P1-07-10: Validation of a Diagnostic Molecular Signature (EHT Dx14) on Fine-Needle Aspirate Samples from Breast Tumors.

Abstract

Abstract Background: A 1228-probeset molecular classifier (EHT Dx14) able to discriminate benign breast lesions from breast adenocarcinoma with a 96% performance was previously identified in a panel of 71 fine-needle aspirate (FNA) samples of nodular breast lesions collected at Institut Gustave Roussy (IGR). The microarray which enabled the signature identification was based on the genomewide RNA analysis using the ExonHit's SpliceArray™ technology. The objective of the current study was to validate the performance of the EHT Dx14 molecular classifier on an independent cohort of FNA samples stored at the Centre de Ressources Biologiques at IGR. Patients and methods: The samples were collected from women undergoing investigations for suspicious nodular breast lesions classified BIRAD ACR 4 or 5 on mammogram and/or ultrasonography. The morphological cytological analysis of FNA samples classified the lesions as either benign, malignant or indeterminate. All samples were associated with a definite clinico-pathological diagnosis of the breast lesions. The first part of the study aims at confirming the sensitivity, specificity and overall performance of EHT Dx14 among lesions clearly classified by morphological cytology. The second part aims at establishing its performance among lesions with indeterminate cytological results. Results: In the first part of the study, the EHT Dx14 transcriptomic signature was applied on a set of 94 FNA consecutive samples categorized as benign (n=47) or adenocarcinoma (n=47) on cytology. The benign/malignant status of the lesions was confirmed by subsequent investigations or follow-up. Results of EHT DX-14 assessment of the samples were respectively 97.9% (95%CI: [88.7%-99.9%]) for sensitivity and 91.5% (95%CI: [79.6%-97.6%]) for specificity. Results on indeterminate FNA samples are being investigated in the second part of the study. Conclusion: This study did confirm the high performance of EHT Dx14 signature in identifying malignant from benign breast lesions in clinically suspect lesions. EHT Dx14 may provide adequate biological diagnosis in the absence of capability of morphological evaluation. Citation Information: Cancer Res 2011;71(24 Suppl):Abstract nr P1-07-10.