• Open Access

CpG island methylation of microRNAs is associated with tumor size and recurrence of non-small-cell lung cancer


To whom correspondence should be addressed. E-mail: dtakai-ind@umin.ac.jp


We investigated whether the CpG island methylation of certain microRNAs was associated with the clinicopathological features and the prognosis of non-small-cell lung cancer. The methylation of mir-152, -9-3, -124-1, -124-2, and -124-3 was analyzed in 96 non-small-cell lung cancer specimens using a combined bisulfite restriction analysis. The median observation period was 49.5 months. The methylation of mir-9-3, -124-2, and -124-3 was individually associated with an advanced T factor independent of age, sex, and smoking habit. Moreover, the methylation of multiple microRNA loci was associated with a poorer progression-free survival in a univariate analysis. Our result enlightens the accumulation of aberrant DNA methylation which occurs in concordance with the tumor progression. (Cancer Sci 2011; 102: 2126–2131)

Lung cancer is the leading cause of cancer-related mortality in the world,(1) and non-small-cell lung cancer (NSCLC) is the most common type. Surgical resection remains the only curative treatment for NSCLC. Identifying factors that are associated with aggressive disease may lead to the development of novel biomarkers and the identification of therapeutic targets that can help reduce the burden of this disease.

MicroRNAs (miRNAs) are a class of small non-coding RNAs that negatively regulate target gene expression by accelerating the degradation of mRNA and translational inhibition, with potentially hundreds of target mRNAs.(2) They influence a variety of cellular functions including proliferation, differentiation, and apoptosis.(3) Specific miRNAs can behave as either tumor suppressor genes or oncogenes, depending on the tissue type and the presence of specific targets.(4,5)

More than 100 species of known miRNAs are embedded within or near the CpG islands of the human genome and are potentially subject to control by epigenetic alterations such as DNA methylation and histone modification. Systematic assessments of miRNA expression and epigenetic modifications among cell lines and primary tumor specimens have revealed the existence of the epigenetic regulation of miRNAs in multiple tumor types.(6–9)

The goals of this study were to explore possible relationships between the methylation profiles of miRNAs and the clinicopathological characteristics of NSCLC patients and to identify new specific methylation markers capable of detecting advanced pathological features.

In the present study, the methylation status of five miRNA loci within five separate CpG islands was determined in 96 NSCLC tissue specimens. The choice of the miRNAs was based on our previous study(9) as well as other published reports.(10,11) Our data show that the CpG island methylation of miRNAs is common in NSCLC, and that the methylation of multiple miRNA loci is associated with an advanced T status as well as the progression-free survival (PFS) of patients with NSCLC.

Materials and Methods

Patients.  We collected cancer tissues and normal lung tissues from NSCLC patients who underwent surgical resection at the University of Tokyo Hospital (Tokyo, Japan) between June 2005 and September 2007 after receiving approval from the Institutional Ethics Review Committee and after obtaining informed consent from all patients. The diagnoses were based on pathological evidence and were classified according to the TNM classification criteria.(12) To circumvent statistical disruption arising from the heterogeneity of significantly advanced diseases, patients whose tumors were pathologically confirmed as T3 or T4 were excluded.

Combined bisulfite restriction analysis (COBRA).  DNA was extracted by the standard proteinase K and phenol method.(13) One microgram of genomic DNA was bisulfite treated and purified, according to the protocol described previously.(14,15) The bisulfite-treated DNA was amplified by PCR using AmpliTaq Gold 360 Master Mix (Applied Biosystems, Foster City, CA, USA). The PCR product was treated with ExoSAP-IT (US Biochemical, Cleveland, OH, USA) to improve the visibility of the band.(16) The product was digested with the restriction enzymes BstUI or BssHII (New England Biolabs, Ipswich, MA, USA). To dissociate the tightly bound enzymes from the DNA, the digested product was purified using phenol–chloroform extraction when necessary. The DNA was visualized by 2% agarose gel electrophoresis with the addition of a loading buffer containing SDS (final concentration, 0.1%) and was identified as methylation-positive if an appropriately sized band was present.

Statistics.  A statistical analysis was carried out using JMP 7 software (SAS Institute, Cary, NC, USA). The difference in frequencies was assessed using a χ2-test. The risks of advanced pathological features were analyzed using a multivariate logistic regression analysis. Progression-free survival curves were calculated using the Kaplan–Meier method and compared using a log–rank test. A P-value of <0.05 was considered significant.


Characteristics of the patients.  A total of 96 cases were analyzed. The baseline characteristics of the patient cohort are shown in Table 1. The median follow-up period was 49.5 months. A total of 30 patients developed recurrent disease during the observation period. The most frequent recurrent site was the brain (46.7%), followed by the lymph nodes (23.3%), bone (13.3%), and the lungs (10.0%).

Table 1.   Baseline characteristics of 96 patients with non-small-cell lung carcinoma
CharacteristicsPatient cohort (n = 96)
  1. EGFR, epidermal growth factor receptor; n, number of cases.

Age, years
 Interquartile range59–72
Sex, n (%)
 Male60 (62.5)
 Female36 (37.5)
Smoking habit, n (%)
 Never33 (34.4)
 Ever63 (65.6)
Histological type, n (%)
 Adenocarcinoma78 (81.3)
 Squamous cell carcinoma17 (17.7)
 Adenosquamous carcinoma1 (1.0)
Stage, n (%)
 IA41 (42.7)
 IB26 (27.1)
 IIA5 (5.2)
 IIB7 (7.3)
 IIIA17 (17.7)
T factor, n (%)
 T154 (56.3)
 T242 (43.8)
N factor, n (%)
 N067 (69.8)
 N112 (12.5)
 N217 (17.7)
Lymphatic invasion, n (%)
 Negative72 (75.0)
 Positive24 (25.0)
Vascular invasion, n (%)
 Negative58 (60.4)
 Positive38 (39.6)
EGFR mutation, n (%)
 Negative65 (67.7)
 Positive27 (28.1)
 Unknown4 (4.2)

miRNA loci analyzed in COBRA.  We focused on miR-152, the methylation of which has never been documented in lung cancer. In addition, we included miR-9 and miR-124, the methylation of which has been previously reported in lung cancer.(10,11) In the human genome, miR-9 and miR-124 are each located at three distinct loci: mir-9-1, -9-2, -9-3; and mir-124-1, -124-2, and -124-3. MiR-152 is located at a single locus: mir-152. All of these loci are embedded within CpG islands, except for mir-9-2. We first carried out COBRA using tumor and adjacent normal tissues from eight NSCLC cases (Fig. S1). The locus mir-9-1 was excluded from the study because no tumor-specific methylation was observed. The gene maps, primer sequences, annealing temperatures, and restriction enzymes used for COBRA are shown in Figure 1 and Table 2.

Figure 1.

 Gene maps around the CpG islands in relation to mir-152, -9-1, -9-3, -124-1, -124-2, and -124-3. For combined bisulfite restriction analysis, PCR products were restricted using BstUI (vertical downward arrows) or BssHII (closed arrowheads). Gray boxes, CpG islands; horizontal bidirectional arrows, PCR products; stars, both fragments were designed to be equal in size to emphasize the visualization of the restricted PCR products; vertical tick marks, CpG sites.

Table 2.   Primer sequences, PCR characteristics, and restriction enzymes for combined bisulfite restriction analysis
miRNALocusPrimers from 5′ to 3′Product size (bp)Ta (°C)Restriction enzyme
  1. †Phenol–chloroform extraction was needed after the digestion for clear vision of bands. ‡Two kinds of reverse primers, which changed a part of R into G or A, were mixed and used to amplify the bisulfite-modified genomic DNA. Ta, annealing temperature.


Relationships between CpG island methylation of the miRNAs and the clinicopathological features of 96 NSCLC patients.  The methylation profiles of the five miRNA loci were obtained from 96 NSCLC specimens (Figs 2). The prevalence of methylation among the 96 cases was 25.0% (24/96) for mir-152, 64.6% (62/96) for mir-9-1, 27.1% (26/96) for mir-124-1, 50% (48/96) for mir-124-2, and 50% (48/96) for mir-124-3. The frequency of methylation is shown according to the clinical features in Table 3. The methylation of mir-124-2 and mir-124-3 was similarly and preferentially observed in older patients or smokers. No significant correlations were seen between the methylation status of mir-152 or mir-124-1 and the clinicopathological features, whereas the methylation statuses of mir-9-3, -124-2, and -124-3 were correlated with the T classification.

Figure 2.

 Representative combined bisulfite restriction analysis results of non-small-cell lung carcinoma tissue samples. Arrows, unmethylated alleles; arrowheads, methylated alleles; L, DNA size markers; numbers on top, tissue samples; stars, samples with significant restricted fragments from methylated alleles.

Table 3.   Relationships between CpG island methylation of miRNAs and clinicopathological features in 96 patients with non-small-cell lung carcinoma
CharacteristicsMethylated locus
n (%)Pn (%)Pn (%)Pn (%)Pn (%)P
  1. *P < 0.05. EGFR, epidermal growth factor receptor; n, number of cases.

Overall (n = 96)24 (25.0)62 (64.6)26 (27.1)48 (50.0)48 (50.0)
Age (years)
 ≤65 (n = 47)12 (25.5)0.906226 (55.3)0.063112 (25.5)0.737616 (34.0)0.0022*18 (38.3)0.0247*
 >65 (n = 49)12 (24.5)36 (73.5)14 (28.6)32 (65.3)30 (61.2)
 Male (n = 60)18 (30.0)0.144140 (66.7)0.581617 (28.3)0.72234 (56.7)0.091733 (55.0)0.2059
 Female (n = 36)6 (16.7)22 (61.1)9 (25.0)14 (38.9)15 (41.7)
Smoking habit
 Never (n = 33)6 (18.2)0.264220 (60.6)0.55548 (24.2)0.650311 (33.3)0.0181*11 (33.3)0.0181*
 Ever (n = 63)16 (25.4)42 (66.7)18 (28.6)37 (58.7)37 (58.7)
Histological type
 Adenocarcinoma (n = 78)22 (28.2)0.225352 (66.7)0.374324 (30.8)0.140236 (46.2)0.116736 (46.2)0.1167
 Others (n = 18)2 (11.1)10 (55.6)2 (11.1)12 (66.7)12 (66.7)
 IA, IB (n = 67)20 (29.9)0.095342 (62.7)0.554819 (28.4)0.669231 (46.3)0.266433 (49.3)0.8241
 IIA–IIIA (n = 29)4 (13.8)20 (69.0)7 (24.1)17 (58.6)15 (51.7)
T factor
 T1 (n = 54)11 (20.4)0.234929 (53.7)0.0115*12 (22.2)0.224320 (37.0)0.0040*19 (35.2)0.0010*
 T2 (n = 42)13 (31.0)33 (78.6)14 (33.3)28 (66.7)29 (69.0)
N factor
 Negative (n = 67)20 (29.9)0.095342 (62.7)0.554819 (28.4)0.669231 (46.3)0.266433 (49.3)0.8241
 Positive (n = 29)4 (13.8)20 (69.0)7 (24.1)17 (58.6)15 (51.7)
Lymphatic invasion
 Negative (n = 72)18 (25.0)1.000044 (61.1)0.217919 (26.4)0.790934 (47.2)0.345833 (45.8)0.1573
 Positive (n = 24)6 (25.0)18 (75.0)7 (29.2)14 (58.3)15 (62.5)
Vascular invasion
 Negative (n = 58)15 (25.9)0.809636 (62.1)0.524515 (25.9)0.739427 (46.6)0.403828 (48.3)0.6764
 Positive (n = 38)9 (23.7)26 (68.4)11 (28.9)21 (55.3)20 (52.6)
EGFR mutation
 Negative (n = 65)15 (23.1)0.307742 (64.6)0.594917 (26.2)0.732938 (58.5)0.0118*36 (55.4)0.2007
 Positive (n = 27)9 (33.3)19 (70.4)8 (29.6)8 (29.6)11 (40.7)

Risk of advanced pathological features according to CpG island methylation status of five miRNAs.  To estimate the risk of advanced pathological features among the methylation statuses of the miRNA loci, compared with that among the unmethylated statuses, a multivariate logistic regression analysis was carried out (Table 4). The odds ratio for each miRNA locus was adjusted for age, sex, and smoking habit. The methylation status of mir-9-3, -124-2, and -124-3 were significantly associated with the T factor. When mir-9-3, -124-2, or -124-3 was methylated, the adjusted odds ratio for a T2 status was 3.11 (95% CI, 1.26–8.22; = 0.0130), 3.39 (95% CI, 1.39–8.67; P = 0.0066), or 4.17 (95% CI, 1.69–10.89; P = 0.0017), respectively. The status of at least two methylated miRNA loci was also examined to determine the association with advanced pathological features. When the methylation of two or more miRNA loci was observed, the estimated odds ratio for a T2 status was 4.75 (95% CI, 1.83–13.59; P = 0.0011).

Table 4.   Multivariate odds ratios (OR) and 95% confidence intervals (CI) for the estimated risk of advanced pathological findings in non-small-cell lung cancer specimens according to miRNA methylation status
CategoryT factor (T2 vs T1)N factor (Positive)Lymphatic invasion (Positive)Vascular invasion (Positive)
OR† (95%CI)POR† (95%CI)POR† (95%CI)POR† (95%CI)P
  1. *P < 0.05. †Odds ratios are reported on the basis of a multivariate logistic regression model adjusted for age, sex, and smoking habit.

 Methylated1.76 (0.67–4.68)0.24280.35 (0.09–1.09)0.07191.04 (0.32–3.02)0.94320.76 (0.27–1.99)0.5815
 Methylated3.11 (1.26–8.22)0.0130*1.36 (0.53–3.72)0.52121.77 (0.64–5.45)0.27661.38 (0.57–3.48)0.4717
 Methylated1.69 (0.67–4.29)0.25910.80 (0.27–2.18)0.68011.08 (0.36–2.97)0.87761.18 (0.45–3.05)0.7215
 Methylated3.39 (1.39–8.67)0.0066*1.98 (0.76–5.47)0.16321.43 (0.53–3.94)0.47661.37 (0.56–3.42)0.4815
 Methylated4.17 (1.69–10.89)0.0017*1.25 (0.48–3.33)0.63711.84 (0.68–5.21)0.22901.12 (0.45–2.76)0.7995
0 or 1 methylated locus1.
2 or more methylated loci4.75 (1.83–13.59)0.0011*0.99 (0.38–2.68)0.99941.98 (0.69–6.27)0.20211.59 (0.63–4.11)0.3222

Kaplan–Meier estimates.  The 5-year PFS of the cohort was 67%. In a univariate analysis of the known clinicopathological risk factors, the T classification was the most significant prognostic factor. Patients with a T2 status had a significantly lower PFS (< 0.0001), compared with those with T1 disease (Fig. 3A). The presence of at least two methylated miRNA loci, which was shown to be a strong predictor of a T2 status, was also a significant predictor of a poor PFS in a univariate analysis (= 0.0177, Fig. 3B).

Figure 3.

 (A) Kaplan–Meier analysis of progression-free survival among 96 non-small-cell lung carcinoma patients according to T classification. (B) Kaplan–Meier analysis of progression-free survival among 96 non-small-cell lung carcinoma patients, according to the number of methylated loci.


Our study showed that the CpG island methylation of mir-152 was common in NSCLC clinical specimens. The prevalence was 25%. However, no correlation was seen between the methylation status of mir-152 and clinicopathological features. The methylation of mir-152 was reportedly associated with an advanced tumor stage and a poorer PFS in patients with bladder cancer.(17) The regulation of miRNAs is believed to be highly tissue specific;(18) thus, the role of miR-152, if any, may not be dominant in lung cancer.

Our results showed that the methylated status of mir-9-3 was a significant risk factor for an advanced T status in a multivariate regression analysis. In contrast, no correlation was seen between the methylation of mir-9-3 and pathological parameters other than the T classification. The methylation of mir-9-3 has been reported in a wide range of primary tumors, such as colorectal, head and neck, breast, and lung cancers, and melanoma.(11,19) Lujambio et al.(11) reported the frequency of methylation in lung cancer to be 53%, and the methylation of mir-9-3 was associated with lymph node metastasis. Although we observed a similar proportion of methylated mir-9-3, no correlation was seen between the methylation of mir-9-3 and lymph node metastasis in our cohort. This discrepancy may be explained by inconsistencies between the patient cohorts, as the characteristics of the patients were not described in the previous report. Other possible reasons could include the methods used to detect methylation (i.e. methylation-specific PCR versus COBRA), or the precise CpG site targeted for analysis by these methods.

One remarkable finding regarding miR-124 was that mir-124-2 and mir-124-3 showed similar methylation profiles that were distinct from that of mir-124-1 (Table 3, Fig. S2), even though mir-124-2 and mir-124-3 are on separate chromosomes (8q12 and 20q13, respectively). The methylation of miR-124 has been reported in various tumor types including colon, breast, lung, and hepatocellular carcinomas, leukemias, and lymphomas.(10,20,21) However, whether the methylation of miR-124 occurs independently or coordinately remains uncertain.

The methylation of the promoter regions of certain protein-coding genes is reportedly associated with poor clinical outcomes in lung cancer.(22–24) A subset of the CpG island methylator phenotype (CIMP), originally established in colorectal carcinoma,(25,26) is characterized by the synchronous hypermethylation of multiple promoter CpG island regions and peculiar clinicopathological findings. Although several reports have indicated that CIMP is also distinguishable in NSCLC and is associated with a poor prognosis,(27,28) we have failed to distinguish the phenotype in the previous study regarding the methylation of the promoter regions of numerous protein-coding genes.(29) In the present study, we analyzed whether the concordant hypermethylation of multiple miRNA loci results in characteristics similar to those of CIMP. The methylated status of two or more miRNA loci was found to be associated with a significantly higher risk for advanced T status independent of age, sex, and smoking habit. Moreover, the methylation of multiple miRNA loci was associated with a poorer PFS.

Whether the aberrant methylation that occurs in cancer is a cause or a consequence is a difficult question.(30) The tumor size of NSCLC can be assumed to reflect the number of cancer cell divisions that have occurred since the initiation of carcinogenesis. However, the pathological findings of lymph node metastases or vessel invasions may be regarded as traits that have acquired a certain aggressive character at some point in time. Our result that only the T factor was associated with miRNA methylation suggests that the methylation of individual miRNA loci did not trigger any drastic phenotypic change in the NSCLCs and that what we observed was the accumulation of DNA methylation events during tumor progression. Indeed, according to the neutral theory of molecular evolution established in population biology,(31) it is most natural to suppose that the great majority of somatic mutations or aberrant methylations observed in cancer are neutral with regard to phenotypic alterations. Alterations that truly disrupted the cell cycle and affected the cell phenotype would be lethal for those cells, and such cells would probably not be visible in the tissue specimens.

In the present study, we did not carry out an expression analysis of miRNAs. Degrading of RNAs is inevitable during the sample preparation. Moreover, detecting suppression of aberrantly methylated genes is vulnerable to minimal contamination of highly expressed transcripts in the normal tissues. In this concern, detecting the accumulation of DNA methylation in the promoter region is one of the simplest and most consistent approaches to looking at the natural course of carcinogenesis, and our concept of focusing on a small number of examined loci among a small miRNA population (less than one-tenth of the protein coding genes) has the advantages of being cost-effective.

Although our data showed the impact of miRNA methylation on PFS, the T classification was an outstandingly strong determiner for PFS. Obtaining a greater number of study subjects and redesigning the study into a comparison within a single T status would be essential to assess the capability of miRNA methylation as a potential biomarker for cancer prognosis.

In summary, we analyzed the CpG island methylation statuses of mi-152, -9-3, -124-1, -124-2, and -124-3 in NSCLC. We found that the methylation of multiple miRNA loci was associated with an advanced T status as well as a poor PFS in a univariate analysis. Our result enlightens the accumulation of aberrant DNA methylation that occurs in concordance with the tumor progression.


This work was supported by grants from the Smoking Research Foundation, the Takeda Science Foundation, and the Japan Society for the Promotion of Science, Grants-in-Aid for Scientific Research (KAKENHI) (20659129, 22790749).

Disclosure Statement

Daiya Takai is supported by research grants from the Smoking Research Foundation and the Takeda Science Foundation.