Prognostic Value of LC3B and p62 Expression in Small Intestinal Adenocarcinoma

Autophagy, a mechanism that maintains cellular homeostasis, is involved in tumor cell growth and survival in cancer, and autophagy inhibitors have been tested clinical trials for anticancer therapy. To elucidate the clinical and prognostic implications of autophagy in small intestinal adenocarcinoma (SIAC), we assessed the expression of autophagy markers, LC3B and p62, in 171 surgically resected primary SIACs using automated quantitative analysis. Positive LC3B, p62 nuclear (p62Nu), and p62 cytoplasmic (p62Cy) expression was observed in 23 (13.5%), 52 (30.4%), and 43 (25.1%) carcinomas, respectively. LC3B+ expression was correlated with undifferentiated carcinoma (p < 0.001) and high histologic grade (p = 0.029). The combined expression of LC3B and p62Nu (LC3+/p62Nu+) was related to the older age of patients (p = 0.017), undifferentiated carcinoma (p < 0.001), and high grade (p = 0.031). LC3B+ (p = 0.006), p62Cy+ (p = 0.041), or p62Nu+ (p = 0.006) expression were associated with worse survival. In addition, SIAC patients with either LC3B+/p62Nu+ (p = 0.001) or LC3B+/p62Cy+ (p = 0.002) expression had shorter survival times. In multivariate analysis, LC3B expression remained an independent prognostic factor (p = 0.025) for overall survival. In conclusion, autophagy may play a role in the tumorigenesis of SIACs, and LC3B and p62 could be used as prognostic biomarkers and potential therapeutic targets for SIACs.


Introduction
Small bowel cancer is rare in Korea, with 976 new cases occurring in 2018 [1]. However, its incidence is increasing; it reached 1.9 per 100,000 in 2018 compared to 0.9 per 100,000 in 2003 [2]. The proportion of small intestinal cancers among gastrointestinal (GI) malignancies has also increased from 0.69% in 2003 to 1.06% in 2018 [1,2]. Small intestinal adenocarcinoma (SIAC) is the most common histologic type of cancer originating in the small bowel, which accounts for an estimated 30% to 40% of small intestinal cancer diagnoses [3]. Although the mucosal surface of the small bowel covers >90% of the digestive canal, the overall prevalence of SIAC among carcinomas of the tubular GI tract is <2% [1]. This prompts the hypothesis of a protective environment associated with enzymes specific to the small bowel; however, it may be related to the shorter transit time of dietary carcinogens [4]. The small intestine is long and heterogeneous, leading to the development

Study Population
We used human SIAC tissue microarrays (TMAs), as previously reported [27,28]. The TMAs were from 197 surgically resected primary SIAC cases collected from the surgical pathology archives of 22 Korean institutions by the Korean Small Intestinal Cancer Study Group [27,28]. The tumor was considered a primary SIAC when the tumor was solitary or predominantly involved the mucosa of duodenum, jejunum and ileum regardless of extension into the serosa, without considering the presence of peritumoral dysplasia. Carcinomas extending from the surrounding gastrointestinal tract organs, such as the stomach, ampulla of Vater, pancreas, cecum or appendix, into the small bowel were excluded. Neuroendocrine tumors and mesenchymal tumors arising in small intestine was also not included. Clinical and pathologic data collected and analyzed in previous studies were used [27,28]. Cancer stages were determined according to the American Joint Committee for Cancer (AJCC), eighth edition [29]. Histological subtypes and grades were classified according to the 2019 World Health Organization (WHO) classification [30]. Information on predisposing conditions, such as Crohn's disease, familial adenomatous polyposis (FAP), Lynch syndrome, Peutz-Jeghers syndrome, Gardner syndrome, glutensensitive enteropathy, intestinal duplication, Meckel's diverticulum, or heterotopic pancreas was obtained through a review of medical records. This retrospective study was approved by the Institutional Review Board of Incheon St. Mary's Hospital (OC13SISI0162), and informed consent was waived because the study used leftover specimens. All procedures were conducted in accordance with the principles of the Declaration of Helsinki.

IHC
TMAs were constructed as described previously [31]. They included three invasive adenocarcinoma tissue cores of one-millimeter diameter per patient and one corresponding normal small intestinal mucosa tissue core.
Tissue specimens were cut and used on 5 µm thick paraffin sections. After deparaffinization and rehydration, heat-induced antigen retrieval was performed for 20 min by incubating the samples in antigen retrieval buffer pH 9.0 (DAKO, Carpinteria, CA, USA) for LC3B and pH 6.0 (DAKO) for p62 using a steam pressure cooker (Pascal; DAKO). Endogenous peroxidase activity was quenched with 3% hydrogen peroxide (H 2 O 2 ) for 10 min. The sections were incubated with rabbit polyclonal anti-LC3B (Cat no., ab48394; 1:2000; Abcam, Cambridge, MA, USA) and mouse monoclonal anti-p62 (clone D5L7G; 1:1500; Cell Signaling Technology, Danvers, MA, USA) for 1 h at room temperature. Antigenantibody reactions were detected with EnVision + Dual-HRP (DAKO) and visualized with 3,3-diaminobenzadine (DAB; DAKO). Peripheral nerve tissue with ganglion cells served as an internal positive control for LC3B [16]. Incubation with immunoglobulin G (IgG) or without the primary antibody was performed to generate the negative controls. Finally, the stained sections were lightly counterstained with hematoxylin.

Evaluation of IHC
Stained TMA slides were scanned using a NanoZomer XR Digital Pathology (NDP) system (Hamamatsu, Hamamatsu, Japan) at × 40 objective magnification with a singlefocus layer. The tissue on the slides was automatically detected with focus points to obtain the optimal image. Digitalized images were automatically analyzed using Visiopharm software v6.9.1 (Visiopharm, Hørsholm, Denmark) as previously described [32]. In brief, a pathologist (JWK) blinded to patient outcome and other clinical findings generated screenshots of single representative areas of the regions of interest. Blue-colored (hematoxylin) tumor cell nuclei were initially defined, and then brown-colored (DAB) nuclei and cytoplasm were separated spectrally. The brown cytoplasmic intensity (weak and strong) of LC3B and p62 was obtained, and each proportion was analyzed using a predefined algorithm and optimized settings. For p62, nuclear staining was also evaluated. The brown nuclear staining intensity (0 = negative, 1 = weak, 2 = moderate, and 3 = strong) and the percentages of stained cells were obtained, and histoscores were calculated by multiplying the percentage of positive cells with their staining intensity. The average of the three tumor cores was calculated as the final value. Expression values for cytoplasmic staining and histoscores were dichotomized (negative vs. positive), with cutoff values showing the most discriminative power. The cutoff value of LC3B and p62 cytoplasmic expression was set at 1.8% with a strong punctate pattern and 68.6% with weak intensity, respectively. The cutoff histoscore for p62 nuclear expression was 112.6.

Microsatellite Instability (MSI) Analysis
MSI data were obtained from a previous study in the same cohort [28]. Briefly, the five microsatellite loci (BAT25, BAT26, NR21, NR24, and NR27) were amplified in a single multiplex polymerase chain (PCR) reaction. PCR products were analyzed by capillary electrophoresis using an ABI 310 Genetic Analyzer (Applied Biosystems, Foster City, CA, USA). High-frequency MSI (MSI-H) was defined as the presence of two or more loci showing instability, whereas instability at only one locus was defined as low-frequency MSI (MSI-L). Tumors with no instability were defined as microsatellite stable (MSS).

Statistical Analysis
Categorical data were assessed using χ 2 or Fisher's exact tests and Mann-Whitney or unpaired Student's t test, and correlation analyses were applied to compare continuous 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. OS curves with log-rank tests were generated using the Kaplan-Meier method. Univariate and multivariate survival analyses were performed using the Cox proportional hazard regression model. The statistical significance was set at p < 0.05. The SPSS Statistics for Windows, version 21 (IBM Corp., Armonk, NY, USA) was used for the analyses.

Study Population and Tumor Characteristics
Out of the enrolled 197 primary SIACs, 171 (86.8%) with interpretable immunohistochemical and molecular results were analyzed in this study ( Table 1). The median age of the patients was 59 years (range, 23-86 years), and they were predominantly men (male:female = 1

Association between Clinicopathologic Features and p62 and LC3B Expression
The relationships between the clinicopathological features of SIACs and LC3B and p62 expression are summarized in Table 3. LC3B+ expression was significantly associated with undifferentiated carcinoma (p < 0.001) and high histologic grade (p = 0.029). Positive p62 Cy expression tended to be associated with undifferentiated carcinoma (p = 0.056) and observed in patients aged ≥60 years (p = 0.056). Tumors with predisposing conditions showed frequent p62 Nu + (p = 0.044) and p62 Cy + (p = 0.023) expression. No association was observed between LC3B and p62 expression and other clinicopathological variables including gender, growth type, tumor location and stage, Lynch syndrome (not shown), and MSI status.  The relationship between the combined expression of LC3B and p62 and clinicopathological variables was also examined (Table 4). LC3B+/p62 Nu + expression was correlated with older age (p = 0.017), undifferentiated histologic type (p = 0.031), and higher histologic grade (p < 0.001). LC3B+/p62 Cy + expression was associated with an undifferentiated histology (p < 0.001). SIACs with age ≥60 years and high grade showed frequent LC3B+/p62 Cy + expression; however, the difference did not reach statistical significance (p = 0.056 and 0.057, respectively).

Discussion
The present study highlights the prognostic impact of autophagy-related markers, LC3B and p62, in SIACs. First, we described the immunohistochemical expression patterns of LC3B and p62 proteins in our cohort of SIACs. There are various cellular assays for assessing autophagy, such as transmission electron microscopy, western blotting, flow cytometry, and fluorescence microscopy [18]. In formalin-fixed paraffin-embedded human tissue, IHC is a valuable modality that may provide information about the static level of autophagy-related proteins and prove to be useful in identifying patients for targeted therapy for modulating autophagy in clinical applications [16,18]. Considering the roles of LC3B and p62 in the formation of autophagosomes, overexpression of LC3B and p62 may be observed mainly after the activation of autophagy, but they do not necessarily indicate high levels of active, ongoing autophagy [16,33]. The inhibition of autophagy may prevent the degradation of autophagosomes and cause their accumulation. Therefore, high levels of LC3B and p62 may reflect a defective autophagy pathway [16].
We observed high levels of LC3B and p62 selectively in cancer cells, as previously described for colon and gastric cancers [33,34]. With regard to staining pattern, LC3B exhibited nuclear staining, whereas p62 showed definite nuclear expression in addition to cytoplasmic staining with punctate or diffuse patterns [26,35]. The significance of p62 Cy and p62 Nu expression for the assessment of autophagy remains unclear [16]. However, both staining patterns have been interpreted as surrogates of autophagy [36][37][38][39]. We found a strong association between p62 Cy + and p62 Nu + expression (p < 0.001, Table 2). In addition, the level of p62 Cy expression positively correlated with the level of p62 Nu expression (Pearson correlation coefficient, 0.922; p < 0.001). Thus, the significance of the p62 Nu staining pattern may be comparable to that of p62 Cy staining. Nucleocytoplasmic shuttling of p62 has been reported in an in vivo study [40]. In addition, LC3B+ expression was related to p62 Cy + expression (p = 0.003, Table 2) and the level of LC3B expression was correlated with each the levels of p62 Cy or the levels of p62 Nu expression (correlation coefficients, 0.240 and 0.304; p = 0.002 and p < 0.001, respectively). Hence, we conclude that p62 protein, in combination with LC3B, might be used as an ancillary marker of autophagy regardless of the staining pattern. The close relationship between LC3B, p62 Cy , and p62 Nu expression has been described in gastric cancer [33].
Our data demonstrated that autophagy marker expression was associated with the aggressive behavior of SIACs. LC3B+, p62 Nu +, or p62 Cy + expression was associated with shorter survival of patients with SIAC. Moreover, LC3B expression was found to be an independent prognostic factor in the multivariate analysis. Similar to our results, in previous studies on gastric [33] and colorectal [39] cancers, LC3B expression was related to worse prognosis. However, it has been reported that LC3B expression is inversely correlated with poor prognosis in esophageal [37] and colon [34] cancers. The discrepancies can be attributed to various factors, such as organ specificity, characteristics of the tumor itself, genetic factors, antibody clone used in the study, IHC conditions, and cutoff for expression. In our study, strong cytoplasmic staining with a punctate pattern was regarded as positive LC3B staining, and many researchers suggested that only punctate staining pattern correlated with autophagy induction and poor prognosis as opposed to diffuse cytoplasmic staining [26,34].
The prognostic impact of p62 expression in GI adenocarcinomas is also controversial [33,34,39,41]. Yoon et al. [41] and Masuda et al. [33] reported that high p62 Nu and p62 Cy expression were associated with worse OS in gastric adenocarcinoma. On the contrary, Schmitz et al. [39] and Niklaus et al. [34] showed that high p62 Cy expression was significantly correlated with favorable OS in colorectal cancer, while p62 Nu expression level was not associated with OS. Regarding the combination of LC3B and p62 staining, we found that SIAC patients with either LC3B+/p62 Nu + or LC3B+/p62 Cy + expression had shorter survival times than those with other combinations of IHC phenotypes. However, Niklaus et al. showed that tumors with high LC3B/high p62 Cy expression had the best OS, whereas tumors with high LC3B/low p62 Cy expression showed the worst outcome [34]. This inconsistency might be due to the wide range of functions of p62, including those in the autophagy pathway, the regulation of cell death, and the activation of transcription factor NF-kB [18]. Furthermore, the levels of p62 can be transcriptionally regulated by non-autophagic stimuli, such as the mitogen-activated protein kinase (MAPK) signaling pathway [18] and some p62 positive structures might not reflect autophagosomes. We observed more frequent p62 expression (25.1% for p62 Cy + and 30.4% for p62 Nu +) than LC3B expression (13.5%) in SIACs.
Autophagy was previously thought to play a protective role against malignant transformation. Recently, the role of autophagy in cancer progression and resistance to therapy has gained increased attention. Sakanshi et al. showed that LC3B was significantly associated with high pT category and lymphatic and perineural invasion in colorectal cancers [22]. Masuda et al. reported that age ≥60 years, intestinal type, and lymphatic and vascular invasion were positively related to the expression of autophagy markers in gastric cancers [33]. In addition, it has been reported that autophagy promotes the progression of cancer of the upper GI tract at an early clinical stage [33,42]. We found that either LC3B+ or LC3B+/p62 Nu + expression was correlated with the undifferentiated type and high histologic grade. LC3B+/p62 Nu + expression was also more frequently observed in SIAC patients aged ≥60 years. However, subgroup analysis by stage did not reveal an association between stage and the expression of autophagy markers in our study cohort. Sena et al. showed that the expression of LC3B in MSS colorectal cancer cells was higher than that in MSI cancer cells [43]. However, we did not identify any differences in the expression of autophagy-related proteins between MSI-and MSS-SIAC patients. Lynch syndrome was also unrelated to the expression of autophagy markers in SIACs, although p62 Cy + and p62 Nu + expression was frequent in cases with predisposing conditions. A limitation of this study is that patients with stage IV disease were not included because only surgically resected SIAC specimens were collected. In addition, Crohn's disease is a well-known predisposing factor for SIACs in the Western population, but rarely in Korean patients [32]. Indeed, only one (0.6%) Crohn's disease-associated SIAC was observed in our study cohort. Recent advances in genetics have revealed that polymorphisms in autophagy-related 16-like 1 (ATG16L1) gene, which is essential for LC3 lipidation and autophagosome formation, are a genetic risk factor for Crohn's disease [44]. Hence, further studies in larger cohorts might provide insight into the roles of autophagy in the carcinogenesis of SIACs with heterogeneous clinical characteristics.
In conclusion, the high levels of expression of LC3B and p62 proteins selectively in tumor cells of SIACs suggests that the autophagic process is related to tumorigenesis. The correlation between LC3B, p62 Nu , and p62 Cy expression indicates that p62 protein is a surrogate marker of autophagy, irrespective of the staining pattern. Of note, we observed that LC3B and p62 expression, as well as the combined expression of LC3B and p62, have an impact on cancer progression and are related to patient survival. Moreover, LC3B was frequently expressed in tumors with an undifferentiated type and a higher histologic grade. Therefore, LC3B and p62 are potential prognostic biomarkers and promising candidate targets for the treatment of SIACs. Further investigation into the detailed mechanism of