An algorithmic approach utilizing CK7, TTF1, beta-catenin, CDX2, and SSTR2A can help differentiate between gastrointestinal and pulmonary neuroendocrine carcinomas

Primary gastrointestinal neuroendocrine carcinoma (GI-NEC) cannot be distinguished morphologically from pulmonary neuroendocrine carcinoma (P-NEC). This can present a significant diagnostic challenge in cases where site of origin cannot be readily determined. To identify immunohistochemical (IHC) markers that can be used to reliably distinguish between GI-NECs and P-NECs, we constructed 3-mm tissue microarrays, one containing 13 GI-NECs and one containing 20 P-NECs. IHC was performed on both microarrays using 21 stains: AE1/AE3, CK7, CK20, synaptophysin, chromogranin, CD56, INSM1, SSTR2A, CDX2, SATB2, TTF1, Napsin A, PR, GATA3, PAX8, ISL1, beta-catenin, AFP, SMAD4, Rb, and p53. For GI-NEC, the most strongly expressed marker was synaptophysin (mean H-score 248), while AE1/AE3 was the most strongly expressed in P-NEC (mean H-score 230), which was stronger than in GI-NEC (p = 0.011). Other markers that were stronger overall in P-NEC than in GI-NEC included CK7 (p < 0.0001) and TTF1 (p < 0.0001). Markers that were stronger overall in GI-NEC than in P-NEC included SSTR2A (p = 0.0021), SATB2 (p = 0.018), CDX2 (p = 0.019), and beta-catenin (nuclear; p = 0.029). SMAD4, Rb, and p53 showed similar rates of abnormal protein expression. Based on these results, a stepwise algorithmic approach utilizing CK7, TTF1, beta-catenin, CDX2, and SSTR2A had a 91% overall accuracy in distinguishing these GI-NEC from P-NEC. This was tested on a second cohort of 10 metastatic GI-NEC and 10 metastatic P-NEC, with an accuracy in this cohort of 85% and an overall accuracy of 89% for the 53 cases tested. Our algorithm reasonably discriminates GI-NEC from P-NEC using currently available IHC stains.