Correspondence re: Feyler et al., Point: Myeloperoxidase ⁻⁴⁶³G → A Polymorphism and Lung Cancer Risk. Cancer Epidemiol. Biomark. Prev., 11: 1550–1554, 2002, and Xu et al., Counterpoint: The Myeloperoxidase ⁻⁴⁶³G → A Polymorphism Does Not Decrease Lung Cancer Susceptibility in Caucasians. 11: 1555–1559, 2002

Since 1997, nine studies on MPO-463 genotype and lung cancer risk in Caucasians have been published. A significant reduction of the levels of benzo(a)pyrene diol epoxide-DNA adducts in human skin after coal tar treatment for MPO-A allele carriers has also been shown (1). This indicates that this polymorphism has functional significance for metabolism and DNA binding of carcinogens also present in tobacco smoke. However, the possible inverse association of the A allele with lung cancer risk has remained controversial. A recent case-control study by Feyler et al. (Ref. 2; Point) found a decreased risk of 45% for A allele carriers [OR¹ = 0.55 (95% CI 0.3–0.9)], which is in agreement with a meta-analysis of earlier studies (2). On the other hand, the largest study to date by Xu et al. (Ref. 3; Counterpoint; 988 cases), revealed no MPO genotype-related association for lung cancer overall [OR = 1.03 (95% CI 0.8–1.3)]. In our own recent hospital-based study (4), which was not included in the meta-analysis, of 625 cases (36% adenocarcinoma, 36% SCC, 22% SCLC, and 6% others) and 340 controls (all cases and controls were current or ex-smokers), we concentrated on a separate analysis of different histological types of lung tumors. We found a 25% reduced risk for lung cancer overall [OR = 0.75 (95% CI 0.55–1.01)] approaching significance for A allele carriers, whereas a significant inverse association was only observed for SCLC [OR = 0.58 (95% CI 0.36–0.95)], but not for other histological types of lung cancer (4). At first glance, these results seem to be compatible with both the Point and the Counterpoint because the study by Feyler et al. (2), which also finds A allele carriers to be at decreased risk, includes 35% SCLC, whereas Xu et al. (3) contains only 9% SCLC among cases and finds no association. However, the situation remains unclear for SCC; where Feyler et al. (2) find a significant inverse association, we find a nonsignificant decreased risk, whereas Xu et al. (3) find no inverse association for A allele carriers.

Given the different etiology of histological types of lung tumors, and the possible impact of the MPO promoter polymorphism on both tobacco-carcinogen metabolism and levels of oxidative stress, two further points seem important: Xu et al. (3) included 36% of never-smokers in their control group and only 6% in the case group, whereas in the two studies [Feyler et al. (2) and our study (4)] that found an inverse association of the A allele, all cases and controls were either current or ex-smokers. Whereas both Point and Counterpoint find no association between genotype and smoking, such an association might also be dependent on histological tumor type. Our study (4) further suggests that the MPO genotype frequencies may differ in nonmalignant lung diseases. Because lung tumor development is frequently preceded by chronic inflammation of the lung (5), with recruitment of large numbers of neutrophils to the lung and local release of MPO (6), future case-control studies investigating MPO genotype and lung cancer risk would ideally include information on previous lung diseases for both cases and controls. In conclusion, additional (large) case-control studies should preferentially analyze smokers and include a separate analysis of histological types of lung cancer, and in such studies, clinical assessment of and statistical adjustment for inflammatory nonmalignant lung diseases would be desirable.