RESPONSE: Re: Promoter Methylation and Silencing of the Retinoic Acid Receptor-  Gene in Lung Carcinomas

Lamy et al. examined short-term cultures of 71 bronchial biopsy specimens (including 30 dysplastic or carcinoma in situ lesions) from 38 subjects with asbestos and/or smoking exposure for aberrant methylation of the RARβ P2 promoter. All samples were negative for methylation of RARβ. However, aberrant promoter methylation of the CDKN2A/p16INK4a gene was present in 18% of the samples. Consequently, Lamy et al. suggest that methylation of RARβ is not an early event in lung carcinogenesis. Our preliminary data, supplemented with indirect published evidence, do not support their conclusions. First, testing for the status of five genes frequently methylated in lung cancers (1), we examined induced sputum specimens from 55 current or former smokers with more than 30 pack-years of exposure. The methylation frequency of RARβ was the highest (27%) (Zöchbauer-Müller S, Lam S, Minna JD, Gazdar AF: unpublished data). We also tested nonmalignant peripheral lung samples from patients with resected NSCLC. The methylation frequency in these samples was 15 (14%) of 104 (1). Of six short-term cell cultures started from nonmalignant bronchial epithelium of patients with resected NSCLC, three were methylated for RARβ (Kurie J, Virmani A, Gazdar AF: unpublished data). Second, Lamy et al. tested for RARβ methylation by using identical primer sequences and an assay similar to ours. Thus, the possible reasons for our very different results need to be discussed. In our experience, even minor changes in methylation-specific polymerase chain reaction assay conditions may result in widely differing methylation frequencies. Assays need to be validated by use of stringent criteria (2). We demonstrated a high concordance between methylation of the RAR P2 promoter and silencing of its specific transcripts (3). Methylation was absent in control tissues from healthy nonsmoking subjects (3). Our unpublished data (Gazdar AF, Virmani A, Toyooka S) indicate that our assay conditions can detect one methylated cell admixed with 1000 unmethylated cells. Thus, we have validated that our assay is both specific and sensitive. Lamy et al. should determine whether RARβ is expressed in their cell cultures. Third, expression of retinoic acid receptors, including RARβ is reduced in the bronchial epithelium of many smokers, and this reduction may be related to the increasing severity of the histologic changes (4,5). High frequencies of allelic loss at the RARβ locus (at chromosome 3p24) have been reported in bronchial preneoplastic lesions (6). Finally, gene silencing by methylation may be temporarily reversed by exposure to demethylating agents. Conversely, if the mechanism of gene silencing of a receptor was via methylation, exposure to its ligand would not be expected to increase expression of the gene. Lamy et al. suggest that retinoids, the ligands for retinoid receptors, may be more suitable than demethylating agents for chemoprevention of lung cancer. Retinoids have been the major or sole therapeutic agent in many lung cancer chemoprevention trials, most of which have reported negative results. Treatment of smokers with 13-cis-retinoic acid reduced the percentage of subjects with decreased (“aberrant”) expression of RARβ in one or more bronchial biopsy specimens from 86% to 63% (5). Thus, retinoid treatment did not reverse aberrant expression of RARβ in most subjects. Exposure to retinoic acid did not restore RARβ expression or inhibit growth of lung cancer cell lines (7). Considered together, these findings are consistent with the concept that methylation, combined with allelic loss, is the major mechanism of decreased RARβ expression in lung cancers and in smoking-damaged bronchial epithelium and peripheral lung tissue. These findings may also explain the disappointing results from clinical trials using retinoids for the chemoprevention of lung cancer.

Récupéré en direct depuis OpenAlex et désinversé. Les résumés ne sont pas conservés dans cette base de données : les index inversés représentent 8,6 Go des 9,3 Go de texte de la base, et le serveur dispose de 13 Go libres.