Vijayvel Jayaprakash, Nestor R. Rigual, Kirsten B. Moysich, Thom R. Loree, Maureen A. Sullivan Nasca, Ravi J. Menezes, Mary E. Reid. Chemoprevention of Head and Neck Cancer With AspirinA Case-Control Study. Arch Otolaryngol Head Neck Surg. 2006;132(11):1231–1236. doi:10.1001/archotol.132.11.1231
To evaluate the chemopreventive potential of aspirin against head and neck cancer.
Hospital-based case-control study.
National Cancer Institute–designated comprehensive cancer center.
Individuals who received medical services at the Roswell Park Cancer Institute, Buffalo, NY, between 1982 and 1998 and who completed a comprehensive epidemiologic questionnaire.
Main Outcome Measure
Aspirin use among 529 patients with head and neck cancer and 529 hospital-based control subjects matched by age, sex, and smoking status.
Aspirin use was associated with a 25% reduction in the risk of head and neck cancer (adjusted odds ratio, 0.75; 95% confidence interval, 0.58-0.96). Consistent risk reductions were also noted in association with frequent and prolonged aspirin use. Further, a consistently decreasing trend in risk was noted with increasing duration of aspirin use (Ptrend = .005). Risk reduction was observed across all 5 primary tumor sites, with cancers of the oral cavity and oropharynx exhibiting greater risk reduction. When analyzed by smoking and alcohol exposure levels, participants moderately exposed to either showed a statistically significant 33% risk reduction (adjusted odds ratio, 0.67; 95% confidence interval, 0.50-0.91), whereas participants exposed to both heavy smoking and alcohol use did not benefit from the protective effect of aspirin. The reduction in risk was relatively more significant in women.
Aspirin use is associated with reduced risk of head and neck cancer. This effect is more pronounced in individuals with low to moderate exposure to cigarette smoke or alcohol consumption.
Head and neck cancer (HNC) refers, in general, to carcinoma of the upper aerodigestive tract and is primarily a smoking- and alcohol-related malignancy. It is a complex disease arising in various sites including the nasopharynx, oropharynx, oral cavity, hypopharynx, and larynx. The most common form of HNC is squamous cell carcinoma (SCC). Non-SCCs of the head and neck, such as salivary gland tumors, are rare. Head and neck cancer remains an important cause of morbidity worldwide, with more than 630 000 new cases each year (oral cavity, 270 000; nasopharynx, 80 000; other pharyngeal, 130 000; and larynx, 159 000).1 In the United States, HNC is diagnosed in approximately 27 000 men and 12 000 women each year, representing 3.2% of all newly diagnosed cancers and accounting for 2.1% of cancer-related deaths.2 Tobacco use and alcohol intake, independently, are major risk factors for the development of HNC.3 The synergistic effect of alcohol and smoking results in a much increased risk of developing a malignancy compared with these exposures individually.3 Additional risk factors such as human papilloma virus, betel nut use, poor oral hygiene, and environmental carcinogens also seem to have a role in HNC carcinogenesis.4
Despite great advances in treatment methods, HNC is associated with severe disease and treatment-related morbidity. Only about 50% of patients survive at least 5 years after initial diagnosis, and this rate has not improved in more than 2 decades.2 One of the reasons for the poor prognosis is the advanced stage of disease at diagnosis. More advanced disease (stages III and IV) may be cured in only a small fraction (30%) of patients despite extensive multimethod therapy. To decrease the mortality from HNC, an effective chemopreventive agent is needed.
In recent years, convincing laboratory and epidemiologic evidence has emerged to demonstrate the chemopreventive potential of aspirin against various cancers.5 Intake of aspirin and similar nonsteroidal anti-inflammatory drugs (NSAIDs) has been associated with a decreased risk of developing colon, stomach, ovarian, and breast cancer.5,6 Studies have also shown risk reduction associated with aspirin use in smoking-related cancers such as lung7 and esophageal carcinoma.6 Although the mechanisms of HNC chemoprevention seem to be similar to those of nonsmoking-related cancers, the role of aspirin in preventing tobacco-induced carcinogenesis has not been comprehensively studied. Only a few studies have examined the chemopreventive effect of NSAIDs on head and neck lesions, and the results have been inconsistent.8,9 In light of experimental evidence that aspirin might be used as an effective chemopreventive agent in other malignant neoplasms and because of the scarcity of studies, the present hospital-based case-control study was conducted to examine the effect of aspirin on the risk of developing HNC.
Between 1982 and 1998, a comprehensive Patient Epidemiology Data System (PEDS) questionnaire was offered to every individual who received medical services at the Roswell Park Cancer Institute (RPCI), Buffalo, NY. The present study included patients who completed this PEDS questionnaire. The case group consisted of 529 patients with primary HNC and no other concurrent primary cancer and no history of malignancy. Cases included carcinoma of the nasopharynx (4.1%), oral cavity (31.9%), oropharynx (19.2%), hypopharynx (5.8%), larynx (26.7%), and salivary glands (8.5%). Cancers of skin of the lip and malignant melanoma were excluded from the analysis. Cancer history was obtained from the RPCI tumor registry and diagnostic index. The control group included 529 subjects selected from a pool of 6828 eligible persons without a diagnosis of any malignant conditions or benign tumors. The control group was frequency matched 1:1 with the case group by 5-year age group, sex, and smoking status (ie, never smoked, former smoker, or current smoker). The diagnoses of not having cancer in the control group were distributed across 39 disease classifications, including genitourinary or gynecologic disorders, 18.9%; infectious or parasitic diseases, 17.8%; circulatory or hematologic disorders, 14.2%; lower gastrointestinal tract or peritoneal disorders, 13.2%; respiratory tract or other thoracic disorders, 7%; and oral and upper gastrointestinal tract disorders, 5.9%. No other diagnosis individually accounted for more than 5% of diagnoses in the control group.
The PEDS questionnaire was an institutional review board–approved 16-page comprehensive epidemiologic questionnaire that inquired about demographic data, lifestyle, and medical history. Lifestyle information included smoking history, alcohol consumption, physical activity, and key occupational and environmental exposures. Medication history covered many drugs and vitamins, and questions were asked about the frequency and duration of drug use and the year when the drug was first taken. Information was not collected on the dosage used and the reason for use of the drug. Subjects who were ill when completing the questionnaire were instructed to indicate how often they took the medication before the onset of illness. Questions were asked specifically about both aspirin and other NSAID use, although the use of other NSAIDs was infrequent during the period. Approximately 50% of the patients completed the questionnaire. No other information was available for those who did not complete the questionnaire.
In the data analysis, patients who reported to have ever taken aspirin before the onset of the present illness were classified as aspirin users and those who reported no history of aspirin use before the onset of the present illness were classified as nonusers. The frequency of aspirin use was classified as follows: less than 1 tablet per week, 1 tablet per week, 2 to 6 tablets per week, or 7 tablets or more per week. Duration of aspirin use was stratified as follows: less than 10 years, 10 to 20 years, 21 to less than 40 years, and 40 years or more. An index was computed for cumulative use (tablet-years) by multiplying the number of tablets used per day times the number of years used. Cumulative use was categorized as follows: less than 2 tablet-years, 2 to less than 4 tablet-years, 4 to less than 9 tablet-years, and 9 tablet-years or more. All ranges for categorical variables were computed based on the distribution in the control group and using logical cutoff points.
Characteristics of case and control groups were compared using either 2-tailed t tests (continuous) or χ2 tests (categorical). The adjusted odds ratio (AOR) and 95% confidence interval (CI) were calculated using unconditional logistic regression. Potential confounders and significant covariates such as age, sex, packs of cigarettes smoked per day, and alcohol intake per week were included in the final regression model. P values for trend were calculated using the nonexposed groups as reference. The statistical package SPSS version 13.0 (SPSS Inc, Chicago, Ill) was used for all analyses.
Characteristics of HNC in the case and control groups are given in Table 1. Since the 2 groups were matched for age (±5 years), sex, and smoking status (ie, never smoked, former smoker, and current smoker), these variables showed minimal difference between case and control groups. Level of education was lower in the case group; only 32.3% had attended college vs 37.2% in the control group (P = .09). As expected, patients in the case group were significantly heavier smokers compared with the control group (P<.001); 44.8% smoked more than 1 pack of cigarettes per day compared with 29.5% in the control group. A similar distribution was noted for alcohol use (P<.001); 32.7% of patients in the case group reported drinking more than 10 alcoholic drinks per week compared with 19.1% in the control group. Combined smoking and alcohol use in this population was calculated and stratified into 3 exposure categories: never exposed (never smoked cigarettes or drank alcohol), moderately exposed (either <1 pack of cigarettes per day or ≤1 drink of alcohol per day), and highly exposed (≥1 packs of cigarettes per day and >1 drink of alcohol per day). Results showed that, among men, 35.1% of the case group were in the high-exposure category compared with 20.4% of the control group (P<.001). In women, the percentages in the high-exposure category in the case and control groups were 21.2% and 12.2%, respectively (P = .008). Data also show that men were heavier smokers and drinkers compared with women. Table 1 gives results that are in concordance with well-established risk factors in the literature, thus strengthening the possibility that the aspirin results might also be generalizable to the population at-large.
The adjusted association between aspirin use and HNC risk in the total sample is given in Table 2. Aspirin users showed a significant 25% reduction in risk of HNC compared with nonusers (AOR, 0.75; 95% CI, 0.58-0.96; P = .02). Various estimates of quantifying exposure were calculated; the magnitude of the observed risk reduction was consistent across all quantifying methods. When the frequency of aspirin use was stratified into categories, results showed an approximately 30% to 40% reduction in risk in all categories except in the group who used 2 to 6 tablets per week. Duration of aspirin use categories showed a consistent trend in reduction of risk (Ptrend = .005), with more than 10 years of use associated with a statistically significant 33% to 37% reduction in risk. Cumulative aspirin use of more than 2 tablet-years also showed a consistent reduction in risk of approximately 21% to 39% (Ptrend = .05). To examine the effect of aspirin use on SCC of the head and neck only, similar analysis was tried after excluding all non-SCCs. The results were similar to the overall results in Table 2 (data not shown). When stratified by sex, the risk reduction was more significant in women (data not shown).
Aspirin use was associated with a decreased risk of cancer at all 5 major head and neck sites (11%-32%), with risk reduction for cancers of the oral cavity and oropharynx approaching significance at P = .08 (Table 3). Risk reduction was also noted in nasopharyngeal carcinoma, a nonsmoking- and alcohol-related cancer.
The effect of aspirin use on HNC in smokers and drinkers is given in Table 4. Smoking and drinking exposures were classified into 3 groups, as follows: group 1, never exposed (never smoked cigarettes or drank alcohol); group 2, moderately exposed (either <1 pack of cigarettes per day or ≤1 drink of alcohol per day); and group 3, highly exposed (≥1 packs of cigarettes per day and >1 drink of alcohol per day). There was a nonsignificant 37% reduction in risk in group 1 (AOR, 0.63; 95% CI, 0.26-1.53). When aspirin users and nonusers were compared in group 2, a significant 33% reduction in risk was observed (AOR, 0.67; 95% CI, 0.50-0.91). There was no reduction in risk associated with group 3, the group highly exposed to smoking and alcohol use.
The reduction of risk in group 2 (moderate users) was statistically significant in women (AOR, 0.51; 95% CI, 0.30-0.85) but not in men (AOR, 0.77; 95% CI, 0.54-1.12) (data not shown). In addition, the subgroup analysis of group 2 revealed that individuals who either smoked heavily (≥1 packs per day) or drank heavily (>1 drink per day of alcohol), but not both, had a stronger reduction in risk (AOR, 0.50; 95% CI, 0.32-0.78) (data not shown).
Results from this hospital-based, case-control study support the hypothesis that aspirin use is associated with reduced risk of HNC. A plausible explanation is that aspirin, a nonselective cyclooxygenase (COX) inhibitor, blocks the action of COX-1 and COX-2, in turn inhibiting prostaglandin (PG) synthesis, particularly PGE2.10 Expression of COX-2 and PGE2 in tumors has been proved to be associated with suppressed immune response, enhanced inflammation, increased angiogenesis, augmented cellular proliferation, and inhibited apoptotic pathway, thereby promoting tumor progression and invasion.5 This theory is further supported by studies that demonstrate an increased level of COX expression in various other malignant lesions.11,12 Numerous laboratory investigations have also revealed up-regulated COX-2 and PGE2 expression in premalignant and malignant head and neck lesions.13,14 Immunohistochemical analyses have also shown a gradient increase in expression of COX-2 depending on the grade of the oral mucosal lesion, from hyperplasia to dysplasia, with the highest expression in SCC.15 In addition, administration of aspirin and other NSAIDs has been shown to inhibit HNC in animal models.16,17
Despite these studies, not much research has been done on the chemopreventive potential of aspirin on HNC in human beings. Thun et al8 examined the relative risk of death from buccal cavity and pharyngeal cancers in aspirin users and found a 30% nonsignificant risk reduction in some categories of aspirin use. Bosetti et al9 examined the effect of aspirin use on 965 upper aerodigestive cancers (393 oral and pharyngeal, 225 esophageal, and 347 laryngeal) and noted a 67% reduction in risk in users who took aspirin for more than 5 years. None of these studies were designed specifically to study the chemopreventive effect of aspirin on HNC.
In the current study, aspirin users showed a statistically significant 25% reduction in risk of HNC compared with nonusers. This study also showed consistent risk estimates below unity for most methods of quantifying aspirin exposure, with long-term use (>10 years) associated with a 33% to 37% reduction in risk (Ptrend = .005). These results are consistent with previous studies, further strengthening the hypothesized association between aspirin use and reduction in HNC risk. Similar risk reduction was also noted in all the major sites of HNC, with head and neck SCCs, especially cancers of the oral cavity and oropharynx, showing a relatively more significant risk reduction. Nasopharyngeal cancer, a nonsmoking- and alcohol-related tumor, also showed a 12% reduction in risk. Risk reduction appeared to be greater in women at all strata.
Tobacco and alcohol are important risk factors for HNC. Studies have shown that tobacco smoke has an effect on increased COX-2 expression in cancer tissues.18 The COX enzymes have also been found to convert polycyclic aromatic hydrocarbons present in tobacco smoke into reactive metabolites, which, in turn, form metabolic DNA adducts, thereby increasing the chances of carcinogenesis.19 Another case-control study examined the relationship between smoking and aspirin use in patients with lung cancer and reported a reduction in the protective effect of aspirin among heavier smokers, which suggests an effect modification.7 Ethanol functions as a cancer promoter by solubilizing the phospholipid component of cellular membranes, thereby enhancing cellular uptake of carcinogens such as tobacco smoke.20 The risk of oropharyngeal cancer has been noted to increase by more than 35-fold in individuals who are both heavy smokers and drinkers compared with abstainers.3 In the United States, the combined effects of tobacco and alcohol are estimated to be responsible for about 73% of oral and pharyngeal cancers in non-Hispanic white persons and 83% of African American persons.21 It is conceivable that the protective effect of aspirin might be modified by the synergistic carcinogenic effect of cigarette smoke and alcohol on the tissues. However, the combined effects of smoking and alcohol consumption on the chemopreventive action of aspirin have not been documented. The present study examined the combined effects of smoking and alcohol consumption on chemopreventive effects of aspirin on HNC. Aspirin use was associated with a significant 33% reduction in risk in the group composed of either moderate smokers or moderate drinkers. No reduction in risk was noted in the group with high exposure to both tobacco and alcohol, which suggests a possible effect modification.
A few limitations must also be considered when interpreting these results. The study population was restricted to patients at RPCI. This selection may not be representative of the general population. Despite this limitation, inherent in all hospital-based, case-control studies, the characteristics of the case and control groups (Table 1) show that the distribution of the risk factors in the study population is comparable to the well-established results noted in the literature.3,22 Moreover, to minimize the bias from overrepresentation of a particular group, the control subjects were randomly selected from a pool of 6828 eligible subjects who visited RPCI.
The analysis was restricted only to aspirin use and did not address the use of other NSAIDs. However, since over-the-counter non-aspirin NSAIDs were available only during the last few years of the study period, the risk of misclassification might be minimal. Another limitation is that only about 50% of the eligible subjects completed the PEDS questionnaire. The information on characteristics of individuals who did not complete the questionnaire is unavailable to compare them with those who did. Nevertheless, the PEDS database has been used previously to estimate the effect of aspirin use on various cancers and statistically significant inverse associations have been previously replicated.7,23,24
In summary, the results of this study support the hypothesis that aspirin use reduces the risk of HNC. The reduction in risk is observed across all the primary tumor sites studied, with comparatively higher risk reduction for head and neck SCCs, especially oral cavity and oropharyngeal cancers. The most significant risk reduction was noted in the subset who were either moderate smokers or moderate drinkers. This protective effect was greater in women than in men. However, larger clinical trials are needed to confirm the strength of the association and to best define the target population for aspirin chemoprevention.
Correspondence: Mary E. Reid, PhD, Department of Epidemiology, Division of Cancer Prevention and Population Sciences, 308 Carlton House, Roswell Park Cancer Institute, Elm and Carlton streets, Buffalo, NY 14263.
Submitted for Publication: January 18, 2006; final revision received April 20, 2006; accepted May 6, 2006.
Author Contributions: All authors had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Jayaprakash, Rigual, Moysich, Loree, Sullivan Nasca, and Reid. Analysis and interpretation of data: Jayaprakash, Moysich, Menezes, and Reid. Drafting of the manuscript: Jayaprakash, Rigual, and Reid. Critical revision of the manuscript for important intellectual content: Jayaprakash, Rigual, Moysich, Loree, Sullivan Nasca, Menezes, and Reid. Statistical analysis: Jayaprakash, Moysich, Menezes, and Reid. Clinical expertise: Rigual, Loree, and Sullivan Nasca. Study supervision: Reid.
Financial Disclosure: None reported.