients that demonstrate increased transcript levels (termed HER2 Enriched) have tended to be slightly larger than the clinical HER2+ category. A more clinically useful measure might be proof of HER2 downstream activity that might help separate tumors being driven significantly by HER2 from ones where its role is supportive. One of the many consequences of HER2 over-expression is activation of the oncomiR, miR-21 via the MAPK pathway. miR-21 in turn is known to epigenetically regulate multiple targets including the tumor suppressors PTEN and PDCD4. While these molecular mechanisms have been demonstrated convincingly in breast cancer cell lines, clinical studies of these alterations in large numbers are yet to be reported. In this study we have examined the relationship between clinical HER2 positivity and miR-21 levels in 124 surgically excised breast cancer specimens. Methods: We selected 124 surgically excised specimens of primary breast cancers from our cohort that by HER2 immunohistochemistry (IHC) comprised 42 positive, 62 negative and 20 equivocal. Relative abundance of miR-21 was assessed using a TaqMan qRT-PCR, with normalization by RNU48. Relative transcript abundance of a set of 6 genes (HER2, GRB7, MLN64 and 3 reference genes) were evaluated by SYBR Green real time qPCR. Results: The majority of tumors that were clinically HER2+ over expressed miR-21. A concordance with an AUC of 96% at 100% sensitivity and 85% specificity was noted. There is a highly significant differential expression of miR-21 between HER2 positive, negative and equivocal samples (P < 0.0001). HER2 enriched score determined by using the expression levels of 3 genes (HER2, GRB7, MLN64) identified 35% (44/124) of the samples to be HER2 enriched. 72% of these (32/44) were also clinical HER2 positive by IHC. As expected, miR-21 was significantly over expressed in these tumors as well (P<0.0001). To identify all samples which might show HER2 downstream activity, a logistic regression model was built using expression of miR-21, HER2, MLN64 and GRB7 as the determinants of HER2 status. The best fitting model classified 91% (38/42) of HER2 +, 95% (59/62) of the HER2 negative accurately with 94% specificity and an AUC of 0.96. The model helped identify 10% of clinical HER2 negative samples (6/20 equivocal & 3/62 HER-2 negative) to have a high probability of being HER2+. Conclusion: Identification of HER2+ tumors with evidence of downstream activity may help identify patients with tumors being driven significantly by HER2 from ones where its role is supportive. The possibility of targeting miR-21 raises the tantalizing prospect of effecting change by altering the epigenetic regulation of multiple targets including tumor-suppressors. Citation Information: Cancer Res 2013;73(24 Suppl): Abstract nr P4-07-09.