• nasopharyngeal carcinoma;
  • diet;
  • Tunisia;
  • Algeria;
  • Morocco


  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. Conclusions
  7. Acknowledgements
  8. References

North Africa is one of the major Nasopharyngeal Carcinoma (NPC) endemic regions. Specific food items unique to this area were implicated to be associated with NPC risk, but results were inconsistent. Here we have performed a large-scale case-control study in the Maghrebian population from Tunisia, Algeria and Morocco. From 2002 to 2005, interviews were conducted on 636 cases and 615 controls. Controls were hospitalized individuals from 15 non-cancer hospital departments, or friends and family members of non-NPC cancer subjects, matched by center, childhood household type (rural or urban), age and sex. Conditional logistic regression is used to evaluate the risk of factors. In results, consumption of rancid butter, rancid sheep fat and preserved meat not spicy (mainly quaddid) were associated with significantly increased risk of NPC, while consumption of cooked vegetables and industrial preserved fish was associated with reduced risk. Other foods such as fresh citrus fruits and spicy preserved meat (mainly osban) in childhood, industrial made olive condiments in adulthood, were marginally associated. In multivariate analyses, only rancid butter, rancid sheep fat and cooked vegetables were significantly associated with NPC. In regard to possible causative substances, our results implicate the involvement of butyric acid, a potential Epstein-Barr virus (EBV) activator. © 2007 Wiley-Liss, Inc.

Nasopharyngeal carcinoma (NPC) is rare in most areas of the world but is much more common in certain populations. In Hong Kong, the incidence rates were estimated to be 21.4/105/year in males and 8.3/105/year in females,1 where the majority of the population is Cantonese. In Guangdong province of China, the highest incidence rates in males and females were 30.94/105/year and 13.00/105/year in Sihui county,2 where Cantonese is also the primary ethnic group. In Malaysia, in addition to the ethnic Chinese who demonstrate intermediate risk (16.5 and 7.2 per 105 per year),3 an indigenous population, the Bidayuh, shows the highest incidence rate recorded hitherto (32/105 for males and 12/105 for females per year).4 People from the countries of North Africa also demonstrate high risk of NPC; these include Algeria, Morocco and Tunisia. In Sétif, Algeria, the incidence rates were 8/105/year for males and 3/105/year for females5 in 1990–1993, while in Morocco and Tunisia, they may be slightly lower (The age and year adjusted incidence per 100,000 person years in Tunisia is 7.5 for men and 3.3 for women).6 In contrast, the incidence for most other countries is less than 0.5/105/year. In Asian high-risk populations, the incidence rises in adolescence and peaks at 45–55 years, while in North Africa, the Maghrebian population is characterized by a bimodal age distribution, with one peak occurring in the teens and the other at age 40–50 years.7

Major risk factors for NPC include Epstein-Barr virus reactivation, diet and genetic susceptibility.8, 9 It has been reported that lower socio-economic level is an important confounding factor for NPC in China, Macau, Greenland and Tunisia.9 In intermediate to high risk Chinese populations, consumption of salted fish, especially during weaning in childhood, is associated with elevated risk of NPC.10, 11, 12, 13, 14, 15, 16 Similar results have been found in other endemic populations, such as Thai17 and Inuits.18 In North African countries, associated foods include quaddid (dried mutton stored in oil), harissa (very spicy condiment prepared with red pepper, olive oil, garlic, caraway) and toklia (basic stewing preparation, contain red pepper, black pepper, garlic, salt, oil, caraway and coriander) in Tunisia,19 pickled fruit and vegetables, dried and salted fat, and rancid butter in Sétif, Algeria,20 and rancid butter and khelii (dried meat, salty, spicy, cooked and preserved in a mixture of oil and melted bovine greases) in Morocco.21 However these results have to be considered with caution as none of the 3 studies included more than 80 cases and, detailed adjustment for possible confounders and multivariate analysis was not performed in the Algerian and the Moroccan studies.

Despite the divergent distribution of the high-risk populations both geographically and ethnically, they show some common aspects of diet. Salted fish is common and confirmed in Chinese patients from many regions, such as Tianjin,11 Shanghai,15 Guangxi,14, 22 Guangzhou,13 Hong Kong,12 Malaysia,10, 23, 24 as well as other populations including Thai17 and Inuits.18 Among Cantonese, Arab from the Maghreb populations, and Inuits, the common factor is that patients tend to consume more preserved foods and dried meats in childhood than controls.25 In addition, various epidemiological studies consistently found that fresh fruit and leafy vegetables are protective factors against NPC in Chinese13–15, 22, 24, 26 as well as some low risk populations.27 A recent meta-analysis has confirmed the harmful effect of adulthood consumption of preserved vegetables and protective effect of fresh vegetables in various populations from South-East Asia, North Africa and the United States.28

Besides the common features of studies in different populations, there are some specific findings, unique to North Africa. Associated foods, such as toklia, harissa, quaddid, dried and salted fat, and rancid butter,19 are not common in any other NPC endemic regions. However, none of the above food items have been confirmed by all 3 studies from the same region. Therefore, confirmation of these findings is necessary by a larger study with sufficient power.

To address these problems, we have performed a large-scale multi-centric case-control study in the Maghrebian population from Tunisia, Algeria and Morocco. In this article, we examine the roles of dietary factors, in order to validate previously hypothesized food items, and to uncover additional NPC risk factors which may have been missed in previous studies due to lack of power and detailed adjustment for confounders.


  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. Conclusions
  7. Acknowledgements
  8. References

Study population

All incident cases diagnosed between 2001 and 2004 in 5 hospitals were identified by the clinicians in the oncology and radiotherapy departments, and were invited for interview. These hospitals are public oncology centers (Tunis in Tunisia, Casablanca in Morocco, and Blida in Algeria) or the oncology service of public hospitals (Sétif in Algeria and Sousse-Monastir in Tunisia). In Morocco and Tunisia, it is estimated that roughly 10% of all cases are treated in private clinic rather than public hospitals, and therefore are not able to be recruited in this study; In Algeria, cancer patients are not seen in the private sector. The cases from private clinic have a potentially higher socio-economic status (SES) than the majority of cases seen in the public sector. In addition to the incident cases, prevalent cases were also recruited, in order to increase the size for a genetic association study. We designate those cases recruited within 1 year of diagnosis as “incident cases,” otherwise “prevalent cases.” For both cases, subjects less than 15 years old were excluded. A total of 636 cases were enrolled in the study (475 incident; 160 prevalent; 1 unknown). The median lag time between interview and diagnosis was 2 months for incident cases and 32 months for prevalent cases. Regarding histological type, almost all (92%) were undifferentiated carcinoma. The 615 controls were hospitalized individuals from 15 non-cancer hospital departments (61%), including gastroenterology, endocrinology, rheumatology, orthopedics, neurology, nephrology, pulmonary, hematology, cardiology, emergency medicine, surgical, ophthalmology, dermatology, gynecology and infection disease departments; or friends and family members of non-NPC cancer subjects (39%) who came to the hospitals as to visit or to take care of the patients. Controls were frequency matched by center, age, sex and household type (urban/rural) in childhood. The possible bias of higher SES and reduced geographical area in those visitors has been seriously evaluated prior to the recruitment and has been anticipated to be minor. Informed consent was obtained from each participant, and the IARC ethical committee has approved the study protocol. Among the individuals invited to participate (both cases and controls), more than 90% were successfully interviewed. The primary reason for non-participation was old age.

Data collection

Interviews were conducted by trained personnel using the identical questionnaires in each center. The questions covered demographic and ethnic information, medical history, consanguinity, familial cancer history, living conditions, exposure to chemicals and smokes, alcohol and tobacco consumption, and dietary consumption in both childhood and adulthood. For food items, subjects were asked to choose from 6 intake frequency categories: “1–2 times per day,” “3–4 times per week,” “1–2 times per week,” “1–3 times per month,” “less than 10 times per year” and “never”.

Socio-economic variables

In this study, SES was inferred by household type, category of lodging, occupation and educational level (from 1: no education to 5: university). Occupations were classified into 3 levels ((i) manual-basic occupation; (ii) artisan; (iii) professional or managerial job) based on the literal and intellectual requirement of the job, and therefore it is subject to classification bias. However, this classification was blind to the case/control status, so that this bias should be non-differential. “Household type” was designated as rural or urban residency. In contrast with developed countries, the countries of the Maghreb have a large economic disparity between rural and urban areas, where rural areas are less developed and more traditional in lifestyle, so that the likelihood of misclassification is very low. To capture information about household density we constructed a variable, “category of lodging” with 3 levels: “≤2 persons/room”; “2.1–4 persons/room”; and “>4 persons/room or gourbi (makeshift lodging in rural areas).” This variable is thought to be able to capture better the SES than the original types of housing (gourbi, dar or apartment). In the analysis of the dietary exposures, we use these 6 variables for adjustment in the conditional logistic regression analyses, i.e. “category of lodging” in childhood and adulthood, “household type” in childhood and adulthood, “occupation” and “education.” We also adjusted for exposure to toxic substances (chemical fertilizer, pesticides, fumes, dust, formaldehyde or other chemical substances) as an ever/never variable.

Statistical analysis

All statistical analyses were carried out using STATA 9.0 (STATA Corp. College Station, TX). Conditional logistic regression with strata defined by sex and centers was used to evaluate the association of specific dietary factors with NPC, adjusting for age, the 6 SES measures and exposure to toxic substances. Odds ratios and corresponding 95% confidence intervals were calculated. For foods that were rarely consumed, frequency categories were pooled such that every cell in the contingency table had more than 4 observations. If this still could not be reached after pooling, the variable was dropped from the analyses. Reference levels were assigned to the least exposed level. For exposures with more than 2 levels, the Armitage trend test was used to assess significance. Multivariate analysis was performed using stepwise forward selection, with inclusion/exclusion significance level of 0.01. Adjustment variables such as SES, age and exposure to toxic substances were always kept in the model.


  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. Conclusions
  7. Acknowledgements
  8. References


From January 2002 to March 2005, a total of 636 cases and 615 controls were interviewed. The demographic breakdown by sex, age and study center are shown in Table I. The vast majority (89%) of subjects were of Arab ethnicity, with others belonging to Berber or other ethnic minority groups. There was no significant difference between cases and controls regarding ethnic groups, countries or recruitment centers. Seven hundred seventy (63.5%) subjects were living in urban areas at the time of recruitment with a smaller proportion reporting urban environment in childhood (41%). No differences in household type were found between cases and controls. The mean age at onset of the cases is 42.6 years old, comparable to the mean age at interview of the controls (43.5 years old; p = 0.34). The age at diagnosis of the cases ranged from 11 to 81 years, with an apparent bimodal distribution with peaks at ∼20 years and ∼40 years, similar to previous reports from cancer registry data in this population.7 This may suggest heterogeneous etiology of the NPC development in this population. There was no significant difference between the proportion of cases and controls reporting consanguinity in parents (18.3% vs. 19.8%, p = 0.48).

Table I. Demographic Characteristics of the Study Population
 Case (%)Control (%)
  • 1

    Age at diagnosis for cases and age at interview for controls.

  Tunis152 (24)147 (24)
  Sousse115 (18)83 (14)
  Casablanca207 (33)202 (33)
  Sétif51 (8)67 (11)
  Blida111 (17)116 (19)
  <25102 (16)83 (13)
  25–3478 (12)92 (15)
  35–44143 (22)158 (26)
  45–54178 (28)130 (21)
  55–6496 (15)93 (15)
  ≥6539 (6)59 (10)
  Male450 (71)419 (68)
  Female186 (29)196 (32)


Among the 6 variables of SES, “household type” (urban vs. rural) was not associated with NPC as it was used as a matching variable for the cases and controls. However, “category of lodging” was significantly associated with NPC (Table II), thus confirming the association between SES and NPC risk in previous publications. Although the association between occupation and education level with NPC was not significant (Table II), we included all 6 variables as adjustment variables in the remaining analyses to capture the maximum possible confounding effect.

Table II. Association of socio-economic status (SES) with NPC
SES variablesCase (%)Control (%)OR95% CIp-trend
  1. Analyses were stratified by sex and center, adjusted for age.

Household type in childhood     0.3125
 Urban (ref)269 (42)241 (39)1.0 
 Rural367 (58)374 (61) 
Category of lodging in childhood     0.0209
 ≤=2 persons/room (ref)169 (27)189 (31)1.0 
 2.1–4 persons/room195 (31)194 (32) 
 >4 persons/room or gourbi261 (42)228 (37) 
Household type in adulthood     0.6031
 Urban (ref)393 (64)377 (63)1.0 
 Rural220 (36)223 (37) 
Category of lodging in adulthood     0.0076
 ≤=2 persons/room (ref)374 (62)381 (66)1.0 
 2.1−4 persons/room155 (26)152 (26) 
 >4 persons/room or gourbi77 (13)48 (8) 
Education     0.2543
 None (ref)204 (32)211 (35)1.0 
 Primary school203 (32)158 (26) 
 Secondary school114 (18)114 (19) 
 High school70 (11)86 (14) 
 University45 (7)41 (7) 
Occupation     0.1260
 Manual (ref)394 (62)338 (56)1.0 
 Litterary/artisan156 (25)199 (33) 
 Professional/Managerial81 (13)72 (12) 


Table III presents the relationship between dietary factors and NPC risk. Consumption of rancid butter and rancid sheep fat are significant risk factors for NPC, while consumption of cooked vegetables was inversely associated with NPC. While all 3 of these associations were observed in both adulthood and childhood, their effect in adulthood (ORs for consumption frequency ≥ 3 times/week: 3.0, 3.2, 0.6) are much stronger than those in childhood (ORs for consumption frequency ≥ 3 times/week: 1.8, 2.1, 0.8). Preserved meat including quaddid (dried meat stored in oil) and osban (stomach lining filled with garlic and other spices, which may be either fried or boiled), citrus fruits and industrial olives condiments (olives stored in vinegar and water with salt, aged for at least 8 weeks, stored for months to years) were also associated with increased NPC risk. High consumption of industrial preserved fish (unsalted canned fish) in both childhood and adulthood appears to be inversely associated with NPC. In the multivariate logistic regression analyses (Table IV), dietary factors that remained significant were rancid butter, rancid sheep fat and cooked vegetables in adulthood. Restriction of the analyses to the incident cases only, did not materially change the results (data not shown). Stratified or separate analyses in males or females, age group ≤30 or >30 years old, also did not find significant evidence of heterogeneity in the ORs for the dietary components that were significant in the overall analysis, although the female and ≤ 30 strata were quite small and consequently had low power.

Table III. Association of dietary factors with NPC
Exposure levelsChildhoodAdulthood
CaseCtrlOR95% CITrend-pCaseCtrlOR95% CITrend-p
  1. Analyses were stratified by sex and center, adjusted for age, socio-economic status variables, and exposure to toxic substances.

Rancid butter     0.0023     1E–05
 <10 times/year (ref)3123291.0 3503621.0 
 <3 times/week1901941.20.91.7 1731481.61.22.3 
 ≥3 times/week129871.81.22.6 57263.01.75.2 
Rancid sheep fat     0.0003     2E–07
 <10 times/year (ref)4274781.0 3874211.0 
 <3 times/week1441041.61.22.2 129702.21.53.1 
 ≥3 times/week54292.11.23.5 45183.21.75.9 
Cooked vegetables     0.0588     0.0017
 <3 times/week (ref)2001551.0 148941.0 
 ≥3 times/week4304550.80.61.0 4344640.60.40.8 
Citrus fruits     0.0308     0.1499
 <10 times/year (ref)80931.0 38521.0 
 <3 times/week3853801.30.91.8 3353121.40.82.2 
 ≥3 times/week1661381.61.02.5 2091891.50.92.6 
Toklia     0.9442     0.6455
 <10 times/year (ref)2552771.0 2122151.0 
 <3 times/week1431491.00.71.4 1481431.00.71.5 
 ≥3 times/week1411321.00.71.5 1201190.90.61.3 
Harissa     0.9203     0.7566
 <10 times/year (ref)4574731.0 3813581.0 
 <3 times/week1401081.10.81.6 128881.20.81.7 
 ≥3 times/week24260.80.41.5 28310.70.41.3 
Spicy preserved meat (osben)     0.0251     0.3472
 <10 times/year (ref)5645641.0 5094691.0 
 ≥10 times/year37162.21.14.6 1781.50.63.8 
Preserved meat not spicy (Quaddid)     0.0024     0.005
 <10 times/year (ref)4204961.0 4204641.0 
 ≥10 times/year110701.81.22.7 55292.01.23.4 
Merguez, khelli     0.4654     0.4009
 <10 times/year (ref)5845701.0 5194931.0 
 ≥10 times/year35291.20.72.2 33201.30.72.4 
Home made olives condiments     0.7566     0.9283
 <10 times/year (ref)2552721.0 2402281.0 
 <3 times/week2041961.10.81.4 1931751.00.71.3 
 ≥3 times/week1731431.10.81.5 1351041.00.71.5 
Industrial olives condiments     0.6892     0.0285
 <10 times/year (ref)3713461.0 2462301.0 
 <3 times/week2032290.90.61.1 2592851.00.71.3 
 ≥3 times/week56331.50.92.5 74372.01.33.3 
Industrial preserved fish     0.0264     0.0045
 <10 times/year (ref)6215851.0 5394851.0 
 ≥10 times/year10230.40.20.9 20380.40.20.7 
Table IV. Multivariate analyses of dietary factors
Dietary factorsOR95% CI
  1. Analyses were stratified by sex and center, adjusted for age, socio-economic status variables, and exposure to toxic substances. p values were calculated by likelihood ratio test. Model was built by stepwise forward with inclusion criteria p ≤ 0.01.

Rancid butter
 <10 times/year1.0
 <3 times/week1.51.12.2
 ≥3 times/week2.51.44.5
Rancid sheep fat
 <10 times/year1.0
 <3 times/week1.91.32.7
 ≥3 times/week2.61.44.9
Cooked vegetables
 <3 times/week1.0
 ≥3 times/week0.60.40.8


  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. Conclusions
  7. Acknowledgements
  8. References

In this study, 636 cases and 615 controls from 3 Maghrebian countries were investigated for dietary association with NPC. When dietary components were analyzed individually, we found that consumption of rancid butter and rancid sheep fat were significantly associated with NPC risk. Cooked vegetables had an inverse association, especially in adulthood. Other foods such as fresh citrus fruits and spicy preserved meat in childhood, industrial made olive condiments in adulthood, preserved meat not spicy and industrial preserved fish in both age periods were associated with NPC. In this questionnaire, both spicy preserved meat and preserved meat not spicy include quaddid and osban. Although the level of spice varies among districts, generally osban is much more spicy than quaddid, and therefore, preserved meat that is not spicy consists primarily of quaddid, while spicy preserved meat mainly consists of osban.

Compared with the other 3 studies from the same region, this study confirms the reported increased risk of NPC associated with rancid butter, rancid sheep fat and quaddid. In addition, we have newly identified potential associations of osban, industrial made olive condiments, and cooked vegetables with NPC. In contrast to previous findings, toklia was not associated with NPC in our study, and we can exclude an effect of greater than a 50% increased risk. We estimated that the power to detect an increased risk similar to that previously reported for toklia with our sample size and observed consumption prevalence (40%) is almost 100%, so that it is unlikely that this is because of lack of power. The same is observed for harissa, which had been reported to be associated with a 5-fold increased NPC risk if eaten more than twice a week.19 Because the preparation of Harissa could vary a lot between Morocco, Tunisia and Algeria, it was analyzed separately for each country, with no effect found in any individual country. Therefore, it is unlikely that the variation in harissa preparation is the explanation of its lack of association. Other differences in our study compared to previous reports are that preserved fish was found to be inversely associated in this article, rather than positively associated as expected; high consumption of fresh citrus fruits was found to be a risk factor of NPC, which contrast with many other reports.13, 14, 15, 22, 24, 26, 27 It should be noted that, the preparation of industrial preserved fish in Maghrebian countries (unsalted canned fish) are quite different from those in South-East Asia, where sea fish are highly salted and sun dried.

Although these associated foods are significant in terms of 95% confident intervals, they must be interpreted carefully because of the multiple testing problem. In regard of this, p values for trend tests were also calculated for these variables. The association with rancid butter and rancid sheep fat have very significant p values (p = 0.00001, p = 0.0000002, respectively), and thus are unlikely to happen by chance even in 50 independent tests. Cooked vegetables, quaddid and industrial preserved fish have intermediate significant levels, whereas fresh citrus fruits, osban and olive condiments have marginal p values. In light of this, the associations of fresh citrus fruits, osban and olive condiments could be spurious findings resulted from multiple testing.

In multivariate analyses, when we included all nominally significant variables, significance level of most factors decreased remarkably, except cooked vegetables, which was even more significant. Variables that remained in the model are rancid butter, rancid sheep fat and cooked vegetables, all consumed in adulthood. The association with other preserved foods found in the adjusted univariate analyses disappeared with the addition of rancid butter, rancid sheep fat and cooked vegetables into the model, as did the association with fresh citrus fruits. It is interesting to see that although rancid butter and rancid sheep fat are positively correlated with each other (individuals who consume rancid butter also tend to consume rancid sheep fat, with similar frequencies, spearman's correlation rho = 0.35, p < 0.0001), they both remained in the model with very significant p values.

Dose-response relationship is an important consideration in causality inference. In this study, from the odds ratios of different level of consumptions in childhood and adulthood, a clear dose-response pattern was observed for rancid butter and rancid sheep fat, both in univariate and multivariate analyses. This confirms the validity of the findings of rancid fat. Other factors with more than 2 levels had not shown a clear monotonic dose response pattern, especially harissa and toklia.

Possible confounding, validity and bias

Since SES is highly influential on diet, it could be a strong confounder in dietary association analyses. In this article, the SES confounder effect was controlled in 2 ways. First, the cases and controls were matched by household type in childhood. The reason for choosing this in the matching is that, exposures during childhood are more important for disease development than those in adulthood, and that household type is highly related to the childhood SES status. This matching will reduce the difference of childhood SES between cases and controls, leading to less confounder effect. Second, a series of variables that can infer SES status were used in the adjustment in conditional logistic regression, including the “category of lodging,” a strong SES index. Some other SES indices such as occupation and education were not significantly associated with NPC, which could be partly explained by the matching in the study design (data not shown). Nevertheless, we used all of these variables in the adjustment, to minimize any possible confounding; however, it is still likely that not all of the confounding effect was captured in our analyses. In fact, some of the associations found between NPC and consumption of specific food items could partly be explained by residual SES confounding.

Another concern of the validity of the study design is that controls in this study were taken from hospitalized individuals and family members of non-NPC cancer patients. These subjects may have higher chance than the general population to share a common environmental exposure with NPC patients, if any other disease has similar etiology with NPC. However, we believe that given the diversity of clinical services from which the controls were recruited, such effect has been highly diluted. In addition, the sharing of a common environmental exposure between cases and controls could only decrease the power of the study, rather than causing a spurious result. Therefore, this is unlikely to be a major threat to the validity of our conclusions.

Similarly, the fact that some of the cases were recruited some time after their NPC diagnosis could result in survival bias. However, 75% of our cases were interviewed within 1 year of diagnosis, and almost all within 3 years. Further analyses restricted only to incident cases did not materially change the results.

Lastly, as in any retrospective study, our results could be at least partly explained by recall bias, although this problem should be minor, since this is a relatively unsophisticated population in terms of perceived relationship between diet and health, and particularly since the cases in our study would be expected to have no knowledge at all of previously postulated NPC risk factors.

Responsible ingredients

Main ingredients in the foods that have a potential role in the carcinogenic process have been one of the study concerns in previous publications. Investigators have found that N-nitrosamines (N-nitrosodimethylamine (NDMA), N-nitrosopyrrolidine (NPYR) and N-nitrosopiperidine (NPIP)), compounds that are classified as probably or possibly carcinogenic to humans (International Agency for Research on Cancer, monograph Vol.17 Suppl. 7, 1987), were detected in toklia, quaddid, Chinese salted fish, brine fermented vegetables and Greenland dried fish,29, 30, 31, 32 or could be fermented within human stomach from their precursors contained in Chinese salted fish and Tunisian spice.33 Besides N-nitrosamines, lignin-containing high molecule complexes were isolated from harissa and were shown to be strong EBV inducers.34

In this article, neither N-nitrosamine-containing toklia, nor EBV-inducer-containing harissa were significantly associated with NPC, in comparison with rancid butter and rancid sheep fat. Importantly, higher level of N-nitrosamines in rancid fat has not been demonstrated, which suggests some other disease causing chemicals in this population. A possible compound is butyric acid, which is also named n-Butanoic Acid. The glyceride form of butyric acid makes up 3 to 4% of butter, and is released into free butyric acid by hydrolysis when it becomes rancid. Butyric acid is known to be able to activate Epstein-Barr virus in the B-lymphoid cells into lyric cycle,35 and therefore, could be related to NPC.

Vegetables and fruits are well established as protective factors for most epithelial cancers. Here we can now confirm this for NPC also. It is interesting that only cooked vegetables were significantly associated, while raw vegetables were not; if anything, their effects were to increase NPC risk. These 2 dietary components could be different in terms of the specific vegetables consumed, the portion sizes, or their correlation with SES. Further analysis shows that while consumption of raw vegetables is positively associated with higher SES (p = 0.0001), consumption of cooked vegetables is not (p = 0.12).


  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. Conclusions
  7. Acknowledgements
  8. References

In conclusion, this study is the first large-scale epidemiology study that has been performed in the Maghrebian countries of North Africa to infer associations between dietary factors and the risk of Nasopharyngeal Carcinoma. It has provided a clearer view of the dietary risk factors in the NPC etiology, particularly with regard to rancid butter and sheep fat. Although the specific substances contained in these foods that cause NPC are not yet known, results of this study have provided the key information necessary to design a primary preventive strategy aimed at reducing risk factor exposure for nasopharyngeal carcinoma, especially in Maghrebian countries, where NPC has a strong social impact as it is more affected in young, male individuals, typically the main economic support of a family.


  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. Conclusions
  7. Acknowledgements
  8. References

We thank our colleagues, Ms. Valérie Gaborieau for efficient data management, Ms. Colette Bonnardel for data entry and checking, Ms. Hélène Renard and Mr. Othman Yaqoubi for database design, Dr. Dominique Jeannel and Dr. Annie Sasco for help in designing the questionnaire, and Dr. Mohamed Maamer for its efficent help to the study in Tunis.


  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. Conclusions
  7. Acknowledgements
  8. References
  • 1
    Parkin DM,Whelan SL,Ferlay J,Teppo L,Thomas DB. Cancer incidence in five continents, vol. VIII. Lyon, France: International Agency for Research on Cancer, 2002.
  • 2
    Jia WH,Huang QH,Liao J,Ye W,Shugart YY,Liu Q,Chen LZ,Li YH,Lin X,Wen FL,Adami HO,Zeng Y, 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.
  • 3
    Armstrong RW,Kannan KM,Dharmalingam SK,Ponnudurai JR. Incidence of nasopharyngeal carcinoma in Malaysia, 1968–1977. Br J Cancer 1979; 40: 55767.
  • 4
    Devi BC,Pisani P,Tang TS,Parkin DM. High incidence of nasopharyngeal carcinoma in native people of Sarawak, Borneo Island. Cancer Epidemiol Biomarkers Prev 2004; 13: 48286.
  • 5
    Parkin DM,Whelan SL,Ferlay J,Storm H. Cancer incidence in five continents, vol. VII. Lyon, France: International Agency for Research on Cancer, 1997.
  • 6
    Ellouz R,Cammoun M,Zaouche A,Ben Attia R. [Cancers of the rhinopharynx in Tunisia. Epidemiological and anatomo-clinical aspects]. Bull Cancer 1975; 62: 295306.
  • 7
    Ellouz R,Cammoun M,Attia RB,Bahi J. Nasopharyngeal carcinoma in children and adolescents in Tunisia: clinical aspects and the paraneoplastic syndrome. IARC Sci Publ 1978: 20; 11529.
  • 8
    Hildesheim A,Levine PH.Etiology of nasopharyngeal carcinoma: a review.Epidemiol Rev 1993; 15: 46685.
  • 9
    Jeannel D,Bouvier G,Hubbert A. Nasopharyngeal carcinoma: an epidemiological approach to carcinogenesis. Cancer Surveys 1999; 33: 12555.
  • 10
    Armstrong RW,Armstrong MJ,Yu MC,Henderson BE. Salted fish and inhalants as risk factors for nasopharyngeal carcinoma in Malaysian Chinese. Cancer Res 1983; 43: 296770.
  • 11
    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: 29196.
  • 12
    Yu MC,Ho JH,Lai SH,Henderson BE. Cantonese-style salted fish as a cause of nasopharyngeal carcinoma: report of a case-control study in Hong Kong. Cancer Res 1986; 46: 95661.
  • 13
    Yu MC,Huang TB,Henderson BE. Diet and nasopharyngeal carcinoma: a case-control study in Guangzhou, China. Int J Cancer 1989; 43: 107782.
  • 14
    Yu MC,Mo CC,Chong WX,Yeh FS,Henderson BE. Preserved foods and nasopharyngeal carcinoma: a case-control study in Guangxi, China. Cancer Res 1988; 48: 19549.
  • 15
    Yuan JM,Wang XL,Xiang YB,Gao YT,Ross RK,Yu MC. Preserved foods in relation to risk of nasopharyngeal carcinoma in Shanghai, China. Int J Cancer 2000; 85: 35863.
  • 16
    Zheng X,Yan L,Nilsson B,Eklund G,Drettner B. Epstein-Barr virus infection, salted fish and nasopharyngeal carcinoma. A case-control study in southern China. Acta Oncol 1994; 33: 86772.
  • 17
    Sriamporn S,Vatanasapt V,Pisani P,Yongchaiyudha S,Rungpitarangsri V. Environmental risk factors for nasopharyngeal carcinoma: a case-control study in northeastern Thailand. Cancer Epidemiol Biomarkers Prev 1992; 1: 34548.
  • 18
    Lanier A,Bender T,Talbot M,Wilmeth S,Tschopp C,Henle W,Henle G,Ritter D,Terasaki P. Nasopharyngeal carcinoma in Alaskan Eskimos Indians, and Aleuts: a review of cases and study of Epstein-Barr virus, HLA, and environmental risk factors. Cancer 1980; 46: 210006.
  • 19
    Jeannel D,Hubert A,de Vathaire F,Ellouz R,Camoun M,Ben Salem M,Sancho-Garnier H,de The G. Diet, living conditions and nasopharyngeal carcinoma in Tunisia—a case-control study. Int J Cancer 1990; 46: 42125.
  • 20
    Laouamri S,Hamdi-Cherif M,Sekfali N,Mokhtari L,Kharchi R. [Dietary risk factors of nasopharyngeal carcinoma in the Setif area in Algeria]. Rev Epidemiol Sante Publique 2001; 49: 14556.
  • 21
    Ammor S,Baali A,Hubert A,Cherkaoui M. Epidemiological approach of the nasopharyngeal carcinoma in the region of Marrakech. Antropo 2003; 5: 3947.
  • 22
    Zheng YM,Tuppin P,Hubert A,Jeannel D,Pan YJ,Zeng Y,de The G. Environmental and dietary risk factors for nasopharyngeal carcinoma: a case-control study in Zangwu County, Guangxi, China. Br J Cancer 1994; 69: 50814.
  • 23
    Armstrong RW,Eng AC. Salted fish and nasopharyngeal carcinoma in Malaysia. Soc Sci Med 1983; 17: 155967.
  • 24
    Armstrong RW,Imrey PB,Lye MS,Armstrong MJ,Yu MC,Sani S. Nasopharyngeal carcinoma in Malaysian Chinese: salted fish and other dietary exposures. Int J Cancer 1998; 77: 22835.
  • 25
    Hubert A,de The G. Dietary behavior, way of life, and nasopharyngeal cancer. Bull Cancer 1982; 69: 47682.
  • 26
    Chen DL,Huang TB. A case-control study of risk factors of nasopharyngeal carcinoma. Cancer Lett 1997; 117: 1722.
  • 27
    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: 6759.
  • 28
    Gallicchio L,Matanoski G,Tao XG,Chen L,Lam TK,Boyd K,Robinson KA,Balick L,Mickelson S,Caulfield LE,Herman JG,Guallar E, et al. Adulthood consumption of preserved and nonpreserved vegetables and the risk of nasopharyngeal carcinoma: a systematic review. Int J Cancer 2006; 119: 112535.
  • 29
    Zou XN,Lu SH,Liu B. Volatile N-nitrosamines and their precursors in Chinese salted fish—a possible etological factor for NPC in china. Int J Cancer 1994; 59: 1558.
  • 30
    Poirier S,Hubert A,de The G,Ohshima H,Bourgade MC,Bartsch H. Occurrence of volatile nitrosamines in food samples collected in three high-risk areas for nasopharyngeal carcinoma. IARC Sci Publ 1987; 84: 41519.
  • 31
    Poirier S,Ohshima H,de The G,Hubert A,Bourgade MC,Bartsch H. Volatile nitrosamine levels in common foods from Tunisia, south China and Greenland, high-risk areas for nasopharyngeal carcinoma (NPC). Int J Cancer 1987; 39: 29396.
  • 32
    Zou X,Li J,Lu S,Song X,Wang X,Guo L,Li J,Lin Q,Ye J. Volatile N-nitrosamines in salted fish samples from high- and low-risk areas for NPC in China. Chin Med Sci J 1992; 7: 2014.
  • 33
    Poirier S,Bouvier G,Malaveille C,Ohshima H,Shao YM,Hubert A,Zeng Y,de The G,Bartsch H. Volatile nitrosamine levels and genotoxicity of food samples from high-risk areas for nasopharyngeal carcinoma before and after nitrosation. Int J Cancer 1989; 44: 108894.
  • 34
    Bouvier G,Hergenhahn M,Polack A,Bornkamm GW,de The G,Bartsch H. Characterization of macromolecular lignins as Epstein-Barr virus inducer in foodstuff associated with nasopharyngeal carcinoma risk. Carcinogenesis 1995; 16: 187985.
  • 35
    Bauer G. Quantitative analysis of the cooperative effect between inducers of Epstein-Barr virus antigen synthesis. J Gen Virol 1983; 64(Part 6): 133746.