Aberrant Promoter Methylation in Chinese Patients with Non-Small Cell Lung Cancer: Patterns in Primary Tumors and Potential Diagnostic Application in Bronchoalevolar Lavage¹

Lung cancer is the most frequent cause of cancer death worldwide (1). Histologically, about 80% are non-small cell and 20% are small cell lung cancers (2). Among the NSCLCs,³ adenocarcinoma and SCC are the most common histological subtypes. Often grouped together in clinical trials for treatment purposes (3), adenocarcinoma and SCC may have biological differences. Although both subtypes are related to cigarette smoking (4), adenocarcinoma is more prevalent in non-smokers (5). Furthermore, Chinese women in Hong Kong have one of the highest mortality rates of lung cancer in the world (25/100,000 population) with a preponderance of adenocarcinoma, and approximately 80% of these patients have never smoked (6). Therefore, whether there are differences in the pathogenesis and genetic aberrations within the NSCLC remains undefined.

Recently, epigenetic silencing of gene expression by promoter CpG island hypermethylation has been shown to be important in cancer formation (7). DNA methylation may be an alternative mechanism to mutations or deletions in disrupting tumor suppressor gene function. Aberrant gene methylation has also been frequently found in NSCLC. This includes the p16, O⁶-methylguanine-DNA methyltransferase (MGMT), death-associated protein (DAP) kinase, retinoic acid receptor-β (RARβ), Ras association domain family 1A (RASSF1A), and adenomatous polyposis coli (APC) genes (8, 9, 10, 11, 12, 13, 14, 15). Furthermore, aberrant methylation of some of these genes can be found in bronchial epithelial cells in people at high risk of lung cancer (16, 17, 18, 19, 20), suggesting that this might be an earlier step in carcinogenesis.

Most studies of aberrant gene methylation in lung cancers have been conducted in the West. Mindful that Western patients may have different etiological and genetic factors as compared with Asian ones, as well as our unusual epidemiological observation of frequent lung cancer in non-smoking women, we investigated retrospectively a series of Chinese NSCLC patients for aberrant promoter methylation with a candidate gene approach, selecting genes that have been reported to show high frequencies of methylation in Western patients: p16 at 25% (14), RARβ at 40% (11, 14), MGMT at 25–29% (9, 14), and DAP kinase at 19–44% (10, 14). We also studied prospectively the potential use of gene methylation as a surrogate tumor marker in BAL in patients with suspected lung cancer.

Primary tumor specimens were obtained from surgically resected lung cancers. Specimens collected prospectively were snap-frozen at −20°C until use. Archival specimens were paraffin embedded. Before DNA extraction, cancer tissues were micro-dissected from surrounding normal tissue on 5-μm-thick sections as described (21).

BAL samples were obtained during bronchoscopic evaluation of suspected lung cancer, with informed consent. After centrifugation at 3000 rpm for 10 min, the cell pellet was harvested and frozen at −20°C until use.

DNA was extracted from primary tumor/BAL samples with standard phenol-chloroform protocols. For archival paraffin sections, the QIAamp DNA mini kit (Qiagen, Hilden, Germany) was used for DNA extraction. DNA modification was by the bisulphite reaction using the CpGenome DNA Modification Kit (Intergen Co., Purchase, NY), as reported (22). MSP was performed as described (23), with primers for the methylated (M_F and M_R) and unmethylated (U_F and U_R) alleles of p16, RARβ, MGMT, and DAP kinase as shown in Table 1. MSP was carried out in a final volume of 50 μl containing 5 μl of bisulphite-modified DNA, 250 μm dNTP (Life Technologies, Inc., Gaithersburg, MD), 1 μm of each primer (Genosys, Cambridgeshire, United Kingdom), 1.5–3 mm MgCl₂, 1× PCR gold buffer, and 2 units of AmpliTaq Gold (PE Biosystems, Foster City, CA), in an MJ PTC-200 thermocycler (M. J. Research, Inc., Cambridge, MA) with the following cycling parameters: 95°C for 12 min; 32–40 cycles of 94°C for 1 min, specific annealing temperature (Table 1) for 1 min, 72°C for 1 min; and a final extension step at 72°C for 10 min. Normal and methylated DNA (Intergen) were used to optimize the MSP conditions, and included as normal and positive controls in every experiment.

All MSPs were performed in duplicate. As controls, amplifications for the unmethylated/methylated alleles on bisulphite-modified normal/methylated DNA and template-blank controls were performed. To test the sensitivity of the MSP, methylated DNA (Intergen) was serially diluted in normal DNA, bisulphite treated, and amplified with primers for the methylated allele. To confirm the specificity of the MSP, PCR products were gel purified and sequenced in both directions with the same primers used for MSP (dRhodamine Terminator Cycle Sequencing Ready Reaction Kit; PE Biosystems), and analyzed by an automated DNA sequencer (ABI Prism 377; PE Biosystems).

MSP for p16, RARβ, DAP kinase, and MGMT showed the expected results for the positive and negative controls in all of the experiments. The PCR products were sequenced and showed the expected changes induced by bisulphite treatment (Fig. 1). MSP had a sensitivity of 10⁻² for MGMT, 10⁻³ for p16 and DAP kinase, and 10⁻⁴ for RARβ (Fig. 1).

A total of 44 adenocarcinomas and 31 SCCs were studied (Table 2). There were significantly more female patients with adenocarcinoma than SCC (11 of 44 versus 2 of 31, P = 0.04, χ² test). Methylation of one or more genes occurred in 63 of 75 (84%) of cases. Adenocarcinoma and SCC had similar frequencies of gene methylation (p16, 55–68%; RARβ, 70–71%; DAP kinase, 34–35%; MGMT, 14–16%). However, in adenocarcinoma, female patients had a significantly higher frequency of p16 methylation as compared with males (10 of 11 versus 17 of 33, P = 0.02, Fisher’s exact test). This sex difference in p16 methylation was not demonstrable in SCC, probably because of the small number of female patients (2 of 2 versus 19 of 29, P > 0.05). However, when both adenocarcinoma and SCC were considered together, p16 methylation was still significantly more frequent in female than male patients (12 of 13 versus 36 of 62, P = 0.02, Fisher’s exact test). This sex difference was not found in RARβ, MGMT, and DAP kinase methylation in both histological subtypes (Fig. 1).

The results and clinicopathological correlations are shown in Table 3. Concordant results were observed in 25 of 68 samples. Fifteen specimens showed malignant cells on cytological analysis, and MSP was positive in at least one gene tested. Ten samples did not show malignant cells on cytological analysis, and the MSP results were correspondingly negative. On follow-up of the 10 cytology-negative and MSP-negative cases, 6 were shown to be infectious in nature. However, four cases were finally found to be NSCLC by further investigations.

The results of cytology and MSP were discordant in 43 or 68 samples. Forty-one samples were cytologically negative, but MSP positive in one or more genes. Of these, 25 cases underwent further BAL and transbronchial biopsies and were shown to have malignant diseases, 22 of pulmonary origin and 3 from the large bowel or uterine cervix. Ten cases had evidence of systemic metastases, and the patients declined further histological evaluation. Interestingly, four cases with positive MSP have not yet had evidence of a malignancy. Two were chronic smokers, one had systemic sclerosis and fibrosing alveolitis, and one had a positive culture for Mycobacterium tuberculosis and was receiving treatment. All still have persistent lung shadows at the latest follow-up. Finally, there were two cases where the cytology was positive for malignant cells, but the MSP was negative in all of the four genes tested.

In this study, we have observed a preferential increase in female patients with adenocarcinoma as compared with SCC, similar to previous studies in Chinese patients with lung cancer (6). The more frequent occurrence of adenocarcinoma in females, who are non-smokers, suggests that etiological or carcinogenic factors for adenocarcinoma may be different from those of SCC. Previous studies from Hong Kong showed that this could not be explained by different K-ras or p53 mutations (24, 25). In this study, the difference between adenocarcinoma and SCC was also not reflected in the frequencies of methylation of p16, RARβ, DAP kinase, and MGMT, because results were comparable in both histological types. However, an interesting observation was that p16 was methylated in nearly all (92%) of the female patients studied. This sexual difference in p16 methylation has not been previously observed in Western patients, in whom the risk factor for male and female patients alike is cigarette smoking. This frequency of p16 methylation was one of the highest known for any gene in any malignancy, suggesting that an inherent predisposition to p16 methylation might have existed in these Chinese women. Recently, it has been observed that polymorphisms of the glutathione S-transferase P1 gene (A to G at bp 104) and the NADPH quinone oxidoreductase gene (C at bp 909) are associated with increased risks of p16 methylation in bronchial epithelial cells of subjects at high risk of lung cancers (19). Hence, whether polymorphisms in genes involved in modulating or repairing DNA, or other genetic predispositions, may explain our observations will need to be investigated further.

Our results also showed that RARβ was methylated at a frequency (around 70%) considerably higher than that (around 40%) found in Western patients (11, 14), whereas DAP kinase and MGMT were methylated at frequencies (35% and 15%) similar to those reported (9, 10, 14). The biological implications of this observation will need to be studied further.

We have also examined the potential application of MSP as a molecular marker for the detection of cancer cells in BAL samples. For MSP to be a useful clinical test, it has to be shown to be more sensitive than conventional cytological analysis. Such a comparison has been performed in 21 samples in one study, where p16 and MGMT methylation was shown to be more sensitive than cytological analysis (12). In another study, molecular markers including aberrant methylation of p16 and p53/K-ras mutations were detected at high frequencies in 50 BAL samples (26). However, whether these samples were also cytologically positive was unknown. To our knowledge, our study is the first to systematically compare head-to-head the sensitivity of MSP versus conventional cytological analysis in detecting cancer cells in BAL samples. We showed that 15 of 17 (88%) cytologically positive BAL samples showed aberrant methylation of at least one gene in an MSP panel comprising p16, RARβ, DAP kinase, and MGMT, a frequency that was similar to that of 84% in primary tumors. More importantly, MSP was positive in 35 cases in which the cytology was negative, but the pulmonary lesions were shown to be malignant by later investigations. In these cases, a positive MSP has most likely identified the presence of neoplastic cells. It would be interesting to investigate the neoplastic lung lesions in these cases, to define whether the patterns of aberrant methylation in the BAL samples matched those of the primary tumors. Unfortunately, the diagnostic materials in these cases were either cytological fluid or small transbronchial biopsies, which were inadequate or not available for further analysis. There were, however, four cases where MSP was positive, with the patients not having evidence yet of a neoplastic lesion. Two were chronic smokers with persistent lung shadows, and it has been shown previously that MSP might be positive in these subjects at high risk of lung cancers (12, 17, 18, 19, 20, 26). Another patient had scleroderma, a disease well known to be associated with lung cancer, particularly adenocarcinoma (27, 28). These patients with negative cytology and positive MSP will have to be followed closely for further evidence of a malignancy. Therefore, our results confirm that MSP is a useful and sensitive molecular adjunct for the detection of small numbers of cancer cells in BAL.

Finally, another interesting observation was that in MSP of the BAL the positive rate of p16 methylation (25%) was lower than that (64%) in primary tumors whereas the frequencies of methylation of RARβ (88%), DAP kinase (25%), and MGMT (11%) were comparable with those in primary tumors (71%, 35%, and 15%, respectively). For the BAL cases, we have not tested the corresponding tumors, because most of them were unresectable. However, it has been shown that the patterns of methylation in bronchial epithelial cells do not necessarily parallel those of the primary tumors (12, 26). Furthermore, BAL presumably contained small numbers of tumor cells. The lower detection rate of p16 methylation (as compared with, for example, RARβ methylation) may be related to the sensitivity of the MSP as well as the respective clone sizes of the neoplastic cells harboring methylation of these genes, which may become significant when a small number of cells are tested. Therefore, prospective comparison of the methylation patterns of the cancer cells in exfoliative cytology against those of the primary tumor will be both important and interesting. Finally, quantitative studies of gene methylation (15) in BAL samples to quantify the amount of neoplastic cells will be needed.

The cytologic evaluation of BAL was performed in the Department of Pathology, University of Hong Kong.