Detection of HER-2/neu (c-erb B-2) DNA amplification in primary breast carcinoma. Interobserver reproducibility and correlation with immunohistochemical HER-2 overexpression.

Abstract

Fluorescent in situ hybridization (FISH) has been shown to be one of the most reliable methods with which to estimate the status of the HER-2/neu (or c-erb B-2) oncogene at the DNA level. To study interobserver reproducibility and to determine more clinically correlated criteria for HER-2/neu alterations, two observers independently estimated HER-2/neu DNA status. The correlation between the consensus HER-2/neu DNA status by FISH and HER-2/neu protein status detected by immunohistochemistry (IHC) using a polyclonal antibody was studied in 216 surgically resected breast carcinomas and 34 noncancerous tissues. According to the HER-2/CEP17 ratio and mean HER-2 copies per nucleus, agreement level of HER-2/neu amplification was shown to be nearly perfect between two observers (kappa statistic (kappa) = 0.94 and kappa = 0.84). Finally, 40 tumors (19%) were judged to have HER-2/neu DNA amplification, with 6 having low-level amplification (> or = 2 but < 3 folds) and 34 having high-level amplification (> or = 3 folds). One hundred seventy-six other tumors, including 3 tumors that only 1 of the observers determined to be low-level amplifiers, and 34 noncancerous tissues had no detected amplification. The DNA amplification status was concordant between invasive and intraductal components in 14 carcinomas. HER-2/neu protein overexpression of moderate (2+) or high (3+) intensity based on IHC was detected in 51 carcinomas (24%), and was 2+ in 20 carcinomas and 3+ in 31 carcinomas. The HER-2/CEP17 ratio of > or = 2 was concordant with IHC findings of 2+/3+ in 91% of carcinomas (195 of 215 carcinomas), with a sensitivity of 70% (35 of 50 carcinomas) and a specificity of 97% (160 of 165 carcinomas). High-level amplification was detected in 29 of 31 IHC 3+ cases (94%), but in only 5 of 20 IHC 2+ cases (25%) and 0 in 165 IHC 0/1+ cases. All 34 cases with high-level amplification showed an IHC score of 3+ (29 cases) or an IHC score of 2+ (5 cases), but only 1 case was found to have an IHC score of 3+ and the remainder were IHC 0/1+ in 6 low-amplification cases. The concordance rate of the high-level amplification with an IHC score of 3+ was 97% (208 of 215 cases), with a sensitivity of 94% (29 of 31 cases) and a specificity of 97% (179 of 184 cases). The results of the current study indicated that high-level HER-2/neu amplification and an IHC score of 3+ nearly optimally identified breast carcinomas with clinically and biologically significant HER-2/neu activation. Conversely, it was confirmed that careful interpretation of test results is required in the case of low-level amplification and/or an IHC score of 2+.