Differential Diagnostic Value of SATB2 in Primary Pulmonary Enteric Adenocarcinoma and Pulmonary Metastases of Colorectal Carcinoma

Aims: Pulmonary enteric adenocarcinoma (PEA) is composed mainly of tall columnar cells arranged in an irregular acinar or cribriform pattern with extensive central necrosis, closely resembling the appearance of intestinal epithelial and colorectal carcinomas under the microscope. PEA is usually positive for CK7. However, some cases lack CK7 expression and are positive for intestinal differentiation markers, such as CDX2, villin, and CK20. It is dicult to distinguish between PEA and pulmonary metastases of colorectal carcinoma (MCRC), so new identication methods need to be explored. Methods and Results: We found that SATB2 was highly tissue-specic by bioinformatics analysis. It is highly expressed in intestinal tissues, and its expression level in lung tissues is very low. Then, according to the WHO diagnostic criteria, the cases of lung tumor were preliminary screened to obtain PEA and MCRC cases. In the preliminary screening process, immunohistochemical staining of common lung adenocarcinoma immunomarkers (CK7, TTF-1, and NapsinA) and common intestinal cancer immunomarkers (CK20, CDX2, and villin) were performed to aid identication. Then, after combined with clinical history, imaging examination and colonoscopy, 28 primary PEA specimens and 42 MCRC specimens were nal included. The above process was independently reviewed by two pathologists. PEA and MCRC had differences in the number of tumor nodules. The positive rates of SATB2 in PEA and MCRC were 0.00% and 92.86%, respectively. The sensitivity, specicity and diagnostic accuracy of SATB2 for distinguishing MCRC from PEA were 92.86%, 100% and 95.71%, respectively. The diagnostic accuracy of CK7 was 88.57%, which was slightly inferior to SATB2, and the results of CK7 could be used as a reference for SATB2 differential diagnosis. Conclusions: Our study showed that SATB2 can be viewed as the best immunomarkers for distinguishing PEA from MCRC. At the same time, CK7 could be used as references for SATB2 differential diagnosis.


Introduction
Primary enteric adenocarcinoma (PEA) is a rare histological subtype of lung adenocarcinoma. PEA is mainly composed of tall columnar cells arranged in an irregular acinar or cribriform pattern with extensive central necrosis, closely resembling the appearance of intestinal epithelial and colorectal carcinomas under the microscope [1]. Therefore, it is di cult to distinguish between PEA and pulmonary metastases of colorectal carcinoma (MCRC) by morphology.
CK20 is an intermediate lament protein that is selectively expressed in glandular cells of the gastrointestinal (GI) tract. CK20 is stably expressed in colorectal cancer, and the positive rate is greater than 90%. It also retains positive characteristics in metastasis [2]. However, the speci city of CK20 alone is relatively low because this keratin is also expressed in the gastric epithelium, urothelium, and epidermal Merkel cells [3]. CDX2 is a highly sensitive immunomarker for gastrointestinal (GI) adenocarcinomas but may also be expressed in tumors from other organs, such as the pancreas, bile ducts, bladder, uterine cervix, endometrium, and ovary [4]. Numerous previous studies reported that CK7 was positively expressed in almost all PEA patients, but the lack of CK7 expression was also reported in a small number of patients with PEA [5,6]. CK7 is one of the cytokeratin subtypes and is mainly expressed in glandular epithelial cells. The positive expression rate of CK7 in colorectal cancer tissues was 10%~27% [7]. So, in clinical practice, CK20 is often combined with other markers, such as the presence of CDX2 or the absence of CK7, to improve the accuracy of the diagnosis of colorectal carcinoma (CRC). However, some PEA cases lack CK7 expression and are positive for intestinal differentiation markers, such as CDX2, villin, and CK20 [8]. In this case, the combined diagnosis of immune indicators cannot also get accurate results.
As mentioned above, the morphology of PEA and MCRC are similar, and it is also di cult to make a differential diagnosis with commonly used immunomarkers. However, the lung presents a common site of metastasis for colorectal carcinoma, and the treatment strategies and prognosis outcomes differ considerably for patients with these malignancies [9]. So, it is of vital importance for pathologists and clinicians to make a clear distinguish between these entities. Therefore, we need to nd new differential diagnosis methods to distinguish between the two types of cancer.
The special AT-rich sequence binding-protein (SATB2) is a DNA-binding protein, which contains 733 amino acids and speci cally binds to nuclear matrix attachment regions of DNA. SATB2 expression is tissue-speci c, and the only epithelial cells expressing this protein in adult tissues are the glandular cells lining the lower GI tract [10]. SATB2 is a protein involved in transcription regulation and shows nuclear staining, thus, it could have some advantages over cytoplasmic/membrane staining markers such as cytokeratin. In a recent study, SATB2 was detected in more than 1,800 cases of CRC and more than 600 cases of other tumors. The results suggested that SATB2 was highly speci c for primary and metastatic lesions of CRC, and showed a similar level of sensitivity to CRC as CK20, CDX2, and CDH17. Besides, SATB2 in combination with CK20 can identify more than 95% of all CRC [11]. The above studies indicate that SATB2 may provide valuable information for clinical decision-making to con rm or exclude colorectal-derived tumors. In view of the high sensitivity and speci city of SATB2 for the pathological diagnosis of primary and metastatic CRC, it can be used to distinguish colorectal cancer from other cancers. This study collected 28 primary PEA specimens and 42 MCRC specimens to detect the speci city and sensitivity of SATB2 in PEA and MCRC, and to explore its differential diagnostic value in the two types of carcinoma.

Bioinformatics analysis
The expression level of SATB2 in different tissues was searched by pubmed (https://pubmed.ncbi.nlm.nih.gov/). In an article [12] with biological project number PRJEB4337, RNA sequencing was performed in tissue samples from 95 individuals representing 27 different tissues. We got RNA information of SATB2 from this article.

Patients and clinicopathological factors
After obtaining approval from the Ethical Committee of Renmin Hospital of Wuhan University (WDRY2019-K010), we preliminary screened the samples of patients with lung tumor who were admitted to Renmin Hospital of Wuhan University from January 2015 to June 2020 according to the WHO diagnostic criteria [1]. The specimens were independently reviewed by two pathologists, and immunohistochemical staining of common lung adenocarcinoma immunomarkers (CK7, TTF-1, and NapsinA) and common intestinal cancer immunomarkers (CK20, CDX2, and villin) was performed to aid identi cation. Finally, after carefully analyzing clinical history, lower abdomen CT and colonoscopy, we recruited 28 primary PEA specimens and 42 MCRC specimens. For cases that are still di cult to diagnose, follow up on the latest clinical diagnosis for at least 6 months to determine the nal diagnosis. All patients provided signed informed consent.
The clinicopathological data of all patients were collected from the department of pathology and department of thoracic surgery. These clinicopathological data include age, gender, tumor size, tumor number, smoking history, pleura invasion, bronchus invasion, lymph node metastasis, vessel invasion and nerve invasion.

Immunohistochemical staining
All surgically resected specimens were xed with 10% buffered formalin. We selected a representative slide of each case. After pretreatment with PT-link (Dako), immunohistochemistry analysis was performed using the Autostainer Link 48 (Dako).
Primary antibodies against cell markers used in this study included SATB2, CK20, CK7, CDX2, villin, TTF-1, NapsinA, MLH1, MSH2, MSH6 and PMS2, which were purchased from DAKO (Carpinteria, CA, USA). The dilution concentration of the antibodies is 1:1000. A medium-intensity colored tissue was used as a positive control, and PBS was used as a negative control instead of the primary antibody. As a result, it was determined that the positive cells were accurately positioned to be effective.

Staining Evaluation
We use the binary classi cation method to evaluate the immunostaining results. For CK7, CK20, villin, NapsinA, TTF-1 and CDX-2, tumors were considered positive if staining was found in ≥ 10% of the neoplastic cells. For SATB2, because the immunostains of SATB2 in PEA specimens all showed negative, while the immunostains of SATB2 in MCRC specimens almost all showed strong positive. We used the 2tiered scoring (all or nothing) method to evaluate the immunostains of SATB2. Tumor cells were considered to be positive for SATB2 when brownish-yellow nuclear particles were seen in tumor cells.
For MLH1, PMS2, MSH2, and MSH6, preserved expression was de ned as nuclear staining within tumor cells, using in ltrating lymphocytes as a positive internal control. Loss of MLH1, PMS2, MSH2, and MSH6 protein expression was de ned as the complete absence of nuclear staining in tumor cells with concurrent positive labeling in internal non-neoplastic tissues. Pro cient MMR protein expression was de ned as preserved nuclear expression of all four MMR proteins. De cient MMR protein expression was de ned as loss of protein expression of at least one of the 4 MMR proteins.
2.5 Sensitivity, speci city and diagnostic accuracy Sensitivity, speci city, and diagnostic accuracy were calculated to assess test performance. The formula for the calculation is as follows:

BRAF mutation analysis
Tumor targets were manually microdissected from 4-µm unstained histologic sections. DNA was extracted from para n sections, using the DNeasy tissue kit (Qiagen, Valencia, CA), according to the manufacturer's instructions. Real-time polymerase chain reaction (PCR) was performed using allele speci c primers designed to selectively amplify the wild-type (T1799) and mutant (A1799) BRAF alleles. After ampli cation, samples were subjected to a temperature ramp from 60 °C to 99 °C, increasing 1 °C at each step. For wild-type samples, single peaks were observed at 80 °C, whereas samples containing mutant alleles produced single peaks at 85 °C.

Statistics analysis
For analysis of clinicopathological factors, the enumeration data analysis between PEA group and MCRC group was carried out using the χ 2 -test and Fisher's exact test on IBM SPSS statistics 23 (IBM, New York, USA). P-values less than 0.05 were de ned as signi cant (*P < 0.05, **P < 0.01).

SATB2 is highly Tissue-Speci c and has Prognostic Value
We got the expression level of SATB2 in 27 different tissues from an article [12] with biological project number PRJEB4337 by pubmed, thereby obtaining Fig. 1. It can be seen that SATB2 expression was tissue-speci c. It was highly expressed in intestinal tissues and brain tissues, and its expression level in lung tissues was very low. Page 6/17

The Performance of Common Immunomarkers in Diagnosis
As stated in Sect. 2.2 of the article, we preliminary screened (lung tumor vs PEA and MCRC) the patients with lung tumor according to HE and common immunohistochemistry in the WHO [1]. We secondary screened (PEA vs MCRC) the cases obtained from the preliminary screening after carefully analyzing clinical history, lower abdomen CT, colonoscopy and 6 months follow-up. We nal obtained 28 cases of PEA and 42 cases of MCRC (Fig. 2).
In the preliminary screening, SATB2, CK7, CK20, CDX2, villin, TTF-1 and NapsinA play a great role. In order to show the effect of these immunomarkers, we randomly selected 60 cases of lung adenocarcinoma and 60 cases of CRC, and combined the nal PEA and MCRC to evaluate the positive rate of these immunomarkers ( Table 1). It suggested that PEA and MCRC had much higher positive rate of villin than lung adenocarcinoma. There were some differences in the positive rates of TTF-1, NapsinA, CK20 and CDX2 between lung adenocarcinoma and PEA. Between lung adenocarcinoma and MCRC, the positive rates of CK7, CK20, CDX2, villin, TTF-1 and NapsinA were also very different. In addition, we compared the positive rate of immunomarkers between CRC and MCRC. There was almost no difference between the two.

Differences between PEA and MCRC in Clinical Characteristics
We divide patients into PEA group and MCRC group. We performed the chi-square test to detect whether there were differences in age, gender, tumor size, tumor number, pleura invasion, bronchus invasion, lymph node metastasis, vessel invasion and nerve invasion between PEA and MCRC. As shown in The expression rates of SATB2 in PEA and MCRC were 0.00% (0/28) and 92.86% (39/42), respectively.

Mutation analysis results
In our results, there were 3 cases of MCRC specimens that showed negative for SATB2 staining. Protein expression in CRC is often in uenced by underlying molecular alterations such as KRAS and BRAF mutations, and DNA mismatch repair (MMR) protein de ciency [13][14][15][16][17][18]. In Ma's research [19], loss of SATB2 expression was more commonly seen in CRC with MMR protein de ciency and BRAF mutation. However, in our study, there were no BRAF mutations and MMR protein de ciency in the 3 cases of MCRC specimens with negative SATB2 ( gure omitted). This may be due to the poor differentiation of their cancer cells, so the SATB2 staining of the cancer cells can't show positive.

Discussion
PEA is characterized by the features of growth architectures of moderate-to-well differentiated glands, sometimes with a cribriform pattern formed, lined by tall columnar tumor cells with nuclear pseudostrati cation, dirty necrosis in the lumen and occasionally, prominent nuclear debris [1]. However, some of these morphologic features such as the tall columnar cells with brush-border and eosinophilic cytoplasm arranged in garland-like patterns can also be seen in MCRC. In our research, the expression rate of CK7 in PEA was very high; CK20, CDX2 and villin, which were not often expressed in lung adenocarcinoma, were also expressed in some PEA cases; while TTF-1 and NapsinA, which were often expressed in lung adenocarcinoma, was rarely expressed in PEA. As for MCRC, many cases expressed intestinal cancer immunomarkers (CK20, CDX2 and villin) and few cases expressed lung adenocarcinoma immunomarkers (TTF-1, NapsinA). And the expression rate of CK7 in MCRC was very low. These immunomarkers have limited value in the differential diagnosis of PEA and MCRC. These have caused di culties in diagnosis, but the treatment strategies and prognosis outcomes differ considerably for patients with the two malignancies [20]. Therefore, there is an urgent need to develop new markers to differentiate PEA and MCRC.
SATB2 is a nuclear matrix-associated transcription factor and a reliable marker of intestinal differentiation [11]. It serves as an important immunomarker to distinguish primary CRC from metastatic tumors [20]. SATB2 is expressed in 86%-93% of primary CRC and 81%-94% of MCRC [11], and the frequency of SATB2 expression in reported conventional lung adenocarcinoma is about 10% [7]. Here, we used bioinformatics to analyze the expression of SATB2 in the lung and colorectum. SATB2 is highly expressed in the colorectum and hardly expressed in the lung. The tissue-speci c of SATB2 illustrates a theoretical basis of our research.
Although several studies on SATB2 expression in PEA have been published, our conclusions have once again updated our understanding of the differential value of SATB2. We expanded the number of PEA and MCRC patients included in the study. Previous studies have shown that SATB2 could distinguish between PEA and MCRC [21][22][23], but the identi cation value is limited, and we even need to combine it with other immunomarkers in order to achieve a clear identi cation effect. In Mark's research [23], they showed that the differential diagnosis value of SATB2 in pulmonary and colorectal adenocarcinomas. SATB2 was positive in 80 (8%) non-small cell lung cancers (NSCLC), among various types of NSCLC, the positive rate of SATB2 is the highest in pleomorphic carcinomas. And 78 (98%) CRC were SATB2-positive. In my opinion, PEA is a rare histological subtype of lung adenocarcinoma. There will be some differences in the immunohistochemical performance of PEA and lung adenocarcinoma. However, due to changes in the tumor microenvironment, the immunohistochemical performance of MCRC and CRC also showed differences. This article can only provide us with limited reference value. In Jun's research [21], the case 4 was a PEA sample, but its SATB2 immunohistochemical staining was positive. It is worth noting that the mutation analysis of this case showed that its exon 2 of KRAS had a G12V mutation. In Bian's research [22], 2 of 13 PEA specimens showed that SATB2 immunohistochemical staining was positive. But they found that the SATB2 staining was weakly positive in the two cases. Figure 3 is a diagram of our research ideas, through this research, we found that SATB2 can largely identify the PEA and MCRC. The expression rates of SATB2 in PEA and MCRC were 0.00% and 92.86%, respectively. There were 3 cases of MCRC specimens that showed negative SATB2 staining, and these three cases were poorly differentiated and atypical intestinal adenocarcinoma. This may due to the poor differentiation of their cancer cells, leading to the failure of SATB2 staining. In Ma's research [19], loss of SATB2 expression is more commonly seen in CRC with MMR protein de ciency and BRAF mutation. But, in our study, there were no BRAF mutations and MMR protein de ciency in the 3 cases of MCRC specimens with negative SATB2.
The sensitivity, speci city and diagnostic accuracy of SATB2 were respectively 92.86%, 100% and 95.71%, which is much higher than common lung adenocarcinoma immunomarkers (CK7, TTF-1, Napsin A) and intestinal cancer immunomarkers (CK20, CDX2 and villin). We can view that SATB2 is the best immunomarker for identifying PEA and MCRC. The diagnostic value of CK7 is slightly inferior to SATB2, and the results of CK7 can be used as a reference for SATB2 differential diagnosis. The sensitivity of TTF-1 and Napsin A is 95.24% and 97.62%, respectively. They are also quite excellent and can be used as a reference for SATB2 differential diagnosis.