Prognostic implication of SOX2 expression in small intestinal adenocarcinoma


 Background: The presence of KRAS mutation enhance the stem cell features of colorectal carcinoma cells containing mutant adenomatous polyposis coli (APC). However, their potential role in small intestinal adenocarcinoma remains elusive. Here, we aimed to investigate the clinical significance of cancer stem cell markers in the context of small intestinal adenocarcinoma with the KRAS genotype. Methods: SOX2, NANOG, and OCT4 expression were assessed by immunohistochemistry and digital image analysis, and their potential association with KRAS was further examined in 185 small intestinal adenocarcinoma patients. Results: Positive expression of SOX2, NANOG, and OCT4 was detected in 65 (35.1%), 94 (50.8%), and 82 (44.3%) of patients, respectively. SOX2-/wild-type KRAS (KRASWT) was often observed in low-grade carcinoma (P = 0.048). SOX2+ and SOX2+/mutant KRAS (KRASMT) were significantly associated with shorter overall survival relative to SOX2- and others (P < 0.001, both). Multivariate analysis revealed SOX2+ (HR=1.929 [95% CI, 1.320-2.819], P = 0.001) as an independent prognostic factor of worse overall survival among small intestinal adenocarcinoma patients. Conclusions: These results suggest that SOX2 expression, in conjunction with KRAS, is a potential prognostic marker for small intestinal adenocarcinoma.


Background
Small intestinal adenocarcinoma is a relatively rare malignancy, accounting for 1-3% of all gastrointestinal tumors [1,2]. Small intestinal adenocarcinomas are most commonly found in the duodenum, followed by the jejunum and ileum [3]. Moreover, duodenal adenocarcinomas are often observed in patients over the age of 60 years [3]. The early diagnosis of small intestinal adenocarcinoma is uncommon because of its rarity and non-speci c symptoms. Clinical symptoms may differ depending on tumor location, size, or polypoid growth pattern [4]. Unfortunately, approximately 90% of patients are only diagnosed once the cancer has reached stage III and IV [5,6]. The prognosis for patients with disseminated disease remains signi cantly poor, with a median survival of 2 to 14.4 months after diagnosis, and less than 5% of patients surviving for 5 years [3,4,6]. The mortality rate from small intestinal adenocarcinoma has not improved over recent decades. There is no standard adjuvant chemotherapy due to the lack of conclusive data. Thus, there is a strong need for a more effective and systematic individualized treatment option for advanced-stage, metastatic small intestinal adenocarcinoma.
Cancer stem cells (CSCs) are a small subpopulation of tumor cells that have the capacity for self-renewal, differentiation, and tumorigenicity [7,8]. Furthermore, CSCs are proposed to be responsible for metastasis, relapse of cancer cells, and drug resistance [8,9]. Thus, targeting CSCs is a promising therapeutic strategy because it allows complete eradication of CSCs and prevents recurrence.
Identi cation and eradication of CSCs are not easy because they are highly plastic and hidden within the tumor cell population, usually in specialized hypoxic microenvironments [7,10]. CSC identi cation generally relies on CSC markers, such as sex-determining region Y-box 2 (SOX2), NANOG, octamerbinding transcription factor 4 (OCT4), and DNA methyltransferase 1 (DNMT1). The SOX2, NANOG, and OCT4 genes play important roles in regulating pluripotency [11] and have been spotlighted in association with the treatment responses to, and prognoses of, various cancers [12]. High SOX2 expression has also been correlated with colorectal cancer metastasis and lymph node in ltration [13]. Interestingly, NANOG expression is regulated by the OCT4/SOX2 complex, and its high expression is positively correlated with tumor progression and poor prognosis among patients with colorectal cancer [14]. OCT4 expression has been reported in colorectal cancer cells undergoing epithelial-mesenchymal transition (EMT) [15], and its expression is associated with poor clinical outcomes [16]. Although there is increasing evidence of the importance of CSCs in cancer progression, the clinical signi cance of NANOG, OCT4, and SOX2 expression in small intestinal adenocarcinoma remains unknown.
KRAS encodes a protein that is linked with the extracellular-regulated kinase (ERK) and phosphatidylinositol 3-kinase (PI3K)-AKT signaling pathways. The activation of these pathways leads to cellular growth and proliferation. Indeed, KRAS mutations are one of the most prevalent oncogenic mutations found in human cancers, with approximately 40% of colorectal cancer patients exhibiting the mutated genotype [17]. In genetic models of colorectal cancer, KRAS mutations with mutated adenomatous polyposis coli (APC)-induced colorectal cancers led to tumor growth and liver metastasis [18,19]. Notably, Moon et al. also demonstrated that KRAS mutations induce the stemness of colorectal cells harboring APC mutations [20]. In addition, Jun et al. reported that KRAS mutations were found in one-thirds of patients with small intestinal adenocarcinoma and that these mutations were associated with a poor prognosis in the early stages [21]. However, a recent study demonstrated that small intestinal adenocarcinoma presents lower APC mutation frequency (26.8%) in comparison to that of colorectal cancer (75.9%) [22]. In this context, it is unclear whether KRAS mutation, in conjunction with low APC mutation rates play a role in the regulation of CSCs associated with small intestinal adenocarcinoma tumorigenesis We investigated the clinical values of SOX2, NANOG, and OCT4 and expression in surgically resected small intestinal adenocarcinoma specimens by immunohistochemistry (IHC) and quantitative image analysis. Furthermore, we examined the potential association between the expression of CSC markers and KRAS genotypes in patients with small intestinal adenocarcinoma.

Patients specimens
A total of 197 surgically resected primary small intestinal adenocarcinomas were collected from the surgical pathology archives of 22 South Korean institutions by the Korean Small Intestinal Cancer Study Group, as previously reported [4]. Only carcinomas originating from the mucosa of the small intestine, including the duodenum, jejunum, and ileum, were included in the present study. This retrospective study was approved by the Institutional Review Board of Incheon St. Mary's Hospital (OC14OIMI0133). All procedures were conducted in accordance with the Declaration of Helsinki.
Clinical and pathologic data that were collected as part of our previous study were used again in this study [4]. Clinical data included patient sex, age, tumor location, operation date, T-and N-categories and stage grouping according to the eighth edition of the American Joint Committee on Cancer (AJCC) cancer staging system, most recent follow-up examination, survival status, presence or absence of synchronous or metachronous malignancies, and presence or absence of conditions predisposing patients to small intestinal adenocarcinomas (including Crohn's disease, familial adenomatous polyposis, Lynch syndrome, Peutz-Jeghers syndrome, Gardner's syndrome, gluten-sensitive enteropathy, intestinal duplication, Meckel's diverticulum, or heterotopic pancreas). Pathological data obtained from the gross examination included tumor size and growth pattern. The macroscopic growth patterns of the small intestinal adenocarcinomas were divided into 3 groups, including polypoid pattern, exophytic with predominantly intraluminal growth; nodular pattern, endophytic/ulcerative with intramural growth; and in ltrative pattern, annular with circumferential involvement or diffusely in ltrative [23]. The microscopic characteristics included histologic subtype, tumor grade, depth of invasion, peritoneal seeding, pancreatic and other intestinal loop invasions, nodal metastasis, and perineural and lymphovascular invasion.
Histologic types and differentiations were classi ed according to the 2019 World Health Organization (WHO) classi cation [23]. Tumors were graded as low-grade (well and moderately differentiated, > 50% gland formation) and high-grade (poorly differentiate and undifferentiated, < 50% gland formation) carcinomas following the criteria for the histological grading of colorectal adenocarcinoma, which was described in the 2019 WHO classi cation [23].
Tissue microarray (TMA) and IHC TMAs were constructed from the archived formalin-xed and para n-embedded tissue blocks, as previously described [24]. Brie y, representative areas with invasive adenocarcinomas and normal small intestinal mucosa were identi ed on the corresponding hematoxylin and eosin-stained slides. Three cores from each tumor and 1 matched core from normal mucosa were sampled using a 1.0-mm punch, and 7 TMA blocks were constructed using a manual tissue arrayer (Beecher Instruments, Sun Prairie, WI).
TMA slides sectioned at 5-µm thickness were depara nized in xylene and rehydrated with a series of graded ethanols. Heat-mediated antigen retrieval was performed using a pressure chamber (Pascal; Dako, Carpinteria, CA, USA) with pH 6.0 citrate buffer (Dako) for SOX2 and OCT4, and pH 9.0 citrate/EDTA (Dako) for NANOG. To block endogenous activity, slides were incubated with 3% H 2 O 2 for 10 minutes with additional protein serum block (Dako) applied to NANOG to reduce background staining for 20 minutes at room temperature. Primary antibodies were incubated with the following conditions: rabbit polyclonal anti-OCT4 (Abcam, Cambridge, MA; cat# ab19857) at 1:1200 for 1 hour; rabbit monoclonal anti-SOX2 antibody (Cell Signaling Technology, Danvers, MA; clone D6D9; cat# 3579) at 1:500 for 1 hour, and rabbit monoclonal anti-NANOG (Cell Signaling Technology; clone D73G4; cat#4903) at 1:200 for 1 hour at room temperature, respectively. Antigen-antibody reactions were detected with Envision + Ms/Rb HRP dual link secondary (Dako) and visualized with 3,3'-diaminobenzidine (Dako) followed by hematoxylin counterstain, dehydration, and clearing to coverslip prior to examination by light microscopy. The negative control was performed by omitting the primary antibody and rabbit immunoglobulin, and human testis tissue with seminoma was used as a positive control in each staining run.

Quantitative evaluation of immunostaining
All stained slides were digitalized using an Aperio AT2 digital scanner (Leica Biosystems, Vista, CA) in × 40 objective magni cation. Subsequently, the images were automatically analyzed using Visiopharm software v6.9.1 (Visiopharm, Hørsholm, Denmark). In brief, screenshots of single relevant areas of regions of interest were generated by a pathologist (JWK) who was blind to clinicopathological information. Blue-colored (hematoxylin) tumor nuclei were initially de ned, and then brown-colored (DAB) nuclei and cytoplasm were separated spectrally. For SOX2 IHC, a brown nuclear staining intensity (0 = negative, 1 = weak, 2 = moderate, and 3 = strong) and the percentage were obtained using a prede ned algorithm and optimized settings. Histoscores, calculated by the percentage of positive cells multiplied by their staining intensity, were assigned to evaluate SOX2 IHC results. For NANOG and OCT4 IHC, a brown cytoplasmic intensity (weak and strong) with or without nuclear staining was obtained, and each proportion was analyzed. Expression values for histoscores (SOX2) and cytoplasmic staining (NANOG and OCT4) were dichotomized (negative vs. positive), with cut-off values showing the most discriminative power. Cut-off values of NANOG and OCT4 were 5.7% and 40.0% with strong cytoplasmic staining with and without nuclear expression, respectively, and the SOX2 cut-off histoscore was 2.5.

KRAS mutation
KRAS mutations were previously evaluated in the same cohort. KRAS mutations in codons 12 and 13 of KRAS exon 1 were identi ed by cycle sequencing, as previously described. In brief, 10 sections (each 10 µm in thick) from formalin-xed para n-embedded tissue blocks were used to extract genomic DNA with a QIAmp DNA Mini Kit (Qiagen, Valencia, CA) following the manufacturer's instructions. The KRAS genes were polymerase chain reaction (PCR)-ampli ed with primers for KRAS (F: 5′-TGACATGTTCTAATATAGTCAC-3′, R: 5′-ACAAGATTTACCTCTATTGTT-3′). PCR reactions were run in a total volume of 25 µl with 0.3 µM of each primer using AmpliTaq Gold PCR Master Mix (Applied Biosystems, Foster City, CA, USA). Samples were subjected to initial denaturation at 95 °C for 15 minutes, 40-45 cycles at 95 °C for 50 seconds, annealing for 50 seconds, and elongation at 72 °C for 1 minute, followed by nal elongation at 72 °C for 7 minutes. PCR products were column-puri ed using a QIAquick PCR Puri cation Kit (Qiagen) or enzymatically treated with ExoSAP-IT (USB, Cleveland, OH). The sequencing primers were identical to the PCR primers, and all samples were sequenced in both directions using a BigDye Terminator Cycle Sequencing Kit, version 1.1 (Applied Biosystems). The sequencing reactions were analyzed on an ABI Prism 3100 Genetic Analyzer with Sequencing Analysis software, version 3.7 (Applied Biosystems).

Statistical analysis
Data analysis was performed using the SPSS Statistics for Windows, version 21 (IBM Corp., Armonk, NY, USA). Unpaired Student's t-test was applied to compare continuous variables. The χ 2 or Fisher's exact tests were used to characterize relationships between categorical variables. All survival analyses used an overall survival (OS) model, which captured all patient deaths as events and censored other patients at their last visit dates. Survival curves were constructed using the Kaplan-Meier method, and differences between groups were assessed using the log-rank test. Multivariate analysis was performed using Cox proportional hazards modeling to investigate the signi cance of any prognostic factors. P values less than 0.05 were considered statistically signi cant.

Patient characteristics
Out of 197 total patients, 185 (93.9%) with interpretable immunohistochemical and molecular results were included in our study cohort. Eleven cases were not successfully immunostained due to tissue loss, folding during sectioning, or the staining process.
There was also no signi cant difference in OCT4 expression between small intestinal adenocarcinoma (mean ± SD; 6.0 ± 14.7%) and normal intestinal tissues (mean ± SD; 12.4 ± 18.2%) (P = 0.134). As summarized in Supplementary Table 2, positive expression of OCT4 was signi cantly associated with low pT category (P = 0.022) and stage grouping (P = 0.020). In terms of SOX2 and NANOG, there were no meaningful differences between negative and positive expression with respect to the investigated clinicopathological variables. Furthermore, there was no association between CSC marker expression and KRAS genotype.

Discussion
As the CSC model has evolved, many biomarkers have been identi ed in various tumors to better understand CSCs and elucidate their roles in conferring stemness. SOX2, NANOG, and OCT4, contributing to the "core pluripotency network," are transcription factors that regulate the development of embryonic stem cells (ESCs). These transcription factors are thought to regulate pluripotency and lead to selfrenewal in embryonic and induced pluripotent stem cells [25,26]. Moreover, it has been reported that SOX2 and OCT4 play roles in inducing the stemness of various cancer cells, as well as embryonic cells [26][27][28]. However, due to its rarity and a lack of understanding of its pathogenesis, only a few studies have been conducted on the clinical value of stem cell markers in small intestinal adenocarcinoma. Here, we investigated the clinical signi cance of SOX2, NANOG, and OCT4 expression in small intestinal adenocarcinomas.
We rst determined SOX2 (35.1%), NANOG (50.8%), and OCT4 (44.3%) expression in small intestinal adenocarcinoma samples and demonstrated that SOX2 expression is an independent negative prognostic factor. Due to the lack of a standardized method for assessing SOX2 expression [13,25], and the paucity of SOX2 nuclear expression, we used a histoscore method and digital image analysis to objectively quantify the results. In colorectal cancer samples, SOX2 expression has been analyzed with an absolute quantitative or semiquantitative scoring system via microscopy. It has been reported that SOX2 expression ranges from 11-45.6% and that it correlates with lymph node metastasis, tumor grade, TNM categories, and BRAF mutations [13,25].
In this study, we observed that OCT4 is mostly expressed in the cytoplasm and is associated with low pT category and stage grouping. When considering different isoforms of OCT4, OCT4A is a nuclear transcription factor responsible for the pluripotency properties of ESCs, while OCT4B resides in the cytoplasm, where it may respond to cellular stress [29]. Therefore, it has been suggested that OCT4B expression is predominantly found in small intestinal adenocarcinoma cells, which correlates with low tumor stage. Additionally, alternatively spliced OCT4 transcripts may exhibit diverse functions in different tissues, considering that OCT4 expression (mean, 12.4%) is also found in normal small intestinal mucosa.
There is ongoing debate about the contribution of the Wnt signaling pathway to self-renewal and differentiation in human ESCs; however, many members of the Wnt signaling pathway are implicated in stem-cell proliferation [26,30] Table 2).
We found small intestinal adenocarcinomas with KRAS MT and NANOG + expression to be associated with a lack of lymphovascular tumor invasion and small intestinal adenocarcinomas with KRAS MT and OCT4 + expression to be associated with eary stage group. Regarding the combined expression of SOX2, NANOG, and OCT4, we observed that NANOG + /OCT4 + expression was more common in early-stage carcinoma, whereas SOX2 + /OCT4 + and SOX2 + /NANOG + /OCT4 + expression were associated with short OS. It is well known that the transcription factor NANOG is localized to the nucleus [32]. We detected that NANOG is expressed in the cytoplasm of small intestinal cancer cells with and without nuclear accumulation, which corroborates the ndings of other colorectal cancer studies [14,33]. This aberrant expression pattern is frequently found in a variety of cancers, with testicular germ cell tumors being a notable exception [34]. The regulation mechanism for localization is currently unknown and should be elucidated in the future. Recent studies also indicate that NANOG is a negative prognostic factor among colorectal cancer patients. Meng et al. revealed that NANOG expression signi cantly correlates with poor prognosis, lymph node metastasis, and TNM categories [14], and Xu et al. reported that NANOG may be a potential biomarker for the postoperative hepatic metastasis of colorectal cancer [33]. These results suggest that SOX2, NANOG, and OCT4 play complex roles in small intestinal adenocarcinoma. Further studies are needed to clarify the interaction between SOX2, NANOG, OCT4, and KRAS mutations in small intestinal adenocarcinoma.
The main challenge of this study was that it relied on a relatively imbalanced cohort, despite collecting patient samples from multiple institutions. In this study cohort, 57% of the patients had pT4 tumors, and 52% had cancers with AJCC stage group III, even though inoperable stage IV cases with distant metastases were not included. However, this deviation seems to be characteristic of small intestinal adenocarcinoma. Indeed, the ndings of a previous large single-center study agreed with our ndings when comparing just the percentages of stage I, II, and III cases, which were 12%, 45%, and 43%, respectively. Additionally, other epidemiologic characteristics, such as age, sex, and location, also paralleled those of our cohort [3].

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download. SOX2inSIACSupplFig.S1BMC2020.ppt