Abstract
Background/Aim: Head and neck adenoid cystic carcinoma (HNACC) is a rare malignancy of the salivary glands with a tendency to metastasize in lung or liver without lymph node involvement, whereas squamous cell carcinoma (HNSCC) preferentially metastasizes to locoregional lymph nodes. The expression patterns of microRNA, a class of small non-coding RNA transcripts, involved in gene regulation and various developmental processes, could be of influence during the metastatic process. The aim of the present study was to compare mircoRNA expression patterns of HNACC and HNSCC. Materials and Methods: In a total of 21 tissue samples, a genome-wide screening for microRNAs was performed. A microRNA array platform was used for the identification of target microRNA. Results: Five microRNAs, hsa-MiR-214, hsa-MiR-125a-5p, hsa-MiR-574-3p, hsa-MiR-199a-3p/199b-3p and hsa-miR-199a-5p were identified to be over-expressed in HNACC compared to HNSCC, whereas hsa-MiR-452 showed a lower expression level. Conclusion: Our data showed significantly different expression patterns of mircoRNA in HNACC and HNSCC supporting the theory of tumor-specific expression and giving hints for different clinical behavior.
Head and neck adenoid cystic carcinoma (HNACC) is a rare condition compared to squamous cell carcinoma (HNSCC) and the second most common malignancy of salivary glands in the head and neck region. Both tumors have epithelial origin; in HNSCC carcinogenesis is caused by chemical or viral influence, while in HNACC it remains uncertain. A fundamental difference in clinical behavior is the tendency of HNSCC to metastasize lymphogenic into cervical nodes and in HNACC to develop hematogenous lung metastases (1, 2). In HNACC, radical tumor resection with or without adjuvant radiotherapy is considered as the principal way of treatment with curative intent. Local perineural recurrence and distant metastases show initially satisfying surgical results and lead to a significant reduction of progression-free survival and, thus, a mediocre long-term prognosis (3). In case of local or distant failure, classical chemotherapies, e.g. cisplatin or gemcitabine, lead to stable disease in some patients, while partial remission is achieved only in a few cases and progression is observed in up to 50% of patients (2, 4).
Different molecular and genetic features of this entity, like high epidermal growth factor receptor expression (EGFR), have been identified and lead to potential targets for novel therapeutic approaches, like EGFR-blockage with cetuximab. The clinical response in several trials showed at least stable disease in the majority of patients for several months but no tumor remissions (5). Therefore, a better molecular understanding of the biology and genetics of this tumor seems essential.
Recent work has revealed the existence of a class of small non-coding RNA species, called microRNAs, representing a new, rapidly growing class of gene regulators involved in various developmental and physiological processes (6, 7). Accordingly, microRNAs have been implicated in various human diseases, in particular cancer. Some microRNAs regulate the expression of oncogenes or tumor suppressor genes. Recently, it has also been shown that microRNA expression patterns can be used to classify human cancer (6-10). The role of microRNA expression in survival and outcome of patients in HNACC is not very well known so far. A recent analysis by Mitani et al. (2013) showed that there is a dysregulation of certain microRNA subtypes in adenoid cystic carcinoma (ACC of the parotid grand) compared to normal salivary gland tissue (8). In this study on 30 ACCs, an over-expression of miR-17 and miR-20a was found to be associated with a poor outcome. The ACC-typical MYB/NFIB gene translocation (11) was not found to be correlated to microRNA patterns.
The aim of the current study was to compare microRNA expression patterns of HNACC with HNSCC in order to obtain a better insight to specific tumor biology and in potential molecular pathways during metastasis.
Patients and Methods
A total of 21 tissue samples were collected at the University Hospital Dusseldorf, Germany. This study was approved and endorsed by the local Ethics Committee (protocol #2661). All patients gave informed consent to the acquisition of tissue samples and trial participation before operative procedures. A genome-wide screen for microRNAs that are differentially expressed in HNACC and HNSCC was performed.
Fresh-frozen tumor samples from 5 ACC patients and 10 HNSCC patients, 6 human papilloma virus (HPV)-positive and 4 samples of HPV-negative patients and 6 healthy mucosa references were collected over a period of 12 months. Patients were treated surgically under general anesthesia and samples were acquired during definitive surgical therapy procedures. A total of 5 μg RNA per case was extracted from fresh-frozen tissue samples using acidic phenol chloroform extraction (Trizol™; Invitrogen, Carlsbad, CA, USA). RNA was purificated using the RNeasy Mini Kit™ (Qiagen, Hilden, Germany) following the manufactures instructions. After having passed sample quality control on the Bioanalyser 2100™ and RNA measurement on the Nanodrop instrument (Agilent technologies, Santa Clara, USA) the samples were labeled using the miRCURY™ Hy3™/Hy5™ power labeling kit (Exiqon, Vedbaek, Denmark) and hybridized on the miRCURY™ LNA Array (version 9.2; Exiqon).
Analysis of the scanned slides showed that the labeling was successful as all capture probes of the oligo spike-in controls produced signals in the expected range. The quantified signals (no background subtraction) were normalized using the global Lowess (LOcally WEighted Scatterplot Smoothing) regression algorithm (Exiqon, Vedbaek, Denmark), to produce the best within-slide normalization, in order to minimize the intensity-dependent differences between the dyes.
Spike-ins (10 different RNA oligonucleotides) were used in these experiments to assess the overall performance of these experiments. The spike-ins were labeled as normal microRNA species and hybridized to capture probes specifically designed to capture these RNAs.
One-way ANOVA analysis between time point groups was used; all microRNAs with a p-value <0.05 were considered significantly over-expressed. The unsupervised heat map was generated based on an unsupervised two-way hierarchical clustering analysis of the top 50 microRNAs showing the largest variation across all samples (Figure 1). The expert analysis heat map depicts a list of 20 microRNAs based on the ANOVA analysis of the results (Figure 2) and was used for further analysis of expression patterns.
The results of microRNA over- and under-expression of the groups were compared to each other in a cross table analysis. Frequency of expression above and below mean were compared to each other and tested for significance with the chi-square test.
Results
MicroRNA expression in HNACC tumor samples. In the 5 HNACC patients, similar expression patterns were found: microRNAs hsa-miR-125b, hsa-miR-199a-5p, hsa-miR-199a-3p/199b-3p, hsa-miR-574-3p, hsa-let-7d, hsa-miR-214 and hsa-miR-125a-5p showed an expression above the mean value in the expert analysis heat map (Figure 2). Other microRNAs, like hsa-miR-203, hsa-miR-149 and hsa-miR-617, were expressed well below mean. All HNACC were clustered next to each other and showed similar expression patterns (Figure 2).
MicroRNA expression in HNSCC tumor samples. In the microRNA samples from the HPV-positive HNSCC patients, a similar expression pattern within most of the 6 samples was found: hsa-miR-203, hsa-miR-125b, hsa-miR-452 were expressed above mean, whereas hsa-miR-615-3p, hsa-miR-631 and hsa-miR-517* were expressed below mean. Interestingly, the expression patterns were similar to healthy mucosa (Figure 2). The HPV-negative samples of HNSCC patients showed over-expression of hsa-miR-615-3p, hsa-miR-631 and hsa-miR-432* and an expression below mean of hsa-miR-125b, hsa-miR-199a-5b and others. Still, the patients were displayed on the opposite sides of the hierarchical trees of the expert analysis heat map shown in Figure 2 due to differences of the composition of microRNA. This may confirm the fact that HPV-negative HNSCCs display a much wider variety of genetically different tumors in comparison to HPV-positive tumors (12, 13).
In a cross table analysis, 5 of the expert analysis microRNAs were identified to be differentially expressed in HNACC compared to HNSCC (Table II). They were expressed significantly more often above mean in HNACC compared to HNSCC, in which they were expressed below mean. These patterns have not been reported to play a role in these entities so far. MicroRNAs hsa-miR-214, hsa-miR-125a-5p, hsa-miR-574-3p, hsa-miR-199a-3p/199b-3p and hsa-miR-199a-5p were affected. One microRNA, hsa-miR-452, was found to be significantly under-expressed in HNACC compared to HNSCC, where it was expressed above mean (Table II).
Discussion
Different expression patterns and qualitative changes of specific microRNA expression have been found in different cancer types and stages. Some microRNAs seem to have a regulative impact on the expression of oncogenes or tumor suppressor genes; however, their impact on the clinical course is not yet entirely clear. Recently, it has also been shown that microRNA expression patterns can classify human cancer and may be used as targets in novel therapeutic approaches (14, 15). In laryngeal cancer, a microRNA inhibition (miR-196a) has shown anticancer potential in a preclinical mouse model (16).
MicroRNA expression above or below mean in HNACC and HNSCC, according to a heat map ANOVA analysis. Cross table analysis with Chi-square test showed 5 microRNAs with significantly more frequent expression above mean in HNACC and 1 mircoRNA (has-miR-452) with significant expression below mean. Only significant values are displayed and marked with an asterisk.
Summary of differentially expressed miRNAs in HNACC compared to HNSCC (cross table analysis). Down arrow symbol indicates a significant down-regulation compared to up arrow symbol, which indicates a significant over-expression. CRC, colorectal carcinoma; GC, gastric cancer; HCC, hepatocellular carcinoma; LC, lung carcinoma.
The aforementioned 5 significantly over-expressed and the 1 significantly under-expressed miRNAs in HNACC, in comparison to HNSCC, have so far not been reported to play a role in this entity (Figure 2 and Tables I and II). Hsa-miR-214 has been found to play a role in the promotion of metastatic niches in colorectal cancer and additionally seems to have oncogenic effects in osteosarcoma (17-19). On the other hand, hsa-miR-125a-5p displays tumor-suppressing abilities in hepatocellular cancer and induces apoptosis in lung cancer (20, 21). Hsa-miR-574-3p was found to delay cell growth in vitro and play a role in tumor suppression (22, 23). Hsa-miR-452 under-expression promotes stem-like traits and tumorigenicity in glioma (24). Hsa-miR-199a-3p/199b-3p was described in hepatocellular carcinoma and liver cirrhosis leading to up-regulation of transforming growth factor-beta signaling pathway (25). Finally, hsa-miR-199a-5p up-regulation was found to be oncogenic in gastric cancer, while its down-regulation induces drug resistance in vitro (26, 27). In HNACC, little is known on the specific microRNA expression. The results of our analysis support the fact that different cancer entities - even of anatomically close regions - display significantly different expression profiles of microRNA. This fact may play a role in the regulation of different pathways of metastasis, although a clinical correlation in a larger cohort is necessary. Unfortunately, in our group of patients, the follow-up was insufficient and, thus, a correlation of microRNA expression patterns and clinical course was not possible. The results of this pilot study will be followed-up in a larger case series. Our results are not in accordance with the results of Mitani et al. who found a significant over-expression of the miR-17-92 cluster and its paralogs, miR-106b-25 and miR-106a-363. In that study a high expression of miR-17 and miR-20a was found to be significantly associated with poor survival (8). Reasons might lie in the larger samples size or the genetic variability of the tumors in the different samples. In the presented data a different expression of microRNA was found in HNACCs and HNSCC. In 5 HPV-positive HNSCC patients, a relative over-expression of hsa-miR-203, hsa-miR-125b, hsa-miR-452 was registered, whereas hsa-miR-615-3p, hsa-miR-631 and hsa-miR-517* were expressed below mean levels. Interestingly, the microRNA expression of these tumors was similar to healthy mucosa. In the HPV-negative HNSCC-samples, no consistent pattern of microRNA expression was found. This seems in accordance with the fact that HPV-negative HNSCCs display a wider variety of genetically different tumors in comparison to HPV-positive tumors (12, 28, 29). The large variety of up- and down-regulated microRNAs in HNSCC is shown in a meta-analysis of Chen et al. in 2013. They reported 7 consistently up-regulated microRNA in HNSCC (30), a finding that is not in accordance with our results and may be due to higher patient numbers and heterogeneity of tumors.
Unsupervised heat map. Each row represents a microRNA and each column represents a sample. The microRNA clustering tree is shown on the left and the sample clustering tree appears at the top. The color scale shown at the top left illustrates the relative expression level of a microRNA across all samples: red color represents an expression level below mean, green color represents expression above mean. The clustering is performed on log2(Hy3/Hy5) ratios from 50 microRNAs with p-values <0.05 calculated between groups of cell types in one-way ANOVA (healthy mucosa, 1c, 2c, 5c-8c; HNSCC/HPV+, 1-6; HNSCC/HPV-, 7-10; HNACC patients, 11-15).
Expert analysis heat map. In the present analysis the top 20 microRNA with significantly different expression patterns tested by ANOVA analysis through all groups are displayed. Each row represents a microRNA and each column represents a sample. The microRNA clustering tree is shown on the left and the sample clustering tree appears at the top. The color scale shown at the bottom illustrates the relative expression level of a microRNA across all samples: red color represents an expression level below mean, while green color represents expression above mean (healthy mucosa. 1c, 2c, 5c, 6c; HNSCC/HPV+. 1-6; HNSCC/HPV-. 7-10; HNACC patients, 11-15).
HNACC displays significantly different expression patterns of microRNA compared to HNSCC, both HPV-negative and HPV-positive samples. The role of these patterns in carcinogenesis and clinical behavior is not clear until today, although the understanding of certain pathways in tumor suppression and promotion, as well as drug resistance, is increasing. The current study gives a first insight in the complex differential regulation of two typical head and neck carcinomas. However, additional studies are required in order to obtain a high-definition picture with the option to use it for therapeutic interventions.
Footnotes
-
Conflicts of Interest
The Authors declare no financial or ethical conflicts of interest.
- Received October 22, 2014.
- Revision received December 9, 2014.
- Accepted December 12, 2014.
- Copyright© 2015 International Institute of Anticancer Research (Dr. John G. Delinassios), All rights reserved
