BackgroundThe excision repair cross-complementation group 1 (ERCC1) protein is the key enzyme of the nucleotide excision repair (NER) pathway. Loss of protein expression on immunohistochemistry is predictive for platinum-based chemotherapy response. Frequently, the diagnosis of malignancy is made on cytologic effusion samples. Therefore, we evaluated the staining quality of monoclonal anti-ERCC1 antibodies 8F1 and D-10 on microarrays of malignant pleural and peritoneal effusions by automated immunochemistry.MethodsCores from effusion cell blocks of 117 patients with > 40 malignant cell clusters per whole section (pleural n = 75, peritoneal n = 42) were assembled together with 30 histologic control cores from large tissue blocks (lung, breast and ovarian carcinoma, each n = 10) on hybrid cytology-tissue microarrays (C/TMA). Four immunochemistry protocols (Mab 8F1 and D-10, CC1-mono Ventana and H2-60 Bond automat) were performed. Immunoreactivity was semi-quantitatively scored for intensity and intensity multiplied by percentage staining (H-score).ResultsTumors were classified into female genital tract carcinoma (n = 39), lung adenocarcinoma (n = 23), mesothelioma (n = 15), unknown primary (n = 14), breast carcinoma (n = 10), gastro-intestinal carcinoma (n = 12) and other (n = 4). On both platforms, reproducible nuclear ERCC1 immunoreactivity was achieved with both antibodies, although D-10 was slightly weaker and presented more background staining as well as more variation in the low expression range. No significant differences were found between cytologic and histologic cores. Using the 8F1 CC1-mono protocol, lung and breast carcinomas had lower ERCC1 expression in comparison to the other entities (p-value < 0.05).ConclusionsCytology microarrays (CMA) are suitable for investigation of clinical biomarkers and can be combined with conventional TMA's. Dichotomization of ERCC1 immunoreactivity scores is most suitable for patient stratification since definition of negativity is antibody-dependent.