p14ARF genetic polymorphisms and susceptibility to second primary malignancy in patients with index squamous cell carcinoma of the head and neck
Yang Zhang MD,
Department of Head and Neck Surgery, The University of Texas M. D. Anderson Cancer Center, Houston, Texas
Department of Otolaryngology Head and Neck Surgery, Beijing Tongren Hospital, Capital Medical University, Key Laboratory of Otolaryngology Head and Neck Surgery, Ministry of Education, Beijing Institute of Otolaryngology, Beijing, China
p14ARF, an alternate reading frame (ARF) product of the cyclin-dependent kinase inhibitor 2A locus, plays a critical role in crosstalk between the tumor protein 53 (p53) and retinoblastoma (Rb) pathways and in cellular anticancer mechanisms. Therefore, the authors of this report investigated the association between single nucleotide polymorphisms (SNPs) of the p14ARF gene and the risk of developing a second primary malignancy (SPM) after an index squamous cell carcinoma of the head and neck (SCCHN).
The log-rank test and Cox proportional hazards models were used to assess the association of 2 p14ARF SNPs (reference SNP [rs]3731217 and rs3088440) with SPM-free survival and with the risk of developing an SPM among 1287 patients who had SCCHN.
Patients with either p14ARF variant genotypes of the 2 polymorphisms had a significantly reduced SPM-free survival compared with patients with no variant genotypes (log-rank test; P = .006). Compared with the p14ARF thymine-thymine (TT) and guanine-guanine (GG) genotypes, the variant genotypes of p14ARF TG/GG and guanine-adenine (GA)/AA were associated with a significantly moderately increased risk of developing an SPM (p14ARF rs3731217: adjusted hazard ratio [aHR], 1.48; 95% confidence interval [CI], 1.00-2.19; p14ARF rs3088440: aHR, 1.61; 95% CI, 1.07-2.43). Moreover, after combining the variant genotypes of the 2 SNPs, patients who had variant genotypes were at significantly greater risk of developing an SPM compared with patients who had no variant genotypes (aHR, 3.07; 95% CI, 1.54-6.12), and the risk was particularly pronounced in several subgroups.
Squamous cell carcinoma of the head and neck (SCCHN), in which tumors arise from sites of the oral cavity, oropharynx, hypopharynx, and larynx, is the sixth most common cancer worldwide and has a moderately low survival rate, a high recurrence rate, and a high rate of second primary malignancy (SPM).1 There were approximately 48,010 new cases of and 11,260 deaths from SCCHN in 2009 in the United States.2 Nationally, approximately 70% of patients with SCCHN are men, and approximately 91% are white3; whereas, in our institution, approximately 76% of patients with SCCHN are men, and 85% are white. The occurrence of SPMs in patients with SCCHN remains 1 of the major factors contributing to the poor overall survival of patients with SCCHN.4-6 Although cigarette smoking and alcohol use have been associated with the risk of SPM,7, 8 most patients with SCCHN never develop an SPM, suggesting that there is interindividual variation in genetic susceptibility to SPM among these patients.4 Several previous studies reported that single nucleotide polymorphisms (SNPs) of genes involved in carcinogen metabolism, DNA repair, cell cycle control, and apoptosis are associated with the risk of an SPM after primary SCCHN.9-15
The gene p14ARF, a tumor suppressor gene, is located at the cyclin-dependent kinase inhibitor 2A (INK4a) locus on chromosome 9p21, which encodes 2 distinct tumor suppressor proteins, p16INK4a and p14 alternate reading frame (p14ARF), and is 1 of the most commonly mutated regions in approximately 50% of all human cancers, second only to tumor protein 53 (p53).16, 17 p14ARF interacts directly with murine double minute protein (MDM2), thereby suppressing the ubiquitin ligase activity of MDM2 and subsequently inhibiting MDM2-mediated degradation of p53. Such interaction leads to stabilization and accumulation of p53. Thus, genetic alteration of p14ARF may affect cell cycle regulation and apoptosis by disrupting the p53 pathway.18, 19 Therefore, p14ARF plays an important role in the ARF-MDM2-p53 pathway. In addition, independent of p53, p14ARF has multiple other tumor suppressor functions, which involve interaction with several proteins in other cellular activities, such as cell proliferation.20 Furthermore, p14ARF is involved in the ataxia telangiectasia mutated/ataxia telangiectasia and Rad3-related/carboxyl-terminal Src kinase homologous kinase (ATM/ATR/CHK) signaling pathway in response to DNA damage,21 indicating that p14ARF may affect cell cycle and DNA repair. It has been demonstrated that p14ARF also interacts with transcription factors, such as E2F-1 and E2F-2, in the retinoblastoma (Rb) pathway to prevent Rb proteasomal degradation and trigger its antiproliferative function.22-24 Thus, p14ARF acts in maintaining genomic stability by mediating cellular activities in both the p53 pathway and the Rb pathway.
Alterations or mutations of p14ARF are frequent events in the development of SCCHN.25-29 However, to our knowledge, the roles of p14ARF polymorphisms in the etiology of SPM after an index SCCHN have not been investigated. We hypothesized that p14ARF polymorphisms contribute toward a genetic susceptibility to SPM after an index SCCHN and that these polymorphisms may be genetic markers with which to identify subgroups of patients with SCCHN who have a high risk of developing an SPM and may benefit from targeted follow-up, screening for SPMs, and considering tobacco/alcohol cessation and/or chemoprevention protocols. To test this hypothesis, first, we identified 2 tagging SNPs (those with a minor allele frequency [MAF] >5% and a linkage disequilibrium [LD] measure r2 threshold at 0.8) in the p14ARF gene (reference SNP [rs]3731245 and rs3088440), and we included an additional reported SNP30, 31 from a pilot study of 400 individuals. Although the MAF of all 3 SNPs in the overall population is >5%, we observed that p14ARF rs3731245 had a <5% MAF in our study patients. Therefore, we finally genotyped the 2 common SNPs p14ARF rs3731217 and rs3088440 (ie, SNPs with an MAF >5%) in a cohort of 1287 patients with incident SCCHN, and we evaluated the association between each genotype and the 2 polymorphisms combined with the risk of developing an SPM.
MATERIALS AND METHODS
In this study, 1667 patients with incident SCCHN were recruited consecutively from May 1995 through January 2007 at The University of Texas M. D. Anderson Cancer Center as described previously.12-15 This cohort of patients had newly diagnosed, histopathologically confirmed, and untreated SCCHN and completed an Institutional Review Board-approved informed consent without the restriction of age, sex, ethnicity, or clinical stage. The exclusion criteria included any prior cancer history (except for nonmelanoma skin cancer), distant metastases at presentation, primary sinonasal tumors, salivary gland tumors, cervical metastases of unknown origin, and tumors outside the upper aerodigestive tract. Approximately 95% of contacted patients consented to enrollment in the study. Some blood samples for p14ARF genotyping were not available for the patients who were recruited early in the study, and these patients were not included in the analysis. Patients without follow-up and patients who underwent only palliative treatment also were excluded. Therefore, the final analysis included 1287 patients.
At our institution, patients with SCCHN typically are followed and monitored through their treatment and post-treatment courses with regularly scheduled clinical and radiographic examinations. On the basis of the modified criteria of Warren and Gates,32 A second lesion was classified as an SPM if the second lesion was of a different histopathologic type, or if it occurred >5 years after treatment for the index tumor, and/or if it was separated clearly by normal epithelium based on clinical and radiographic assessment. Pulmonary lesions were considered SPMs if they had a nonsquamous histology, or if they were isolated squamous lesions that occurred >5 years after the initial SCCHN, and if the thoracic oncologist and thoracic surgeon believed that the lesions represented SPMs. If there was a discrepancy or a difference of opinion regarding the origin of the tumor (ie, recurrence vs SPM), then the second lesion was classified as a local recurrence rather than an SPM.
Clinical data, including overall stage at presentation of the index tumor, site of the index tumor, and treatment, were obtained at initial presentation and through follow-up examinations. Then, the index cancer stage was dichotomized into early stage disease (stages I and II) and late-stage disease (stages III and IV). Treatment was grouped into 4 categories: surgery only, surgery with radiotherapy and/or chemotherapy, radiotherapy, and radiotherapy plus chemotherapy. Epidemiologic data, including alcohol and smoking history, were obtained from all patients during the visits. Patients who had consumed at least 1 alcoholic beverage daily for at least 1 year during their lifetime were defined as ever drinkers, and those who never had such a pattern of drinking were defined as never drinkers. Those patients who had smoked at least 100 cigarettes in their lifetime were defined as ever smokers; otherwise, patients were considered never smokers.
Genomic DNA was isolated from a leukocyte cell pellet of each blood sample by using the QIAGEN DNA Blood Mini Kit (QIAGEN Inc., Valencia, Calif) according to the manufacturer's instructions. We genotyped for p14ARF polymorphisms by using polymerase chain reaction (PCR)-restriction fragment length polymorphism analysis. The primers we used for the 3 initial SNPs were as follows: 1) p14ARF rs3731217: (forward) 5′-AAAAGG GGGACAACCATTCTC-3′ and (reverse) 5′-CCCCTCTCAAATATGCTGTCC-3′; 2) p14ARF rs3088440: (forward) 5′-TGCTCA CTCCAGAAAACTCCA-3′ and (reverse) 5′-ATGTGCCACACATCTTTGACC-3′; and 3) p14ARF rs3731245: (forward) 5′-CAAAAATGGGTCCACAAGGTT-3′ and (reverse) 5′-CCC AACATAACCCCAAGT GTT-3′. For p14ARF rs3731217, the 280-base pair (bp) PCR products were digested with Mva I (New England BioLabs, Inc., Beverly, Mass) at 37°C overnight, the thymine (T) allele was uncut, and the guanine (G) allele was cut into 126-bp and 154-bp bands; for p14ARF rs3088440 and rs3731245, the 356-bp and 323-bp PCR products were digested with Hae III (New England BioLabs, Inc., Beverly, Mass) at 37°C overnight, the adenine (A) allele was uncut, and the G allele was cut into 215-bp and 141-bp bands (for rs3088440) (Fig. 1) and into 125-bp and 198-bp bands (for rs3731245). There was 100% concordance when at least 10% of the random samples were retested.
In this study, the Statistical Analysis System (SAS) software package (version 9.1.3; SAS Institute, Inc., Cary, NC) was used to perform all statistical analyses. Statistical significance was set at P < .05, and all tests were 2-sided. The occurrence of an SPM was considered the primary endpoint of the study. The Student t test was used to compare the mean age and follow-up of the patients who did and did not develop an SPM. The differences in distributions of demographic, epidemiologic, and clinical variables as well as genotypes between the 2 groups were evaluated using the chi-square test. The time to event was calculated from the date of diagnosis of the index SCCHN to the date of SPM occurrence. Patients who did not have a known event as of the date of last contact and patients who died were censored. Kaplan-Meier curves were used to estimate SPM-free survival, and the log-rank statistic was used to evaluate significant differences (α = .05) in SPM-free survival between the 2 groups with and without variant genotypes of both polymorphisms. The associations between individual epidemiologic risk factors, clinical characteristics, genotypes, and the time to the occurrence of SPMs were assessed using both univariate and multivariate Cox proportional hazards regression models. We used a stepwise search strategy to build the multivariate models. A final multivariate proportional hazards model was built by using the variables that had prognostic potential in the univariate analysis, always retaining the variables age, sex, and ethnicity because of epidemiologic and clinical considerations, as described previously.12-15 We assessed associations using hazard ratios (HRs) and their 95% confidence intervals (CIs) for developing an SPM in the final Cox regression models, which were adjusted for age, sex, ethnicity, and smoking and alcohol status.
The demographics, risk exposure, and clinical variables for all 1287 patients are summarized in Table 1. The 1287 patients were followed for a median of 29.7 months (range, 0-142.4 months), and 1167 patients did not develop an SPM, whereas 120 patients (9.3%) developed an SPM. The mean age at the index cancer diagnosis for all patients was 57.5 years (range, 18-94 years; median, 57 years), and the mean age at the index SCCHN diagnosis of patients who developed an SPM was significantly older compared with the patients who did not develop an SPM (60.8 years vs 57.1 years, respectively; P < .001). Although this patient cohort included predominantly men (75.9%), sex was not associated with SPM development (P = .515). We did not observe significant differences between patients with and without SPMs with regard to smoking history (P = .121), alcohol consumption (P = .345), index cancer site (P = .316), index cancer stage (P = .693), or treatment (P = .889). However, compared with patients who did not develop an SPM, patients who developed an SPM were more likely to be non-Hispanic whites (P = .050).
Table 1. Distribution of Selected Characteristics of the Patient Cohort (n=1287)
Of the 120 patients with SPMs, 81 patients developed SPMs at tobacco-associated sites, including 44 SCCHNs and 37 other tobacco-associated cancers (34 lung cancers, 2 esophageal cancers, and 1 bladder cancer); 35 patients developed SPMs at other sites (10 prostate cancers, 8 papillary thyroid carcinomas, 4 colon adenocarcinomas, 3 lymphomas, 3 hepatic adenocarcinomas, 2 breast cancers, and 1 of each of the following SPMs: sarcoma, renal cell carcinoma, endometrial carcinoma, leukemia, and maxillary sinus adenocarcinoma); and 4 patients developed SPMs at 2 sites (2 patients had both SCCHN and prostate cancer, and 2 patients had both SCCHN and papillary thyroid carcinoma). Of the 44 patients with second SCCHNs, 24 were synchronous SCCHN primaries. Of these 24 patients with synchronous SCCHNs, 2 patients had bilateral oral cavity cancers, 3 patients had bilateral oropharyngeal cancers, 1 patient had bilateral hypopharyngeal cancers, and the remaining patients had simultaneous cancers of more than 1 head and neck subsite.
Association of p14ARF Polymorphisms With the Risk of a Second Primary Malignancy After an Index Squamous Cell Carcinoma of the Head and Neck
Table 2 shows the distribution of the p14ARF rs3088440 and rs3731217 genotypes between patients who did and did not develop SPMs and the associations with the risk of developing an SPM. Distribution of the p14ARF rs3731217 genotypes did not differ significantly between patients who did and did not develop SPMs, whereas a significant difference was observed for patients who had the p14ARF rs3088440 polymorphism (P = .002). For each polymorphism, compared with patients who had the corresponding homozygous wild-type genotypes, patients who had the variant genotypes had a significant approximately 1.5-fold increased risk of developing an SPM after multivariate adjustment for age, sex, ethnicity, and smoking/drinking status (Table 2). To evaluate the combined effect of both p14ARF polymorphisms on the risk of SPM, patients who were wild-type homozygous for both genotypes were included in the “no variant” reference group, and the remaining patients with other combined variant genotypes, including variant homozygous and heterozygous genotypes, were included in the “variant” group (Table 2). We observed that the distribution of the combined genotypes differed significantly between patients who did and did not develop SPMs (trichotomized, P < .001; dichotomized, P = .007). Moreover, patients who had either variant allele (p14ARFG or p14ARFA) had an almost 1.6-fold increased risk of developing an SPM compared with patients who had the combined p14ARF TT and GG wild-type genotypes (Table 2). There was a trend toward an increased risk of developing an SPM with an increasing number of variant genotypes, and this trend in risk was statistically significant in a dose-response manner (P = .002 for trend) (Table 2). Specifically, patients who had 2 variant genotypes had an approximately 3-fold increased risk of developing an SPM compared with patients who did not have any variant genotypes (Table 2). Furthermore, patients who had p14ARF variant genotypes of both polymorphisms had a significantly reduced SPM-free survival compared with patients who had no p14ARF variant genotypes (log-rank test; P = .006) (Fig. 2).
Table 2. The Risk of a Second Primary Malignancy Associated With p14ARF Gene Polymorphisms After an Index Squamous Cell Carcinoma of the Head and Neck
SPM indicates second primary malignancy; HR, hazard ratio; CI, confidence interval; p14ARF, a tumor suppressor gene that is an alternate reading frame (ARF) product of the cyclin-dependent kinase inhibitor 2A (or INK4a) locus; rs, reference polymorphism number; T, thymine; Ref, reference group; G, guanine; A, adenine.
Chi-square tests were used to determine differences in the distribution of p14 genotypes between patients who did and did not develop SPMs.
A Cox model was adjusted for age, sex, ethnicity, tobacco smoking, and alcohol drinking.
This P value was significant at the 5% level after adjusting for multiple comparisons with Bonferroni correction.
For these genotype comparisons, 0 indicates the p14ARF rs3731217 TT and rs3088440 GG genotypes; 1, the p14ARF rs3731217 TT and rs3088440 GA+AA or rs3731217 TG+GG and rs3088440 GG genotypes; 2, the p14ARF rs3731217 TG+GG and rs3088440 GA+AA genotypes.
Stratification Analysis of the Combined p14ARF Variant Genotypes and the Risk of a Second Primary Malignancy
Table 3 shows the association between the combined p14ARFvariant genotypes and the risk of SPM in each subgroup according to age, sex, ethnicity, smoking/drinking status, index cancer site, index cancer stage, and index cancer treatment after adjusting for all other potential confounders. When we considered patients who did not have any combined variant genotypes (p14ARF rs3731217 TT or rs3088440 GG) as the reference group, there was a significant approximately 50% to 70% increased risk of developing an SPM for who had any p14ARF variant genotypes among men, non-Hispanic whites, drinkers, those with late-stage disease, those with nonoropharyngeal cancers, those who received treatment with DNA-damaging agents, and those with tobacco-associated SPMs (Table 3). In addition, there was a significant >2-fold elevated risk of developing an SPM for patients who had any p14ARF variant genotypes among younger patients (aged ≤57 years) (Table 3).
Table 3. Stratification Analysis of Associations Between Combined p14ARF</SUP Gene Polymorphisms and the Risk of a Second Primary Malignancy
Given the critical roles of p14ARF as a tumor suppressor gene in many cellular activities, inactivation or changes in the expression of this gene may deregulate these activities and, consequently, could influence a the risk of cancer. Although many previous studies have focused on the role of alterations in the INK4/AFR or CDKN2A locus in development of SCCHN,25-29 to our knowledge, no previous studies examined the association between genetic variants of p14ARF and the risk of SCCHN, particularly the risk of developing an SPM after an index SCCHN. In the current study, we investigated this association and observed that both p14ARF polymorphisms were associated with a significantly moderately increased risk of developing an SPM for patients who had an index SCCHN, and the risk was significantly greater for patients who had either of the p14ARF variant genotypes of the 2 polymorphisms compared with patients who had the wild-type homozygous p14ARF genotypes. Patients who simultaneously had both variant p14ARF genotypes had an approximately 3-fold increased risk of developing an SPM compared with patients who did not have any variant genotypes. Although the functional relevance of these 2 p14ARF polymorphisms remains unknown, they are within the functional regions of the gene's promoter (p14ARF rs3731217) and 3′untranslated region (p14ARF rs3088440), and these 2 polymorphisms potentially could affect p14 expression levels, leading to interindividual differences in susceptibility to a SPM after an index SCCHN. To date, although no studies have examined the association between these 2 polymorphisms and the risk of cancer, our data suggest that these 2 SNPs may have functional significance and may contribute toward a genetic susceptibility for developing an SPM after an index SCCHN in this patient cohort.
In addition, we observed a greater SPM risk associated with the p14ARF variant genotypes in younger patients (aged ≤57 years) with SCCHN and no significant association among patients aged >57 years, a finding consistent with the concept that genetic susceptibility includes an early age at onset. Furthermore, the risk of developing an SPM was associated significantly with the combined p14ARF variant genotypes in men but not in women. Our results indicated that men with SCCHN were more likely to be ever-smokers than women with SCCHN (P < .001), and it is possible that men with SCCHN who carry the combined p14ARF variant genotypes are more sensitive to tobacco carcinogens, which may have been responsible for both the index cancer and the SPM. This speculation is supported by the finding that the risk of SPM associated with the p14ARF variant genotypes was greater for tobacco-associated SPMs than for nontobacco-associated SPMs.
Although continued exposure to tobacco or alcohol appears to be associated with an elevated risk of developing an SPM compared with avoiding such exposure,7, 33 only 1 study to date has demonstrated that an index treatment modality (radiotherapy) may influence the risk of developing an SPM.34 In the current study, the p14ARF variant genotypes appeared to be risk factors for SPM in patients with SCCHN, but it is possible that that risk was independent of both the radiation therapy and/or chemotherapy they received for the index cancer and their smoking and alcohol status. p14ARF is redistributed in the nucleus in response to DNA damage35; thus, p14ARF polymorphisms may influence the risk of developing an SPM after such treatment exposure. However, the notion that the risk of developing an SPM associated with p14ARF polymorphisms depends on the index treatment received simply may be an artifact of the small sample size of patients who underwent surgery.
Our study also demonstrated that patients who had nonoropharyngeal index SCCHNs had a greater risk of developing an SPM associated with the p14ARF variant genotypes than patients who had oropharyngeal index cancers. This may represent differences in the etiology of index SCCHNs at oropharyngeal and nonoropharyngeal sites in relation to both environmental risk factors and genetic susceptibility. Our previous study suggested that squamous cell carcinomas of the oropharynx more likely are driven by human papillomavirus type 16 (HVP16), whereas squamous cell carcinomas of the oral cavity and larynx are more likely caused by smoking and alcohol use.36 Therefore, the risk of developing a tobacco-induced or alcohol-induced SPM after an index nonoropharyngeal squamous cell carcinoma may be modified by p14ARF genotypes, and these genotypes may play an even greater role in SPM of nonoropharyngeal cancers arising in ever-smokers and ever-drinkers. Supporting this hypothesis was the observation that the p14ARF variant genotypes were associated more strongly with SPMs at tobacco-associated sites than with SPMs at other sites. Although the risk of developing an SPM associated with p14ARF variant genotypes was greater among the observed subgroups, the significant association may have been by chance because of the rather small sample sizes of these subgroups. However, further large studies will be needed to validate our findings.
Although our results support an increased risk of developing an SPM after an index SCCHN associated with both p14ARF polymorphisms (individually and in combination) in a large and well characterized cohort of patients with SCCHN who were treated at a multidisciplinary cancer center, our findings have several inherent limitations. First, there may have been a selection bias for study patients because of the hospital-based nature of this study, and the inclusion of selected SNPs in this analysis based on allele frequency in these patients may have limited the external validity of this study. Although the sample size of our study was relatively large, the small number of SPMs in subgroups, especially when the patients were stratified, may have limited our ability to detect a certain degree of association. The low rate of SPMs in this patient cohort likely reflects the high prevalence of both never-smokers and patients who presented with late-stage disease as well as our strict criteria for defining SPMs. In addition, currently, this cohort has received relatively limited follow-up time (30 months) to develop SPMs. Furthermore, our patient cohort included approximately 85% non-Hispanic whites, and our findings may not be relevant to the risk of developing an SPM after an index SCCHN in other ethnicities. Although demographics, exposure, and clinical data for the cohort were collected prospectively, clinical outcomes, including SPMs, were collected retrospectively with no strictly defined screening or follow-up regimen. Finally, because of the retrospective nature of the original study design, we did not have information on HPV infection or on continued smoking behavior after the index SCCHN diagnosis, and these potential confounders may have biased the observed association. Therefore, our future studies on the association between genetic polymorphisms and risk of SPMs should incorporate HPV tumor status and smoking behavior after index cancer treatment into the study design.
CONFLICT OF INTEREST DISCLOSURES
Supported by a Research Training Award from the American Laryngological, Rhinological, and Otological Society (to E.M.S.); The University of Texas M. D. Anderson Cancer Center Start-Up Funds (to E.M.S.); and National Institute of Health (NIH) grants R01 ES11740 and CA131274 (to Q.W.), P-30 CA16672 (to The University of Texas M. D. Anderson Cancer Center), and CA135679 and CA133099 (to G.L.).