Image cytometric HER2 in gastric carcinoma: Is a new algorithm needed?

BACKGROUND: Algorithms for quantitation of HER2 immunohistochemistry were developed for breast carcinoma, where the membranous stain must be entirely around the cell membrane. For gastric carcinoma, although assessment of intensity of immunostain (0 to 3) is similar, the site and percentage of stain differs by lacking the requirement of entire cell membrane positivity (complete, basolateral, or lateral membranous reactivity is sufficient for a positive result). We quantitated HER2 in gastric cancer specimens visually and by image cytometry, comparing results, and where available, with fluorescence in situ hybridization (FISH). The goal was to assess whether lack of concordance among results, might suggest a requirement for changing the image cytometric algorithm. DESIGN: All gastric carcinoma biopsies, resections, and cell blocks studied for HER2 expression/amplification in the past 2 years were included. Immunostain intensity, percentage, and score 0 to 3+ (0, 1+ negative, 2+ equivocal, 3+ positive), were evaluated visually, and by image cytometry with the ACIS score 0 to 3 (0, 1 negative, 2 equivocal, 3 positive). FISH (<1.8 negative, 1.8 to 2.2 equivocal, >2.2 amplified) was performed on all specimens with scores 2 and 3 by image cytometry. Results were compared. Results: Sixty-eight specimens were studied, including 43 (63.2%) biopsies, 17 (25%) resections, and 8 (11.8%) cell blocks. Forty-seven (69.1%) were primary gastric, esophageal, or gastroesophageal junction adenocarcinoma; 19 (27.9%) were metastatic; 3 (4.4%) were well, 14 (20.6%) moderately (17, 25% tubular), and 51 (75%) poorly differentiated (poorly cohesive). Fourteen (20.6%) of cases were HER2 IHC positive with no significant difference in frequency based on type of specimen, site of carcinoma, or differentiation. Of the 14 visually HER2 IHC positive, 13 were positive by image cytometry (93% concordance), all 13 were amplified by HER2 FISH (100% concordance). Of the 3 cases equivocal both visually and by image cytometry, only 1 was FISH amplified. Fifty-one were negative by IHC visually and 52 by image cytometry (98% concordance). None of the 5 HER2 IHC negative were amplified by FISH. Conclusions: Despite different recommendations for interpretation of HER2 in gastric versus breast cancer, equivocal and positive/amplified results visually, and by image cytometry, and where FISH was performed, are similar. This concordance is noted for biopsy, resection, and cell block specimens, for primary versus metastatic, and for moderately versus poorly differentiated carcinoma; HER2 positivity/amplification is most frequent with poor differentiation, but not significantly so. There seems to be no need for the HER2 image cytometric algorithm used for breast cancer, to be changed when used for assessment of gastric cancers.