Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
2356 Age-Incidence Curves of Nasopharyngeal Carcinoma Worldwide: Bimodality in Low-Risk Populations and Aetiologic Implications Freddie Bray,1,2 Marion Haugen,2 Tron A. Moger,2,3 Steinar Tretli,1 Odd O. Aalen,1,2 and Tom Grotmol1 1 The Cancer Registry of Norway, Institute of Population-Based Cancer Research, and 2Department of Biostatistics, Institute for Basic Medical Sciences, and 3Institute of Health Management and Health Economics, University of Oslo, Oslo, Norway Abstract The distinct geographic variation in the global incidence of nasopharyngeal carcinoma reflects a complex etiology involving viral, environmental, and genetic components. The high to intermediate rates observed in endemic areas contrast markedly with the uniformly low rates seen in much of the world. An interesting epidemiologic observation is the early peak in ageincidence curves observed in certain geographically disparate populations, suggestive of distinct causal entities and the possible exhaustion of susceptible individuals from the population at a certain age. The aim of this study was to systematically evaluate the age-incidence profiles of NPC worldwide on partitioning populations according to level of risk, in an effort to provide clues about the importance of early-in-life factors and genetic susceptibility. Using data from 23 high-quality population-based cancer registries for the period 1983-1997, a key finding was the consistent pattern of bimodality that emerged across low-risk populations, irrespective of geographic location. Continual increases in NPC risk by age up to a first peak in late adolescence/early adulthood (ages 15-24 years) were observed, followed by a second peak later in life (ages 65-79 years). No such early peak in NPC incidence by age group was evident among the high-risk populations studied. These findings are discussed according to existing lines of biological and epidemiologic evidence related to level of population risk, age at diagnosis, and histologic subtype. A modified model for NPC tumor development is proposed on the basis of these observations. (Cancer Epidemiol Biomarkers Prev 2008;17(9):2356 – 65) Introduction The incidence of nasopharyngeal carcinoma (NPC) has a rather distinct and geographically well-defined distribution worldwide. High to intermediate rates are observed in certain populations and regions in China, Southeast Asia, Northern Africa; among the Inuit populations of Alaska, Greenland, and Northern Canada; and in migrants of Chinese and Filipino descent (1-6). Outside of these endemic areas, however, NPC can be considered a rather rare neoplasm. To illustrate, incidence rates in high-risk Hong Kong were between 30 and 60 times those in equivalent national or regional registry populations in Colombia, the United States, Finland, Zimbabwe, India, and Japan over the period 1998-2002 (4). The etiology of NPC, described by several experts as enigmatic (7, 8), has viral, environmental, and genetic components. Whereas the EBV is considered a critical step in the progression of NPC from precancerous lesion to malignancy, only a fraction of the EBV-infected population go on to develop NPC, and early seroepidemiologic surveys showed the virus to be both a lifelong infection and one that is ubiquitous in most areas of the Received 5/20/08; revised 6/18/08; accepted 6/30/08. Grant support: (sfi)2—Statistics for Innovation (M. Haugen). Requests for reprints: Freddie Bray, The Cancer Registry of Norway, Institute of Population-based Cancer Research, Montebello, N-0310 Oslo, Norway. Phone: 47-23-33-39-15; Fax: 47-22-45-13-70. E-mail: [email protected] Copyright D 2008 American Association for Cancer Research. doi:10.1158/1055-9965.EPI-08-0461 world (9). That persons migrating from high- to low-risk countries retained incidence rates that were intermediate between natives of their host country and their country of origin (10) has long implied a role for other contributory factors, including environmental and host (genetic) factors, possibly interacting with EBV. Much of our understanding of the underlying risk determinants of NPC, including the role of a nitrosamine-rich diet, has arisen from studies based on individuals from populations at high to intermediate risk. The two comprehensive epidemiologic reviews of NPC in the recent past have noted the need for further elucidation of the risk factors at play in low-risk populations (8, 11). In particular, there is a need to identify sources of variability that may provide clues to the determinants of NPC in nonendemic populations, and how they contrast with NPC cases in high-risk populations. Some of the recent developments in our understanding of NPC etiology have arisen following stratification by degree of risk—often characterized by geographic location—and histologic subtype of diagnosis, as defined in the recent WHO classification (12). An investigation of the properties of the age-incidence curves of NPC in different populations can be informative. Rates of NPC are often described as constantly increasing with age in the epidemiologic review literature (6, 8), although earlier peaks in both high- and low-risk regions have been reported (8, 13-17). The Cancer Epidemiol Biomarkers Prev 2008;17(9). September 2008 Downloaded from cebp.aacrjournals.org on August 12, 2017. © 2008 American Association for Cancer Research. Cancer Epidemiology, Biomarkers & Prevention Table 1. Characteristics of the registry populations by region and risk group Mean annual population coverage, 1983-1997 Low-risk India Chennai* Mumbai* Japan Miyagi Osaka North America Canada SEER White North West Europe Denmark Estonia c Switzerland, Zurich UK, Birmingham UK, Merseyside UK, North Western b UK, Oxford UK, South Thames UK, Yorkshire UK, Scotland Australia New South Wales South Victoria High-risk Philippines, Manila* Thailand, Chiang Mai* Singapore: Chinesex Singapore: Malayx Mean annual no. cases, 1983-1997 119.8 13.5 3.8 9.8 10.9 2.2 8.7 46.7 27.6 19.1 37.6 5.2 1.5 1.1 5.3 2.4 4.0 2.6 6.7 3.7 5.1 11.6 5.8 1.4 4.3 8.4 4.5 1.3 2.1 0.4 584 51 19 32 55 11 44 262 179 83 145 22 8 4 18 12 16 8 24 11 22 75 42 6 27 464 153 29 269 13 ASR, 1983-1997 Age (y) Males Females 1st peak 2nd peak 0.5 0.6 0.7 0.6 0.6 0.5 0.6 0.7 0.8 0.5 0.4 0.4 0.6 0.4 0.4 0.5 0.4 0.3 0.4 0.3 0.5 0.7 0.9 0.5 0.8 8.4 7.5 2.9 17.4 5.9 0.2 0.3 0.3 0.2 0.2 0.2 0.2 0.3 0.3 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.1 0.2 0.2 0.3 0.1 0.3 3.2 3.0 1.4 6.6 1.9 15-19 15-19 65-69 65-69 15-19 65-69 15-19 65-69 15-19 70-74 15-19 75-79 — — — — — — — — 45-49 55-59 Abbreviations: ASR, age-standardized rate; SEER, Surveillance, Epidemiology, and End Results. * Population data available in 16 age groups (0.4, 5-9,. . ., 70-74, 75+). cIncidence data available for the years of diagnosis 1983-1996. bIncidence data available for the years of diagnosis 1985-1997. x Population data available in 17 age groups (0.4, 5-9,. . ., 75-79, 80+). bimodality of disease rates, a phenomenon commonly observed for Hodgkin lymphoma, suggests the possibility of distinct causal entities (18) and the exhaustion of a pool of susceptible individuals, leaving a group who are, in theory at least, initially nonsusceptible (19). For NPC, a systematic evaluation of the age-incidence profiles according to level of risk may provide clues about the importance of early-life environmental factors and genetic susceptibility. This descriptive study thus critically examines the shape of cross-sectional age-incidence curves in 13 countries across four continents. Data from 23 high-quality population-based cancer registries over the diagnostic period 1983-1997 are dichotomized into those considered at low and high risk of NPC. The results are interpreted using existing lines of biological and epidemiologic evidence, in relation to level of population risk, age, and microscopic subtype. Subjects and Methods The Cancer Incidence in Five Continents (CI5) Vol. I to VIII ADDS database contains global cancer incidence data from consistently high-quality population-based registries that have provided histologic data covering a minimum span of 12 consecutive years (3), as part of their submission to successive CI5 Volumes. We initially extracted diagnoses of nasopharyngeal cancer (ICD-10 C11) for 72 cancer registries, together with the corresponding population data by year of diagnosis, sex, age, and, where available, ethnicity. Whereas we extracted nasopharyngeal ‘‘cancers’’ rather than ‘‘carcinomas,’’ the term NPC is used here to identify the latter, given that carcinomas represent the vast majority of nasopharyngeal tumors, and the subset for which most epidemiologic studies have focused. We restricted analyses to the period 1983-1997 to enable a sufficiently long span of data available that allowed as many registries to contribute as possible. To remove some of the uncertainty in the rates by age and over time, we excluded registries with an average annual population coverage of less than 1 million inhabitants across the 15-y period. We retained registry data on the Malay population of Singapore however, given its relatively high NPC incidence. We used U.S. incidence data (for Whites only) based on the Surveillance, Epidemiology, and End Results 9 registries, and national data for Canada. Because there were only six NPC diagnoses in the Quito Registry in Ecuador 1983-1997, this population was excluded from further analyses. For the remaining 23 registry populations, incidence data were available by 5-y age group and sex for each of Cancer Epidemiol Biomarkers Prev 2008;17(9). September 2008 Downloaded from cebp.aacrjournals.org on August 12, 2017. © 2008 American Association for Cancer Research. 2357 2358 Age-Incidence Curves of NPC Worldwide the years of diagnosis 1983-1997 and available for 18 age groups (0.4, 5-9,. . ., 80-84, 85+; see footnotes of Table 1 for exceptions). The mean number of NPC cases diagnosed annually during the period 1983-1997 varied from 4 in Zurich, Switzerland (mean annual population of 1.1 million) through to 269 among the Chinese population of Singapore (mean annual population of 2.1 million). Regional registries were aggregated to national or larger area levels on the basis of geographic area, thus enabling sufficient numbers for meaningful age-specific analyses. Age-specific rates were calculated and were summarized as age-standardized rates using the weights from the world standard population (20, 21). Those populations with age-adjusted rates of >2.5 were then identified as ‘‘high-risk’’ countries, with the remainder considered as ‘‘low-risk’’ (rates V2.5). Further analyses and the presentation of results reflect this stratification (Table 1; Fig. 1). The low-risk and high-risk age-standardized rates were based on simple averages of the age-standardized rates in the respective registry populations. To assess relative changes over time, trends of age-standardized rates by 5-y period are presented on a semi-log scale by area, sex, and risk group. The aggregated age-specific rates were calculated by weighting the counts and person-years of the constituent registry populations by the square root of the personyears, allowing smaller populations to attain a relatively higher weighting in the rate calculation. Graphs of rates versus age are presented on a semi-log scale to convey the structure and relative magnitude of the age curves by geographic area and sex, stratified by risk status. The overall low-risk age-specific rates are based on 18 age groups, but incidence in the last age group (75+) is excluded for the two registry populations missing population data. The overall age-specific rates in high-risk populations are based on the 16 age groups for which rates could be calculated in each of the four comprising populations. Results Geographic Variations 1983-1997. For the entire study period, the four high-risk populations in Table 1 contribute f44% of the 1,048 mean annual number NPC cases, but only 6.5% of the mean annual personyears at risk (f128 million). In the low-risk areas, the aggregated age-adjusted rates in men and women are uniformly low at 0.5 and 0.2 (per 100,000), respectively, with an f3-fold variation in rates between populations, observed in both sexes. There is greater variation in the rates in the Asian populations comprising the high-risk group (Fig. 1), with the highest rate of 17.4 observed among the Chinese population of Singapore and the lowest in Chiang Mai, Thailand (2.9). Rates in the high-risk group are thus four to eight times higher than those comprising the lowrisk group. In women in the high-risk areas, rates are lower and there is less variation, although the rank ordering of registries (according to the magnitude of their rates) is similar to that of men. The male/female ratio tends to vary uniformly from 2 to 3, irrespective of risk group, with no emerging patterns related to geographic location. Age-Specific Rates 1983-1997 Low-Risk Populations. The age-specific incidences rates of NPC by 5-year age group in the five low-risk countries/regions (and overall) are shown in Fig. 2A, and the observed peaks are noted in Table 1. A consistent pattern across populations emerges from continual increases in NPC risk by age up to a first peak in late adolescence/early adulthood followed by a second peak later in life (ages 65-79 years). This observation tends to be clearer among men, but is still apparent in women, although the reverse is true in North America, with the peak and decline more evident in female rates. The first low-risk peak commonly occurs in the age group 15-19 years, and is followed by a subsequent decline that is Figure 1. NPC age-standardized incidence rates (per 100,000 person-years at risk) according to the studied registry population and sex. The two risk groups are indicated, as is the magnitude of the rates in selected populations. Cancer Epidemiol Biomarkers Prev 2008;17(9). September 2008 Downloaded from cebp.aacrjournals.org on August 12, 2017. © 2008 American Association for Cancer Research. Cancer Epidemiology, Biomarkers & Prevention observed to vary from 5 to 10 years in Australia, Northern and Western Europe, and North America to up to 20 years in the low-risk Asian populations in this study (India and Japan). It is also notable that male/female rate ratios up to the age of 30 years are above unity (closer to 2) in most of the low-risk populations in Fig. 2A. In general, a ‘‘second wave’’ involves rates of NPC continually increasing from the age 30 to 39 years, toward a further peak at an elderly age. This second maximum is observed to vary between countries and regions, occurring at 65 to 69 years in North American men, for example, but some 10 to 15 years later in Australia and Northern Europe. High-Risk Populations. No such early peak in NPC incidence with age is evident in the four high-risk populations (Fig. 2B), although a slight kink in the continual increase in rates by age may be observed, most notably among Thai men. What is more apparent is an earlier decline in age-incidence (in middle-age, ages 45-60 years) than seen in any low-risk population under study. This is most striking among the Chinese population of Singapore in both sexes, with a peak and subsequent decline beginning in the rates of the age group 45-49 years. Such an observation is less discernible in the incidence rates in the Thai and Filipino men and women, although there seems to be an early plateau in the rates in the former population. Time Trends 1983-1997. Figure 3 shows the all-ages standardized rates in consecutive 3-year periods. The time trends for the period 1983 through to 1997 are generally flat in both low- and high-risk populations. There is, however, some evidence of a minor decline among women in the late 1980s and early 1990s, most notably among the Chinese population of Singapore and residents of Chiang Mai, Thailand. Trends restricted to NPC at younger ages (ages under 50 years) are not dissimilar, although declines in both sexes are more visible in recent periods in the Singapore Chinese and Filipino populations (data not shown). Discussion This descriptive study has compared international age curves of NPC incidence 1983-1997 on the basis of high-quality cancer registration data worldwide. On dichotomizing these populations according to the magnitude of their rates, the key finding is that there exists distinctive age-incidence curve for NPC for low- and high-risk groups, respectively, and in particular, that the age-incidence curves seem to be uniformly bimodal in every low-risk population, irrespective of geographic location and sex. This finding is perhaps at odds with the widely held concept that NPC incidence increases monotonically with age in the majority of low-risk populations, as reported in recent comprehensive reviews (6, 8). The low-risk curves shown in this study convey a systematic pattern of continual increases in risk by age up to a first peak in late adolescence/early adulthood (ages 15-24 years), a subsequent decline in risk of up to the ages 30-39 years, followed by increasing risk with age to a second peak later in life (ages 65-79 years). The bimodality property of the NPC incidence rates has previously been described in selected low-risk populations including (in both Whites and Blacks) the United States (13), Malaysia (22), and among a hospital series in Mumbai, India (14). Given the relative rarity of NPC incidence in nonendemic areas and the random variation inherent in the age curves, it is perhaps not surprising that a more general observation of bimodality has not been reported previously. Evidence for such an attribute was achieved only on aggregating incidence and person-years over a number of registries across a 15-year span, and a limitation of this study is the necessity of doing so. Registry data were collapsed on the basis of geographic proximity rather than a priori knowledge of a correspondence in underlying risk determinants for NPC between populations. Further, basing the rates on diagnoses 1983-1997 may have led to a distortion in the age curves where trends diverge by age group during this calendar period, as would be evident if strong birth cohort effects were present, markers of corresponding changes in the prevalence of underlying risk factors. Both the age-adjusted and age-specific trends seemed to be rather stable among the low-risk populations, however, and given the similarity in magnitude and structure of the registry population age-incidence curves within a predefined geographic area, it seemed reasonable to aggregate the data for the predefined purposes. It is also evident that the broad dichotomization of risk levels into only two categories is a rather crude maneuver, particularly given the wide variation observed in rates among populations within the high-risk group, although, however, this was necessary simplification, given the limited data available. The first peak of the bimodal age-incidence curve in low-risk populations is similar to that of clear cell adenocarcinoma of the vagina. This condition is caused by diethylstilboestrol exposure during pregnancy, and there is thus reason to believe that a similar time frame is underlying the early peak of the current NPC age-incidence curve. The first ‘‘hit’’ in individuals diagnosed with NPC in late adolescence/early adulthood in the low-risk group, is most likely a germ-line alteration (major gene transmission) according to the current model for the development of NPC (Chan et al., 2004 in ref. 12). The molecular mechanisms pertaining to these early events in the low-risk populations, however, are not very well characterized. The familial cases compose the majority within this group of NPC, which supports the notion that susceptibility genes play an important role. In adult NPC, EBV infection is well established as a causative agent, being responsible for the second hit in the pathogenesis. On the molecular level, 10 of the 100 EBV genes incorporated in latently infected cells are expressed, the membrane protein LMP1 being the most important with its oncogenic properties. Recent evidence suggests a similar role for EBV infection in the pathogenesis of NPC occurring in late adolescence/early adulthood [reviewed by Ayan et al. (22), and references therein]. The novel finding of an early peak in low-risk populations implies a role for susceptibility genes in the pathogenesis and would hopefully stimulate research aimed at identification of such genes. Several susceptibility genes have already been implicated in the etiology of NPC, such as certain alleles of the human leukocyte antigen classes I and II. Human leukocyte antigen Cancer Epidemiol Biomarkers Prev 2008;17(9). September 2008 Downloaded from cebp.aacrjournals.org on August 12, 2017. © 2008 American Association for Cancer Research. 2359 2360 Age-Incidence Curves of NPC Worldwide Figure 2. A. NPC incidence rates (per 100,000 person-years at risk) versus age (5-y groupings) by registry population in each of the low-risk countries/regions, by sex. The aggregated rates are in bold, and were calculated by weighting the counts and personyears by the square root of the total population 1983-1997. Vertical lines are included to highlight the rates in age groups 15-24 and 65-74 y. plays an important role in presenting the viral antigens to the T cells of the immune system, and it has been shown that alleles less efficient at inducing CTL responses are more frequent in high-risk than in low-risk populations (23). It is plausible that polymorphisms in human leukocyte antigen, as well as in other relevant susceptibility genes, play a role for the development of NPC among adolescents in these populations. Our consistent finding of higher NPC incidence rates among younger men than younger women (aged <30 years), however, could also imply a role for early-in-life environmental factors that are differentially distributed among the sexes. Of recent interest has been the relationship of age and pathologic subtype, and in Table 2, the three sets of age-incidence patterns observed in this article are linked Cancer Epidemiol Biomarkers Prev 2008;17(9). September 2008 Downloaded from cebp.aacrjournals.org on August 12, 2017. © 2008 American Association for Cancer Research. Cancer Epidemiology, Biomarkers & Prevention Figure 2 Continued. B. NPC incidence rates (per 100,000 person-years at risk) versus age (5-y groupings) by registry population in each of the high-risk countries/ regions, by sex. The aggregated rates are in bold, and were calculated by weighting the counts and p e r s o n -y e a r s b y t he square root of the total population 1983-1997. Vertical lines are included to highlight the rates in age groups 15-24 and 4549 y. to established or putative risk factors for the three distinct histopathologic entities developed by the WHO (12). In the WHO system, the first two groups of NPC are moderately differentiated squamous cell carcinomas with keratin production (type I) or nonkeratinizing (type II), with a third group containing undifferentiated carcinomas or lymphoepitheliomas (type III). There are a number of intriguing epidemiologic differences between these subtypes in low-risk populations. Type I is more common in low-incidence populations (13, 24) and nonfamilial cases (25) and is associated with an older age at diagnosis. Type I is also the only subtype that has been associated with alcohol and cigarette consumption (24), and in an occupational study examining subtype, type I (but not types II and III) has been linked with formaldehyde exposure (26). Cancer Epidemiol Biomarkers Prev 2008;17(9). September 2008 Downloaded from cebp.aacrjournals.org on August 12, 2017. © 2008 American Association for Cancer Research. 2361 2362 Age-Incidence Curves of NPC Worldwide Figure 3. Trends in agestandardized (world) NPC incidence rates (per 100,000 person-years at risk) by calendar period in countries/region classified as low or high risk, by sex. Rates are based on 3-y nonoverlapping aggregates of annual data over the period 1983-1997. It may be that the increases in risk in low-incidence populations in later ages—following the first peak to a second maximum during, or soon after, the onset of elderly age—are the result of sporadic NPC development associated with environmental and lifestyle-related risk factors postulated for NPC. Heavy tobacco (24, 27) and alcohol use (24) have been associated with an increased risk of overall NPC, with the association most notable among those diagnosed at ages z50 years (24). A decline in smoking has been hypothesized as a contributor to the overall decline in incidence trends among U.S. men (28). The contribution of occupational exposure such as formaldehyde and NPC risk remains unresolved. Whereas an IARC Monograph Working Group concluded that there was ‘‘sufficient epidemiologic evidence that formaldehyde causes NPC in humans’’ (29), other reviews (e.g., refs. 8, 30) have considered the evidence of an association inconsistent. Perhaps because of the relative rarity of NPC, there has been little breakthrough in explaining younger cases of NPC in low-risk populations, although they tend to be more often familial cases diagnosed with type II or type III NPC, with the latter the more commonly diagnosed type in children (31). Undifferentiated nasopharyngeal carcinoma may thus develop through exposure to factors early in life, in particular EBV infection, affecting genetically susceptible individuals. Types II and III are more commonly associated with elevated EBV titers, Cancer Epidemiol Biomarkers Prev 2008;17(9). September 2008 Downloaded from cebp.aacrjournals.org on August 12, 2017. © 2008 American Association for Cancer Research. Cancer Epidemiology, Biomarkers & Prevention Table 2. Risk factors associated with NPC according to WHO microscopic type and population level of risk WHO type Putative risk factors Type I, keratinizing squamous cell Smoking Type II, nonkeratinizing squamous cell, differentiated Type III, nonkeratinizing squamous cell, undifferentiated Mainly affecting low-risk populations and older ages EBV? Formaldehyde? Nonfamilial factors EBV Familial factors Preserved foods? EBV Risk group/age affected 5% of all cases Mainly affecting high-risk populations, as well as children (irrespective of level of risk in population) Preserved foods Familial factors with EBV detection in type I less consistent (8). A study of NPC cases in low-risk setting (the United States) has also reported an increased risk of types II and III with frequent consumption of preserved meat (32). Among high- to intermediate-risk populations, type III tumors predominate (28, 33). The age-incidence curves in this study exhibit a single peak with a decline observed earlier than in the low-incidence populations. Whereas a bimodal relation has been reported in several intermediate-risk populations in Northern Africa (15) and Western Asia (8, 16, 17), the unimodal age pattern in the highestrisk populations is well established. The decline with age seen in Singapore Chinese, for instance, beginning at the age of 50 years, has been postulated as compatible with either a viral or carcinogenic exposure early in life, or an exhaustion of the pool of genetically susceptible individuals (34). Whereas trends in NPC have been reported as rather stable in Southern China (35), declines have been noted in a number of populations of Chinese origin over the last 20 years (28, 33, 36-38). Interestingly, those studies that were able to stratify by histologic category reported that the observed decreases were largely restricted to type I tumors (28, 33), suggesting that it is the changing prevalence of type I risk factors that dominates the trends. Numerous studies have reported increased risks associated with certain foods eaten in high-risk areas including salted fish, various preserved foods, and hot spices, which are high in nitroso compounds and volatile nitrosamines (see ref. 8), and it is possible that such a chronic dietary exposure over a sustained period of time may explain the age curves seen in high-risk populations. Figure 4. Three pathways leading to the development of NPC, modified from Chan et al., 2004 in chapter 2 of ref. 12. Cancer Epidemiol Biomarkers Prev 2008;17(9). September 2008 Downloaded from cebp.aacrjournals.org on August 12, 2017. © 2008 American Association for Cancer Research. 2363 2364 Age-Incidence Curves of NPC Worldwide Exposure to such foods during the time window that includes weaning and childhood may be particularly important (39, 40). The time of infection with EBV in early life may also be critical; a Danish study reported evidence of children from the high-risk Eskimo population in Greenland becoming infected with EBV at an earlier age than children in low-risk Denmark (41). Conversely, in low-risk populations, the diagnosis of infectious mononucleosis, a marker of late infection with EBV, has been linked with decreased NPC risk (24). It can be hypothesized that bimodality is an inherent feature of NPC age-incidence curves irrespective of level of risk, and that the first peak is simply masked in the Filipino, Singaporean, and Thai populations by genetic and epigenetic factors associated with their high-risk status. That the apparent kink in the age-incidence curves at ages 15-24 years in the intermediate-risk groups (Chiang Mai and Singapore Malay) is not evident among the highest-risk Singapore Chinese would tend to support this. Less speculatively, real differences in the NPC rates between low-, intermediate-, and high-risk populations at younger ages should primarily relate to early events in the tumorigenesis pathway, and would likely comprise differentials in both susceptibility to genetic damage and exposures to environmental causes. Loss of heterozygosity on chromosomes 3p and 9p is such an early event that is linked to dietary habits, including salted fish intake (42). Further, the combined frequency for loss of heterozygosity on 3p and 9p in histologically normal nasopharyngeal epithelium has been shown to be much higher in high-risk than in low-risk populations (43). More generally, there are aspects of the rationale for bimodality in low-risk populations, including germ-line mutations and human leukocyte antigen associations, which would also, given the current evidence, apply in intermediate- and high-risk groups. This descriptive study has presented the age-incidence curves of NPC partitioned into level of risk, with a novel finding that there is clear evidence of uniform bimodality in low-risk populations worldwide, irrespective of place of residence and sex. The early peak in incidence, although unexplained, suggests a role for major susceptibility genes in the pathogenesis, possibly mediated by other factors including EBV infection, for which age at infection may be critical. The second peak in low-incidence populations may owe more to the classic risk factors for NPC, heavy smoking, high alcohol consumption, and possibly long-term exposure to certain occupational carcinogens; other unexplained factors, however, may also be in operation. The observed curves in high-risk populations are unimodal, although a masking of the first peak via early epigenetic events is a possible explanation for the lack of bimodality. In summary, Fig. 4 provides an adapted and simplified model for the development of NPC along three pathways and incorporates level of risk, age, and pathologic subtype. In brief, type I tumors, which predominate in low-risk populations, are considered as mainly occurring through environmental exposures such as smoking, with the presence of latent EBV infection indicated as a potentially important step in the pathway. The pathogenesis of NPC at young ages, which are mainly type II/III tumors, includes EBV infection at a critical later step following either germ-line mutations of individuals with familial predisposition or polymorphic variants involving minor genes. Finally, the pathway for type III tumors, common in high-risk populations, includes an initial genetic polymorphism associated with chronic (dietary) exposure to nitrosamine compounds. EBV infection is also considered a critical step in the pathway of undifferentiated malignancies. Further studies of the age-incidence curves according to WHO histologic type would certainly seem to be warranted. Another aspect of the age-incidence curves that could be further explored involves the concept of ‘‘frailty’’ or, in terms of incidence, ‘‘susceptibility’’ to the development of NPC. A frailty effect is one that describes the influence of individual selection in the general population. As an example, one explanation for the first peak in low-incidence populations could be the exhaustion of a fixed number of susceptible individuals in the population at a given age, with the decline in rates being the result of the remaining group remaining at low risk at given age at diagnosis (in theory, they are nonsusceptible). Frailty models have been developed and successfully applied to better understand the biological and etiologic mechanisms involved in the development of testicular cancer (44, 45), and it would seem to be appropriate to apply such a frailty modeling approach to rates of NPC. Model-based estimates of the proportion susceptible to NPC, and the approximate number of genetic events required for the tumor to develop would better quantify and aid interpretation of the apparent differences in age-incidence by geographical area and sex. Disclosure of Potential Conflicts of Interest No potential conflicts of interest were disclosed. Acknowledgments The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. We thank the numerous population-based cancer registries worldwide that have contributed their data to successive volumes of Cancer Incidence in Five Continents, thus making the present study possible. References 1. 2. 3. 4. 5. 6. 7. 8. 9. Parkin DM, Bray F, Ferlay J, Pisani P. Estimating the world cancer burden: Globocan 2000. Int J Cancer 2001;94:153 – 6. Parkin DM, Bray F, Ferlay J, Pisani P. Global cancer statistics, 2002. CA Cancer J Clin 2005;55:74 – 108. Parkin DM, Whelan SL, Ferlay J, Storm H. Cancer Incidence in Five Continents. Vols. I to VIII. IARC CancerBase No. 7. Lyon: IARC; 2005. Curado MP, Edwards B, Shin HR, Storm H, Ferlay J, Heanue M et al. Cancer Incidence in Five Continents. Vol. IX. IARC Scientific Publications No. 160. Lyon: IARC; 2007. Parkin DM, Whelan SL, Ferlay J, Bah E, Hamdi-Cherif M. Cancer in Africa. IARC Scientific Publications 153. Lyon: IARC; 2003. Yu MC, Yuan J. Nasopharyngeal cancer. In: Schottenfeld D, Fraumeni-JF J, editors. Cancer epidemiology and prevention. 3rd ed. New York: Oxford University Press; 2006. p. 620 – 6. Klein G. Nasopharyngeal carcinoma (NPC) is an enigmatic tumor. Semin Cancer Biol 2002;12:415 – 8. Chang ET, Adami HO. The enigmatic epidemiology of nasopharyngeal carcinoma. Cancer Epidemiol Biomarkers Prev 2006;15:1765 – 77. Henle W, Henle G. The cause of infectious mononucleosis. A review. In: Briggs PM, De-Thé G, Payne LM, editors. Oncogenesis and herpes viruses. Lyon: IARC; 1972. p. 269 – 74. Cancer Epidemiol Biomarkers Prev 2008;17(9). September 2008 Downloaded from cebp.aacrjournals.org on August 12, 2017. © 2008 American Association for Cancer Research. Cancer Epidemiology, Biomarkers & Prevention 10. Buell P. The effect of migration on the risk of nasopharyngeal cancer among Chinese. Cancer Res 1974;34:1189 – 91. 11. Hildesheim A, Levine PH. Etiology of nasopharyngeal carcinoma: a review. Epidemiol Rev 1993;15:466 – 85. 12. Barnes L, Eveson JW, Reichart P, Sidransky D. World Health Organization classification of tumours. pathology and genetics of head and neck tumours. Lyon: IARC Press; 2005. 13. Burt RD, Vaughan TL, McKnight B. Descriptive epidemiology and survival analysis of nasopharyngeal carcinoma in the United States. Int J Cancer 1992;52:549 – 56. 14. Balakrishnan U. An additional younger-age peak for cancer of the nasopharynx. Int J Cancer 1975;15:651 – 7. 15. Nwaorgu OG, Ogunbiyi JO. Nasopharyngeal cancer at the University College Hospital Ibadan Cancer Registry: an update. West Afr J Med 2004;23:135 – 8. 16. Barista I, Varan A, Ozyar E. Bimodal age distribution in Hodgkin’s disease and nasopharyngeal carcinoma. Med Hypotheses 2007;68:1421. 17. Andejani AA, Kundapur V, Malaker K. Age distribution of nasopharyngeal cancer in Saudi Arabia. Saudi Med J 2004;25: 1579 – 82. 18. MacMahon B, Pugh H. Epidemiology: principles and methods. Philadelphia: Lippincott Willams and Wilkins; 1970. 19. Henderson BE. Nasopharyngeal carcinoma: present status of knowledge. Cancer Res 1974;34:1187 – 8. 20. Doll R, Payne P, Waterhouse J. Cancer incidence in five continents: a technical report. New York: Springer; 1966. 21. Segi M, Kurihara M. Cancer mortality for selected sites in 24 countries (1950-1957). 2nd ed. Sendai: Tohoku University of Medicine; 1960. 22. Rothwell RI. Juvenile nasopharyngeal carcinoma in Sabah (Malaysia). Clin Oncol 1979;5:353 – 8. 23. Hildesheim A, Apple RJ, Chen CJ, et al. Association of HLA class I and II alleles and extended haplotypes with nasopharyngeal carcinoma in Taiwan. J Natl Cancer Inst 2002;94:1780 – 9. 24. Vaughan TL, Shapiro JA, Burt RD, et al. Nasopharyngeal cancer in a low-risk population: defining risk factors by histological type. Cancer Epidemiol Biomarkers Prev 1996;5:587 – 93. 25. Levine PH, Pocinki AG, Madigan P, Bale S. Familial nasopharyngeal carcinoma in patients who are not Chinese. Cancer 1992; 70:1024 – 9. 26. Vaughan TL, Stewart PA, Teschke K, et al. Occupational exposure to formaldehyde and wood dust and nasopharyngeal carcinoma. Occup Environ Med 2000;57:376 – 84. 27. Zhu K, Levine RS, Brann EA, Gnepp DR, Baum MK. A populationbased case-control study of the relationship between cigarette smoking and nasopharyngeal cancer (United States). Cancer Causes Control 1995;6:507 – 12. 28. Sun LM, Epplein M, Li CI, Vaughan TL, Weiss NS. Trends in the incidence rates of nasopharyngeal carcinoma among Chinese Americans living in Los Angeles County and the San Francisco metropolitan area, 1992-2002. Am J Epidemiol 2005;162:1174 – 8. 29. Cogliano VJ, Grosse Y, Baan RA, Straif K, Secretan MB, El Ghissassi F. Meeting report: summary of IARC monographs on formaldehyde, 2-butoxyethanol, and 1-tert -butoxy-2-propanol. Environ Health Perspect 2005;113:1205 – 8. 30. Bosetti C, McLaughlin JK, Tarone RE, Pira E, La Vecchia C. Formaldehyde and cancer risk: a quantitative review of cohort studies through 2006. Ann Oncol 2008;19:29 – 43. 31. Ayan I, Kaytan E, Ayan N. Childhood nasopharyngeal carcinoma: from biology to treatment. Lancet Oncol 2003;4:13 – 21. 32. Farrow DC, Vaughan TL, Berwick M, Lynch CF, Swanson GM, Lyon JL. Diet and nasopharyngeal cancer in a low-risk population. Int J Cancer 1998;78:675 – 9. 33. Tse LA, Yu IT, Mang OW, Wong SL. Incidence rate trends of histological subtypes of nasopharyngeal carcinoma in Hong Kong. Br J Cancer 2006;95:1269 – 73. 34. Shanmugaratnam K, Lee HP, Day NE. Cancer incidence in Singapore, 1968-1977. Lyon: IARC Scientific Publications No. 47; 1983. 35. Jia WH, Huang QH, Liao J, et al. Trends in incidence and mortality of nasopharyngeal carcinoma over a 20-25 year period (1978/1983-2002) in Sihui and Cangwu counties in southern China. BMC Cancer 2006;6:178. 36. Lee AW, Foo W, Mang O, et al. Changing epidemiology of nasopharyngeal carcinoma in Hong Kong over a 20-year period (1980-99): an encouraging reduction in both incidence and mortality. Int J Cancer 2003;103:680 – 5. 37. Luo J, Chia KS, Chia SE, Reilly M, Tan CS, Ye W. Secular trends of nasopharyngeal carcinoma incidence in Singapore, Hong Kong and Los Angeles Chinese populations, 1973-1997. Eur J Epidemiol 2007; 22:513 – 21. 38. Hsu C, Shen YC, Cheng CC, Hong RL, Chang CJ, Cheng AL. Difference in the incidence trend of nasopharyngeal and oropharyngeal carcinomas in Taiwan: implication from age-period-cohort analysis. Cancer Epidemiol Biomarkers Prev 2006;15:856 – 61. 39. Yu MC, Huang TB, Henderson BE. Diet and nasopharyngeal carcinoma: a case-control study in Guangzhou, China. Int J Cancer 1989;43:1077 – 82. 40. Ning JP, Yu MC, Wang QS, Henderson BE. Consumption of salted fish and other risk factors for nasopharyngeal carcinoma (NPC) in Tianjin, a low-risk region for NPC in the People’s Republic of China. J Natl Cancer Inst 1990;82:291 – 6. 41. Melbye M, Ebbesen P, Levine PH, Bennike T. Early primary infection and high Epstein-Barr virus antibody titers in Greenland Eskimos at high risk for nasopharyngeal carcinoma. Int J Cancer 1984;34:619 – 23. 42. Young LS, Rickinson AB. Epstein-Barr virus: 40 years on. Nat Rev Cancer 2004;4:757 – 68. 43. Lo KW, Huang DP. Genetic and epigenetic changes in nasopharyngeal carcinoma. Semin Cancer Biol 2002;12:451 – 62. 44. Aalen OO, Tretli S. Analyzing incidence of testis cancer by means of a frailty model. Cancer Causes Control 1999;10:285 – 92. 45. Moger TA, Aalen OO, Halvorsen TO, Storm HH, Tretli S. Frailty modelling of testicular cancer incidence using Scandinavian data. Biostatistics 2004;5:1 – 14. Cancer Epidemiol Biomarkers Prev 2008;17(9). September 2008 Downloaded from cebp.aacrjournals.org on August 12, 2017. © 2008 American Association for Cancer Research. 2365 Age-Incidence Curves of Nasopharyngeal Carcinoma Worldwide: Bimodality in Low-Risk Populations and Aetiologic Implications Freddie Bray, Marion Haugen, Tron A. Moger, et al. Cancer Epidemiol Biomarkers Prev 2008;17:2356-2365. Updated version Cited articles Citing articles E-mail alerts Reprints and Subscriptions Permissions Access the most recent version of this article at: http://cebp.aacrjournals.org/content/17/9/2356 This article cites 35 articles, 6 of which you can access for free at: http://cebp.aacrjournals.org/content/17/9/2356.full#ref-list-1 This article has been cited by 2 HighWire-hosted articles. Access the articles at: http://cebp.aacrjournals.org/content/17/9/2356.full#related-urls Sign up to receive free email-alerts related to this article or journal. To order reprints of this article or to subscribe to the journal, contact the AACR Publications Department at [email protected]. To request permission to re-use all or part of this article, contact the AACR Publications Department at [email protected]. Downloaded from cebp.aacrjournals.org on August 12, 2017. © 2008 American Association for Cancer Research.