Analysis of the molecular features of rectal carcinoid tumors to identify new biomarkers that predict biological malignancy.

Although gastrointestinal carcinoid tumors are relatively rare in the digestive tract, a quarter of them are present in the rectum. In the absence of specific tumor biomarkers, lymphatic or vascular invasion is generally used to predict the risk of lymph node metastasis. We, therefore, examined the genetic and epigenetic alterations potentially associated with lymphovascular invasion among 56 patients with rectal carcinoid tumors. We also conducted a microRNA (miRNA) array analysis. Our analysis failed to detect mutations in BRAF, KRAS, NRAS, or PIK3CA or any microsatellite instability (MSI); however, we did observe CpG island methylator phenotype (CIMP) positivity in 13% (7/56) of the carcinoid tumors. The CIMP-positive status was significantly correlated with lymphovascular invasion (P = 0.036). The array analysis revealed that microRNA-885 (miR-885)-5p was the most up-regulated miRNA in the carcinoid tumors with lymphovascular invasion compared with that in those without invasion. In addition, high miR-885-5p expression was independently associated with lymphovascular invasion (P = 0.0002). In conclusion, our findings suggest that miR-885-5p and CIMP status may be useful biomarkers for predicting biological malignancy in patients with rectal carcinoid tumors.


INTRODUCTION
Carcinoid tumors originate from neuroendocrine cells and grow slowly. Although they are relatively rare in the digestive tract, rectal carcinoid tumors account for a quarter of the gastrointestinal carcinoid tumors. Carcinoid tumors are classified as neuroendocrine tumor-grade 1 (NET-G1) tumors, according to the 2010 revision of the World Health Organization (WHO) guidelines [1]. In the absence of specific tumor biomarkers, the presence of lymphatic or vascular invasion is generally used to predict the risk of lymph node metastasis after endoscopic resection. www.impactjournals.com/oncotarget Therefore, we examined the clinical, pathological, and molecular features of rectal carcinoid tumors to identify new biomarkers that were predictive of malignancy.

The association between lymphovascular invasion and CIMP status in rectal carcinoid tumors
We selected the six markers [CDKN2A (p16), IGF2, MLH1, MINT1, MINT2, and MINT31] that were methylated in more than 5% of the rectal carcinoid tumors as CIMP markers. CIMP-positive status was defined as the presence of four or more of the six methylated promoters and CIMP-negative status as zero to three of the six methylated promoters. The CIMP-positive cases were detected in 13% (8/56) of the rectal carcinoid tumors, and the CIMP status was significantly correlated with lymphovascular invasion (P = 0.036) ( Table 1).

Detection of high microRNA-885-5p expression in rectal carcinoid tumors with lymphovascular invasion on miRNA array analysis
To examine the miRNA expression signature in rectal carcinoid tumors with lymphovascular invasion, 10 carcinoid tumors were randomly selected from the carcinoid tumor specimens for miRNA array analysis.  negative in five. The median expression levels in the carcinoid tumors with lymphovascular invasion were compared with those in the tumors without invasion. The miRNA array data revealed different expression levels in individual miRNAs between the two groups. Ten miRNAs that displayed higher expression levels (≥ 30-fold change) in carcinoid tumors with lymphovascular invasion than those without invasion are shown in Table 2. Of the 760 miRNAs, microRNA-885 (miR-885)-5p was the most up-regulated in the carcinoid tumors with lymphovascular invasion compared with that in those without invasion (878-fold change, P = 0.037).

Multivariate analysis to identify the association with lymphovascular invasion in carcinoid tumors
To account for multiple hypothesis testing in associations between the lymphovascular invasion status and the other five covariates (gender, age, tumor size, CIMP status, and miR-885-5p), the P-value was adjusted by Bonferroni correction to P = 0.01 (= 0.05/5). A backward stepwise elimination with a threshold of P = 0.20 was used to select variables in the final model. Of the variables, only CIMP status and miR-885-5p were included in the final model, which showed that The fold change is expressed as the median of that for the rectal carcinoid tumors with lymphovascular invasion divided by those without lymphovascular invasion for each microRNA. P-values were determined by the Mann-Whitney U test. lymphovascular invasion was only significantly associated with high miR-885-5p expression (P = 0.0002).

DISCUSSION
We examined the genetic and epigenetic alterations associated with lymphovascular invasion and conducted miRNA array analysis among 56 patients with rectal carcinoid tumors and who underwent endoscopic resection. We did not detect KRAS, BRAF, or PIK3CA mutations or MSI-high in the carcinoid tumors. With regard to the epigenetic alterations, CIMP positivity was present in 13% (7/56) of the rectal carcinoid tumors and was significantly correlated with lymphovascular invasion. The miRNA array analysis revealed that miR-885-5p was the most up-regulated miRNA in the rectal carcinoid tumors with lymphovascular invasion compared with that in those without invasion. Our data also showed that high miR-885-5p expression was independently associated with lymphovascular invasion.
In this study, specific miRNA expression associated with lymphovascular invasion was identified in the rectal carcinoid tumors. Using miRNA array analysis, our data showed that miR-885-5p expression was the most up-regulated miRNA in the carcinoid tumors with lymphovascular invasion compared with that in those without invasion (878-fold change). In the current study, we found that high miR-885-5p expression was independently associated with lymphovascular invasion in a sample of 56 patients with rectal carcinoid tumors. To our knowledge, this is therefore the first report that has identified an association between miR-885-5p expression and lymphovascular invasion in rectal carcinoid tumors.
MiR-885-5p is located in the 3p25.3 genomic region. A previous study reported that the up-regulation of miR-885-5p reduced the levels of MMP-9 in glioma cells and that it inhibited cellular invasion [43]. Furthermore, the transfection of miR 885-5p mimics could decrease MMP-9 expression. Because MMP-9 is known to play a pivotal role in regulating invasiveness in various human cancers, miR-885-5p may increase for inhibiting tumor cell invasion via MMP-9 in rectal carcinoid tumors.
Our current study was limited by the cross-sectional design and the potential for bias (i.e., selection bias) that could have confounded the results. Nevertheless, our multivariate regression analysis was adjusted for potential clinical and molecular confounders. Although further study is needed to confirm our results, this is an interesting first step toward the improved understanding and diagnosis of these rare tumors.
With regard to colorectal cancer, a recent miRNA array analysis reported that miR-885-5p is significantly up-regulated in liver metastases compared with that in primary colorectal cancer tissue, and high serum miR-885-5p expression significantly predicts survival and metastasis in patients with colorectal cancer [44]. Although miR-885-5p expression in primary colorectal cancer tissue was not previously associated with clinical and pathological features, high serum miR-885-5p is significantly associated with both poor overall survival and poor disease-free survival [44]. Using in situ hybridization analysis, Hur et al. have also confirmed miR-885-5p expression in metastatic tumor cells of the liver, but not in their adjacent hepatocytes; however, they did not utilize laser capture microdissection, which might decrease the influence of stromal cells [44]. When Iino et al. used laser capture microdissection to isolate cancer cells from formalin-fixed paraffin-embedded (FFPE) primary colorectal cancer samples and corresponding metastatic liver tumors, they identified miR-885-5p as the second most up-regulated miRNA in colorectal liver metastases compared with primary tumors [45]. These findings suggest that the up-regulation of miR-885-5p in tumor cells, but not in stromal cells, may play an important role in the development of colorectal liver metastasis.
CIMP is associated with both favorable and unfavorable prognoses, as well as different clinical characteristics, depending on the tumor type. In colorectal cancer, CIMP is associated with a favorable prognosis [35,46]. In contrast, a recent analysis in patients with primary clear cell renal carcinoma showed that CIMP positivity is characterized by tumor clusters that are associated with aggressiveness and patient survival [34]. Regarding early colorectal neoplastic lesions, we recently reported that there was a progressive increase in the methylation status (i.e., CIMP status and MLH1 methylation) from hyperplastic polyps to SSA/Ps to SSA/Ps with cytological dysplasia [3], suggesting that accumulating epigenetic alterations may be associated with biological malignancy in serrated pathway progression. Such findings emphasize the motivation for establishing whether CIMP is universal or cancer specific because of its potential for use as a prognostic marker. There are several possible explanations for the discrepancies identified to date. First, although CIMP has been identified in different tumors, it may simply not be a universal marker of good or bad prognosis. Second, it is possible that the gene panels and cut-off thresholds used to define CIMP are not sufficiently accurate to define the true phenotypes of some cancers.
In contrast to colorectal cancer, research into CIMP in rectal carcinoid tumors is unfamiliar. In the current study, we examined 11 methylated promoters. CIMP positivity was defined as the presence of four or more of the six methylated promoters [CDKN2A (p16) ,  IGF2, MINT1, MINT2, MINT31, and MLH1], which was detected in 13% of the rectal carcinoid tumors and was associated with lymphovascular invasion. These results suggest that some epigenetic alterations were associated with biological malignancy in rectal carcinoid tumors.
In conclusion, we demonstrated that CIMP positivity was significantly correlated with lymphovascular invasion in rectal carcinoid tumors. Furthermore, miR-885-5p was independently associated with lymphovascular invasion in carcinoid tumors. Thus, our findings suggest that miR-885-5p and CIMP status may be useful biomarkers for predicting biological malignancy in patients with rectal carcinoid tumors.

Patients and rectal carcinoid tumor tissue specimens
We collected formalin-fixed paraffin-embedded (FFPE) tissue specimens from 56 cases of rectal carcinoid tumors following endoscopic resection at Sapporo Medical University Hospital or Keiyukai Sapporo Hospital between 2007 and 2014. To minimize selection bias, we collected consecutive FFPE tissue specimens. All lesions were submucosal with smooth surfaces and yellow tones, pathologically corresponding to the WHO NET-G1 classification [1].
Lymphovascular (lymphatic or vascular) invasion was detected in 23% (13/56) of the rectal carcinoid tumors. Of the 13 cases with lymphovascular invasion, surgical resections were treated in 7 cases. Several cases (N = 6) were followed without surgical resection because informed consent was not obtained from those patients. Lymphovascular invasion was used as a surrogate marker of the malignancy potential. No patient exhibited recurrence or death associated with carcinoid tumor until December 2014.
Informed consent was obtained from all the patients before specimen collection. This study was approved by the institutional review boards of the participating institutions and complied with the tenets of the Helsinki Declaration. Our analysis of the rectal carcinoid tumor tissue specimens is fully compliant with the REMARK guidelines [51].

RNA extraction and microRNA array analysis
Total RNA was extracted from FFPE tissues using the miRNeasy FFPE Kit (Qiagen, Valencia, CA, USA). The TaqMan ® Array Human MicroRNA A + B Cards Set v3.0 (Applied Biosystems, Foster City, CA, USA) was used for simultaneous measurement of the expression of 760 miRNAs on a microfluidic PCR platform. In brief, 1 μg of total RNA was reverse transcribed using the Megaplex Pools Kit (Applied Biosystems), following which miRNAs were amplified and detected by PCR with specific primers and TaqMan probes. PCR was run in the 7900HT Fast Real-Time PCR system (Applied Biosystems), and SDS 2.2.2 software (Applied Biosystems) was used for comparative analysis of the cycle threshold (ΔC T ). U6 snRNA (RNU6B; Applied Biosystems) served as an endogenous control. ΔC T was calculated by subtracting the C T values of U6 from the C T values of the gene of interest. Expression of each miRNA in the tumor samples was calculated using the equation 2 −ΔCT , where ΔC T = (C T miRNA − C T U6).

Statistical analysis
The JMP (version 10) software was used for all statistical analyses (SAS Institute, Cary, NC, USA). P values were two-sided. Univariate analysis was performed to determine the clinical and molecular characteristics according to the lymphovascular invasion of the rectal carcinoid tumors. P values were calculated using t-test for age and tumor size and by a chi-square test or Fisher's exact test for all other variables.
A multivariate logistic regression analysis assessing the relationships with lymphovascular invasion status initially included gender (male vs. female), age (continuous), tumor size (continuous), CIMP status (positive vs. negative) and miR-885-5p expression (high expression vs. low expression), considering potential confounding and causal relationships.