Prognostic Impact of MiR-155 in Non-Small Cell Lung Cancer Evaluated by in Situ Hybridization

Lung cancer is the leading cause of cancer-related mortality in both men and women [1]. Despite several new treatment achievements, the consistently poor 5-year survival for lung cancer patients underscores the need for novel modalities for early detection, prognostification and targeted therapies [1, 2].

MicroRNAs (miRNAs) are approximately 19-22 nucleotides single stranded RNAs playing crucial roles in regulating gene expression by either inducing mRNA degradation or inhibiting translation [3, 4]. These non-coding RNAs can simultaneously regulate hundreds to thousands of their target genes or up to one third of the genome, thereby controlling a wide range of biological functions including apoptosis, proliferation and differentiation [3, 5].

To date miR-155 is one of the miRNAs most consistently involved in neoplastic diseases in both hematopoietic malignancies (i.e. Hodgkin's lymphoma, some types of Non Hodgkin's lymphoma, AML and CML) and solid tumors (e.g. breast, colon, cervical, thyroid, pancreatic and lung cancer) [6‐16]. MiR-155 is also involved in other biological processes like hematopoiesis, inflammation and immunity [6]. The frequently detected up-regulation of miR-155 in malignant cells indicates a major role as an oncogene, however, a possible tumor suppression function has also been suggested [17]. In non-small cell lung cancer (NSCLC), miR-155 has so far been considered as an oncogene and been associated with a poor prognosis [13, 16], though a recent large scale study did not find miR-155 to have any prognostic or predictive impact [18].

NSCLC classification according to histology and nodal status are two of the most important determinants for NSCLC treatment strategies [13, 19]. However, a considerable variability in prognosis has been observed for subsets of patients with the same clinical features. Consequently, the clinical incorporation of predictive and prognostic molecular biomarkers with traditional cancer staging should improve the management of patients with NSCLC.

Squamous cell carcinomas (SCCs) and adenocarcinomas (ACs) are the major histological subtypes of NSCLC. During recent years, treatment responses and side effects by novel therapies have been correlated to NSCLC subgroups according to histology, gender, ethnicity and smoking status. The vascular endothelial growth factor (VEGF) monoclonal antibody, bevacizumab, is only given to non-SCCs due to the risk of fatal bleeding in SCCs [20]. Further, mutations in epidermal growth factor receptor (EGFR) and response to EGFR tyrosine kinase inhibitors appear related to ACs, female gender, Asian ethnicity and non-smokers, and the new antifolate agent pemetrexed appears to have better response in non-SCC patients and females [21, 22]. Consequently, ACs and SCCs are increasingly recognized as different diseases instead of one.

In an unselected NSCLC cohort of 335 patients [23] we aimed to explore, using in situ hybridization on a high throughput platform, possible prognostic roles by miR-155 in all NSCLC cases and subgroups according to histology and stage.

MiRNAs are well preserved in formalin-fixed tissue, making them attractive candidates for use in routinely processed material [34, 35]. Most of the previous studies on miRNA expression were done on microarrays using RNA extracted from human cancer tissues samples and containing a mixture of neoplastic tumor cells and tumor related stromal cells. A major advantage of in situ hybridization is to precisely identify positive signals at the cellular level. For instance, recent data have demonstrated that some miRNAs had high expression levels in stromal cells but not in tumor cells [36]. Using RNA extracts from whole tumors, this finding would easily be missed.

Strengthening the relevance of our miR-155 expression data, there was a significant inverse correlation with PTEN. This corroborates a study by Yamanaka et al. showing that reduced expression of miR-155 led to up-regulation of PTEN in NK lymphoma cell lines [37].

Several studies have shown miR-155 to be overexpressed in NSCLC [13, 14, 16]. But, to our knowledge, only three studies have investigated the prognostic impact of miR-155 in NSCLC, all using quantitative RT-PCR as the principal method [13, 16, 18]. Yanaihara et al. [16], also using the median value as cut-off, found high miR-155 expression to be an independent negative prognostic factor in 64 stage I adenocarcinomas, corroborating our results.

Recently, Voortman et al. studied the prognostic and predictive values of a panel of miRs by quantitative real-time PCR in formalin-fixed paraffin-embedded tumor specimens from 639 resected NSCLC patients participating in the International Adjuvant Lung Cancer Trial (IALT) [18]. In the total cohort they found, consistent with our results, miR-155 to have no significant prognostic impact. However, subgroup analysis on the prognostic impact with regard to nodal status and histology was not reported. Raponi and coworkers identified 15 miRNAs that were differently expressed between epithelial cells in normal lung and stage I-III SCC, among them miR-155 [13]. Analysis of 54 SCC patients (63% N0) showed that high miR-155 expression tended to have a significant effect on survival (P = 0.06), while it was an unfavorable independent variable in multivariate analysis (HR 2.3, CI 95% 1.0 - 5.6). We found the same tendency (P = 0.15) in our N0 patients. More surprisingly, we found the opposite association in our SCC lymph node positive patients. This may indicate that the oncogenic miR-155 effect may become inhibited or overridden by other mechanisms in SCC patient with nodal metastasis. Though, as the number of cases in this subanalysis is limited (n = 30 in each arm) the result has to be interpreted carefully. There is always a danger of false positive results when stratifying in multiple subgroups. However, we have only stratified for histological classification and nodal status which are considered to be the two most important clinicopathological variables in NSCLC treatment strategies.

As an independent prognostic factor, miR-155 may be a relevant addition to clinicopathological variables in predicting outcome in adenocarcinoma patients. As a prognosticator, however, miR-155 expression appears more interesting in SCCs with nodal metastasis, as none of the clinicopathological variables were significant prognosticators in this subgroup. In the clinic, valid prognostic marker in the subpopulation of N+ patients is warranted and miR-155 seems to be a potentially interesting candidate, though further prospective validation studies are needed to confirm these results. Potential microRNA-based therapy is now being exploited in cancer, attempting to modulate their expression, reintroducing microRNAs lost in cancer, or inhibiting oncogenic microRNAs by using anti-micro oligonucleotides [38]. In a novel approach to inhibit microRNA function, synthetic mRNAs, called microRNA sponges, are able to bind up the microRNA, preventing its association with endogenous targets [39]. MiR-155 has also been suggested as a possible target in future treatment strategies. Indeed, as miR-155 (together with let-7a, miR-21 and miR17-92 cluster) is aberrantly expressed in a wide variety of hematological and solid malignancies, it has been speculated that strategies to silence miR-155 may have impact on multiple groups of cancer patients [40]. But according to our results, the miR-155 effect is apparently context specific, and though it may be relevant for a diversity of malignancies, an "individualized" approach is needed.