Augmented CD133 expression in distal margin correlates with poor prognosis in colorectal cancer

Abstract Pathological assessment of excised tumour and surgical margins in colorectal cancer (CRC) play crucial role in prognosis after surgery. Molecular assessment of margins could be more sensitive and informative than conventional histopathological analysis. Considering this view, we evaluated the distal surgical margins for expression of cancer stem cell (CSC) markers. Cellular and molecular assessment of normal, tumour and distal margin tissues were performed by flow cytometry, real‐time q‐PCR and immuno‐histochemical analysis for CRC patients after tumour excision. CRC patients were evaluated for expression of CSC markers in their normal, tumour and distal tissues. Flow cytometry assay revealed CD133 and CD44 enriched cells in distal margin and tumour compared to normal colorectal tissues, which was further confirmed by immunohistochemistry. Most importantly, immunohistochemistry also revealed the enrichment of CSC markers expression in pathologically negative distal margins. Patients with distal margin enriched for CD133 expression showed an increased recurrence rate and decreased disease‐free survival. This study proposes that although distal margin seems to be tumour free in conventional histopathological analysis, it could harbour cells enriched for CSC markers. Further CD133 could be a promising molecule to be used in molecular pathology for disease prognosis after surgery in CRC patients.

It is known that around 20% to 40% of patients undergoing curative resection for CRC develop loco-regional recurrence or distant metastasis. 4 Although chemoradiotherapy has radically enhanced the treatment efficacy in CRC, still surgery remains the conclusive treatment regimen given to most patients. Removal of all residual cells in and around tumour has been suggestive of an effective surgery. In CRC, excision of surgical margins along with tumour is critical for curative surgery practice, as studies have shown that positive surgical margins correlate with disease recurrence and distant metastasis. 5,6 Presently, biopsies are evaluated visually by microscopic examination of hematoxylin and eosin stained slides by a pathologist. However, molecular alterations and residual tumour cells may go unnoticed in conventional pathology. Molecular evaluation of surgical margins has been described to be more sensitive and informative compared to conventional histopathological assessment in many cancers. 7 Currently, insights on molecular signatures of surgical margins are lacking, which otherwise could aid in defining tumour-free resectable margin in CRC. Criteria for precise selection of distal margin in a sphincter preserving surgery need more research as distal resection length still remains a topic of debate. 8 Molecular assessment includes many advanced and precise techniques, which enable to assess many risk parameters and could be of help in improved prognosis of patients.
Presently, there is a need for robust, reliable prognostic and predictive molecular markers in CRC, which could help oncologist to decide patient-specific treatment regimens as well as help in better patient's follow-up post surgery. Accumulating evidence on a sub-population of cells in tumours, demonstrated to have high tumourigenic proficiency compared to the bulk of the tumour cells and were designated as cancer stem cells (CSCs). CSCs are reported to be the possible culprits behind therapy resistance 9,10 and also responsible for disease recurrence. 11 CD133, CD44 and Ep-CAM have been shown to be specific for CSC cells in CRC and Ep-CAM high / CD44 + positivity has been reported to be associated with metastasis in CRC. 12,13 CD133 has been shown to be an independent prognostic marker for lower survival rate in CRC patients. 15 However, there are hardly any studies evaluating CSCs-associated molecular signatures in surgical margins and their role in CRC prognosis.
CRC distal margin remains as the least resolved entity in regard to its molecular signatures and prognostic values. Our study explored distal surgical margin conjointly with tumour using established molecular markers specific for CSCs in CRC patients undergoing curative surgical resection.

| Patients and samples
Biopsies were collected from CRC patients undergoing curative surgery between 2013 and 2017 at the Regional Cancer Centre, Trivandrum after human ethics committee approval and sanction from Institutional Review Board. All patients gave written informed consent in accordance with the Declaration of Helsinki. Patient's details and clinical information were collected from medical records of the same institution. In this study, only adenocarcinoma was included, excluding patients exhibiting gastrointestinal tumours.
Patient's demographic details are given in Table S1. Briefly, patients presented with rectal cancer had undergone a long course of radiotherapy of 50 Gy in 25 fractions followed by low anterior resection and abdominoperineal resection for upper and lower rectal cancer respectively. On the other hand, colon cancer patients went through hemicolectomy. Surgery involved removal of tumour, lymphnodes and associated margins followed by anastomosis for retaining the intestinal continuity.
The biopsies were collected from normal, tumour and distal resection margins from each patient. Biopsies were collected in HBSS (Invitrogen, USA) or in RNA stabilizing buffer (Ambion, Invitrogen, USA) immediately after surgery and transported under cold condition to the laboratory. and kept at 37°C in temperature-controlled shaker with mild agitation for 4 hours. After that, DMEM media containing cell suspension was passed through 40 µm cell strainer (BD Bioscience, USA) and pelleted at 1600 g for 5 minutes at 4°C. RBCs were lysed by incubating cell pellet in RBC lysis solution for 5 minutes at room temperature. Cell suspension was centrifuged at 1600 g for 5 minutes at 4°C and supernatant was discarded. Finally, cell pellet was suspended in DMEM media supplemented with 10% serum and kept at 37°C for 3 hours for recovery.

| Magnetic assorted cell depletion
Lymphocytes were depleted from isolated cells using CD45 Dynabeads (Invitrogen, USA), according to the manufacturer's protocol. Briefly, Dynabeads were added to cell suspension in PBS and incubated at 4°C for 30 minutes in rotary mixing. After incubation, cells with beads were placed on a magnet for 2 minutes and bound cells were discarded. Finally, supernatant was used for further analysis.

| Immunophenotyping
Cell suspension obtained after CD45 depletion was then washed three times with HBSS buffer and pelleted by centrifugation at 1600 g for 5 minutes. Washed cells were suspended in 1 mL HBSS and 1 × 10 6 cells were used for antibody staining according to the manufacturer's protocol. The details of antibodies used are provided in Table S3. After antibody staining, cells were again washed with ice-cold HBSS and resuspended in HBSS containing 2% FBS.
Cells were kept on ice till flow cytometry analysis was carried out.
Flow Cytometry analysis was carried out using BD FACS ARIA II (BD Bioscience, USA).

| RNA extraction
Total RNA was isolated from approximately 50 mg tissues (normal, tumour and distal margin) by using Trizol reagent (Invitrogen, USA) following the manufacturer's protocol. Isolated RNA quality and quantification were assessed using gel electrophoresis and Nanodrop 1000 (ThermoScientific, USA).

| mRNA sequencing and identification of mutations
Isolated RNA from normal, tumour and distal tissues collected from a CRC patient was quality checked using bioanalyzer. Whole mRNA sequencing was performed with Illumina Nextseq500 (Genotypic Technology, Bangalore, India). The NextSeq500 paired-end raw reads were quality checked using FastQC. NextSeq 500 raw reads were processed by in-house script for adapters and low-quality ncifcrf.gov/tools.jsp). Driver mutations were selected by comparing top 20 mutated genes in CRC from COSMIC datasets (https://cancer.sanger.ac.uk/cosmic) available online with our observed data. 16 Further, the driver mutant genes fate was analysed by Mutation Taster bioinformatics tool. 17

| Quantitative real-time-PCR
A total of 1 µg of isolated RNA was converted to cDNA from each RNA sample using PrimerScriptcDNA conversion kit (TAKARA, Japan), according to the manufacturer's protocol. Quantitative realtime PCR was performed with SYBR-Green (TAKARA, Japan) based on fluorescence detection method and HT9700 detection system (AB, Life Technologies, USA). Target genes specific primers were designed using NCBI-Primer blast tool for an amplicon size of maximum 150 bp. Analysed genes and the primers used are shown in Table   S2. PCR was performed in a final volume of 5 µL containing SYBR-Green PCR master mix, 50 ng of cDNA as template and forward and reverse primers. Relative changes in mRNA expression were obtained using ΔΔCt method keeping normal as reference sample and L19 as endogenous control using Data Assist software (AB Life Technologies, USA).

| Immunohistochemistry
The tissues received after surgery were fixed with 4% paraformaldehyde (Sigma-Aldrich, USA) for 16 hours at 4°C and processed with xylene and alcohol gradient followed by embedding with paraffin to make paraffin blocks. Five-micrometre thick sections were mounted on poly L-lysine-coated StarFrost glass slides (Leica, Germany). The  Table S3. After incubation, sections were washed with PBS and subsequently incubated with AB (avidin-biotin-peroxidase complex) (Vecta stain kit, USA) for 30 minutes. Diaminobenzidine (Sigma-Aldrich, USA) was used as a chromogen and haematoxylin (Merck Millipore, Germany) was used as a counter stain. Sections without incubation of primary antibody served as negative control. Semiquantitative analysis was carried out by counting three independent microscopic fields (n = 100-200 cells/field) for staining of specific antigen using upright microscope (Leica DM1000, Germany). Finally, the mean value was taken for each antigen to get the percentage of positive cells.

| Patient stratification
On the basis of expression levels, patients enriched for CSC markers were grouped in four categories: CD133 (Tumor), CD44 (Tumor), CD133 (Distal) and CD44 (Distal). The enrichment was decided by comparing the expression of both markers to the median value of corresponding normal tissue samples.

| Statistical analysis
All statistical analyses in this study were performed using GraphPad Prism 5 and SPSS version 25. For flow cytometry and immunohistochemistry data, we used un-paired t test for mean comparison and Pvalue determination between normal, tumour and distal tissues. For real-time PCR data, one sample t test was performed to determine the P-value comparing log 2 fold change among tumour and distal samples. Kaplan-Meier plots were used to determine the diseasefree survival (DFS) in followed up patients and P-value was obtained by Gehan-Breslow-Wilcoxon test. Binary logistic regression model was used for multivariate analysis and Hosmer and Lemeshow statistical test was used to derive the P-value and ODDS ratio.

| Enrichment of CD133 and CD44 positive cells in colorectal tumour and distal margin
CD133 and CD44 antigens happen to be the most widely studied CSCs markers in CRC and combination of these markers could be helpful in identifying CSCs. 18 In this study, we investigated CRC  Figure 3F).
In addition to that, we also looked into the expression of β-catenin and Oct4, which are core molecules in Wnt-mediated stem cell signalling. 19 Oct4 showed significantly increased nuclear expression in tumour and distal tissues compared to normal tissue with a mean percentage of cells (34.91 ± 9.74, P = 0.0445) and (26.48 ± 5.42, P = 0.0329) vs (8.749 ± 3.43) respectively ( Figure 3G). However, β-catenin expression was seen in both cytoplasm and nucleus of epithelial cells and a higher expression was observed in tumour and distal margin compared to normal tissues with % positive cells (46.41 ± 11.27, P = 0.0256) and (37.03 ± 7.35, P = 0.0322) vs (11.08 ± 7.41) respectively ( Figure 3G). Overexpression of Oct4 and β-catenin proteins in tumour and distal margin further strengthened our observation on the presence of cells with stem cell signatures.
Put together, these results suggest the involvement of altered Wnt signalling behind the enrichment of CD133 and CD44 in tumour as well as distal tissues.
We further analysed for mRNA levels of CD133, CD44, OCT4, SOX2 and BMI1, in tumour and distal margin tissues compared to the matched normal tissues expression using real-time PCR ( Figure   S2). mRNA levels did not show any significant expression for the above-mentioned genes unlike protein expression. However, few cases showed enrichment of CD133 and CD44 expression in distal margin tissues as well and CD44 showed a significantly enriched expression in tumours (Log 2 FC of 1.09, P = 0.0044) ( Figure S2B).

| Enriched CD133 expression in distal margin associated with a higher recurrence rate and lower DFS in CRC patients
To investigate the clinical significance of the enrichment of CD133 and CD44 markers in CRC and distal margin tissues, all the patients quantified for CD133 and CD44 expression were followed-up to a period of 24 months from the day of tumour excision. From follow-up data, we observed that among all the enriched groups, CD133 (Distal) patients showed a significantly higher recurrence rate of 50% (Table 1). To delineate the DFS in the above-mentioned groups of patients, we used Kaplan-Meier plot. CD133 (Distal) and CD44 (Distal) showed an early recurrence with first event occurring at 9 months compared to 15 months in negative groups without any effect on overall DFS ( Figure 4C,D). However, no significant difference in DFS was found among CD133 (Tumour), CD44 (Tumour) and CD44 (Distal) groups compared to control group ( Figure 4A,B,D). Interestingly, patients enriched for CD133 in distal margin showed significantly lower DFS of 50% compared to CD133 control group with 82% (P = 0.02, HR = 0.18)( Figure 4C). This is significantly lower than other enriched groups with a DFS of more than 70%.
Further, multivariate analysis results also supported the relation between high expression of CD133 in distal margin with disease recurrence with a ODDS ratio of 15.30 and P = 0.03 ( Table 2).
However, we could not find any significant association of recurrence with status of tumour stage, lymphnode and distal margin resection length (Table 2). These results clearly suggest that CD133 expression in distal margin could have crucial role in disease recurrence in CRC.

| D ISCUSS I ON
In this study, we investigated the expression of putative CSC markers and Wnt signalling molecules associated with regulation of CSC in CRC and distal margins in patients enduring curative surgery. Surgical margins have been assessed for identification of residual cells, probably responsible for disease relapse if left untreated. Circumferential and distal margins represent important prognostic factors in CRC and are routinely checked for presence of residual cells after surgery using conventional histology. 5,20,21 Previous reports suggest a relation between local recurrence and survival with resection margin status. 22,23 In some cases of patients treated with pre-chemoradiotherapy, tumours and margins may not depict any residual cells (also known as minimal residual cancer) during pathological examination; however, these undetectable tumour cells or clusters could possibly be identified using molecular markers. A recent study in breast cancer has shown that the residual cells that persisted after chemoradiotherapy were found to be enriched in CSCs. 24 As CSCs are thought to be the most nasty cells in a tumour, there is a possibility of their escape upon chemoradiotherapy and manifesting minimal residual disease. 25 Recent reports have shown that CD133 marker could be a crucial prognostic factor as higher expression of CD133 was found to be associated with poor clinical outcomes in CRC. 26 Another study has revealed an association between loco-regional recurrence and expression of CD133 and CD44 CSC markers in rectal cancer suggesting their importance in disease prognosis. 27 However, molecular studies on margins associated with CRC are lacking; hence we studied the tumour and its associated distal margin for identification and quantification of putative CSC markers and regulatory genes.
Ep-CAM + /CD44 + and Ep-CAM + /CD133 + cells happen to be the most studied CSC markers in CRC. 14,18 Higher expression of Ep-CAM has already been reported in CRC and revealed to have high tumourigenic potential in combination with CD44 expression. 14,28,29 Our flow cytometry analysis showed significant enrichment of Ep-CAM + /CD44 + and Ep-CAM + /CD133 + cells in distal margin after   Figure 1B). In concordance with above findings, immunohistochemistry also showed a significant enrichment for CD133 and CD44 in tumour and distal tissues ( Figure 2B). CD133 staining was found confined to luminal surface of crypts whereas CD44 stained epithelial and stromal cells (Figure 2A). However, Ep-CAM showed no significant variation across normal, tumour and distal margin tissues, although the staining was more intense in tumour and distal margin (Figure 2A). This heterogeneity in abundance of CSCs in tumor and distal tissues could have significant clinical importance as Ep-CAM + /CD44 + cells are invasive in nature and possess higher metastatic potential. 14 Therefore, the presence of these cells at distal margin raises the concern for the presence of invasive cancer cells, which could lead to distant metastasis if left unidentified.
On the other hand, enrichment of Ep-CAM + /CD133 + cells are known to be therapy resistant 32,33 and abundance of these cells in tumour and distal margin suggests the presence of drug-resistant cells.
Interestingly, the enrichment of CD133 and CD44 positive cells were not evident in all cases of patients sample analysed (Table S4), showing inter-tumour heterogeneity among CRC patients. Enriched expression of CD44 and CD133 in tumour and distal margin could suggest the presence of disseminated cells from the tumour (residual cells) or may be due to the molecular alteration caused by treatment given.
A recent study has reported that the expression of CSC markers after preoperative chemoradiotherapy has independent role in predicting recurrence-free survival in rectal patients. 34 Yet, most importantly the question which needs to be addressed happens to be the causal reason behind this heterogeneity in CSC abundance in tumour and distal margin upon therapy and their role in disease prognosis.
To explore the mechanism of CSCs enrichment in pathologically negative distal margin, we performed RNA sequencing of tumour and distal tissue keeping normal tissue as reference. This study revealed a significant number of cancer-associated genes, which were found to be mutated in both tumour and distal margins ( Figure 3A-D).
This finding suggests that histopathologically normal cells in distal tissue possess genetic alterations similar to tumour. In addition, a significant number of mutated genes were found to be involved in cell proliferation, migration, drug response, adhesions, ECM organizations needed for cancer cell growth and invasion. Importantly, we found Wnt signalling genes to be enriched among the top 20 bioprocess in our analysis ( Figure S4). Enrichment of mutated genes associated with stem cell maintenance bioprocess in both tumour as well as distal margin supports our observation on presence of CSCs in distal margin ( Figure 3E). However, exposure to chemoradiotherapy has been already studied to induce stemness characteristics in various cancers. So it is obvious that it can affect normal surrounding tissues of tumour as well. Interestingly, RNA sequencing study revealed distinct genetic differences between normal tissue compared to distal and tumour. This further shows that CSC signatures in distal margin cannot be solely due to chemoradiotherapy and distal adjacent tissue could share a tumour field effect unlike normal tissue in CRC. Among altered CSC maintaining genes, we observed a significant number of Wnt signalling components ( Figure 3E). Studies have shown the regulatory role of Wnt signalling in the expression of CD133 and CD44 in CRC. 35,36 Oct4 and β-catenin are the two major Wnt pathway genes which are involved in stem cell maintenances. 37 In our IHC studies, we observed an increased expression of Oct4 and β-catenin in tumour and distal margins, whereas normal tissues showed minimal expression ( Figure 3G). Higher Oct4 and β-catenin expression thus again supports the enrichment of CSC in tumour and distal margin tissues. In addition to that, these observation implies that the active involvement of Wnt signalling could be the possible driver behind the enrichment of CD133 and CD44 markers.
Real-time q-PCR analysis of CRC tissues did not show significant enrichment of CSCs markers in case of tumour and distal margin like immunophenotyping ( Figure S2). This could be due to limitation of real-time PCR in specificity as involvement of stromal cells could influence the results.
The expression of CD133 in CRC has already been reported to be involved with drug resistance and disease recurrence and also associated with lower patients survival. 15 (Table 2). Interestingly, low distal resection length also did not show any significant relation with recurrence. However, higher expression of CD133 in distal margin found to have significant association with recurrence (P = 0.03), but CD44 expression did not show similar result (P = 0.82) ( Table 2). Altogether these results advocate the potential of CD133 marker in CRC prognosis.
We further propose that CD133 expression in distal margin should be evaluated in a larger cohort of sample to validate its potential role as an independent prognostic marker in CRC.
As distal resection margin length remains a crucial parameter in cases of rectal cancer due to challenges associated with sphincter preserving surgery, 20,41 our study suggests molecular evaluation of distal margins along with histopathology for better prognostication of patients with CRC.

| CON CLUS ION
In conclusion, our study for the first time showed that in CRC, distal margin harbouring cells with CSCs characteristics can affect the patient's prognosis. In addition, enrichment of CD133 and CD44 in distal margin are independent of its pathological status.
Although we accept the limitation in our study, for lacking a larger cohort, we further propose that CD133 expression status in distal margin could be a crucial factor in disease progression in CRC patients and could be used as a prognostic marker. As residual cells could be the sole cause behind local recurrence and metastasis, and histological analysis alone may not provide adequate information regarding the heterogeneity of tumour cell types present or molecular alternation set in the tumour and its margins. Based on these observations, we propose that molecular assessment of margin in CRC could be an added tool supplementing histopathology for better treatment and management of CRC patients in the long run.

E TH I C A L A PPROVA L A N D CO N S E NT TO PA RTI CI PATE
All human tissue samples were obtained from Regional Cancer Centre, Trivandrum, Kerala, India after approval of the Regional Cancer Centre Ethical committee (HEC No. 43/2011) and donors that signed written informed consent.

ACK N OWLED G EM ENT
We

CO N FLI C T O F I NTE R E S T
The authors declare no conflict of interest.