SATB2 as a Diagnostic Marker for Tumors of Colorectal Origin
SATB2 as a Diagnostic Marker for Tumors of Colorectal Origin
The study was approved by the Department of Surgical Pathology, Uppsala University Hospital, Uppsala, Sweden, and all cases were prospectively selected from the clinical cases registered between June 2010 and December 2011. The inclusion criteria were defined as tissue specimens submitted to the department of pathology for diagnosis, where the pathologist in charge had performed an immunohistochemical staining of CK20 as part of the clinical diagnostic procedure. The total number of cases immunohistochemically stained for CK20 during this period was 840. The corresponding paraffin blocks from these cases were, in parallel with CK20, sectioned and immunostained for SATB2 expression. In accordance with current diagnostic guidelines, the pathologist in charge used several other antibodies in addition to CK20 (such as CK7, CDX2) and other ancillary methods to reach the final diagnosis. However, the result of SATB2 staining was disregarded in the diagnostic process when cases were signed out by the pathologists.
IHC was performed after routine antigen retrieval using an automated IHC stainer (Autostainer, DakoCytomation, Glostrup, Denmark) as previously described. Primary antibodies against CK20 (mouse monoclonal M7019; DakoCytomation) and SATB2 (mouse monoclonal antibody AMAb90635, clone CL0276, Atlas Antibodies AB, Stockholm, Sweden) were used at a dilution of 1:200. All other antibodies deemed necessary for the diagnostic procedure were stained using the same instrumentation as specified before and at dilutions routinely established in the surgical pathology laboratory.
The CK20 and SATB2 immunohistochemically stained slides in all cases were reevaluated by two independent observers, one pathologist (A.D.) and one specially trained technician (C.J.). All disagreements were resolved by evaluating the slides with a double-headed microscope. The annotation process included a semiquantitative, categorical estimation of the fraction (%) of tumor cell nuclei positive for SATB2 regardless of intensity (nuclear fraction [NF]), which was scored 0 for less than 1%, 1 for 2% to 25%, 2 for 26% to 75%, and 3 for more than 75%. For the purpose of statistical analyses, SATB2 nuclear staining was further dichotomized into negative tumors (score 0) and positive tumors (scores 1–3). For CK20, a binary expression of the cytoplasmic staining was used, scored 1 for positive staining (any number of tumor cells) and 2 for negative staining, regardless of intensity. For the remaining immunohistochemical markers, the results were retrieved from the pathology records and were scored in a similar binary mode as for CK20. Representative images of SATB2 staining are illustrated in Image 1.
(Enlarge Image)
Image 1.
Typical examples of CK20 and SATB2 staining in primary colorectal tumors. A, CK20 positive. B, CK20 negative. C, SATB2 diffusely positive (more than 75% nuclei stained, nuclear fraction [NF] = 3). D, SATB2 partly positive (26%–75% nuclei stained, NF = 2). E, SATB2 sparsely positive (2%–25% nuclei stained, NF = 1). F, SATB2 negative (0%–1% nuclei stained, NF = 0). (All images ×200.)
The final diagnosis was retrieved from the pathology record. No independent review of the diagnosis was attempted, because the authors had access only to a fraction of the diagnostic material, limited to the tissue sections selected for CK20 and SATB2 immunostaining. The final diagnosis was classified into one of four categories: "CRC," defined other non-CRC cancer ("other C"), "CUP," and benign tumor ("benign"). Information on whether the tumor was a primary cancer or a metastasis and the anatomic origin was recorded for cases in which such information was available. For the cases classified as CUP, the pathology database was assessed 6 months after the inclusion time, resulting in a follow-up time of 6 to 24 months. For cases in which new diagnostic data were available, the new diagnosis of the tumor type was recorded separately.
Statistical analyses were carried out using the SPSS software program (SPSS version 20, IBM, Armonk, NY). The distribution of cases is presented either as absolute values or, in parentheses, as relative frequencies. The Pearson χ test (two-sided) was used to investigate the relationship between SATB2 and other markers, and a P value of less than .05 was considered statistically significant. The sensitivity, specificity, and predictive power of various immunohistochemical markers were computed from binary contingency tables, in which the reported diagnosis in the pathologic record was defined as the "true" diagnosis.
Materials and Methods
Case Selection and IHC
The study was approved by the Department of Surgical Pathology, Uppsala University Hospital, Uppsala, Sweden, and all cases were prospectively selected from the clinical cases registered between June 2010 and December 2011. The inclusion criteria were defined as tissue specimens submitted to the department of pathology for diagnosis, where the pathologist in charge had performed an immunohistochemical staining of CK20 as part of the clinical diagnostic procedure. The total number of cases immunohistochemically stained for CK20 during this period was 840. The corresponding paraffin blocks from these cases were, in parallel with CK20, sectioned and immunostained for SATB2 expression. In accordance with current diagnostic guidelines, the pathologist in charge used several other antibodies in addition to CK20 (such as CK7, CDX2) and other ancillary methods to reach the final diagnosis. However, the result of SATB2 staining was disregarded in the diagnostic process when cases were signed out by the pathologists.
IHC was performed after routine antigen retrieval using an automated IHC stainer (Autostainer, DakoCytomation, Glostrup, Denmark) as previously described. Primary antibodies against CK20 (mouse monoclonal M7019; DakoCytomation) and SATB2 (mouse monoclonal antibody AMAb90635, clone CL0276, Atlas Antibodies AB, Stockholm, Sweden) were used at a dilution of 1:200. All other antibodies deemed necessary for the diagnostic procedure were stained using the same instrumentation as specified before and at dilutions routinely established in the surgical pathology laboratory.
IHC Scoring
The CK20 and SATB2 immunohistochemically stained slides in all cases were reevaluated by two independent observers, one pathologist (A.D.) and one specially trained technician (C.J.). All disagreements were resolved by evaluating the slides with a double-headed microscope. The annotation process included a semiquantitative, categorical estimation of the fraction (%) of tumor cell nuclei positive for SATB2 regardless of intensity (nuclear fraction [NF]), which was scored 0 for less than 1%, 1 for 2% to 25%, 2 for 26% to 75%, and 3 for more than 75%. For the purpose of statistical analyses, SATB2 nuclear staining was further dichotomized into negative tumors (score 0) and positive tumors (scores 1–3). For CK20, a binary expression of the cytoplasmic staining was used, scored 1 for positive staining (any number of tumor cells) and 2 for negative staining, regardless of intensity. For the remaining immunohistochemical markers, the results were retrieved from the pathology records and were scored in a similar binary mode as for CK20. Representative images of SATB2 staining are illustrated in Image 1.
(Enlarge Image)
Image 1.
Typical examples of CK20 and SATB2 staining in primary colorectal tumors. A, CK20 positive. B, CK20 negative. C, SATB2 diffusely positive (more than 75% nuclei stained, nuclear fraction [NF] = 3). D, SATB2 partly positive (26%–75% nuclei stained, NF = 2). E, SATB2 sparsely positive (2%–25% nuclei stained, NF = 1). F, SATB2 negative (0%–1% nuclei stained, NF = 0). (All images ×200.)
Establishing the Final Diagnosis
The final diagnosis was retrieved from the pathology record. No independent review of the diagnosis was attempted, because the authors had access only to a fraction of the diagnostic material, limited to the tissue sections selected for CK20 and SATB2 immunostaining. The final diagnosis was classified into one of four categories: "CRC," defined other non-CRC cancer ("other C"), "CUP," and benign tumor ("benign"). Information on whether the tumor was a primary cancer or a metastasis and the anatomic origin was recorded for cases in which such information was available. For the cases classified as CUP, the pathology database was assessed 6 months after the inclusion time, resulting in a follow-up time of 6 to 24 months. For cases in which new diagnostic data were available, the new diagnosis of the tumor type was recorded separately.
Statistics
Statistical analyses were carried out using the SPSS software program (SPSS version 20, IBM, Armonk, NY). The distribution of cases is presented either as absolute values or, in parentheses, as relative frequencies. The Pearson χ test (two-sided) was used to investigate the relationship between SATB2 and other markers, and a P value of less than .05 was considered statistically significant. The sensitivity, specificity, and predictive power of various immunohistochemical markers were computed from binary contingency tables, in which the reported diagnosis in the pathologic record was defined as the "true" diagnosis.
