SATB2 as an Immunohistochemical Marker for Colorectal Adenocarcinoma: A Concise Review of Benefits and Pitfalls

SATB2 was first identified in a complementary DNA sequencing project in 2002, and was then characterized as a gene involved in isolated cleft palate defects by FitzPatrick et al⁴  in 2003. Recently, a genetic syndrome associated with SATB2 defects has been described, with a phenotype including intellectual disability, craniofacial abnormalities including cleft palate and dental malformations, and osteopenia.⁵ 

Characterization of the transcription factor mechanism of SATB2 showed it to be part of the matrix attachment region–binding transcription factor family, which consists of transcription factors binding to AT-rich regions in the nuclear matrix (matrix attachment regions).⁶  These transcription factors are capable of altering chromatin structure over large distances and affecting multiple genes.⁶ 

SATB2 has high levels of expression in the brain, including in the cerebral cortex and the spinal cord, and has a role in central nervous system development.⁷  It is highly expressed during the period of cortical neuron differentiation, as well as in specific spinal cord neuron populations.^7,8 

SATB2 has also been shown to have a role in osteoblast differentiation via interactions with transcriptional regulator Runx2 and a microRNA network.^9,10  Dobreva et al⁹  showed that mice with SATB2 defects had global defects in bone formation, as well as craniofacial abnormalities thought to result from the effects on bone formation.

More recently, expression of SATB2 has been investigated as a prognostic factor in both colorectal and laryngeal carcinoma. A retrospective analysis of 146 and a prospective cohort of 557 colorectal carcinoma patients both showed that decreased SATB2 expression by immunohistochemistry was associated with a worse prognosis, as well as decreased sensitivity to chemotherapy and radiation, for colonic but not rectal carcinoma.^11,12  Mansour et al¹³  showed a tumor-suppressive function for SATB2 in vitro and a potential mechanism of this suppression via inactivation of tumor-promoting factor ERK5. Decreased immunohistochemical expression of SATB2 has also been shown to be a poor prognostic factor for laryngeal squamous cell carcinomas.¹⁴ 

The expression pattern of SATB2 was first described in mouse tissues, and was found to be present in the developmental first branchial arch, developing cortical and spinal cord neurons, and developing bones, as well as in mature pre-B cells, kidney, brain, testis, and thymus.^4,6,7,9 

SATB2 was then identified as a potential immunohistochemical marker of human colorectal epithelium through screening of the Human Protein Atlas database by Magnusson et al.³  The authors characterized the expression profile of SATB2 in normal human tissues using tissue microarrays, and found it to be highly expressed in the epithelium of the lower gastrointestinal tract (including appendix, colon, and rectum), as well as specific neurons (in the cerebral cortex and hippocampus), non–germinal center lymphoid cells, and the ductal epithelium of the testis and epididymis.³ 

The malignancy expression profile of SATB2 was initially characterized using a tissue microarray of 216 malignancy samples designed to represent the 20 most common human malignancies.³  Colorectal adenocarcinoma showed strong expression of SATB2 in the majority of samples, and strong expression was also seen in 1 renal cell carcinoma.³  Weak to moderate SATB2 expression was also seen in sinonasal carcinomas (56%; 5 of 9) and a small percentage of ovarian, breast, lung, and urothelial carcinomas (<7%).³ 

Since this initial screen for expression in malignancies, the finding of SATB2 expression in renal carcinoma has been replicated^15,16 ; in the study by Guo et al,¹⁵  26% (19 of 73) of clear cell renal cell carcinomas showed moderate to strong SATB2 staining. In addition to colorectal adenocarcinoma, SATB2 has also been found to be expressed in colorectal medullary carcinoma (16 of 18 cases).¹⁷  SATB2 has also recently been reported to be expressed in Merkel cell carcinomas, with 75% (15 of 20) of cases in one study expressing SATB2, the same number as were expressing CK20 in the characteristic perinuclear pattern.^16,18 

SATB2 has also been shown to be highly expressed in tumors of osteoblastic differentiation, including osteosarcomas and several benign bone tumors, and in some soft-tissue malignancies, including liposarcomas.^19,20  Incisional biopsies of osteoblastic tumors (which had no or brief decalcification) stained more diffusely and at a higher rate (56% versus 45%) than resection specimens that had been subjected to decalcification, indicating that SATB2 immunohistochemistry is sensitive to decalcification and ideally would be performed on the biopsy.²⁰ 

Identification of the primary location of adenocarcinoma presenting as a metastasis is a common diagnostic problem in anatomical pathology, and immunohistochemistry can be a helpful tool in addition to morphologic features, especially in poorly differentiated adenocarcinomas. A common approach to the immunohistochemical identification of metastatic colorectal adenocarcinoma is the use of the typical immunostaining pattern^21,22 CK7⁻, CK20⁺, and CDX2⁺ (see Figure 1, A through D). However, CDX2 and CK20 have been shown to be positive in up to 21%/21% of gastric and 14%/21% of ovarian mucinous adenocarcinomas, respectively, and CK7 was not particularly helpful in this differential, as it was positive in only about 50% of gastric and ovarian mucinous adenocarcinomas.²¹  This combination was more helpful in differentiating colorectal from pancreatic adenocarcinoma, which was only 2% CDX2⁺, 15% CK20⁺, and predominantly CK7⁺ (94%), and with only 3% of colorectal adenocarcinoma being CK7⁺.²¹  Bayrak et al²²  compared the use of the CK7⁻CK20⁺ pattern, which had a sensitivity of 64% and specificity of 97%, with use of CDX2 positivity, which had a 78% sensitivity and 85% specificity in differentiating colorectal from gastric and pancreatic adenocarcinoma. These and other studies have demonstrated the need for a better marker in differentiating colorectal from gastric, pancreatic, and ovarian mucinous adenocarcinoma.

The initial identification of SATB2 as a potential marker of colorectal adenocarcinoma by Magnusson et al³  was followed by validation using colorectal cancer tissue from 9 different research cohorts (1558 localized and 252 metastatic colorectal adenocarcinoma). Using SATB2 as a solitary marker, SATB2 was positive in 92.4% (110 of 119) of stage I, 91.4% (402 of 440) of stage II, and 83.7% (431 of 515) of stage III/IV colorectal adenocarcinomas. Based on these results, and the results of the tissue microarray from other common malignancies, Magnusson et al³  suggested that the combination of SATB2 with CK20 is a highly specific marker for colorectal adenocarcinoma.

In a prospective study of 840 cases in which CK20 was being used to reach a final diagnosis, Dragomir et al¹⁶  analyzed the expression of SATB2. As a solitary marker, SATB2 had a 93% sensitivity and 77% specificity in diagnosing colorectal adenocarcinoma, but when used as a combined marker along with CK20 positivity and CK7 negativity the sensitivity was 83% and the specificity was 100%.¹⁶  As a comparison, the sensitivity and specificity of the CK7⁻CK20⁺ immunophenotype were 85% and 99%, respectively, and for the CDX2⁺ immunophenotype these were 96% and 80%. Thus, the addition of SATB2 to the standard CK7 and CK20 panel did not significantly improve the sensitivity or specificity for diagnosis of colorectal carcinoma.

Another potential application of SATB2 is to distinguish adenocarcinomas of colorectal origin from those of gastric and pancreatic origin, which are common differentials. Lin et al¹⁷  looked at SATB2 expression in 1941 malignancy cases using tissue microarrays. They found SATB2 to be highly expressed in colorectal adenocarcinoma (96.8%; 121 of 125), with low expression in gastric adenocarcinoma (0%; 0 of 20) and pancreatic adenocarcinoma (4.2%; 4 of 95). This finding of low SATB2 expression in gastric and pancreatic adenocarcinomas has been replicated in other studies (see Table), and is promising for this particular application of SATB2.^3,17,23  However, in contrast to the results from other studies, Dragomir et al¹⁶  had a 22% SATB2 positivity in both gastroesophageal and pancreatobiliary cancer. Possible reasons for discrepancies include use of a different SATB2 antibody clone (see Table) and a low threshold for positivity (2%), used elsewhere only by Magnusson et al³  (the other referenced studies used 5% to 25%, or a “moderate” cutoff). With regard to gastric cancer, Dragomir et al¹⁶  included only 9 cases, and did not distinguish between gastric and esophageal cancers or the pathologic subtype. Interestingly, Lin et al¹⁷  showed 6.7% SATB2 positivity in esophageal adenocarcinoma, so esophageal carcinoma may partially account for the high rate in the Dragomir et al¹⁶  study.

The malignancy panel by Lin et al¹⁷  also showed low rates of SATB2 expression in lung and gynecologic malignancies, including 3% (6 of 198) of lung adenocarcinomas, 11% (13 of 121) of lung squamous cell carcinomas, 5% (5 of 55) of endocervical adenocarcinomas, and 4% (5 of 131) of endometrial adenocarcinomas. This finding of SATB2 expression in a low percentage of lung and gynecologic adenocarcinomas, which may be differential diagnoses in adenocarcinomas of unknown primary, was also seen in the Dragomir et al¹⁶  study.

A commonly encountered diagnostic problem in day-to-day anatomic pathology is intestinal-type mucinous carcinomas of the ovary, with the differential diagnosis being a primary ovarian mucinous adenocarcinoma or a metastatic colorectal/appendiceal adenocarcinoma. CK7, CK20, and CDX2 have all been used to help distinguish these entities, with ovarian mucinous carcinoma being primarily CK7⁺CK20⁻CDX2⁻ and colorectal/appendiceal carcinoma being primarily CK7⁻CK20⁺CDX2⁺, but there is a significant degree of overlap.²⁴  Prior to the discovery of SATB2, CDX2 was the most specific marker available, present in 60% to 100% of colorectal carcinomas but strongly positive in only up to 18% of ovarian mucinous carcinomas.²⁴  Two recent studies have shown SATB2 to be a specific marker of colorectal or appendiceal metastases versus primary ovarian mucinous carcinomas.^23,24  Montiel et al²⁴  studied immunohistochemical profiles of 51 primary ovarian mucinous tumors (including cystadenomas, borderline tumors, and adenocarcinomas) and 20 colorectal/appendiceal ovarian metastases. They found SATB2 to be expressed in 95% of the metastases and 0% of the ovarian primaries (versus 95% and 13% for CDX2).²⁴  Moh et al²³  examined SATB2 expression in 111 primary ovarian tumors and 46 colorectal/appendiceal ovarian metastases, and found SATB2 expression in 1 borderline tumor and no adenocarcinomas versus 78% of the metastases. In both studies, SATB2 was highly specific for colorectal and appendiceal metastases versus primary ovarian mucinous tumors, with the exception that SATB2 was found to be expressed in associated teratomas of ovarian primaries (see Table).^23,24 

The utility of SATB2 as a marker for metastatic colorectal adenocarcinoma in 111 cytology specimens was assessed as compared with a standard panel of CK7, CK20, and CDX2.²⁵  Use of SATB2 as a solitary marker had a sensitivity of 60% and a specificity of 91%, whereas the standard panel had a sensitivity of 79% and a specificity of 100%. When SATB2 was added to the standard panel, there was no significant difference in predictive value. The conclusion was that SATB2 did not add any additional value to the standard panel for diagnosing colorectal metastases in cytology specimens.²⁵ 

SATB2 had been in use as an immunohistochemical marker at Vancouver General Hospital in Vancouver, Canada, for 6 months when this article was written. The majority of cases expressing SATB2 (23 of 34) have been colorectal or appendiceal adenocarcinomas. Conversely, the majority of colorectal and appendiceal adenocarcinomas have stained positively with SATB2, with only 2 cases of 25 colorectal adenocarcinomas staining negatively for SATB2. Both negative cases were poorly differentiated, which is in line with the original results from Magnusson et al,³  with intact DNA mismatch repair proteins, and one had undergone osseous metaplasia. Thus, in our institution, SATB2 has a sensitivity of 92% and specificity of 75% for diagnosis of colorectal and appendiceal adenocarcinoma, which is comparable with the results for surgical specimens in the study by Dragomir et al¹⁶  (93% and 77%, respectively).

Also in line with previous results, there have been 1 case of papillary renal cell carcinoma (see Figure 2, A through C), 1 urothelial carcinoma with intestinal differentiation, and 1 urothelial villous adenoma that have stained positively with SATB2. Unexpected findings of malignancies staining with SATB2 include a desmoplastic mesothelioma (see Figure 2), a rectal neuroendocrine carcinoma, a poorly differentiated pancreatic adenocarcinoma, and a mucinous adenocarcinoma of the prostate. In addition, there was 1 case of a primary ovarian mucinous adenocarcinoma with intestinal differentiation, arising in a borderline mucinous tumor, which is the first reported ovarian mucinous adenocarcinoma staining with SATB2 (see Figure 2).

SATB2 has been shown to be highly expressed in the lower gastrointestinal tract and adenocarcinomas of colorectal and appendiceal origin, as well as in osteoblasts and tumors of osteoblastic differentiation, and moderately expressed in renal tissues and some renal/urothelial tumors. Many studies have now shown a high specificity for SATB2 in distinguishing adenocarcinomas of colorectal and appendiceal origin, but have not shown sensitivity and specificity improving on the standard panel of CK7, CK20, and CDX2. In addition, the expression of SATB2 seems to decrease in poorly differentiated adenocarcinoma, which is the most common situation in which this marker would be needed, and this diagnostic observation also appears to correspond with prognostic data indicating a poorer prognosis with decreased SATB2 expression.

One potential specific useful application of SATB2 is in differentiating colorectal/appendiceal adenocarcinomas from ovarian mucinous adenocarcinomas, with 2 papers showing a high specificity for SATB2 in this application. However, in the short time SATB2 has been in use in our institution, we have already seen 1 case of SATB2-positive ovarian mucinous adenocarcinoma.

Another potential specific application of SATB2 is in distinguishing adenocarcinoma of colorectal/appendiceal from that of gastric and pancreatic origin, with promising results from studies thus far, and it would be interesting to see a study specifically addressing this question.

Despite these relative shortcomings, SATB2 is a robust marker if used within a panel approach. The data thus far seem to indicate that SATB2 is a less sensitive but more specific marker of colorectal and appendiceal origin than CDX2 for an adenocarcinoma of unknown origin.