- Split View
-
Views
-
Cite
Cite
C. Vanesa Díaz-García, Alba Agudo-López, Carlos Pérez, José A. López-Martín, J. Luis Rodríguez-Peralto, Javier de Castro, Ana Cortijo, Miriam Martínez-Villanueva, Lara Iglesias, Rocío García-Carbonero, Juan A. Fresno Vara, Angelo Gámez-Pozo, José Palacios, Hernán Cortés-Funes, Luis Paz-Ares, M. Teresa Agulló-Ortuño, DICER1, DROSHA and miRNAs in patients with non-small cell lung cancer: implications for outcomes and histologic classification, Carcinogenesis, Volume 34, Issue 5, May 2013, Pages 1031–1038, https://doi.org/10.1093/carcin/bgt022
- Share Icon Share
Abstract
The clinical and functional significance of RNA-interference machinery in lung cancer is poorly understood. Besides, microRNAs (miRNA) have the potential to serve both as biomarkers and therapeutic agents, by personalizing diagnosis and therapy. In this study, we investigated whether the expression levels of DICER1 and DROSHA, components of the RNA-interference machinery, can predict survival, and whether the miRNA expression profiles can differentiate histologic subtypes in non-small cell lung cancer (NSCLC). Levels of DICER1, DROSHA and five different miRNAs were measured in NSCLC specimens ( N = 115) by qRT–PCR assay and correlated with clinical outcomes. Low expression of DROSHA was associated with an increased median survival (154.2 versus 39.8 months, P = 0.016). Also, high DROSHA expression was associated with decreased median survival in the following subgroups: adenocarcinoma ( P = 0.011), grade III tumors ( P = 0.038) and low-stage patients ( P = 0.014). In multivariate analyses, we found two independent predictors of reduced disease-specific survival: high DROSHA expression [hazards ratio = 2.24; P = 0.04] and advanced tumor stage (hazards ratio = 1.29, P = 0.02). In general, the overall tumor miRNA expression was downregulated in our cohort compared with normal tissues. Expression levels of hsa-let-7a ( P = 0.005) and miR-16 ( P = 0.003) miRNA were significantly higher in squamous cell carcinoma than in adenocarcinoma samples. This study supports the value of the expression profiling of the components of the miRNA-processing machinery in the prognosis of NSCLC patients, especially DROSHA expression levels. In addition, differential expression of miRNAs, such as hsa-let-7a and miR-16 may be helpful tools in the histologic subclassification of NSCLC.
Introduction
Lung cancer is the leading cause of cancer-related death worldwide ( 1 ). Pulmonary neoplasias are classified as small or non-small cell lung cancer (NSCLC), the latter comprising a heterogeneous group that is also subclassified in different morphological varieties mainly, squamous cell carcinoma (SCC), adenocarcinoma (ADC) and large cell carcinoma (LCC).
The emerging treatment revolution determined by new targeted therapies stresses the importance of accurate tumor subtyping as a mandatory step in the clinical work-up of these patients. Moreover, the discovery that gene expression can be altered through RNA interference ( 2 ) has stimulated research on the role of RNA interference in the development of cancer. Regulation of gene expression through RNA interference occurs by means of microRNA (miRNA) or small interfering RNA (siRNA).
Expression profiling of many miRNAs in various normal and diseased tissues has demonstrated unique spatial and temporal expression patterns. Some miRNAs have been functionally characterized as oncogenes or tumor suppressor genes ( 3 , 4) . Also, miRNAs are involved in tissue differentiation during both normal development and carcinogenesis, and miRNA expression profiles have been seen not only as a new class of diagnostic and prognostic tools but also as candidates for pharmacological targeting ( 3 , 5–10 ). Most importantly, miRNA expression signatures in both tumor cells and peripheral blood cells from cancer patients can predict outcome ( 11–14 ). The biogenesis of miRNA has been reviewed extensively ( Supplementary Figure 1 , available at Carcinogenesis Online) and its pathway components could be either misexpressed or mutated in tumors. A clear consequence is that miRNA overexpression could result in downregulation of tumor suppressor genes, whereas their underexpression could lead to oncogene upregulation. Their functional association with cancer, small gene size and potential to simultaneously affect a multitude of genes make them unique candidate loci for conferring cancer susceptibility, as a small genetic change in a miRNA sequence can theoretically lead to widespread phenotypic effects.
The RNases DICER1 and DROSHA serve as key regulatory proteins in miRNA biogenesis pathway and their alterations may contribute to widespread miRNA deregulation in cancer ( 15 ). Thus, we examined whether altered levels of DICER1 and DROSHA mRNA, components of the RNA-interference machinery, are associated with clinical outcome in lung cancer. Furthermore, we checked out the role of several miRNAs in differentiating between the two major NSCLC subtypes in human tumor samples, and their potential as biomarkers for lung cancer risk stratification in order to validate the reliability of these novel markers in routine diagnostic samples.
Materials and methods
Tumor samples
Patients were eligible if they had been diagnosed NSCLC, surgically resected, never received radiotherapy or chemotherapy and if there was evaluable tumor tissue available. From 2000 to 2006, we selected 115 patients at 3 centers for participation in the study: ‘12 de Octubre’ University Hospital ( N = 66), ‘La Paz’ University Hospital ( N = 44) and ‘Virgen del Rocio’ University Hospital ( N = 11). In addition, we obtained eight benign epithelial lung tissues from ‘12 de Octubre’ University Hospital Tissue Bank. Specimens were snap-frozen at –80°C until used. Histologic preparations of tumor sections were examined by pathologists without any information about the outcome, and only tissue samples with tumor content >70% were selected. The study was approved by the local Ethics Committees and Commissions of Research (Comité Ético de Investigación Clínica del Hospital Universitario 12 de Octubre), and permission was obtained for the use of all samples.
Real-time quantitative reverse transcription–PCR
Samples were blinded and coded before any laboratory procedure was started. Quantitative reverse transcription (qRT)–PCR was performed with the use of the TaqMan gene expression assay and TaqMan MicroRNA assay kits (Applied Biosystems) according to the manufacturer’s instructions. Amplifications were carried out on the Applied Biosystem 7500 RT–PCR system, and relative gene expression values were calculated by the ΔΔCt method (Sequence Detection System 2.0.5). Results from analysis, done in triplicate, are expressed relative to expression levels of TATA box-binding protein for DICER1 and DROSHA and small nuclear RNA U6 (RNU6B) for miRNAs. The final mRNA levels were converted to ratios of decreased expression (≤1) or increased expression (>1) relative to levels of respective mRNA or miRNA in normal lung epithelium.
Immunohistochemical analyses
The immunohistochemical experiments were performed on 5 μm thick formalin-fixed and paraffin-embedded tissue sections. All steps were performed on the Leica Bond III automated system (Leica Microsystems) according to the manufacturer’s instructions. In brief, specimens were deparaffinized and antigen was retrieved on the instrument. All slides were incubated with primary antibodies against DICER1 (diluted 1:200, Abcam) or DROSHA (1:100, Abcam) for 20min, followed by incubation with a mouse-rabbit-horseradish peroxidase polymer and 3,3′-diaminobenzidine substrate. The stained slides were scored by two investigators who were unaware of the RT–PCR results. Staining intensity was given as no or very weak staining (score 0), weak to moderate (score 1) and deep staining (score 2) at ×100 magnification. The percentage of positive cells was scored as follows: 0 (0–10%), 1 (11–25%), 2 (26–50%) and 3 (51–100%). We defined the immunohistochemical staining score as the sum of the intensity and percentage scores. Tumors were categorized into low-expression group (score = 0–2) and high-expression group (score = 3–5). The scores from two independent investigators were compared and disagreements were resolved by consensus.
Validation analysis
The relation between levels of DICER1 and DROSHA mRNA and survival among patients with lung cancer was examined in existing microarray data sets of samples (GEO accession number: GSE3141 and GSE31210) that have been profiled with an Affymetrix GeneChip assay (HG-U133 Plus 2.0) ( 16 , 17) . Probe sets 212888_at and 218269_at were used to measure DICER1 and DROSHA expression, respectively.
Statistical analysis
Statistical analysis was performed using SPSS 15.0 (Chicago, IL). Kaplan–Meier plots were constructed and log-rank test was used to determine differences in survival and disease-free survival curves. Multivariate analyses with a Cox proportional hazard model were used to examine the effects of DICER1, DROSHA and miRNA expression on death while adjusting for other covariates. Student’s t -test was used for comparison between two groups and Mann–Whitney U -test was used to compare groups that did not conform to the assumption of normality (Kolmogorov–Smirnov test). Contingency tables and Fisher exact test were used to statistically evaluate the relationship between death and categorical variables. All the statistical tests were conducted at the two-sided 0.05 level of significance.
Results
DICER1 and DROSHA are deregulated in NSCLC
Levels of DICER1 mRNA in cancer specimens followed a normal distribution ( P = 0.252). However, the distribution of DROSHA mRNA levels was bimodal ( P = 0.000) with two peaks in the ratio (0.85 and 1.61). DICER1 and DROSHA expression levels varied among cancer specimens; 66% had decreased DICER1 mRNA and 53% had increased DROSHA mRNA ( Supplementary Table 1 , available at Carcinogenesis Online). We did not find a significant correlation between DICER1 and DROSHA expression levels in NSCLC tumors ( P = 0.085).
The protein levels of DICER1 and DROSHA were also examined by immunohistochemical staining in a subgroup of 20 samples ( Supplementary Figure 2 , available at Carcinogenesis Online). The results were consistent with the mRNA expression levels for both DICER1 (Spearman’s rho, ρ = 0.796, P = 0.000) and DROSHA (Spearman’s ρ = 0.701, P = 0.002).
Analysis of clinical associations
Table 1 lists the baseline characteristics of all 115 patients with NSCLC. Clinical outcome data were obtained from patient records. At a median follow-up of 37 months (range 1–154), overall 5-year survival was 62.6% and 39 patients (33.9%) had relapsed. At the end of the study, most patients were alive without disease (51.3%).
Characteristic ( N = 115) . | N (%) . |
---|---|
Age at diagnosis (years) | |
Mean | 65.9 |
Range | 37–85 |
Gender | |
Male | 103 (89.6) |
Female | 12 (10.4) |
Tobacco ( N = 70) | |
Non-smokers | 7 (10.0) |
Former smokers | 41 (58.6) |
Current smokers | 22 (31.4) |
Histology | |
Squamous | 56 (48.7) |
Adenocarcinoma | 45 (39.1) |
Large cell | 8 (7.0) |
Other | 6 (5.2) |
Histologic differentiation | |
Well differentiated (grade I) | 21 (18.3) |
Moderately differentiated (grade II) | 22 (19.1) |
Poorly/undifferentiated (grade III) | 39 (33.9) |
NOS (not otherwise specified) | 33 (28.7) |
Tumor stage | |
I (A and B) | 54 (47.0) |
II (A and B) | 27 (23.4) |
III (A and B) | 27 (23.4) |
IV | 6 (5.2) |
NA | 1 (0.9) |
Disease status | |
Alive with disease | 12 (10.4) |
Alive without disease | 59 (51.3) |
Dead from disease | 44 (38.3) |
Characteristic ( N = 115) . | N (%) . |
---|---|
Age at diagnosis (years) | |
Mean | 65.9 |
Range | 37–85 |
Gender | |
Male | 103 (89.6) |
Female | 12 (10.4) |
Tobacco ( N = 70) | |
Non-smokers | 7 (10.0) |
Former smokers | 41 (58.6) |
Current smokers | 22 (31.4) |
Histology | |
Squamous | 56 (48.7) |
Adenocarcinoma | 45 (39.1) |
Large cell | 8 (7.0) |
Other | 6 (5.2) |
Histologic differentiation | |
Well differentiated (grade I) | 21 (18.3) |
Moderately differentiated (grade II) | 22 (19.1) |
Poorly/undifferentiated (grade III) | 39 (33.9) |
NOS (not otherwise specified) | 33 (28.7) |
Tumor stage | |
I (A and B) | 54 (47.0) |
II (A and B) | 27 (23.4) |
III (A and B) | 27 (23.4) |
IV | 6 (5.2) |
NA | 1 (0.9) |
Disease status | |
Alive with disease | 12 (10.4) |
Alive without disease | 59 (51.3) |
Dead from disease | 44 (38.3) |
NA: not available.
Characteristic ( N = 115) . | N (%) . |
---|---|
Age at diagnosis (years) | |
Mean | 65.9 |
Range | 37–85 |
Gender | |
Male | 103 (89.6) |
Female | 12 (10.4) |
Tobacco ( N = 70) | |
Non-smokers | 7 (10.0) |
Former smokers | 41 (58.6) |
Current smokers | 22 (31.4) |
Histology | |
Squamous | 56 (48.7) |
Adenocarcinoma | 45 (39.1) |
Large cell | 8 (7.0) |
Other | 6 (5.2) |
Histologic differentiation | |
Well differentiated (grade I) | 21 (18.3) |
Moderately differentiated (grade II) | 22 (19.1) |
Poorly/undifferentiated (grade III) | 39 (33.9) |
NOS (not otherwise specified) | 33 (28.7) |
Tumor stage | |
I (A and B) | 54 (47.0) |
II (A and B) | 27 (23.4) |
III (A and B) | 27 (23.4) |
IV | 6 (5.2) |
NA | 1 (0.9) |
Disease status | |
Alive with disease | 12 (10.4) |
Alive without disease | 59 (51.3) |
Dead from disease | 44 (38.3) |
Characteristic ( N = 115) . | N (%) . |
---|---|
Age at diagnosis (years) | |
Mean | 65.9 |
Range | 37–85 |
Gender | |
Male | 103 (89.6) |
Female | 12 (10.4) |
Tobacco ( N = 70) | |
Non-smokers | 7 (10.0) |
Former smokers | 41 (58.6) |
Current smokers | 22 (31.4) |
Histology | |
Squamous | 56 (48.7) |
Adenocarcinoma | 45 (39.1) |
Large cell | 8 (7.0) |
Other | 6 (5.2) |
Histologic differentiation | |
Well differentiated (grade I) | 21 (18.3) |
Moderately differentiated (grade II) | 22 (19.1) |
Poorly/undifferentiated (grade III) | 39 (33.9) |
NOS (not otherwise specified) | 33 (28.7) |
Tumor stage | |
I (A and B) | 54 (47.0) |
II (A and B) | 27 (23.4) |
III (A and B) | 27 (23.4) |
IV | 6 (5.2) |
NA | 1 (0.9) |
Disease status | |
Alive with disease | 12 (10.4) |
Alive without disease | 59 (51.3) |
Dead from disease | 44 (38.3) |
NA: not available.
In univariate analyses, neither DICER1 nor DROSHA mRNA levels were significantly associated with age, tobacco consumption, tumor histology, tumor grade or tumor stage ( Table 2 ). Death from any cause was associated with high levels of DROSHA mRNA (1.46±0.8 versus 1.16±0.9, P = 0.036; 63.64% of dead patients), but we found no relationship to the levels of DICER1 mRNA ( P = 0.589).
Variable . | DICER1 mRNA level . | DROSHA mRNA level . | ||||
---|---|---|---|---|---|---|
Low ( N = 76) . | High ( N = 39) . | P value . | Low ( N = 54) . | High ( N = 61) . | P value . | |
Age (year) | 66.8±9.8 | 64.1±9.4 | 0.21 | 67.2±8.9 | 64.7±10.3 | 0.19 |
Tobacco, N (%) | ||||||
Non-smokers | 4 (9.1) | 3 (11.5) | 0.81 | 4 (11.8) | 3 (8.3) | 0.64 |
Former smokers | 25 (56.8) | 16 (61.5) | 18 (52.9) | 23 (63.9) | ||
Current smokers | 15 (34.1) | 7 (26.9) | 12 (35.3) | 10 (27.8) | ||
Histology, N (%) | ||||||
Squamous | 39 (51.3) | 17 (43.6) | 0.23 | 30 (55.6) | 26 (42.6) | 0.29 |
Adenocarcinoma | 31 (40.8) | 14 (35.9) | 17 (31.5) | 28 (45.9) | ||
Large cell | 4 (5.3) | 4 (10.3) | 3 (5.6) | 5 (8.2) | ||
Others | 2 (2.6) | 4 (10.3) | 4(7.4) | 2 (3.3) | ||
Histologic differentiation, N (%) | ||||||
Well diff. (grade I) | 11 (21.2) | 10 (33.3) | 0.43 | 11 (28.2) | 10 (23.3) | 0.78 |
Moderately diff. (grade II) | 14 (26.9) | 8 (26.7) | 11 (28.2) | 11 (25.6) | ||
Poorly/undiff. (grade III) | 27 (51.9) | 12 (40.0) | 17 (43.6) | 22 (51.2) | ||
Tumor stage, N (%) | ||||||
I | 33 (43.4) | 21 (55.3) | 0.27 | 29 (53.7) | 25 (41.7) | 0.26 |
II | 19 (25.0) | 8 (21.1) | 12 (22.2) | 15 (25.0) | ||
III | 18 (23.7) | 9 (23.7) | 9 (16.7) | 18 (30.0) | ||
IV | 6 (7.9) | 0 | 4 (7.4) | 2 (3.3) |
Variable . | DICER1 mRNA level . | DROSHA mRNA level . | ||||
---|---|---|---|---|---|---|
Low ( N = 76) . | High ( N = 39) . | P value . | Low ( N = 54) . | High ( N = 61) . | P value . | |
Age (year) | 66.8±9.8 | 64.1±9.4 | 0.21 | 67.2±8.9 | 64.7±10.3 | 0.19 |
Tobacco, N (%) | ||||||
Non-smokers | 4 (9.1) | 3 (11.5) | 0.81 | 4 (11.8) | 3 (8.3) | 0.64 |
Former smokers | 25 (56.8) | 16 (61.5) | 18 (52.9) | 23 (63.9) | ||
Current smokers | 15 (34.1) | 7 (26.9) | 12 (35.3) | 10 (27.8) | ||
Histology, N (%) | ||||||
Squamous | 39 (51.3) | 17 (43.6) | 0.23 | 30 (55.6) | 26 (42.6) | 0.29 |
Adenocarcinoma | 31 (40.8) | 14 (35.9) | 17 (31.5) | 28 (45.9) | ||
Large cell | 4 (5.3) | 4 (10.3) | 3 (5.6) | 5 (8.2) | ||
Others | 2 (2.6) | 4 (10.3) | 4(7.4) | 2 (3.3) | ||
Histologic differentiation, N (%) | ||||||
Well diff. (grade I) | 11 (21.2) | 10 (33.3) | 0.43 | 11 (28.2) | 10 (23.3) | 0.78 |
Moderately diff. (grade II) | 14 (26.9) | 8 (26.7) | 11 (28.2) | 11 (25.6) | ||
Poorly/undiff. (grade III) | 27 (51.9) | 12 (40.0) | 17 (43.6) | 22 (51.2) | ||
Tumor stage, N (%) | ||||||
I | 33 (43.4) | 21 (55.3) | 0.27 | 29 (53.7) | 25 (41.7) | 0.26 |
II | 19 (25.0) | 8 (21.1) | 12 (22.2) | 15 (25.0) | ||
III | 18 (23.7) | 9 (23.7) | 9 (16.7) | 18 (30.0) | ||
IV | 6 (7.9) | 0 | 4 (7.4) | 2 (3.3) |
Plus-minus values are means ± SD.
P values were calculated after missing values were excluded.
Variable . | DICER1 mRNA level . | DROSHA mRNA level . | ||||
---|---|---|---|---|---|---|
Low ( N = 76) . | High ( N = 39) . | P value . | Low ( N = 54) . | High ( N = 61) . | P value . | |
Age (year) | 66.8±9.8 | 64.1±9.4 | 0.21 | 67.2±8.9 | 64.7±10.3 | 0.19 |
Tobacco, N (%) | ||||||
Non-smokers | 4 (9.1) | 3 (11.5) | 0.81 | 4 (11.8) | 3 (8.3) | 0.64 |
Former smokers | 25 (56.8) | 16 (61.5) | 18 (52.9) | 23 (63.9) | ||
Current smokers | 15 (34.1) | 7 (26.9) | 12 (35.3) | 10 (27.8) | ||
Histology, N (%) | ||||||
Squamous | 39 (51.3) | 17 (43.6) | 0.23 | 30 (55.6) | 26 (42.6) | 0.29 |
Adenocarcinoma | 31 (40.8) | 14 (35.9) | 17 (31.5) | 28 (45.9) | ||
Large cell | 4 (5.3) | 4 (10.3) | 3 (5.6) | 5 (8.2) | ||
Others | 2 (2.6) | 4 (10.3) | 4(7.4) | 2 (3.3) | ||
Histologic differentiation, N (%) | ||||||
Well diff. (grade I) | 11 (21.2) | 10 (33.3) | 0.43 | 11 (28.2) | 10 (23.3) | 0.78 |
Moderately diff. (grade II) | 14 (26.9) | 8 (26.7) | 11 (28.2) | 11 (25.6) | ||
Poorly/undiff. (grade III) | 27 (51.9) | 12 (40.0) | 17 (43.6) | 22 (51.2) | ||
Tumor stage, N (%) | ||||||
I | 33 (43.4) | 21 (55.3) | 0.27 | 29 (53.7) | 25 (41.7) | 0.26 |
II | 19 (25.0) | 8 (21.1) | 12 (22.2) | 15 (25.0) | ||
III | 18 (23.7) | 9 (23.7) | 9 (16.7) | 18 (30.0) | ||
IV | 6 (7.9) | 0 | 4 (7.4) | 2 (3.3) |
Variable . | DICER1 mRNA level . | DROSHA mRNA level . | ||||
---|---|---|---|---|---|---|
Low ( N = 76) . | High ( N = 39) . | P value . | Low ( N = 54) . | High ( N = 61) . | P value . | |
Age (year) | 66.8±9.8 | 64.1±9.4 | 0.21 | 67.2±8.9 | 64.7±10.3 | 0.19 |
Tobacco, N (%) | ||||||
Non-smokers | 4 (9.1) | 3 (11.5) | 0.81 | 4 (11.8) | 3 (8.3) | 0.64 |
Former smokers | 25 (56.8) | 16 (61.5) | 18 (52.9) | 23 (63.9) | ||
Current smokers | 15 (34.1) | 7 (26.9) | 12 (35.3) | 10 (27.8) | ||
Histology, N (%) | ||||||
Squamous | 39 (51.3) | 17 (43.6) | 0.23 | 30 (55.6) | 26 (42.6) | 0.29 |
Adenocarcinoma | 31 (40.8) | 14 (35.9) | 17 (31.5) | 28 (45.9) | ||
Large cell | 4 (5.3) | 4 (10.3) | 3 (5.6) | 5 (8.2) | ||
Others | 2 (2.6) | 4 (10.3) | 4(7.4) | 2 (3.3) | ||
Histologic differentiation, N (%) | ||||||
Well diff. (grade I) | 11 (21.2) | 10 (33.3) | 0.43 | 11 (28.2) | 10 (23.3) | 0.78 |
Moderately diff. (grade II) | 14 (26.9) | 8 (26.7) | 11 (28.2) | 11 (25.6) | ||
Poorly/undiff. (grade III) | 27 (51.9) | 12 (40.0) | 17 (43.6) | 22 (51.2) | ||
Tumor stage, N (%) | ||||||
I | 33 (43.4) | 21 (55.3) | 0.27 | 29 (53.7) | 25 (41.7) | 0.26 |
II | 19 (25.0) | 8 (21.1) | 12 (22.2) | 15 (25.0) | ||
III | 18 (23.7) | 9 (23.7) | 9 (16.7) | 18 (30.0) | ||
IV | 6 (7.9) | 0 | 4 (7.4) | 2 (3.3) |
Plus-minus values are means ± SD.
P values were calculated after missing values were excluded.
We examined whether the expression levels of DICER1 and DROSHA were associated with patient survival after surgery: Kaplan–Meier survival curves show that even though the median overall survival (OS) was substantially reduced among patients whose tumor had low level of DICER1 mRNA, it did not reach statistical significance ( P = 0.275, Figure 1A ). On the other hand, Kaplan–Meier curves demonstrated that the probability of survival was significantly lower for the group of patients with high levels of DROSHA mRNA (39.8 versus 154.2 months, P = 0.016, Figure 1B ). Comparison among other subgroups of patients is shown in Supplementary Figure 3 , available at Carcinogenesis Online.
We did not find any significant difference in the expression levels of DICER1 or DROSHA between SCC and ADC histology ( P = 0.870 and P = 0.133, respectively). Nevertheless, the median OS was significantly reduced among ADC patients with high levels of DROSHA expression (not reached versus 27.7 months, P = 0.011; Figure 1C ), and thus, risk of death was associated with high levels of DROSHA mRNA in ADC patients (1.72±0.21 versus 1.26±0.25, P = 0.024; 80% of dead ADC patients). However, these relationships were not found in SCC patients (data not shown).
Similar to the histology, there were no differences in DICER1 or DROSHA expression levels between high- and low-grade tumors ( P = 0.195 and P = 0.956, respectively). Nevertheless, the median OS was significantly reduced among grade III patients with high levels of DROSHA expression (not reached versus 29.1 months, P = 0.038; Figure 1D ). Neither were there differences between high- and low-stage tumors ( P = 0.186 for DICER1 and P = 0.402 for DROSHA) although the OS was significantly reduced among low-stage (I or II) patients with high levels of DROSHA expression ( P = 0.014; Figure 1E ).
To validate our findings, we used previously reported microarray data to compare expression levels of DICER1 and DROSHA with OS in two different cohorts of lung cancer patients. In the first validation cohort ( 16 ) and similar to our initial findings, increased survival was associated with high expression of DICER1 mRNA (not reached versus 31.9 months, P = 0.016) and low DROSHA mRNA (51.3 versus 28.9 months, P = 0.079; Figure 1F ). Also, in the second validation cohort ( 17 ), increased OS was associated with high expression of DICER1 mRNA ( P = 0.0001) and low expression of DROSHA mRNA ( P = 0.007; Figure 1G ). The latter validation cohort consisted of patients with very early ADC stage (I or II) and the median survival time was not reached.
In multivariate analyses (including age, sex, tumor stage, grade, histology and DICER1 and DROSHA mRNA levels), only tumor stage (III and IV versus others, P = 0.02) and DROSHA mRNA expression ( P = 0.04) were selected as independent prognostic variables ( Table 3 ). An increased DROSHA mRNA level was a significant predictive factor for poor prognosis [hazards ratio = 2.24; 95% CI: 1.01–4.95, P = 0.04], whereas DICER1 expression level did not show a significant association with survival (hazards ratio = 0.87; 95% CI: 0.40–1.91, P = 0.74).
Factor . | Hazard ratio for death (95% CI) . | P value . |
---|---|---|
High DROSHA expression | 2.24 (1.01–4.95) | 0.04 |
High DICER1 expression | 0.87 (0.40–1.91) | 0.74 |
Increased age | 1.03 (0.99–1.08) | 0.15 |
Male | 2.79 (0.58–13.35) | 0.20 |
Adenocarcinoma histology | 1.01 (0.22–4.68) | 0.99 |
Tumor stage III or IV | 2.34 (1.13–4.86) | 0.02 |
High-grade tumor | 1.29 (0.56–2.92) | 0.55 |
Factor . | Hazard ratio for death (95% CI) . | P value . |
---|---|---|
High DROSHA expression | 2.24 (1.01–4.95) | 0.04 |
High DICER1 expression | 0.87 (0.40–1.91) | 0.74 |
Increased age | 1.03 (0.99–1.08) | 0.15 |
Male | 2.79 (0.58–13.35) | 0.20 |
Adenocarcinoma histology | 1.01 (0.22–4.68) | 0.99 |
Tumor stage III or IV | 2.34 (1.13–4.86) | 0.02 |
High-grade tumor | 1.29 (0.56–2.92) | 0.55 |
Factor . | Hazard ratio for death (95% CI) . | P value . |
---|---|---|
High DROSHA expression | 2.24 (1.01–4.95) | 0.04 |
High DICER1 expression | 0.87 (0.40–1.91) | 0.74 |
Increased age | 1.03 (0.99–1.08) | 0.15 |
Male | 2.79 (0.58–13.35) | 0.20 |
Adenocarcinoma histology | 1.01 (0.22–4.68) | 0.99 |
Tumor stage III or IV | 2.34 (1.13–4.86) | 0.02 |
High-grade tumor | 1.29 (0.56–2.92) | 0.55 |
Factor . | Hazard ratio for death (95% CI) . | P value . |
---|---|---|
High DROSHA expression | 2.24 (1.01–4.95) | 0.04 |
High DICER1 expression | 0.87 (0.40–1.91) | 0.74 |
Increased age | 1.03 (0.99–1.08) | 0.15 |
Male | 2.79 (0.58–13.35) | 0.20 |
Adenocarcinoma histology | 1.01 (0.22–4.68) | 0.99 |
Tumor stage III or IV | 2.34 (1.13–4.86) | 0.02 |
High-grade tumor | 1.29 (0.56–2.92) | 0.55 |
miRNA expression profiling
Several miRNAs have recently been linked to tumorigenesis in many types of tumors ( 18 ). In view of this, we decided to study in samples from our cohort in which sufficient RNA was available ( N = 48), the expression of some of the most relevant lung cancer-related miRNA described in the literature. In particular, we studied hsa-let-7a, miR-200c, miR-16, miR-21 and miR-155 expression levels by qRT–PCR.
The overall tumor miRNA expression was downregulated compared with normal tissue ( Supplementary Table 2 , available at Carcinogenesis Online). The qRT–PCR analysis revealed that hsa-let-7a ( P = 0.005) and miR-16 ( P = 0.003) expression levels were significantly higher in SCC than in ADC ( Figure 2A ). The only miRNA with higher expression levels in ADC was miR-21 ( P = 0.148), and at least 3-fold upregulation of miR-21 expression was found in 45.8% of cases. There were no significant differences in miRNA expression levels depending on the tumor stage or tumor grade.
Given the different miRNA profiles by histology, we conducted survival analyses in the two histology groups separately. We did not find significant differences in neither OS nor disease-free survival according to the miRNA expression levels in both histologic subtypes. However, OS in our total population was higher in those patients with higher hsa-let-7a (40.91±4.67 versus 120.57±14.25 months, P = 0.067) and higher miR-200c (41.94±4.21 versus 108.75±9.84 months, P = 0.048) expression levels.
In our study, increased DROSHA expression was associated with poor prognosis in lung cancer patients. Curiously, despite a 1.3-fold increase in DROSHA expression in lung cancer samples, all studied miRNA expression levels were decreased in this subpopulation. Interestingly, when we divided our population according to DICER1 and DROSHA expression levels (low and high), the miRNA expression differences between the two histologic types occurred only when DICER1 or DROSHA levels were below benign epithelial lung tissues levels ( Figure 2B ).
We also calculated pairwise correlation values for the miRNAs and DICER1 and DROSHA: hsa-let-7a showed a high correlation value with miR-16 (ρ = 0.901), miR-200c (ρ = 0.696) and miR-155 (ρ = 0.420). miR-200c and miR-16 showed a Spearman’s correlation value of 0.626 ( Supplementary Table 3 , available at Carcinogenesis Online). There was no correlation between DICER1 and DROSHA expression levels (ρ = 0.161); however, there was an inverse correlation between DICER1 and hsa-let-7a (ρ = –0.335), miR-200 (ρ = –0.490) and miR-16 (ρ = –0.289). Also, DROSHA showed an inverse correlation value with hsa-let-7a (ρ = −0.417) and miR-16 (ρ = −0.446) ( Supplementary Table 3 , available at Carcinogenesis Online).
Discussion
The production of mature miRNA involves a cascade of events that are closely linked to the functions of DICER1 and DROSHA. Likewise, it is possible that deregulated miRNA expression, observed in several types of tumor ( 19 ), is secondary to defective RNA silencing machinery. In our study, we found that DICER1 and DROSHA expression levels are variable among NSCLC specimens, and play a significant role in OS, especially in DROSHA mRNA expression. Levels of gene and proteomic expression were examined by qRT–PCR and immunohistochemical, with high values of correlation.
We found that low DICER1 expression levels were associated with poor outcome although this was not statistically significant. Previously, downregulation or lower expression of DICER1 was found to be significantly associated with NSCLC survival ( 20 ), advanced tumor stages of ovarian carcinomas ( 21 ) or epithelial skin cancer ( 22 ), whereas upregulation of DICER1 has been described for prostate ADCs and lung carcinomas ( 23 , 24) . Chiosea et al . ( 24 ) showed that in lung ADC, DICER1 was downregulated in areas of invasion and in cases of advanced disease.
Similar to Sand et al. ( 22 ), we found a highly significant upregulation of DROSHA expression in NSCLC samples. This result supports the hypothesis that patients with lung cancer may show an alteration in DROSHA expression levels as DROSHA was overexpressed in both the SCC and ADC histologic groups. Moreover, we show an association between high DROSHA expression and poor clinical outcome in both high-grade tumors and early stage tumors, suggesting that DROSHA exerts its influence most profoundly within these subgroups of tumors.
High DICER1 and high DROSHA expression levels were correlated with poor prognostic factors in prostate cancer and esophageal carcinoma ( 23 , 25) . There are several plausible explanations for the divergent expression patterns of DICER1 and DROSHA among different solid tumors and how they relate to clinical and pathologic variables. Direct interactions with other components of the RNA-interference cascade could result in compensatory alterations of DICER1 or DROSHA expression in the presence of mutated cofactors, such as genes for DGCR8, XPO5 and AGO2 ( 24 , 26–28 ).
The findings described in this study indicated that the expression levels of DROSHA appeared to have a significant impact ( P = 0.04) on the postoperative survival of NSCLC patients, independent of disease stage ( P = 0.02). Although there are some differences in the statistical significance levels, we believe that our study has been validated in two independent cohorts of patients because Kaplan–Meier survival analysis of these validation cohorts showed a significant survival disadvantage for patients whose tumors expressed low DICER1 and high DROSHA mRNA levels. Thus, increased expression of DROSHA appears to be clinically useful for the prognosis of lung cancer patients. In addition, other factors may underlie the potential biological effects of increased DROSHA expression in lung cancer cells.
Several large-scale miRNA expression studies have indicated that global miRNA levels are reduced in tumor samples in comparison with normal tissues ( 29 ). Our results are consistent with this, and the five miRNAs studied here show lower expression levels in our set of tumor samples than in normal tissues. Interestingly, the rate of reduction was much more pronounced in ADC than in SCC tumors, with statistically significant differences in the expression levels of hsa-let-7a and miR-16 between both histologic types. This pattern suggests that these miRNAs may be more relevant for ADC tumorigenesis because they exhibit poor expression in this tumor type. Even more interesting is that these differences of expression between the two histologic types only hold when DICER1 or DROSHA expression levels are below normal tissues values. That is, different miRNA expression profiles between ADC and SCC subtypes are evident mainly when the miRNA machinery is defective. On the other hand, we must take into account the inverse correlation between DICER1 and hsa-let-7a, miR-200 and miR-16 expression levels and also between DROSHA and expression levels of hsa-let-7a and miR-16. A possible explanation could be that these miRNAs suppress, somewhat, the expression of DICER1 and DROSHA, but this assertion must be tested further and new experiments should be designed. We must consider that an examination of any individual or small series of miRNAs may not reflect the complexity of changes in miRNA expression participating in clinical tumor biology ( 9 ). In addition, miRNA could have varying regulatory effects independent of alterations in the RNA-processing machinery ( 30–32 ). Posttranscriptional regulation of miRNA processing also plays an important role in the regulation of miRNA expression ( 3 , 33) , but a connection of these modifications to tumorigenesis has not yet been definitive. Moreover, one-half of the miRNAs have been aligned to genomic fragile sites or regions associated with cancer. Finally, several studies suggest that miRNA function is determined by the specific cellular context ( 28 ).
More important, miRNA overexpression could result in downregulation of tumor suppressor genes, whereas their underexpression could lead to oncogene upregulation. For this reason, some miRNAs have been shown to function as oncogenes, whereas others function as tumor suppressors: miR-21 and miR-155 are examples of miRNA described as oncogenes, and hsa-let-7a and miR-200c as tumor suppressor miRNA ( 34 ). In our set of patients, OS was higher in patients with hsa-let-7a or miR-200c overexpressed and these results are in accordance with the function as tumor suppressors of these miRNAs.
We found higher miR-16 expression levels in SCC compared with ADC samples, and this difference was highly significant statistically. Similarly, Bandi et al. ( 35 ) detected overexpression of miR-16 in 26% of NSCLC tumors. Tan et al. ( 36 ) described a 5-microRNA classifier that can distinguish cancerous SCC lesions from adjacent normal tissues; however, none of the miRNA described match those studied here. They also showed that hsa-miR-31 directly targets the DICER1 3′-untranslated regions and repressed their expression. Moreover, high expression of hsa-miR-31 was associated with poor survival, which is consistent with our data, given that low levels of DICER1 correlates with shortened postoperative survival. Voortman et al. ( 37 ) studied a panel of miRNA by qRT–PCR in formalin-fixed and paraffin-embedded tumor specimens from 639 International Adjuvant Lung Cancer Trial (IALT) patients. No significant association was found between any of the tested miRNAs and survival, and expression patterns examined were neither predictive nor prognostic in this large patient cohort.
In conclusion, this study supports the value of the expression profiling of the miRNA-processing machinery components in the prognosis of NSCLC patients, especially DROSHA expression levels. Whether alterations in the miRNA machinery are associated with the risk of lung cancer or whether DICER1 and DROSHA expression levels are concomitantly altered due to lung cancer remains unclear and needs to be addressed further in future studies. In recent years, no biomarker has generated as much interest as miRNAs, which have been considered for a variety of purposes, and we have proved that differential expression of some miRNA such as hsa-let-7a and miR-16 are reliable tools in the subclassification of NSCLC. Thus, miRNAs could become powerful therapeutic tools in the near future. However, the validation of these findings in multiple cohorts and the testing of their applicability to different ethnic populations are also required.
Supplementary material
Supplementary Figures 1–3 and Supplementary Data can be found at Supplementary Data
Funding
Instituto de Salud Carlos III, Spain (PI042641, PI112688 to L.P.-A., PI042535 to R.G.-C., PI041091 to J.P.); Ministerio de Sanidad y Política Social, Spain (TRA-151 to A.A.-L. and C.P.); Fundación Mutua Madrileña, Spain (2010/018 to V.D.-G.).
Acknowledgement
We thank Juan Carlos Rubio of Genomic Department (Instituto de Investigación Hospital 12 de Octubre, Madrid) for his invaluable technical assistance.
Conflict of Interest Statement: None declared.
Abbreviations:
- ADC
adenocarcinoma
- LCC
large cell carcinoma
- miRNA
microRNA
- NSCLC
non-small cell lung cancer
- OS
overall survival
- SCC
squamous cell carcinoma.
References