Associations of lifestyle and diet with the risk of nasopharyngeal carcinoma in Singapore: a case–control study

Nasopharyngeal carcinoma (NPC) is a commonly diagnosed cancer in Southeast Asia. Indeed, NPC is endemic in China (especially South China, Hong Kong, and Taiwan) and Southeast Asia countries such as Malaysia and the Philippines. For example, in 2010, South China was reported to have an age-standardized rate (ASR) of incidence of 2.44 per 100,000 people and an ASR of mortality of 1.18 per 100,000 people [1]; the Philippines was reported to have an ASR of incidence of 1.79 per 100,000 people [2]. In Singapore, NPC is most commonly diagnosed in men; in 2012, the ASR of incidence of NPC reported in men was 8.8 per 100,000 people [3].

Numerous studies have been conducted in Southeast Asia to identify risk factors related to the development of NPC [4–14]. Some of these studies suggested that NPC results from an interaction of genetic and environmental factors [7, 10, 11, 15]. Studies conducted in a Chinese population in Hong Kong, China found that over 90% of people younger than 35 years consumed salted fish during childhood and that they have continued to consume salted fish throughout adulthood [11, 16]. In addition, studies of the Chinese population in China reported that multiple food items, including preserved and fermented foods, were significantly associated with NPC during weaning and children exposure before 2 years of age [12]. Furthermore, risk factors such as occupational exposure to smoke and dust [15] and genetic variants [7] were shown to be associated with NPC. In 1994, Lee et al. [10] studied Singaporean Chinese patients with NPC who were younger than 45 years and found an increased risk of NPC in adults who frequently consumed salted food. In 2006, a Singaporean Chinese cohort study found that smoking duration, smoking intensity, age of smoking initiation, and alcohol consumption were strongly associated with an increased risk of other oropharyngeal carcinomas; however, they found that the association of these factors with NPC was rather weak, except for those who had smoked for 40 or more years [17]. This previous cohort study did not specifically focus on the risk factors for NPC in a Singaporean population, and the non-significant finding was likely the result of the small sample size of NPC cases.

Tobacco smoking and alcohol consumption have been widely studied in Southeast Asia [7, 15, 18, 19] and in Western countries such as the United States [20–22] and Italy [23], with no consistent relationships found between lifestyle factors and the risk of NPC. Although many studies have reported inconsistent results, in 2012 the World Health Organization (WHO) reported that tobacco use was an established risk factor for many cancer types, including NPC [24]. However, the relative importance of tobacco exposure in different NPC histologic subtypes remains largely unexplored [25–27], and additional prospective studies are needed. Because of smoking’s modifiable nature and causal relationship in the development of various cancers, it is an important lifestyle risk factor to target. In 2012, WHO in the Western Pacific region reported that China (which is a high-risk area for NPC) had more than 300 million smokers—one-third of the world’s total [28]. In our study, we postulated that given the higher incidence of smoking as countries (e.g., China) become further developed, given the increased accessibility of treatment, given the rapid aging of populations, and given the increasing median age of survival, the burden of cancer (prevalence) will increase.

The nitrosamines in tobacco are known to be active carcinogenic metabolites that cause DNA damage and chronic inflammation in nasopharyngeal mucosa [25]. Organs in direct contact with smoke—the oral cavity, esophagus, and lungs—have high risks of developing cancer. Similarly, the nasopharynx, which is located between the nasal cavity and the larynx, is also directly exposed to tobacco smoke. As tobacco smoking is a modifiable risk factor, indeed, 30% of all cancers reported in the United States could be prevented if smoking were eliminated [29]. Tobacco contains carcinogens that may interact with other exposures, such as alcohol, resulting in an increased risk of cancer [29]. Therefore, smoking cessation is the single most important and effective way to reduce a person’s risk of cancer [29, 30].

In the present study, we examined the lifestyle factors (smoking, alcohol consumption, various salted and preserved food consumption, pre-chewing of food, and weaning practices) involved in the development of NPC in the Singaporean population. Since Singapore is a high-risk area for the development of NPC, studying the risk factors in this population provides more data for future studies of NPC in other Southeast Asian countries. The National Cancer Centre of Singapore (NCCS) is the country’s largest tertiary referral center for NPC patients; as such, NPC patients recruited from there are representative of all Singaporean patients diagnosed with NPC.

In our study, we found that tobacco smoking (by both current and former smokers) was a statistically significant risk factor for developing NPC. Compared with never smokers, current smokers had four times the risk of developing NPC, and those who had ever smoked had double the risk. Studies conducted in Taiwan, China [5], in Thailand [6, 7], in Wuhan, China [8], in Shanghai, China [14], and in the United States [22] also found that smoking was a risk factor for NPC, although other studies in Singapore [17], Malaysia [33], Serbia [34], and China [35] found the association to be less clear but did suggest that the inhalation of passive smoke during childhood could affect the risk of developing NPC [34]. A recent meta-analysis found that ever smokers had a significantly higher risk of developing NPC than never smokers, but these associations were more relevant to squamous cell carcinoma and less relevant in undifferentiated carcinoma [27]. Since most patients in our study (80.0%) had undifferentiated carcinoma type of NPC, our results seem to suggest that exposure to tobacco smoke is equally relevant to undifferentiated carcinoma and the Singaporean population.

The literature offers several possible explanations regarding the carcinogenic mechanism. Tobacco smoke comes in direct contact with the nasopharynx and, therefore, might have direct action. However, it has also been proposed that tobacco is just one of many risk factors for NPC and may contain Epstein-Barr virus (EBV)-activating substances [26]. It is known that EBV infection is strongly associated with undifferentiated carcinoma, and in high-risk areas 95% of NPC patients have this type of carcinoma [27]. Several epidemiological studies have suggested that high concentrations of volatile nitrosamines are found in tobacco smoke [12], and these compounds were positively associated with cancer in other sites, such as the esophagus [36] and possibly the nasopharynx [12]. With regard to alcohol drinking, our results were similar to those of other studies, finding no strong association between alcohol drinking and NPC risk [4–6, 14, 17].

Additionally, we observed that people who consumed salted vegetables at least once a week had four times the risk of developing NPC compared with those who never or rarely consumed salted vegetables. This observation is consistent with the results of studies conducted in 1994 in Singapore [10], in 1998 in Malaysia [4], and in 2000 [14] and 2010 in China [9], all of which suggested that frequent consumption of salted vegetables was associated with risk of developing NPC. Salted vegetables are widely consumed in Singapore, with 18.3% of the control population in this study being regular consumers. These products are readily available in the supermarket, and people generally consume salted vegetables as part of a meal. The salted vegetables called suan cai is a traditional Chinese pickled cabbage—a unique type of cabbage and mustard greens because of the ingredients used and the method of production [37]. It is usually prepared first by being pressed slowly and fermented, followed by pickling with salt and brine [38]. It is similar to sauerkraut, which is a common food in Central and Eastern Europe [37]. On the other hand, preserved vegetables referred to other types of fermented vegetables or fruits or marinated in mixtures based on soy sauce and soybean paste [38]. Little is known about particular carcinogenic compounds found in salted vegetables that affect the nasopharynx; more research on this relationship is warranted. On univariate analysis, we found, in contrast to previous studies [4, 9, 14, 35], no clear significant association between adult consumption of salted fish and NPC risk. This could be due to the broad decreased consumption of salted fish in populations with high risk of NPC such as Singaporeans [25]. Indeed, since 1980, a markedly decreasing trend in salted fish consumption has been observed in similar high-risk areas such as Hong Kong, China [25, 39]. Not surprisingly, the Singaporean population, the majority of which is of Chinese ethnicity, has also changed their diet to a more Westernized and international one, replacing the traditional Chinese diet. Furthermore, the harms associated with the frequent consumption of salted fish, although not salted vegetables, was widely publicized in Singapore [40].

Our findings may have several limitations, particularly those inherent to retrospective case–control studies. First, case–control studies are vulnerable to recall bias when information can be collected only through the recall of study participants—this was the case for our study. We tried minimizing recall bias by asking confirmation questions and, on dietary questions, having several options in terms of consumption frequency. Second, our data collection was limited by the questions in the questionnaire asked directly about the study participants, particularly with regards to lifestyle, dietary information, and weaning practices. Data on weaning are ideally obtained from the mother of the study participant, rather than the participant himself or herself. But given that cancer is a late-onset disease, the mother may no longer be alive or may not remember this information, so the information was self-reported by the participant where possible. Notably, the amount of missing data on weaning was similar across cases and controls. Nonetheless, none of the weaning variables was conclusive, and these variables were not included in the multivariate model. Our results must be interpreted in the context of these constraints. Third, we recruited controls from our staff, friends of patients, and various organizations (both hospital and community). We acknowledge that there could be selection bias in our controls because controls were people who understood the importance of such studies and were willing to participate for the benefit of society. Also, they were more likely to have higher education. In Singapore, the general population’s participation rate in scientific studies is nearly low, and often significant effort is needed to recruit enough controls. We tried minimizing selection bias by recruiting controls from many different organizations; at the same time we recruited cases of approximately the same age, sex, and ethnicity. Because of the fairly homogenous population in Singapore, recruiting controls from various sources did not create significant biases. Fourth, we adjusted for educational level when analyzing the risk factors of NPC. Fifth, interviewer bias was one additional concern because knowledge and understanding of the risk factors by the interviewers and study participants may cause skewed results. When the study participants and interviewers are aware of the risk factors, the patients are more likely than the controls to pay greater attention to risk factor information during data collection. Therefore, we reduced the risk of interviewer bias by having only two trained interviewers who employed a standardized interview technique, using the same lifestyle questionnaire for both patients and controls. Sixth and finally, the lack of information regarding the EBV status of each participant was another potential limitation. We are aware that EBV status could be a main risk factor for the development of NPC and could potentially have influenced the results of our study. However, EBV is fairly common in Singapore, and seroconversion occurs at an early age, with a rate of EBV seropositivity of 40%–60% in children younger than 5 years.

In summary, we found that certain modifiable risk factors, namely smoking and frequent consumption of salted vegetables, contribute significantly to the development of NPC. This finding is important because understanding the risk factors of NPC could potentially lead to changes in cancer prevention campaigns. The burden of NPC is a major public health concern in Singapore, and by knowing that the risk factors are modifiable, public health programs might be better able in the future to reduce the burden of NPC by allocating resources and formulating population strategies accordingly. Although it may take a long time to adopt the cancer prevention strategies that could reduce the incidence of NPC, our findings could help tailor such important strategies, which hold significant future promise.