Value of SATB2, ISL1, and TTF1 to differentiate rectal from other gastrointestinal and lung well-differentiated neuroendocrine tumors

Abstract Background Management of neuroendocrine tumors (NETs) depends on the primary site, but the location of many well-differentiated (WD) NETs is elusive. Organ-specific markers are required for pathological diagnosis from biopsy. Transcription factors with good organ specificity include TTF1 (thyroid transcription factor 1; lung), CDX2 (caudal type homeobox transcription factor 2; midgut), and ISL1 (ISL LIM homeobox 1) and PAX8 (paired box 8) for the pancreas and rectum. SATB2 (SATB homeobox 2) has shown high sensitivity and specificity in colorectal adenocarcinoma. This study determined the viability of SATB2 and other transcription factors as markers, single or in combination, for WD-NETs of various sites. Methods Immunohistochemical staining of 81 WD-NETs from 8 organ sites was performed to identify SATB2, TTF1, CDX2, ISL1, and PAX8. Receiver operating characteristic (ROC) curves were constructed for different combinations of the 5 markers to determine sensitivity and specificity. Results Among the WD-NETs, SATB2 was predominantly found in those of the rectum; TTF1 in the lung, larynx, and esophagus; and ISL1 in the duodenum and rectum. PAX8 and CDX2 showed poor organ specificity. ROC profiles showed 50% sensitivity and 96% specificity to lung for TTF1 + ISL1–; and 65% sensitivity and 100% specificity to rectum for SATB2 + ISL1– TTF1–. ISL1 + SATB2– TTF1– showed 83% sensitivity and 85% specificity to the duodenum, and 44% sensitivity and 87% specificity to the pancreas. A literature search showed that there was no significant difference in the expression rates of the five transcription factors (TTF1, CDX2, SATB2, PAX8 and ISL1) between primary and metastatic WD-NETs at the same organ when there was a large sample size. Conclusion Among the 5 transcription factors tested, SATB2 may be a viable marker of WE-NETs of the rectum. The combination of SATB2, ISL1, and TTF1 may help estimate the locations of WD-NETs of unknown origin.