Fax: (011) 45-3268-3165
A prospective study of tobacco and alcohol use as risk factors for pharyngeal carcinomas in Singapore Chinese
Article first published online: 21 FEB 2007
Copyright © 2007 American Cancer Society
Volume 109, Issue 6, pages 1183–1191, 15 March 2007
How to Cite
Friborg, J. T., Yuan, J.-M., Wang, R., Koh, W.-P., Lee, H.-P. and Yu, M. C. (2007), A prospective study of tobacco and alcohol use as risk factors for pharyngeal carcinomas in Singapore Chinese. Cancer, 109: 1183–1191. doi: 10.1002/cncr.22501
- Issue published online: 7 MAR 2007
- Article first published online: 21 FEB 2007
- Manuscript Accepted: 4 DEC 2006
- Manuscript Revised: 28 NOV 2006
- Manuscript Received: 12 OCT 2006
- National Cancer Institute. Grant Numbers: R01 CA55069, R35 CA53890, R01 CA80205
- The Danish Cancer Society
- The Danish Agency for Science
- The Hans and Emma Skoubys Foundation
- nasopharyngeal carcinoma;
- pharyngeal carcinoma;
Nasopharyngeal carcinoma (NPC) is a rare disease in most populations; however, in areas of Southeast Asia and North Africa and in the Arctic, undifferentiated NPC is the most frequent pharyngeal malignancy. Although smoking and alcohol have been established firmly as synergistic risk factors for other pharyngeal carcinomas, previous studies on the association between these risk factors and NPC have not been consistent. Therefore, the authors analyzed this relation in a cohort of Singapore Chinese, which is a population with a high incidence of NPC.
From 1993 to 1998, a population-based cohort of 61,320 Singapore Chinese ages 45 years to 74 years who were free of cancer completed a comprehensive interview on living conditions and dietary and lifestyle factors. By linkage to Singapore population-based registries, the cohort was followed through 2005, and cancer occurrence was determined. The relative risk of NPC and other oropharyngeal carcinomas in the cohort was investigated by using a Cox proportional hazards model.
In total, 173 NPCs and 75 other oropharyngeal carcinomas were observed during 601,879 person-years of follow-up. Smoking for >40 years was associated with a doubled risk of NPC (relative risk, 2.0; 95% confidence interval, 1.2–3.3), whereas smoking intensity, age at smoking initiation, and alcohol consumption were not associated with NPC risk. In contrast, smoking duration, smoking intensity, age at smoking initiation, and alcohol consumption all were associated with an increased risk of other oropharyngeal carcinoma (P for trend, <.0001).
Smoking and alcohol influenced the risk of NPC and other oropharyngeal carcinomas differently in a high-incidence NPC population. Long-term smoking was a risk factor for NPC, but alcohol consumption was not. Cancer 2007. © 2007 American Cancer Society.
With incidence rates <1 per 100,000 population per year, nasopharyngeal carcinoma (NPC) is a rare disease in most populations, although it is diagnosed frequently among populations in the southern part of China, in Southeast Asia, in North Africa, and in the Arctic.1, 2 In these NPC-endemic areas, NPC constitutes the most frequent pharyngeal malignancy, and its incidence rates exceeds 40 per 100,000 per year in men residing in certain Cantonese-speaking counties of southern China.3 In addition to its remarkable geographic distribution, NPC differs from other pharyngeal carcinomas in its distinctive pathogenesis. NPC is associated with Epstein-Barr virus (EBV), and viral genomes are detected in the vast majority of NPC tumors from high-incidence areas.4, 5 The involvement of a genetic factor in the development of the disease also is widely accepted, and the familial risk of NPC is among the highest of any malignancy.6–8 Other major risk factors for NPC include the consumption of preserved foods, in particular, Chinese salted fish, which has been associated consistently with an increased risk of NPC.9, 10
Although cigarette smoking and alcohol consumption are firmly established as synergistic risk factors for other pharyngeal carcinomas,11, 12 previous studies on the association between smoking and NPC in NPC-endemic areas have not been consistent. The majority of case-control studies have implicated the nasopharynx as a tobacco-susceptible cancer site, and 2- to 4-fold increased risks of NPC have been observed among heavy smokers compared with nonsmokers.1, 13–15 This is much less pronounced than in other pharyngeal sites, and a number of other studies from endemic areas have reported no difference in the risk of NPC between smokers and nonsmokers.16–19
Studies on alcohol consumption and the risk of NPC have suggested a limited influence of alcohol on the predominantly undifferentiated NPC observed in Chinese populations.1 However, a significantly increased risk of NPC has been observed in heavy drinkers in Western populations, where differentiated forms of NPC are more frequent, indicating a difference in risk between Asian and Western populations.1, 15, 20
Using a large, prospective cohort of Singapore Chinese, we examined and clarified the relationship between smoking, alcohol consumption, and the risk of NPC as well as other oropharyngeal carcinomas (OPCs). Singapore Chinese are at high risk for NPC, where the age-standardized incidence rates of NPC were 16.3 for men and 5.4 for women per 100,000 per year from 1993 to 1997: These rates are comparable to those in reported in Hong Kong.2 The current study represents the first large, prospective study to our knowledge of the association between these common risk factors and NPC in a high-incidence population.
MATERIALS AND METHODS
In 1993, a population-based cohort study among middle-aged and older ethnic Chinese men and women was initiated in Singapore. Participants were either Hokkien or Cantonese, the 2 major dialect groups of Singapore Chinese, who originate from the contiguous provinces of Fujian and Guangdong, respectively, in southern China. The sampling frame consisted of randomly selected, government-built housing estates all over the city-state (86% of Singaporeans resided in such facilities during the enrolment period). Trained interviewers canvassed, from door to door, all residential units within the sampling frame. Eligible individuals (self-identified Cantonese or Hokkien men and women ages 45–74 years) were asked to participate in the cohort study, and 85% consented. The main focus of the Singapore Chinese Health Study was the role of diet and nutrition in cancer etiology; and, from 1993 to 1998, a total of 63,257 individuals ages 45 to 74 years were enrolled.21 Because 1937 individuals had a history of cancer prior to enrolment, those individuals were excluded from the analysis; thus, the final cohort consisted of 61,320 individuals. The study was approved by the Institutional Review Boards of the National University of Singapore and the University of Minnesota.
At the time of recruitment, a face-to-face interview was conducted, and information on demographics, physical activity, tobacco and alcohol use, reproductive history, medical history, and family history of NPC was obtained by using a structured questionnaire. The questionnaire included a validated, semiquantitative food-frequency section, which listed 165 food items, including a comprehensive selection of Chinese preserved foods.21 Because it was reported previously that the intake of Chinese-style preserved foods, especially fish products, and fresh vegetables influenced the risk of NPC,1 we calculated the total intake of Chinese-style preserved proteins and vegetables in adulthood for each individual based on the food-frequency questionnaire.
Questions concerning cigarette smoking covered age at smoking initiation (ages ≤14 years, 15–19 years, 20–29 years, and ≥30 years), years of regular smoking (≤9 years, 10–19 years, 20–29 years, 30–39 years, and ≥40 years), smoking intensity (1–6 cigarettes per day, 7–12 cigarettes per day, 13–22 cigarettes per day, 23–32 cigarettes per day, 33–42 cigarettes per day, and ≥43 cigarettes per day), and, for exsmokers, the time since smoking cessation (<1 year, 1–2 years, 3–4 years, 5–9 years, 10–14 years, 15–19 years, and ≥20 years). Questions on alcohol intake asked about the past year's consumption of beer, rice wine, grape wine, and hard liquor separately. For each of the 4 types of alcoholic beverages, frequency of intake (never or hardly ever, once a month, 2–3 times per month, once a week, 2–3 times per week, 4–6 times per week, once a day, and ≥2 times per day) and the number of drinks per intake were ascertained. One drink was defined as 1 small bottle (375 mL) of beer, 1 cup (30 mL) of rice wine, 1 glass (118 mL) of grape wine, or 1 shot (30 mL) of hard liquor. The amounts of ethanol per drink are comparable across the 4 types of alcoholic beverages, ranging from 10.85 g to 13.5 g. For ease of comparison with published data, the number of drinks per week was used to denote alcohol intake levels in study participants.
Identification of Cancer Patients
In Singapore, each citizen or permanent resident is provided with a unique personal identifier, the National Registration Identity Card (NRIC) number. For the current study, all incident cancer cases and deaths among cohort members were identified through routine record linkage analysis of individual NRIC numbers with the nationwide Singapore Cancer Registry and the Singapore Registry of Births and Deaths. The Singapore Cancer Registry is based on notifications from all sections of the medical profession in Singapore and is supplemented by information from pathology reports, hospital discharge records, and death certificates.22 Tumors are coded by using the International Classification of Disease for Oncology (ICD-O). We used ICD-O topography code C11 to define cancer of the nasopharynx and used ICD-O codes C01 through C06, C09 and C10, and C12 and C13 to define other OPCs, thus excluding cancers of the lip from this group. To date, only 17 patients (0.03%) were lost to follow-up because of migration out of Singapore.
In total, as of December 31, 2005, 176 cases of histopathologically confirmed NPC had occurred, Three cases of primary lymphomas were excluded, which left 173 cases of NPC in the current analysis. Among these, there were 153 were undifferentiated carcinomas, 12 unspecified carcinomas, 5 squamous cell carcinomas, and 3 adenocarcinomas.
During the same period of follow-up, 87 cases of other OPCs occurred among the cohort participants. All 87 cases were histopathologically confirmed. After excluding 12 primary lymphomas, 75 cases of other OPCs were included in the current study.
Person-years of follow-up for each cohort member were counted from the date of the baseline interview to the date of cancer diagnosis or death, the date of last contact for the 17 participants who had migrated out of Singapore, or December 31, 2005, whichever occurred first. The relative risks (RRs) and their 95% confidence intervals (95% CIs) of NPC and other OPCs associated with various exposure variables was estimated by using Cox proportional hazards modeling. The proportionality assumption was tested and was satisfied. All models were adjusted for age at interview, sex, dialect group, year of interview, level of education (no formal schooling, primary school, secondary school or above), summed intakes of protein-rich preserved food items (in quartiles), summed intakes of fresh vegetables (in quartiles), and history of familial NPC. A positive history of familial NPC was defined as having an NPC diagnosis in ≥1 first-degree relative(s) (parents, siblings, or children) before the baseline interview. Models examining the association between smoking and the risk of cancer were adjusted further for alcohol consumption (drinks per week), whereas models examining alcohol consumption and the risk of cancer were adjusted further for the number of years of smoking (nonsmokers, smoking for <40 years, or for ≥40 years). We also examined the associations of age at smoking initiation, number of cigarettes smoked per day, and number of alcoholic drinks per week with the risk of NPC and other OPCs stratified by number of years of smoking.
We performed tests for linear trend on levels of smoking or alcohol intake by using ordinal scores for the 3 or 4 levels that were created. We used a polytomous logistic regression model to directly compare the estimated RR for NPC and other OPCs associated with smoking or alcohol use. Statistical analyses were conducted by using SAS software (version 9.1; SAS Institute Inc., Cary, NC). All P values quoted are 2-sided.
After 12 years of follow-up of 61,320 cohort members who were free of cancer at baseline (601,879 person-years), 173 patients were identified with NPC (Table 1), yielding incidence rates of 45.3 and 16.1 per 100,000 person-years for men and women, respectively. Similarly, there were 75 incident cases of other OPCs (Table 1), corresponding to incidence rates of 21.3 and 5.6 per 100,000 person-years in men and women, respectively. The mean ages at diagnosis of NPC for men and women were 59.3 years and 58.7 years, respectively. The corresponding ages for men and women with other OPCs were 67.2 years and 63.6 years, respectively. Women had a lower risk of NPC (RR, 0.4; 95% CI, 0.2–0.5) and other OPCs (RR, 0.7; 95% CI, 0.3–1.2) than men. The incidence rates of NPC decreased with increasing age, whereas the rate of other OPCs increased with age in the range from 45 years to 79 years (P for trend, <.001 for both). Compared with Cantonese, Hokkiens had a significantly lower risk of NPC (RR, 0.7; 95% CI, 0.5–1.0; P = .04), but they had a higher risk of other OPCs (RR, 1.4; 95% CI, 0.9–2.3), although the latter difference was not significant. A high level of education was associated inversely with the risk of both NPC and other OPCs.
|Characteristic||Total population, N = 61,320||Nasopharyngeal carcinoma, n = 173||Other oropharyngeal carcinomas, n = 75|
|Age at interview, y|
|Level of education|
|No formal schooling||16,661||27.2||32||18.5||17||22.7|
|Secondary school or above||17,436||28.4||53||30.6||16||21.3|
|Frequency of alcohol consumption|
Smoking prevalence in the study population was greater in men (36.4%) than in women (6.2%). The overall RR of NPC was comparable for never smokers, exsmokers, and current smokers. Exsmokers who had quit smoking for <10 years and ≥10 years had a risk of NPC comparable to that for lifelong nonsmokers (Table 2). Compared with never smokers, current smokers with ≥40 years of smoking had doubled their risk of NPC (RR, 2.0; 95% CI, 1.2–3.3) (Table 2). There was only 1 patient with NPC who was an exsmoker with ≥40 years of smoking. Therefore, we were unable to examine the impact of currency of smoking on the association between smoking duration and disease risk. The sex-stratified RR of NPC associated with smoking for >40 years was 1.9 (95% CI, 1.1–3.5) and 3.6 (95% CI, 1.2–10.8) for men and women, respectively. Starting to smoke cigarettes at an early age was associated with an increased risk of NPC, but the association was of statistically borderline significance (P = .08). The number of cigarettes smoked per day was not associated with a significantly increased risk of NPC. We also examined the associations between the age at which individuals started to smoke cigarettes or the number of cigarettes per day and the risk of NPC stratified by the number of years of smoking. Among smokers who had <40 years of smoking, smokers, regardless of age at smoking initiation or smoking intensity, had a risk of NPC comparable to that of never smokers (Table 3). Conversely, among smokers who had ≥40 years of smoking, the significantly increased risk of NPC was not influenced by age at smoking initiation or the number of cigarettes per day.
|Variable||Total no. of individuals, N = 61,320||Nasopharyngeal carcinoma||Other oropharyngeal carcinomas|
|No. of patients N = 173||RR (95% CI)*||P for trend||No. of patients N = 75||RR (95% CI)*||P for trend|
|Former smokers||6680||19||0.8 (0.5–1.4)||10||1.4 (0.6–3.2)|
|Quit >10 y||3600||6||0.5 (0.2–1.1)||3||0.7 (0.2–2.6)|
|Quit <10 y||3080||13||1.2 (0.7–2.3)||7||2.3 (0.9–5.7)|
|Current smokers||12,057||54||1.3 (0.9–1.9)||42||3.5 (1.9–6.4)|
|Age at smoking initiation, y†|
|≥20||5278||17||1.0 (0.6–1.7)||12||2.5 (1.2–5.4)|
|15–19||4476||26||1.6 (1.0–2.5)||12||2.8 (1.3–6.3)|
|<15||2303||11||1.5 (0.8–2.8)||18||7.7 (3.7–15.9)|
|Smoking history, y†|
|1–39||6896||30||1.0 (0.7–1.6)||14||2.5 (1.2–5.3)|
|≥40||5161||24||2.0 (1.2–3.3)||28||4.8 (2.4–9.5)|
|No. of cigarettes per d†|
|1–12||4918||25||1.6 (1.0–2.6)||11||2.6 (1.2–5.7)|
|13–22||5015||21||1.1 (0.7–1.9)||17||3.6 (1.7–7.6)|
|≥23||2124||8||1.0 (0.5–2.0)||14||6.5 (2.9–14.6)|
|Frequency of alcohol consumption|
|Monthly||4494||15||1.0 (0.6–1.6)||3||0.6 (0.2–2.1)|
|Weekly||5018||20||1.0 (0.6–1.7)||14||2.3 (1.2–4.3)|
|Daily||2159||10||1.2 (0.6–2.3)||14||4.1 (2.2–7.7)|
|No. of drinks per wk|
|1–7||8794||30||0.9 (0.6–1.4)||14||1.4 (0.8–2.7)|
|>7||2877||15||1.3 (0.8–2.3)||17||3.8 (2.1–7.0)|
|Variable||Nasopharyngeal carcinoma||Other oropharyngeal carcinomas|
|1–39 years of smoking||≥40 years of smoking||1–39 years of smoking||≥40 years of smoking|
|Age at smoking initiation, y|
|No. of patients||13||4||6||6|
|RR (95% CI)*||0.9 (0.5–1.6)||1.8 (0.6–5.2)||1.8 (0.7–4.8)||4.0 (1.5–11.1)|
|No. of patients||17||20||8||22|
|RR (95% CI)*||1.2 (0.7–2.2)||2.0 (1.1–3.4)||3.8 (1.5–9.6)||5.0 (2.4–10.1)|
|No. of cigarettes per day|
|No. of patients||27||19||9||19|
|RR (95% CI)*||1.1 (0.7–1.8)||2.0 (1.1–3.4)||2.1 (0.9–5.0)||4.2 (2.0–8.7)|
|No. of patients||3||5||5||9|
|RR (95% CI)*||0.5 (0.2–1.7)||1.9 (0.7–4.9)||4.8 (1.6–14.1)||7.8 (3.2–19.2)|
The risk of other OPCs for current smokers was 3.5-fold that of never smokers (Table 2). There was a strong, graded dose-response relation between age at smoking initiation, number of years of smoking, or number of cigarettes per day and the risk of other OPCs (Table 2). The highest RR was for smokers who began to smoke before age 15 years (RR, 7.7; 95% CI, 3.7–15.9). The sex-stratified RR of other OPCs associated with smoking for >40 years was 4.2 (95% CI, 1.9–9.7) and 7.1 (95% CI, 2.1–24.5) for men and women, respectively. When smokers were stratified further according to the number of years of smoking, the RR of other OPCs was increased again by early age at smoking initiation and the number of cigarettes smoked per day (Table 3).
We conducted formal statistical testing of differences in the smoking-NPC versus smoking-OPC associations. Statistically, the influences of smoking status, age at smoking debut, and smoking intensity differed significantly for the 2 cancer sites (2-sided values: P = .02, P = .01, and P = .003, respectively), whereas the impact of the duration of smoking on the risk of the 2 cancer sites was of borderline statistical significance (P = .10).
The prevalence of alcohol drinking on a weekly basis was greater in men (20.8%) than in women (4.4%). For alcohol consumption, neither the frequency nor the number of drinks was associated with the risk of NPC (Table 2). Conversely, the level of alcohol consumption was associated with a statistically significant increase in risk the of other OPCs (P for trend, <.0001).
We also tested for possible differences in the alcohol-NPC versus alcohol-OPC associations. Statistically, the influence of the frequency of alcohol consumption and the number of weekly drinks differed significantly between the 2 cancer sites (P = .004 for both).
Table 4 shows the RR of NPC and other OPCs stratified by the number of drinks per week and the number of years of smoking (for current smokers only). Drinkers who had <40 years of smoking had a risk of NPC comparable to that of individuals who neither smoked cigarettes nor consumed alcohol regularly. Among long-term smokers (ie, ≥40 years of smoking), an increased amount of alcohol consumed did not add further to the risk of NPC. In contrast, alcohol consumption had a significant influence on the risk of other OPCs. Compared with nonsmokers and nondrinkers, the RR was 18.4 (95% CI, 7.5–45.1) for individuals who smoked for ≥40 years and consumed >7 drinks per week.
|No. of drinks per week||Nasopharyngeal carcinoma||Other oropharyngeal carcinomas|
|1–39 years of smoking||≥ 40 years of smoking||1–39 years of smoking||≥40 years of smoking|
|No. of patients||20||16||6||11|
|RR (95% CI)*||1.2 (0.7–2.1)||2.0 (1.1–3.6)||2.2 (0.8–5.9)||2.9 (1.2–6.8)|
|No. of patients||5||5||5||6|
|RR (95% CI)*||0.7 (0.3–1.8)||2.3 (0.9–5.9)||4.8 (1.6–14.3)||6.2 (2.2–17.5)|
|No. of patients||5||3||3||11|
|RR (95% CI)*||1.2 (0.5–3.1)||1.8 (0.5–5.9)||4.9 (1.3–18.5)||18.4 (7.5–45.1)|
Five patients with NPC and 1 patient with other OPC had a first-degree relative who was afflicted by NPC. A familial history of NPC was associated with an RR of 2.1 (95% CI, 0.9–5.2) for NPC but only 1.3 (95% CI, 0.2–9.2) for other OPCs. After the exclusion of 20 patients with NPC who had histologies other than the undifferentiated cell type, the RR for undifferentiated NPC associated with familial NPC increased to 2.8 (95% CI, 1.1–6.9) compared with individuals who did not have a familial history of NPC.
Approximately 11% of the cohort participants were migrants from China. We examined possible differences in the smoking/alcohol-pharyngeal cancer correlations by place of birth (China vs Singapore/Malaysia), and no discernible differences were observed. We also repeated the analyses described above on the subset of patients with NPC who had undifferentiated tumors, and the results were similar.
The current study is the first large, prospective study to our knowledge that addresses smoking, alcohol, and the risk of NPC in a population with a high-incidence of NPC. It documents a significantly increased risk of NPC in long-term smokers.
Our observations of a lower risk of NPC among women compared with men, among Hokkien compared with Cantonese, and with higher education and increasing age ≥45 years are compatible with the literature.1 The observation of a higher risk of other OPCs among men compared with women and with decreasing education and increasing age also is compatible with earlier findings.23 Incidence rates of NPC and other OPCs in the cohort were comparable to rates reported among Singapore Chinese aged ≥45 years in the Singapore Cancer Registry from 1993 to 1997 when the analysis was adjusted for age by using the proportion of observed person-years by age groups within the cohort.2
Tobacco is an established risk factor for the development of pharyngeal carcinomas other than NPC. However, earlier case-control studies on smoking and the risk of NPC have been inconsistent.1, 13–19 Results from case-control studies may be prone to recall bias, but a previous, small cohort study from Taiwan offered no clarification, because the nonsignificant finding of a 4-fold increased risk of NPC among current smokers was impeded by a very low number of cases.24
We observed that smoking for ≥40 years was associated with a 2-fold increased risk of NPC, which is in accordance with results from the largest of the previous case-control studies.10 Although most of the earlier, positive, retrospective studies reported a dose-response relation between the number of daily cigarettes and the risk of NPC, the absence of a dose-response relation was reported by others.15, 24 We observed that the number of daily cigarettes had no influence when the analyses were stratified for the duration of smoking, suggesting that duration more than intensity of smoking determines the risk of NPC associated with smoking in this population. We noticed relatively comparable risks of NPC in exsmokers versus never smokers, suggesting that smoking cessation may exert a substantial effect on NPC risk reduction relative to continued smoking.
The risk associated with smoking is present in both sexes and in both carcinoma sites. However, the moderate effect of smoking and the absence of a dose-response relation between smoking intensity and the risk of disease clearly separate NPC from the other OPCs. The risk of other OPCs, as documented in the same cohort, increased significantly with both the duration of smoking and the number of daily cigarettes, equivalent to the relation between smoking and the risk of developing cancer at other sites along the respiratory tract.12
Reduced exposure of the nasopharyngeal epithelium to tobacco compounds probably cannot explain the difference between NPC and other OPCs. Even with a reduced exposure, a dose-response relation between smoking intensity and the risk of NPC would be expected, and others have reported a high risk of carcinomas of the sinonasal region associated with smoking, indicating a significant amount of tobacco compounds in the nasal cavity.25, 26 It is more likely that the reason for the differentiated risk of smoking in NPC and other OPCs must be sought in the differences among carcinogenesis, as illustrated by the histopathologic differences between the tumors. Although other OPCs in general are differentiated squamous cellular carcinomas, the vast majority of NPCs from high-incidence areas are undifferentiated carcinomas associated with EBV.27 In a Western population, it was demonstrated that the association between smoking and NPC was highly dependent on the histologic type, because a 6-fold increased risk of differentiated squamous cell NPC was observed in smokers, whereas the risk of undifferentiated NPC was comparable to that in nonsmokers.28
Although the exact mechanism by which smoking causes NPC is unknown, >60 carcinogens have been identified in tobacco smoke.29 One possible compound is nitrosamines, a class of tobacco-related chemicals that has been implicated in nasopharyngeal carcinogenesis. Nitrosamines induced nasal cavity tumors in rats, and high concentrations of nitrosamines have been detected in preserved foods associated with NPC development in high-incidence areas.1, 9
However, the close association between NPC and EBV infection may represent an alternative mechanism for the smoking-related risk of the disease. Smoking has a substantial impact on the immune system, and long-term smoking significantly reduces serum levels of immunoglobulins in humans, whereas loss of T-cell proliferation has been observed in animals chronically treated with nicotine.30 Thus, hypothetically, smoking may facilitate NPC development by interacting with the EBV infection. This possibility is supported in part by the observation that smoking is more common among patients with EBV-positive Hodgkin lymphoma than among EBV-negative patients with Hodgkin lymphoma.31, 32 The mechanism by which smoking and EBV interact is unknown. Formaldehyde, a compound that is present in tobacco smoke, may play an interactive role with EBV in nasopharyngeal carcinogenesis. Occupational exposure to formaldehyde has been associated with an increased risk of NPC,14, 33 especially among individuals with positive EBV-specific immunoglobulin A antibodies.33
Tobacco chewing is an established risk factor for OPC23 but has not been associated with an increased risk of NPC.34 In the current study, we could not address this specific tobacco exposure-pharyngeal cancer relation because of the negligible use of this tobacco product in our study population.
On the basis of light-microscopic studies, the World Health Organization has classified NPC into 3 histologic types: 1) keratinizing squamous cell carcinoma, 2) differentiated nonkeratinizing carcinoma, and 3) undifferentiated carcinoma.35 Among the southern Chinese who are at high-risk for NPC, including our study population, virtually all NPCs are undifferentiated tumors. In other words, the current study did not allow for a meaningful analysis of the smoking/alcohol-NPC associations by histologic subcategories.
Our finding that alcohol consumption was not associated with an increased risk of NPC confirmed earlier results from Chinese populations.1 A number of studies performed in Western populations linked alcohol to an increased risk of NPC, especially the highly differentiated type,20, 28 and our results generally support the belief that the factors responsible for NPC development are different in high- and low-incidence populations. The intake of alcohol in this population generally was low compared with that in other populations.36, 37 However, even with such a low alcohol consumption level, a clear linear correlation was observed between the amount of alcohol consumed and the risk of other OPCs.
Familial clustering of NPC in Chinese populations is well described, and earlier studies estimated a 5- to 10-fold increased risk of NPC among first-degree relatives of patients with NPC.10, 13, 38–41 Our current results indicate a 2- to 3-fold increased risk of NPC among first-degree relatives, which is in the lower end compared with earlier studies. However, recall bias in earlier case-control studies may have resulted in overestimated risks, whereas the risk in our study may have been underestimated, because it is believed that the onset of NPC at an early age depends more on genetic factors, and individuals in the current cohort were aged ≥45 years.7 The apparent lack of an increased risk of other OPCs in these families suggests that the familial susceptibility is specific for NPC and may be related to the association with EBV. This is compatible with the observation of a clearly increased risk of EBV-positive salivary gland carcinomas among relatives of NPC patients in the Arctic, which is another high-incidence NPC area.6, 42 Other than genetic susceptibility or an association with EBV, the increased familial risk of NPC may be caused by shared environmental risk factors, including diet, or a shared medical history, including upper respiratory tract infections, which have been linked to NPC risk.1
A potential limitation of the current study is the information on familial cases of NPC, which was based on interview data with no additional confirmation. Given the prospective nature of this study, misclassifications are likely to be nondifferential and, thus, may lead to an attenuation of the familial risk estimate rather than causing spurious associations. Another possible limitation of this study was the age of the original cohort, because individuals aged <45 years were not included. However, although risk factors may differ between early- and late-onset NPC, the majority of patients with NPC in both low- and high-incidence areas were diagnosed at age >45 years. Finally, information on the use of tobacco and alcohol beyond baseline was unavailable, leading to potential misclassification of an individual's exposure status.
Despite these limitations, the current study represents the first, large prospective study to our knowledge that has been performed in a high-incidence NPC population. Along with the prospective design of this study, its advantages include the ability to adjust for multiple confounders and the opportunity to investigate the whole spectrum of pharyngeal carcinomas together. The current results clearly differentiate between the risk of NPC and the risk of other OPCs associated with smoking and alcohol intake and strengthen the notion that tobacco smoking is a risk factor for NPC development in a NPC-endemic population. This result is important, because smoking is among the only preventable risk factors for NPC among adults.
We thank Ms. Siew-Hong Low (National University of Singapore) for supervising the field work of the Singapore Chinese Health Study and Ms. Kazuko Arakawa (University of Southern California) for the development and management of the cohort study database. We also thank the Singapore Cancer Registry for assistance with the identification of cancer outcomes.
- 2Cancer Incidence in Five Continents. Vol VIII. Lyon: IARC; 2002., , , , .
- 9Nasopharyngeal cancer. In: SchottenfeldD, FraumeniJFJr, eds. Cancer Epidemiology and Prevention. New York: Oxford University Press; 1996: 603–618., .
- 22Trends in Cancer Incidence in Singapore, 1968–2002. Singapore: Singapore Cancer Registry; 2004., , , , , .
- 23Cancers of the oral cavity and pharynx. In: SchottenfeldD, FraumeniJFJr, eds. Cancer Epidemiology and Prevention. 3rd ed. New York: Oxford University Press; 2006: 674–696., , .
- 29International Agency for Research on Cancer. Tobacco Smoke and Involuntary Smoking. Lyon: IARC; 2004.
- 35ShanmugaratnamK, SobinL, eds. Histological Typing of Tumours of the Upper Respiratory Tract and Ear. 2nd ed. Berlin: Springer-Verlag; 1991.