ABSTRACT
BACKGROUND: Distinguishing metastatic carcinoma of breast origin (MCBO) to lung from primary lung carcinomas (PLC) is a diagnostic quandary with clinical ramifications. Immunostains CK7, CK20, ER, PR, and Mammaglobin as well as pertinent negative stains are utilized but prove insufficient. We set out to identify stains either alone or as a group that would better discern between these 2 entities. DESIGN: Tissue microarrays of 109 MCBO to lung and 102 PLC were stained with CK7, CK20, ER, PR, AR, Mammaglobin, Napsin A, GATA-3, and TTF-1. An H-score was calculated for each case and stain. RESULTS: The highest area under the receiver-operating characteristic curves for each stain was seen with GATA-3 (0.817), Napsin A (0.817), and TTF-1 (0.854). When all possible combinations were analyzed, GATA-3 and TTF-1 proved to correctly classify with the highest accuracy (0.934). Combinations of GATA-3 and Napsin A (0.920) and GATA-3, Napsin A, and TTF-1 (0.933) were not significantly different from GATA-3 and TTF-1. The odds ratios for each stain and combination of stains showed that those for GATA-3 and TTF-1 were divergent, signifying that cases with higher H-scores for GATA-3 and TTF-1 were more likely to be classified as MCBO and PLC, respectively. CONCLUSIONS: GATA-3 and TTF-1 can correctly classify a case as either MCBO or PLC in 93.4% of cases. Although highly specific and sensitive for PLC, Napsin A in lieu of TTF-1 or as an additional stain did not improve classification accuracy.
