Aberrant promoter methylation of p16INK4a, RARB2 and SEMA3B in bronchial aspirates from patients with suspected lung cancer

Authors


Abstract

Aberrant promoter methylation of normally unmethylated CpG-islands offers a promising tool for the development of molecular biomarkers. We investigated bronchial aspirates of patients admitted for suspected lung cancer with regard to the prevalence of aberrant methylation of potential marker genes. Applying quantitative methylation specific PCR (QMSP) we analyzed bronchial aspirates from 75 patients with primary lung cancer and 64 bronchial aspirates of patients diagnosed with benign lung disease for promoter methylation of 3 candidate marker genes (p16INK4a, RARB2 and SEMA3B). Hypermethylation of p16INK4a detected 18/75 (24%) cases with primary lung cancer and was present predominantly in squamous cell carcinomas (14/25; 56%). RARB2 QMSP at an assay threshold greater than 30 was found in 42/75 (56%) patients with lung cancer without relation to histological subtype. Patients with benign lung disease showed methylation of p16INK4a and a RARB2 QMSP at an assay threshold greater than 30 in 0/64 (0%) and 8/64 (13%) cases, respectively. Combining the 2 methylation markers, p16INK4a and RARB2, yielded a sensitivity of 69% and a specificity of 87% for the diagnosis of pulmonary malignancy. In contrast, SEMA3B displayed frequent promoter methylation (around 90%) both in bronchial aspirates of tumor and nontumor cases and thus was not suited as a biomarker. The results of this study indicate that QMSP analysis of p16INK4a and RARB2 may aid the diagnosis of primary lung cancer in bronchial aspirates. In particular, detection of p16INK4a methylation by QMSP may serve as a highly specific marker of pulmonary squamous cell carcinoma. © 2005 Wiley-Liss, Inc.

More than 1.2 million new cases of lung cancer are diagnosed worldwide every year.1 The overall survival rate is less than 11%, making lung cancer the leading cause of cancer-related deaths in the industrialized world.1 Since there is a strong inverse correlation between prognosis and stage of the disease, efforts aimed at early identification in lung cancer are of major importance.

Aberrant promoter methylation of normally unmethylated CpG-islands is an important mechanism of silencing tumor suppressor and tumor-related genes. Methylation results in transcriptional repression and serves as an alternative way of inactivating gene function.2, 3 The use of aberrant CpG-island methylation in promoter regions as a molecular marker system may offer a powerful approach for early detection of lung cancer.4, 5, 6 Recently, the development of fluorogenic real-time quantitative methylation specific PCR (QMSP) by Eads et al.7 has greatly improved sensitivity and specificity as well as throughput of methylation analyses. In addition, the QMSP assay allows the introduction of thresholds of variable stringency.8

The current study focuses on three genes,p16INK4a, RARB2 and SEMA3B, which have been reported to show frequent aberrant promoter methylation in lung cancer.9, 10, 11, 12, 13, 14, 15 The p16INK4a tumor suppressor gene plays a key role in cell cycle regulation and senescence. The RARB2 gene is thought to limit cell growth by inducing differentiation or apoptosis. SEMA3B belongs to the family of semaphorins, which contribute to axonal pathfinding during neural development, but also appear to more generally mediate cell-cell interactions.16 Recently, SEMA3B located at 3p21.3 was linked to the development of lung cancer and shown to be frequently inactivated by epigenetic changes.10, 11 Promoter hypermethylation of each of these 3 genes was found to be strongly correlated with a lack of gene expression, which could be restored by treatment of cell lines with 5-aza-2′-deoxycytidine.10, 14, 17 Thus, p16INK4a, RARB2 and SEMA3B are potential biomarkers for detection of lung cancer. In the current retrospective case-control study, we employed QMSP to examine bronchial aspirates of patients admitted for suspected lung cancer with regard to the prevalence of aberrant p16INK4a, RARB2 and SEMA3B promoter methylation.

Abbreviations:

AC, adenocarcinoma; CpG, cytosine-phosphorothiolated guanine; MYOD1, myoblast determination protein 1; NSCLC, non small cell lung cancer; p16INK4a, cyclin-dependent kinase inhibitor 2A; QMSP, quantitative methylation specific PCR; RARB2, retinoic acid receptor β-2; SCC, squamous cell carcinoma; SCLC, small cell lung cancer; SEMA3B, semaphorin 3B; UICC, International Union Against Cancer.

Material and methods

Clinical samples

Bronchial aspirates (bronchial washings, bronchoalveolar lavage fluids) were collected at the Institute for Cytopathology (Duesseldorf, Germany) between July 2001 and October 2002. A total number of 75 specimens from cancer patients (25 squamous cell carcinoma [SCC], 25 adenocarcinoma [AC], 25 small cell lung cancer [SCLC]) was investigated. Each histologic subtype was represented by a cohort of subsequent cases with highly suspicious (n = 4; 3 SCC, 1 SCLC) or positive (n = 71) cytology and sufficient clinical data for staging. During the ongoing study, 2 cases with negative cytology originally assigned to the control group turned out to have lung cancer and subsequently were classed with the tumor group. With respect to the first case, simultaneous histology was negative. An SCC was diagnosed 1 month later elsewhere. In the second case, information about a positive histology (SCLC) was delayed. Finally, the control group comprised 64 consecutive specimens from patients with benign lung disease (chronic obstructive bronchitis, n = 30; pneumonia, n = 15; tuberculosis, n = 7; pulmonary embolism, n = 5; lung fibrosis, n = 2; sarcoidosis n = 1). Four patients with benign lung disease underwent bronchoscopy twice within an interval of 7 to 21 days, therefore 2 bronchial aspirates were available for each. Clinicopathological characteristics of patients are compiled in Table I. The patients represent a subset of a previously described cohort, which was investigated for APC promoter 1A hypermethylation.18 Information regarding smoking habits, occupational exposure to asbestos and clinical staging according to the UICC TNM Classification of Malignant Tumours19 was obtained from chart review.

Table I. Characteristics of the Patient Population
 Controls (n = 60)Cases(n = 75)
  • 1

    Median (range).

  • 2

    Mean ± SD.

  • 3

    Smoking characteristics and stage data are missing for 10 and 5 patients, respectively.

Age167 (20–82)64 (26–84)
Sex
 Female2420
 Male3655
Pack-years2335 ± 2046 ± 27
Smoking status3
 Ex-smoker3027
 Current smoker2544
Stage3
 I 5
 II 1
 III 14
 IV 25
 Limited disease 5
 Extensive disease 20
Histology
 NSCLC
  Adenocarcinoma 25
  Squamous cell carcinoma 25
 SCLC 25

Bronchial aspirates were fixed in Saccomanno fixative (50% ethanol, 2% polyethylene glycol 1.500, 60 mg/l rifampicin) immediately after bronchoscopy. For routine diagnostic purposes an aliquot from each sample was used to prepare 4 conventional Papanicolaou stained smears.20 Residual material was stored for subsequent investigations.

Quantification of tumor cells

The number of tumor cells present in a bronchial aspirate was estimated semiquantitatively as described previously.18 Tumor positive cases were subdivided into 2 distinctive groups: (i) few tumor cells/clusters, i.e., tumor cells or small clusters of tumor cells were present in up to 3 loci; (ii) multiple tumor cells/clusters, i.e., tumor cells or small clusters of tumor cells were present in at least 4 loci or presence of large tumor cell clusters.

Cell lines

The human cell lines T24 (ATCC no.: HTB-4, bladder carcinoma) and Wi-38 (ATCC no.: CCL-75, fibroblast, lung) were grown and harvested according to the instructions of the American Type Culture Collection (Manassas, VA).

Bisulfite treatment and real-time QMSP

The genomic DNA of cell lines and cells in bronchial aspirates was isolated using the Puregene DNA Isolation Kit (Gentra Systems, Minneapolis, MN). DNA extracted from Wi-38 cells was treated with SssI methyltransferase (New England Biolabs, Frankfurt, Germany) according to the vendor's instructions. A total of 1 μg of DNA per sample was modified by sodium bisulfite treatment according to Herman et al.21 Promoter methylation analysis was performed by fluorescence-based, real-time quantitative PCR using a LightCycler (Roche Diagnostics GmbH, Mannheim, Germany) as previously described.7, 18, 22 Primers, TaqMan probes and annealing temperatures are given in Table II. MYOD1 was used as an internal reference to control for input DNA.23 Each set of PCR reaction was run with 3 positive controls and 1 negative control: for RARB2 DNA extracted from Wi-38 cells and treated with SssI methyltransferase and for p16INK4a and SEMA3B DNA isolated from the cell line T24 served as positive control. A negative control was provided by a sample without DNA. Results were confirmed in 30% of arbitrarily chosen samples by repeated QMSP assays.

Table II. Sequence of PCR Primers and Probes for QMSP
GenePrimer sequencesAnnealing temperatureCpGsSize (bp)
p16INK4aforward: 5′-TGGAGTTTTCGGTTGATTGGTT-3′58°C770
reverse: 5′-AACAACGCCCGCACCTCCT-3′
probe: 5′-6FAM-ACCCGACCCCGAACCGCG-TAMRA-3′
RARB2forward: 5′-CGAGAACGCGAGCGATTC-3′60°C11146
reverse: 5′-GACCAATCCAACCGAAACGA-3′
probe: 5′-6FAM-CCTTCCGAATACGTTCCGAATCCTAC-TAMRA-3′
SEMA3Bforward: 5′-GGTTGTTTGTTCGGTTATTCGATT-3′58°C1083
reverse: 5′-CCCACCAAACCAACCACCG-3′
probe: 5′-6FAM-CACGCACCGCCCGCCGTACC-TAMRA-3′
MYODIforward: 5′-CCAACTCCAAATCCCCTCTCTTTAT-3′58°C0161
reverse: 5′-TGATTAATTTAGATTGGGTTTAGAGAAGGA-3′
probe: 5′-6FAM-TCCCTTCCTATTCCTAAATCCAACCTAAATACCTCC-TAMRA-3′

The sensitivity of QMSP was tested for each gene. For this purpose, DNA from Wi-38 cells treated with SssI methyltransferase or DNA from T24 cells was serially diluted in unmethylated salmon sperm DNA, bisulfite treated, and amplified with primers for the p16INK4a, RARB2, SEMA3B promoter as specified above.

Promoter methylation levels were calculated as the ratio of the fluorescence emission intensity values of the gene of interest to those of the MYOD1 products obtained by LightCycler analysis multiplied by 100. The ratio [gene of interest/internal reference] × 100 was used as a measure for representing the relative level of methylated p16INK4a, RARB2 and SEMA3B DNA in the particular sample.24 Assessment of promoter hypermethylation was done blinded for the diagnosis of the underlying lung disease.

Statistical analysis

Observed frequency distributions were compared with a theoretical one by χ2 test. The comparison of 2 groups was performed by the Mann-Whitney U test. Results of χ2 test and Mann-Whitney U test are given as χ2 and U values, respectively. All statistical tests are 2-sided. The level of significance was set to p < 0.05.

Results

Quantification of tumor cells

The semiquantitative analysis of tumor cell number in bronchial aspirates of lung cancer patients disclosed 2 cases without any tumor cells/clusters visible by light microscopy (false negative cytology), 33 cases with few tumor cells/clusters (highly suspicious or positive cytology) and 43 cases with multiple tumor cell/clusters (positive cytology). There was no significant correlation between tumor cell number and methylation frequency of any of the genes investigated (χ2 ≤ 1.60; p > 0.1).

Sensitivity and specificity of methylation assay

Methylation analysis was performed by QMSP using a LightCycler. Positive and negative controls showed the expected results in each analysis. Dilution experiments with DNA from T24 cells or Wi-38 cells treated with SssI methyltransferase showed that the methylation assay was able to detect 1 methylated allele among 5,000 unmethylated alleles.

p16INK4a, RARB2 and SEMA3B promoter hypermethylation detected in bronchial aspirates

The prevalence of p16INK4a promoter methylation detected in bronchial aspirates is shown in Figure 1. Aberrant promoter methylation of p16INK4a was observed predominantly in patients diagnosed with SCC, i.e., in 56% of cases (χ2 ≥ 10.8; p < 0.01). No case with benign lung disease was found to be positive. Aberrant RARB2 methylation was found in 60/75 (80%) of bronchial aspirates of tumor patients. The frequencies of RARB2 hypermethylation in patients diagnosed with SCC (19/25; 76%), AC (20/25; 80%) and SCLC (21/25; 84%) showed no major differences (χ2 ≤ 0.32; p > 0.5). Methylation of the RARB2 promoter was significantly less frequent in bronchial aspirates of nontumor patients than in tumor patients (48 vs. 80%; χ2 ≥ 16.0; p < 0.001). Methylation of SEMA3B was detected in bronchial aspirates of tumor and nontumor cases in 88 and 92%, respectively (χ2 < 0.76; p > 0.25). Therefore, testing of this marker in patients with benign lung disease was limited to 25 specimens.

Figure 1.

Percentage of cases with p16INK4a (black), RARB2 (dark gray; cutoff at a methylation level >30) and SEMA3B (light gray) hypermethylation in bronchial aspirates from patients with benign lung disease and primary lung cancer. NT = benign lung disease, SCC = squamous cell carcinoma, AC = adenocarcinoma, SCLC = small cell lung cancer.

The results of quantitative evaluation of promoter methylation are depicted in Figure 2. Statistical analysis of cases with aberrant methylation demonstrated significant lower methylation levels for RARB2 in bronchial aspirates of the nontumor group as compared to those belonging to patients diagnosed with AC and SCLC (U ≥ 2.82; p < 0.005). No further statistically significant differences of quantitative results were detected.

Figure 2.

Promoter methylation level ([gene of interest /MYOD1] × 100) detected in bronchial aspirates from patients with benign lung disease and primary lung cancer. The dotted line refers to a cutoff point at a methylation level >30, indicating high-level hypermethylation of RARB2. NT = benign lung disease.

Analysis of simultaneously occurring methylations disclosed 8/75 bronchial aspirates with promoter methylation of p16 and RARB2. All cases also showed methylation of SEMA3B. The subgroup with triplicate methylation comprised predominantly males (7 males, 1 female) with SCC (7 SCC, 1 AC) and was slightly older than the remaining study population (median age, 68 years; range 52–78 years; U = 2.03; p < 0.05). Tumor stage was comparable to the whole study population (1 with IIB, 1 with IIIB, 4 with IV, 2 with stage not available). All of them were smokers. No aberrant methylation at all was found in 4/75 bronchial aspirates. Clinicopathological data were similar to whole study population with respect to sex (3 males, 1 female), age (median age, 64 years; range, 60–66 years), histologic type (3 AC, 1 SCLC), staging (1 with IIIB, 2 with IV, 1 with extensive disease) and smoking habits (2 smokers, 2 former smokers).

Discrimination between benign and malignant lung disease

Analysis of p16INK4a methylation (methylation level > 0) in bronchial aspirates allowed a complete discrimination between tumor and nontumor cases (Fig. 1). The sensitivity of aberrant p16INK4a methylation for the detection of patients diagnosed with SCC was 56% with a specificity of 100%. Since methylation of RARB2 was frequently found in bronchial aspirates of nontumor patients, after testing of 70 samples we defined a RARB2 QMSP assay threshold greater than 30 as a biomarker of malignancy. Using this cutoff an improved distinction between tumor and nontumor cases was achieved, resulting in a sensitivity of 56% and a specificity of 87% (Fig. 1). A combination of the marker genes p16INK4a and RARB2 (QMSP assay threshold > 30) identified 69% (52/75) of patients diagnosed with lung cancer and 13% (8/64) of patients with benign lung disease. Therefore, a sensitivity of 69% with a specificity of 87% was achieved. Both cases with false negative cytology showed a positive methylation assay (SCC: p16 positive; SCLC: RARB2 positive). In addition, all 4 cases with highly suspicious cytology presented with a positive QMSP assay (3 SCC and 1 SCLC: RARB2 positive).

A subset of tumor cases (n = 24) presented with a positive cytology but negative simultaneous histology (3 with stage I, 4 with stage III, 11 with stage IV, 1 with limited disease, 5 with extensive disease). In 7 (29%) of these cases the proof of p16 hypermethylation confirmed the cytologic tumor diagnosis (6/12 SCC, 1/6 AC, 0/6 SCLC). With respect to RARB2 this was true of 13 (54%) cases (6/12 SCC, 5/6 AC, 2/6 SCLC). The combined analysis of p16 and RARB2 was able to confirm cytology in 17/24 (71%) cases with a negative simultaneous histology (10/12 SCC, 5/6 AC, 2/6 SCLC).

SEMA3B did not allow a discrimination between the tumor and nontumor group (Fig. 1).

Clinicopathological correlation

No correlation was found between p16INK4a, RARB2 and SEMA3B methylation status in bronchial aspirates and age, gender, smoking habits or tumor stage. A total of 4 of 5 tumor patients with asbestos exposure showed methylation of the RARB2 promoter.

Discussion

The goal of the present study was to evaluate whether QMSP analysis of p16INK4a, RARB2 or SEMA3B promoter methylation status in bronchial aspirates may aid diagnosis of lung cancer. Since there are only limited QMSP data on bronchial aspirates available in the literature,25 we evaluated the prevalence of aberrant promoter methylation of these genes mainly in cases with positive cytology as well as in a larger set of controls to determine specificity. We found p16INK4a methylation to be frequent in patients diagnosed with SCC and at a lower rate in patients with AC and SCLC (56 vs. 12 and 4%, respectively; p < 0.01) This is in line with the majority of previous studies on resected tumor tissues, which observed methylation of p16INK4a in 21–68% (median, 44%) of SCC of the lung, in 9–55% (median, 24%) of pulmonary AC and in 5–9% of SCLC.9, 12, 13, 25, 26, 27, 28, 29, 30 Surprisingly, the p16INK4a methylation rate reported for sputum samples of patients with non small cell lung cancer (NSCLC)5, 27, 31, 32 was even higher, ranging from 35–90% (median, 69%). In contrast, studies which investigated bronchial aspirates of NSCLC patients25, 26, 31, 33 detected considerably lower rates of p16INK4a hypermethylation (median, 23%). In our study, p16INK4a methylation was not detectable in bronchial aspirates of patients without evidence for lung cancer even though the nontumor patients did not represent an actual control group but underwent bronchoscopy for suspected lung cancer or its exclusion. Most of the nontumor patients were current or former smokers. This underlines the high specificity of QMSP assay. Similar to our study, others reported a low rate (0–3%) of aberrant p16INK4a methylation both in lung tissue and in cytologic specimens of cases without lung cancer.25, 34, 35 However, some authors5, 31, 36, 37, 38 found a higher prevalence of aberrant p16INK4a methylation in benign lung disease, reaching up to 35% (median, 14%). The majority of these studies applied a nested methylation specific PCR (MSP), which is more sensitive than QMSP and, thus, might have detected p16INK4a methylation as an early change in lung cancer. To some extent, the differences could be attributed to the positions of the examined CpG sites.

Promoter methylation of the RARB2 gene was generally reported in around 35% and 55% of NSCLC and SCLC, respectively.12, 13, 14, 15 Aberrant RARB2 methylation was also detected in up to 14% of nonmalignant lung tissues of lung cancer patients and sputa from controls.13, 14, 15, 35 In contrast, our investigations showed methylated RARB2 at a considerable higher frequency both in bronchial aspirates of tumor and nontumor patients (80 and 40%, respectively). Chan et al.9 reported a comparable high frequency of RARB2 methylation in tumor tissue and bronchoalveolar lavage fluid of Chinese patients and controls, as in the current study, and attributed this finding to ethnic differences in aberrant methylation. However, Toyooka et al.13 compared methylation frequencies of multiple genes from 4 countries (United States, Australia, Japan and Taiwan) and found no geography-related differences with respect to RARB2 hypermethylation. Therefore, it is reasonable to assume methodical factors to account for the variation in observed RARB2 methylation frequencies. Most previous studies on RARB2 methylation in pulmonary specimens utilized conventional MSP. A more sensitive method like QMSP might detect early epigenetic changes of normal-appearing cells, resulting in an increase of cases with methylated RARB2, as observed in this study. Interestingly, Zöchbauer-Müller et al.35 found an increased frequency of methylated RARB2 in sputum samples of smokers with dysplastic changes of bronchial epithelium. Whether the nontumor patients in the current study who show RARB2 promoter methylation have a clinically occult tumor or an increased risk to develop lung cancer remains to be elucidated by further follow-up. Another explanation for varying RARB2 methylation frequencies relates to the CpG sites examined. Even though Topaloglu et al.25 used a QMSP assay similar to this study, they did not detect aberrant RARB2 methylation in bronchoalveolar lavages of patients with NSCLC.

Recently, the semaphorin family member SEMA3B was linked to tumorigenesis of lung cancer and shown to be frequently inactivated by promoter region methylation, especially in SCLC.10, 11, 39 So far, only few data are available regarding SEMA3B methylation status in nonneoplastic bronchial epithelium. Kuroki et al.10 analyzed NSCLC patients and found methylated SEMA3B in 11% of corresponding noncancerous lung tissues. We detected a considerable higher rate of SEMA3B methylation in benign bronchial aspirates (92%). Obviously, bronchial aspirates contain nonneoplastic cells with a hypermethylated SEMA3B gene promoter. These not yet characterized cells appear to overlay methylated SEMA3B alleles originating from lung cancer cells. Therefore, SEMA3B is unsuited as a biomarker of malignancy when applied to bronchial aspirates.

The use of promoter methylation as biomarker is emerging as one of the most promising molecular strategies for early detection of cancer.40 Exfoliated cells from the lung (bronchial aspirates, bronchial brushings, sputa) are well suited for the detection of promoter hypermethylation as a biomarker since the specimens are accessible with minimal invasive or even noninvasive procedures and the DNA is conserved at high molecular weight.20 However, studies which applied conventional MSP or a highly sensitive 2-stage MSP also detected hypermethylation in a considerable portion of patients without evidence of malignancy.4, 5, 31, 35, 36, 37, 38 In this study, QMSP analysis of p16INK4a hypermethylation on bronchial aspirates demonstrated a specificity of 100% and proved to be a well suited biomarker, especially for the diagnosis of pulmonary SCC. The quantitative evaluation of QMSP raised the specificity of RARB2 up to 87%. Thus, QMSP analysis of p16INK4a and RARB2 hypermethylation may offer a promising tool for the diagnosis of lung cancer in bronchial aspirates. It may be used as an independent diagnostic adjunct in cases with a highly suspicious or positive cytology but negative simultaneous histology. In our study, the proof of aberrant p16 and/or RARB2 methylation was able to confirm the solely cytologic diagnosis of SCC and AC in 10/12 cases and 5/6 cases, respectively. This could avoid further invasive diagnostic procedures if there is reluctance to initiate a highly aggressive therapy based on a positive cytology alone. In addition, both cases with a false negative cytology showed a positive methylation assay, demonstrating its potential value to improve the sensitivity of conventional cytology. This has to be confirmed in a larger series of cases with false negative cytology.

Acknowledgements

We thank Dr. W. A. Schulz, Department of Urology, Heinrich-Heine-University, Duesseldorf, Germany for providing the T24 cell line and for critical review of the manuscript.

Ancillary