No Association between the p21 codon 31 Serine-Arginine Polymorphism and Lung Cancer Risk¹

p21 is a downstream target of p53. Wild-type p53 binds to a site 2.4 kb upstream of the p21 coding sequence and stimulates gene expression (1).The p21 (also known as CIP1/WAF1) protein binds tightly to cyclin complexes, inhibiting the function of cyclin-dependent kinases in DNA-damaged cells (2). A p21 codon 31 Ser-Arg polymorphism has been identified. Mousses et al. (3) found that the frequency of the genotype variants in tumors with wild-type p53(20.4%) was significantly higher than in tumors with mutated p53(4.1%; P < 0.01). In two previous association studies, Sjalander et al. (4) found a marginally increased frequency of the p21 codon 31 Arg allele in 144 Swedish lung cancer patients compared with 761 healthy controls, whereas a Taiwanese study found no such association in 155 lung cancer patients compared with 189 noncancer controls (5).

We report the results of a large case-control study designed to test the association between the variant Arg genotypes of the p21 codon 31 Ser-Arg polymorphism and lung cancer risk.

The study was approved by the Human Subjects Committees of Massachusetts General Hospital and the Harvard School of Public Health, both in Boston, Massachusetts. Caucasian patients with histologically confirmed incident lung cancers were recruited between December 1992 and December 2000 at Massachusetts General Hospital. Controls were recruited first among the friends and non-blood-related family members of cases. Details of this population have been published previously (6).

Peripheral blood was collected and interviewer-administered questionnaires collected information on demographic and detailed smoking histories. DNA was extracted from peripheral blood samples using the Puregene DNA Isolation kit (Gentra Systems, Minneapolis, MN). p21 codon 31 Ser-Arg polymorphism was detected using previously reported PCR-RFLP methods (4). For quality control, a random 5% of the samples were repeated. Two authors (L. S., W. Z.) independently reviewed all results.

Logistic regression was used to model the association between the risk of lung cancer and p21 Ser-Arg codon 31 variant allele, using SAS statistical package Version 6.0. All of the adjusted analyses included variables for age, gender, smoking status (current, ex-smoker, nonsmoker), years since smoking cessation for ex-smokers, and the square root of pack-years. Primary analyses combined the Ser/Arg and Arg/Arg genotypes together as the “variant allele.”

We found no significant association between the p21 codon 31 Ser-Arg polymorphism and lung cancer risk (Table 1). Results were unchanged in stratified analyses by age (age <55 years, age ≥55 years), gender, histological subtypes (adenocarcinoma, squamous cell carcinoma), and clinical disease stage (Stages I/II, III/IV). We found a small nonsignificant increase in risk by the variant allele in never-smokers (AOR,³ 1.34; 95% CI, 0.7–2.5), and a small marginally significant decrease in risk in ever-smokers (AOR, 0.77; 95% CI, 0.6–1.0) that was driven mainly by the association in ex-smokers (AOR, 0.69; 95% CI, 0.5–1.0), but not in current smokers (AOR, 1.04; 95% CI 0.6–1.8).

The results presented here agree with the negative study in a Taiwanese population (5). The Swedish study found a strong association between lung cancer and the p21 codon 31 Ser-Arg polymorphism when comparing lung cancer cases to controls with chronic obstructive pulmonary disease but only a marginal association when using healthy controls (4).

The controls in our study were not matched for age, gender, or smoking characteristics, although we did adjust for these variables in our analyses and performed stratified analyses of subgroups. In only one of the subgroups did we find possible association, a possible protective effect of the variant allele in ex-smokers, but not in current or never-smokers, leading to the likelihood that this association was attributable to chance. Another potential limitation was the use of controls that were recruited through the cases, because of possible selection bias. However, we evaluated our controls extensively for other polymorphisms, including GSTP1, NQO1, CYP1A1, NAT2, MPO, MnSOD, and p53. For each polymorphism and with the present p21 polymorphism, we found that the genotype distributions were similar to other United States-based Caucasian controls samples, and met Hardy-Weinberg equilibrium conditions. Because our cases may reflect the referral biases of a tertiary cancer center, we also evaluated subsets of cases by histological subtype and clinical stage, and similarly found no significant associations.

Our study had 80% power at a two-sided α = 0.05 level to detect an OR of 1.45 for the variant allele. We did not find an association between the p21 codon 31 Ser-Arg polymorphism and lung cancer risk. Although we cannot rule out a small effect of the homozygous Arg/Arg variant, the sample size required to detect an OR of 1.5 is enormous (n = 15,000), whereas the low frequency of the homozygous variant would have little impact on the overall burden of lung cancer.