Cantonese-Style Salted Fish As a Cause of Nasopharyngeal Carcinoma: Report of a Case-Control Study in Hong Kong1

[CANCER RESEARCH 46, 956-961, February 1986]

Cantonese-style Salted Fish as a Cause of Nasopharyngeal Carcinoma: Report of a Case-Control Study in Hong Kong1

Mimi C. Yu,2 John H. C. Ho, Shiu-Hung Lai, and Brian E. Henderson

Department of Preventive Medicine, University of Southern California School of Medicine, Los Angeles, California 90033 [M. C. Y., B. E. H]â€žand Medical and Health Department Institute of Radiology and Oncology, Queen Elizabeth Hospital, Kowloon, Hong Kong [J. H. C. H., S-H. L.]

ABSTRACT an important risk factor for NPC in high-risk Chinese even though such exposure in an occupational setting may explain some Two hundred fifty incident cases of nasopharyngeal carcinoma cases of NPC (3,10,11). under age 35 years in Hong Kong Chinese and an equal number Thirteen years ago, Ho (12, 13) first suggested that ingestion of age- and sex-matched friend controls were interviewed. Moth of Cantonese-style salted fish," a favorite food item among ers of cases and controls were interviewed also, if available, to southern Chinese, might be a risk factor for NPC. Since then, obtain information on childhood events concerning the study three case-control studies conducted among southern Chinese subjects. Consumption of Cantonese-style salted fish during all populations have investigated such an hypothesis. All three time periods was significantly associated with nasopharyngeal studies are supportive of the hypothesis. In Los Angeles, signif carcinoma; the association was especially strong during child icantly more Chinese patients reported current consumption of hood. The relative risk for having Cantonese-style salted fish as Cantonese-style salted fish compared to controls (RR for more one of the first solid foods during weaning was 7.5 (95% confi than once a week consumption compared to none, 3.1) (10). In dence limits, 3.9, 14.8), and the relative risk for consuming the Hong Kong, Cantonese-style salted fish was fed to infants during food at least once a week compared to less than once a month weaning significantly more often in households with a NPC case at age 10 years was 37.7 (95% confidence limits, 14.1, 100.4). than in control households (RR = 2.6) (14). In Malaysia, histories It is estimated that over 90% of young nasopharyngeal carci of Cantonese-style salted fish intake in childhood and adoles noma cases in Hong Kong Chinese can be attributed to con cence were significant risk factors for NPC in Chinese; daily sumption of this food during childhood. consumption of this food item in childhood carried a RR of 17.4 (11). There are supportive experimental findings as well: muta- genie activity (15) and volatile nitrosamines (16, 17) were found INTRODUCTION in Cantonese-style salted fish extracts; Wistar albino rats fed Cantonese-style salted fish were found to pass mutagenic urine NPC3 is a rare cancer among whites in Europe and North (18); and in an experiment in which 20 Wistar albino rats were America; the age-standardized incidence rates in these popula fed Cantonese-style salted fish, 4 subsequently developed car tions are considerably less than 1 per 100,000 (1). In contrast, cinomas in the nasal or paranasal regions while none of the 6 NPC is one of the most common cancers among Chinese residing control rats did (19). in the southeastern provinces of China, particularly the province The southern Chinese diet contains a variety of salt-preserved of Guangdong (of which Hong Kong is a part geographically); foods and one which is rich in salted fish tends also to be rich in the age-standardized incidence rate in males is greater than 50 other preserved foods. Therefore, it is important to investigate if per 100,000 in certain parts of the province (2, 3). Southern other food items that are associated with Cantonese-style salted Chinese who migrated to intermediate-risk areas such as south fish intake could also be important in the etiology of NPC. In the east Asia or low-risk areas such as Canada and the United present study, we examined the relationship of lifetime dietary States continue to show a high rate of NPC (4-8). However, habits to the development of NPC in Chinese patients under age succeeding generations of Chinese-Americans in California (3, 6) 35 years in Hong Kong [NPC is the most common cancer in and Hawaii (8) and Chinese in Australia (9) who are more likely Hong Kong Chinese ages 15-34 years (20)]. Other factors under to give up their traditional ways of life than their southeast Asian study were prior ear and nose disease, domestic and occupa counterparts display a decrease in risk for NPC. This suggests tional exposure to inhalants, use of nasal oil, and cigarette that environmental factors inherent in the traditional culture of smoking. south China are responsible for the extraordinarily high rates of this disease in southern Chinese. Inhalation of carcinogens containing smoke or dust as a risk MATERIALS AND METHODS factor for NPC in Chinese has been investigated in many studies. Results suggest that domestic exposure to smoke or dust is not Cases were incident cases of NPC under age 35 years among Chinese residents of Hong Kong. Cases were identified from the files of four Received 7/29/85; revised 10/15/85; accepted 10/29/85. The costs of publication of this article were defrayed in part by the payment of 4 Cantonese-style salted fish are fish preserved in the traditional manner in page charges. This article must therefore be hereby marked advertisement in Southern China. In general, either the fish is not gutted or the guts are drawn out accordance with 18 U.S.C. Section 1734 solely to indicate this fact. through the throat without making an incision in the belly of the fish. Salting is 1Supported in part by Grants CA00884 from the National Cancer Institute and carried out in wooden vats, the length of time ranging from 1-5 days. Afterwards, SIG-2 from the American Cancer Society, donations from Mr. Wong Hoo Chuen of the fish are taken out to dry in direct sunlight for 1-7 days, depending on the size Hong Kong, and the Hong Kong Anti-Cancer Society. of the fish and the weather. During drying, insect infestation is often a serious 2 To whom requests for reprints should be addressed. problem due to the humid weather in south China. Sometimes, the fish is allowed 'The abbreviations used are: NPC, nasopharyngeal carcinoma; RR, relative to soften by decomposition before salting to produce the "soft meat" salted fish, the rest are called "tough meat" salted fish. risk; CL, confidence limits.

CANCER RESEARCH VOL. 46 FEBRUARY 1986 956

Downloaded from cancerres.aacrjournals.org on October 2, 2021. © 1986 American Association for Cancer Research. SALTED FISH AS CAUSE OF NASOPHARYNGEAL CARCINOMA hospitals in Hong Kong; Queen Mary Hospital; Queen Elizabeth Hospital; for multivariate analysis and individual variables with more than two Hong Kong Sanatorium and Hospital; and Baptist Hospital. According to possible outcomes. We also analyzed the data on all mothers, stratifying data collected by the Hong Kong Cancer Registry, these four hospitals by the sex of their child (21). Results of the stratified analysis were similar together identify over 90% of all new cases of NPC in Hong Kong. to those obtained from the matched analysis. All statistics presented in Eligible patients diagnosed after February 28, 1981, were asked to this paper were based on the matched analysis. Pairs in which either the participate in the study until 250 interviews were completed. Of the first case or the control had a missing value were eliminated from the 266 patients contacted, 16 refused to be interviewed. corresponding analysis. For factors with prevalence rates in the general Controls were friends of cases, one per case, who were matched to population between 25 and 75% (all of our primary variables fall within the index case by sex and date of birth (within 5 years). (Age and sex this range), the expected power of this study to detect a RR of 2.0 with are confounding variables which must be controlled for either in the a one-sided significance level of 0.05 is greater than 90%. analysis or in the study design.) At the end of each case interview, the patient was asked to provide the name of a close friend (excluding blood RESULTS relatives) who met the matching criteria. In two instances, the first identified "close" friend refused to participate and a second "close" friend One hundred sixty male and 90 female cases were inter was interviewed. viewed; their mean age at diagnosis was 29.2 years. The diag We attempted to interview all mothers of cases and controls who resided in Hong Kong. Thirty-eight case-mothers had died and 24 did noses of all but five cases were confirmed histologically. Controls not reside in Hong Kong; of the 188 case-mothers contacted, 6 refused had a mean age of 28.6 years at date of diagnosis of the index to be interviewed. For the control-mothers, 27 had died, 50 did not reside case. One hundred seventy-nine (72%) cases were born in Hong in Hong Kong, and 18 refused to be interviewed. In all, we interviewed Kong compared to 151 (60%) controls. Only 9 cases and 10 182 case-mothers and 155 control-mothers; there were 127 matched controls were bom outside of Guangdong Province. Cases and case-mother control-mother pair interviews. controls were similar with respect to ethnic origin (Cantonese, All interviews were conducted in person by S. Lai using a standard Sze Yap, ChiÃ¹Chau, elsewhere in Guangdong, Fukien/Taiwan, questionnaire for all cases and controls and a separate standard ques elsewhere in China), marital status, and level of education. tionnaire for their mothers. The questionnaire for cases and controls Table 2 shows the self-reported frequencies of Cantonese- requested information on place of birth; ethnic origin; dietary habits (3 years prior to diagnosis and at age 10 years); occupational exposure to style salted fish consumption among cases and controls during smoke, dust, or fumes; exposure to incense, antimosquito coils, and childhood (at age 10 years) and as an adult (3 years prior to other domestic inhalants (3 years prior to diagnosis and at age 10); diagnosis). Consumption during both time periods was signifi cigarette smoking; use of nasal oils; lifetime history on type of cooking cantly associated with NPC; the association was especially fuel used; prior ear or nose disease; and history of NPC in first degree strong for childhood consumption. For weekly consumers of relatives. Mothers were asked about childhood events concerning the salted fish at age 10 years, the RR of NPC compared to those cases and the controls. The questionnaire for mothers contained detailed who never or rarely ate this food item was 37.7 (95% CL = 14.1, questions on weaning habits; dietary habits during childhood (between 100.4). ages 1 and 2 years and at age 10 years); exposure to smoke, dust, or The mothers also were asked about their child's consumption fumes at home; chronic illnesses of the ear or nose; and use of nasal pattern of Cantonese-style salted fish at age 10 years. There oils. Information on diet included frequency of consumption on all pre was again a very strong positive association between frequency served foods eaten by southern Chinese. Also included in our food list were the various sauces and condiment pastes used by Chinese in their of consumption of salted fish and NPC. Only one of the re cooking and the cooling soup, a type of medicinal drink unique to the sponses between the mothers and their offspring (a case) dif southern Chinese (Table 1). Other food items were fresh meats and fish, fered by more than one frequency category (the categories were milk, bread, oranges and other fruits, carrots, tomatoes, and fresh leafy rarely, once a month to less than once a week, and one or more green vegetables. times a week). We used matched-pair methods (21) to analyze the interview data. Table 2 also shows the frequencies of Cantonese-style salted Study variables were examined singly and then jointly for confounding fish intake among cases and controls during other times in their and interaction effects. We used the exact binomial test on individual childhood as reported by their mothers. Significantly more case- dichotomous variables and the multivariate logistic regression method mothers reported feeding salted fish, salted fish head soup, and Tabtel broth from steamed salted fish to their offspring between ages Traditionalfood items included in the study questionnaires 1 and 2 years compared to control mothers. Foods that were Chinese name considered as typical meals for the subject when he or she was Food item between ages 1 and 2 years were coded as eaten "often" by the Salted foods: fish, pork, Harm yu, harm ju yoke, mui vegetables,roots, ol choi. harm choi, choi po, subject. Compared to individuals who never ate salted fish ive, duck egg Chungchoi, lanmgok, between ages 1 and 2 years, those who were fed this food harm dann "often" had a RR for NPC of 20.2 (95% CL = 6.8,60.2). Similarly, Cured meats: pork, Laap yoke, laap arp, laap duck, pork sausage, Cheung,gone Cheung more cases than controls were given salted fish as one of their liver sausage first four solid foods during weaning (RR = 7.5, 95% CL = 3.9, Dried foods: shrimp, Ha mei, bark guor, hung joe, gingko seed, fruits chan pai mui, gar ink gee 14.8). Sauces and condi See yao, yu lo. ho yao, Table 3 presents the salted fish intake frequencies for male ments: soy sauce, harm ha, poun kay go, fish sauce, oyster tau si, min si, hoi sin and female cases and controls. The consumption patterns be sauce, shrimp paste, Cheung tween male and female controls were very similar, and the risk crab paste, bean ratios for corresponding levels of consumption are quite com pastes Cooling soup Leung char parable between the two sexes. Moldy bean curd Fu yu Cases and controls who ate Cantonese-style salted fish reg Fermentedduck egg Pay dann ularly were similar with respect to the current usual kind of salted

CANCER RESEARCH VOL. 46 FEBRUARY 1986 957

Table 2 Table 3 Frequency of salted fish consumption among cases and controls during various Frequency of salted fish consumption among male and female cases and controls time periods"

of Ofcontrols1646619124371081013923116833481213882799631Matchedconfidence of of Of of cases107100376227231012511319057196541992527583925102No.RR1.02.33.27.51.04.21.015.037.71.05.11.06.120.21.08.31.05.112.31.07.595%interval1.5,3.51.7,6.10.9, cases657124881711043241663MaleNo.controls1093813687121561746019MatchedRR1.03.53.31.014.157.11.08.125.61.07.3No.cases4229192444292217939FemaleNo.controls55287403018271743612MatchedRR1.01.33.61.016.825.11.03.613.71.07.8 3 yragoSalted 3 yragoRarely1 fishRarely1/mo than1/wk1-h/wkAt/mo to less 1/wk1 to less than dailyDailySalted+/wk but less than 65.31.6,11.76.0, age 10yrRarely1 soupRarely1+/moAtfish head than1/wk1+/wkBetween/mo to less

age10yrSalted fishRarely1/mo and2yrNeverSometimesOftenDuringages 1

1/wk1+/wkSaltedbut less than 37.214.1,100.42.6,

soupRarely1+/moBetweenfish head weaningNeverEverNo. 10.43.0,

ages 1 and 2 yr (re mother)Saltedported by fishNeverSometimesOftenSalted date, cooling soup, and dried fruits at least once a week at age 10 years. Moldy bean curd and salted mustard green were 12.56.8, 60.22.5, consumed at least weekly by 18 and 25% of cases, respectively, at age 10 years. On the other hand, 46% of cases ate Cantonese- soupNeverEverSaltedfish head style salted fish at least once a week at age 10 years. 32.52.6, A history of NPC among first degree relatives (parents and siblings) was a significant risk factor for NPC; 18 cases compared brothNeverSometimesOftenDuringfish to 4 controls reported such a history. Eight case-mothers, 1 10.04.6, case-father, and 9 case-siblings had NPC; among controls there 32.83.9, were 1 father and 3 siblings who had NPC. All except one of bymother)Saltedweaning (reported these 18 cases (and 2 of the 4 controls) ate Cantonese-style salted fish regularly during childhood. fishNeverEverNo. Table 4 presents the RRs for domestic exposure to smoke, 14.8 dust, or fumes among cases and controls. A significantly in 'Total may not add up to 250 case-control pairs or 127 case-mother/control- creased risk (RR = 1.7) was observed for ever use of wood as mother pairs due to missing values. cooking fuel. The association was not statistically significant after adjusting for Cantonese-style salted fish intake. All remain fish eaten ("soft meat" versus "tough meat") and the current ing RRs in Table 4 (either unadjusted or after adjusting for salted usual method of cooking it (over 90% of both case and control fish intake during weaning and at age 10 years) had 95% groups reported steaming the fish over rice). As expected, confidence intervals that included unity. Significantly elevated consumption patterns during the different time periods were risks to NPC were observed for occupational exposure to smoke, highly correlated. Logistic regression analysis was used to study dust, or fumes (Table 5). However, after taking consumption of the joint effect of consumption of salted fish in the various time salted fish into account, none of the adjusted RRs was signifi periods. Consumption during weaning and at age 10 years were cantly different from one. On the other hand, the RRs for salted independent significant risk factors for NPC; each remained fish intake remained highly significant after these other significant significant after adjustment was made for the other factor. There factors were taken into account. was no significant residual effect for the remaining salted fish Cases and controls were similar in their cigarette smoking habits and in prior ear or nose disease. Very few cases and variables. Besides Cantonese-style salted fish, consumption (3 years controls were exposed to nasal oil at age 10 years. However, prior to diagnosis or at age 10 years) of a number of preserved significantly more cases used nasal oil 3 years prior to diagnosis foods including moldy bean curd, salted pork, pork sausage, compared to controls; reasons given for the use indicate that liver sausage, salted mustard green (mui choi), dried shrimp, such use was probably related to the onset of their disease. dried red date, cooling soup, and dried fruits (chan pai mui, gar ink gee) were significantly associated with NPC. None of these DISCUSSION foods remained significantly associated with NPC after adjusting for salted fish intake. Most of the above foods were not regularly This study strongly suggests that consumption of Cantonese- consumed by the cases. Only 2-7% of cases reported eating style salted fish is the primary cause of NPC in young Hong salted pork, pork sausage, liver sausage, dried shrimp, dried red Kong Chinese. Due to the possibility of bias in every observa-

CANCER RESEARCH VOL. 46 FEBRUARY 1986 958

Table 4 Control Selection. A comparison group nonrepresentative of Domestic exposureto smoke, dust, or fumes among cases and controls the population at risk can lead to a spurious association or mask Unad a real association. The controls in this study are healthy individ No. of No. of justed Adjusted8 cases controls RR RR uals explicitly comparable to the cases in confounding variables Wood as cooking fuel such as age, sex, and socioeconomic status. Friend controls Never 67 89 1.0 1.0 pose the risk of being too similar to the cases; they are unlikely Ever 183 161 1.7* 1.4 to produce spurious associations. More case-mothers (182/250 = 73%) than control-mothers 3 yr ago Incense (155/250 = 62%) were available for interview. There is no Never 64 80 1.0 1.0 Ever 186 170 1.4 1.4 evidence, however, to indicate that these mothers were not representative of the source population. Comparisons of all case- Antimosquito coils mothers to all control-mothers were very similar to those be Never 189 192 1.0 1.0 Ever 61 58 1.1 2.4 tween pairs of mothers of cases and their matched controls. Since the interviewer was aware of the case status of the Atage10yr Incense respondent and the interview took place after the diagnosis of Never 42 57 1.0 1.0 NPC, we had taken special precautions against possible inter Ever 208 193 1.6 0.9 viewer bias and recall bias on the part of the cases and their mothers. All questions were in closed form, were highly struc Antimosquito coils Never 172 174 1.0 1.0 tured, and did not require probing by the interviewer. Salted fish Ever 78 76 1.0 1.8 was included with other food items in a long list with specific At birth (reported by mother) frequencies for each food as the only requested response. The Incense respondent was asked to choose one from among two, three, NeverEverAntimosquito or four frequency categories depending on whether the time period of interest was early or late childhood or adulthood. There coilsNeverEverCarried was close correlation between the subject's chosen category of salted fish consumption frequency at age 10 years and his or her mother's response. Only one of 337 pairs of subject versus cookingNeverOccasionallyOften1910898293552402010790372363411.01.11.00.61.00.50.61.00.61.00.81.00.80.7subject while mother responses differed by more than one frequency category. Thus, it is unlikely that systematic bias from the interviewer or the respondent could completely account for the dramatic differ " Adjusted for salted fish intake during weaning and at age 10 years. ence in salted fish consumption frequency between cases and 6 Lower 95% confidence limit >1.0. controls. Could salted fish be a confounder for some etiological variable Table 5 not measured in this study? We think it is highly unlikely. If an Occupationalexposureto smoke, dust, or fumesamong cases and controls observed association is not causal but simply the reflection of a of of causal association between some other factor and the disease, cases240 controls247 RR10*4.5"1.0 RR1.0 then this latter factor must be more strongly related to the SmokeNever disease (that is possessing a higher relative risk) than is the EverDustNever 10191591753209 0.61.0 former factor (21). None of the other risk factors (genetic or viral) not investigated in this study has been shown to exhibit such a strong association with NPC. 1.8"1.0 EverChemical 41201 1.21.0 Aside from lack of alternative explanations, Breslow and Day (21) stated the strength and specificity of the association, a dose- fumesNever response effect, temporal relation of risk to exposure, consist 2.06Adjusted8 EverNo. 75No. 49Unadjusted 0.7 ency with other studies, and biological plausibility as criteria for Adjusted for salted fish intake during weaning and at age 10 years. assessing causality. 3 Lower 95% confidence limit >1.0. The strength of the association between Cantonese-style salted fish intake and NPC found in this study is among the tional study, we shall discuss in some detail the alternative largest ever reported between an exogenous factor and a human explanations to our findings. cancer. A dose-response effect is observed in both childhood Case Selection. Misleading findingsfrom a case-control study and adulthood consumption. The stronger association with child can be the result of inclusion of a nonrepresentative or even a hood consumption is compatible with the observed age pattern wrong sample of cases. In the present study, all but five cases of NPC among southern Chinese. No other factors examined in (98%) were confirmed histologically, and none of the tumors this study showed an association with NPC that is of the same were incidental pathological findings. The 266 cases we con magnitude as that with intake of Cantonese-style salted fish. As tacted represent over 90% of all new cases of NPC in Hong a matter of fact, all other significant associations could be Kong and only 16 of them (6%) refused to participate in the explained by the consumption of Cantonese-style salted fish. study. The distributions of cases by sex, age, or histological Three earlier case-control studies of NPC in Chinese have classification are similar to the population-based Hong Kong investigated the possible role of Cantonese-style salted fish cancer registry data. intake in the etiology of NPC. Henderson and Louie (10) studied

CANCER RESEARCH VOL. 46 FEBRUARY 1986 959

Downloaded from cancerres.aacrjournals.org on October 2, 2021. © 1986 American Association for Cancer Research. SALTED FISH AS CAUSE OF NASOPHARYNGEAL CARCINOMA only current consumption; their relative risk estimates are very Cantonese-style salted fish intake. Numerous studies have dem similar to ours for the comparable time period. Armstrong ef al. onstrated an association between specific immune response to (11) examined childhood and adolescent consumption in Malay Epstein-Barr virus and NPC independent of race and geography sian Chinese; again their results are comparable to our findings. (31), suggesting an etiological role for the virus. However, it is Geser ef al. (14) studied weaning habits of babies in NPC versus unlikely that the virus is capable of inducing the disease by itself. control households in Hong Kong Chinese and observed that The dissimilarity in geographical and racial distribution between NPC households were 2.6 times more likely to feed Cantonese- Epstein-Barr virus infection and NPC incidence indicates that style salted fish to babies after weaning. Such an indirect mea some critical cofactor(s) must be present. Epstein-Barr virus is a sure of the actual consumption pattern of the case or control ubiquitous virus; there is serological evidence in Asia and Africa after weaning could explain the much lower magnitude of their to suggest that practically everyone in those regions is infected reported relative risk. very early in life (31). On the other hand, NPC has one of the The hypothesis that Cantonese-style salted fish intake causes most marked geographical and racial distribution of all cancers. NPC in Chinese is consistent with many aspects of the observed pattern of this disease in Chinese. Cantonese-style salted fish is ACKNOWLEDGMENTS a favorite food among southern Chinese especially those in We thank Malcolm C. Pike and Ronald K. Ross for their editorial suggestions, Guangdong Province, but the food is rarely eaten by northern Neeltje M. Mack for her technical assistance, Qing-Sheng Wang for editing the Chinese whose incidence rates of NPC are around 1 per 100,000 data, and Beth Woodin and Celia Cadillofor preparationof this manuscript. (2). NPC in southern Chinese displays an inverse relationship with social class (3), and Cantonese-style salted fish is tradition REFERENCES ally one of the cheapest foods available to supplement rice in 1. Waterhouse, J., Shanmugaratnam, K., Muir, C., and Powell, J. Cancer Inci south China. NPC is the most common cancer in southern dence in Five Continents, Vol. 4. IARC Scientific Publications No. 42, Lyon, France: InternationalAgency for Researchon Cancer, 1982. Chinese ages 15-34 years (20) suggesting that exposure to the 2. The Editorial Committee (eds.). Atlas of Cancer Mortality in the People's etiological agent(s) occurs early in life. Consumption of Can Republicof China. Shanghai:China Map Press, 1979. tonese-style salted fish early in life would also explain the high 3. Yu, M. C., Ho, J. H. C., Ross, R. K., and Henderson, B. E. Nasopharyngeal rates observed among migrants residing in low-risk areas such carcinoma in Chineseâ€”saltedfishor inhaledsmoke? Prev. Med., 10:15-24 1981. as North America (3, 7). 4. Armstrong, R. W., Kannan Kutty, M., and Dharmalingam,S. K. Incidence of Volatile nitrosamines, including dimethylnitrosamine and dieth- nasopharyngealcarcinoma in Malaysia,with special referenceto the state of Selangor. Br. J. Cancer, 30: 86-94,1974. ylnitrosamine, were detected in samples of Cantonese-style 5. Muir, C. S., and Shanmugaratnam,K. The incidenceof nasopharyngealcancer salted fish (16, 17). A/-Nitroso compounds are potent animal in Singapore.UICC Monogr. Ser., 1: 47-53,1967. 6. Buell, P. The effect of migration on the risk of nasopharyngealcancer among carcinogens; specifically, diethylnitrosamine has been shown to Chinese.Cancer Res., 34:1189-1191, 1974. induce carcinomas of the nasal cavity in rats (19, 22), gerbils 7. Gallagher,R. P.,and Elwood,J. M. CancermortalityamongChinese,Japanese (23, 24), hamsters (25-27), and bushbabies (28). More signifi and Indiansin British Columbia, 1964-73. Nati. Cancer Inst. Monogr., 53: 89- cantly, rats fed Cantonese-style salted fish have developed 94, 1979. 8. Zippin, C., Tekawa, I. S., Bragg, K. U., Watson, D. A., and Linden, G. Studies carcinomas of the nasal cavity (19). Thus, there is strong evi on heredity and environment in cancer of the nasopharynx. J. Nati. Cancer dence to suggest the presence of preformed nasopharyngeal Inst., 29: 483-490,1962. 9. Worth, R. M., and Valentine, R. Nasopharyngealcarcinoma in New South carcinogens or their precursors in Cantonese-style salted fish. Wales, Australia. UICCMonogr. Ser., 1: 73-76, 1967. In southern Chinese, NPC is twice as common among males 10. Henderson,B. E., and Louie, E. Discussionof risk factors for nasopharyngeal carcinoma. IARCSci. Pubi.,20: 251-260,1978. than females. Interestingly, we did not find any difference in 11. Armstrong, R. W., Armstrong, M. J., Yu, M. C., and Henderson,B. E. Salted frequency of consumption of salted fish between the two sexes fish and inhalants as risk factors for nasopharyngealcarcinoma in Malaysian and the association with salted fish holds equally strong for both Chinese.Cancer Res., 43: 2967-2970,1983. 12. Ho, J. H. C. Genetic and environmentalfactors in nasopharyngealcarcinoma. sexes. We did not investigate portion size, which may account In: W. Nakahara,K. Nishioka,T. Hirayama,and Y. Ito (eds.),Recent Advances for the male preponderance, or there may be constitutional in Human Tumor Virology and Immunology, pp. 275-295. Tokyo: Tokyo UniversityPress, 1971. differences between the two sexes in their susceptibility to NPC. 13. Ho, J. H. C. Nasopharyngealcarcinoma (NPC).Adv. Cancer Res., 75: 57-92, We estimate, based on population attributable risk computed 1972. from the data on frequency of consumption at age 10 years, that 14. Geser, A., Chamay, N., Day, N. E., Ho, H. C., and De-ThÃ©,G.Environmental factors in the etiology of nasopharyngealcarcinoma: report on a case-control over 90% of NPC occurring in young Hong Kong Chinese can study in Hong Kong. IARC Sci. Pubi.,20: 213-229,1978. be attributed to consumption of Cantonese-style salted fish 15. Ho, J. H. C., Huang, D. P., and Fong, Y. Y. Salted fish and nasopharyngeal during childhood. carcinoma in southern Chinese. Lancet, 2: 626,1978. 16. Fong Y. Y., and Walsh, E. 0. Carcinogenic nitrosamines in Cantonese salt- There are other factors that have been shown to be associated dried fish. Lancet 2:1032,1971. with an increased risk of NPC. Exposure to wood fires for 17. Huang, D. P., Ho, J. H. C., and Gough, T. A. Analysis for volatile nitrosamines in salt-preservedfoodstuffs traditionallyconsumedby southern Chinese.IARC cooking has been implicated in the development of NPC in Sci. Pubi.,20: 309-314,1978. Indonesia (29) and Malaysia (30). Our data suggest that it is not 18. Fong, L. Y. Y., Ho, J. H. C., and Huang, D. P. Preserved foods as possible a significant risk factor for NPC in Hong Kong Chinese. Exposure cancer hazards: WA rats fed salted fish have mutagenic urine. Int. J. Cancer, 23: 542-546,1979. to inhalants in an occupational setting has been shown to be 19. Huang, D. P., Ho, J. H. C., Saw, D., Teoh, T. B. Carcinomaof the nasal and related to NPC (3, 10, 11). Our study was unsuited for the paranasal regions in rats fed Cantonese salted marine fish. IARC Sci. Pubi., 20:315-328,1978. evaluation of the association between NPC and occupational 20. Ho, J. H. C. Some epidemiologieobservations on cancer in Hong Kong. Nati. exposure; these young cases had worked only 10-15 years. Cancer Inst. Monogr., 53: 35-47, 1979. Even though our young cases were more likely to be exposed 21. Breslow, N. E., and Day, N. E. Statistical Methods in Cancer Research, Vol. 1. IARC Scientific PublicationsNo. 32. Lyon, France: InternationalAgency for to smoke, dust, or fumes on their jobs, these relatively weak Researchon Cancer, 1980. associations did not persist after adjustment was made for 22. Ujinsky, W., and Taylor, H. W. Relative carcinogenic effectiveness of deriva-

CANCER RESEARCH VOL. 46 FEBRUARY 1986 960

lives of nitrosodiethylaminein rats. Cancer Res., 38:2391-2394,1978. simultaneouslytreated with benzo(a)pyreneor diethylnitrosamine.J. Toxicol. 23. Cardesa, A., Pour, P., Haas, H., Althoff, J., and Mohr, U. Histogenesis of Environ. Health,3: 379-394,1977. tumors from the nasalcavities induced by diethylnitrosamine.Cancer37:346- 28. Adamson,R. H., and Sieber,S. M. The use of nonhumanprimatesfor chemical 355, 1976. carcinogenesisstudies.In: F. Coulston(ed.),Ecotoxicologyand Environmental 24. Green, U., and Ketkar, M. The influence of diazepam and thiouracil upon the Quality, Vol. 2, pp. 275-296. New York: Academic Press, 1979. carcinogenic effect of diethylnitrosamine in gerbils. Z. Krebsforsch., 92: 55- 29. Djojopranoto, M., and Soesilowati. Nasopharyngealcancer in East Java (In- 62, 1978. donesia).UICCMonogr. Ser., 7: 43-46, 1967. 25. Herrold, K. Induction of olfactory neuroepithelialtumors in Syrian hamsters by 30. Armstrong, R. W., Kutty, M. K., and Armstrong, M. J. Self-specific environ- diethylnitrosamine.Cancer (Phila.),77:114-121,1964. ments associated with nasopharyngealcarcinoma in Selangor, Malaysia.Soc. 26. Mohr, U., Althoff, J., and Page, N. Tumors of the respiratory system induced Sci. Med., 72:149-156,1978. in the common European hamster by N-diethylnitrosamine.J. Nati. Cancer 31. De-ThÃ©,G.,Ho, J. H. C., and Muir, C. S. Nasopharyngealcarcinoma. In: A. S. Inst., 49: 595-597,1972. Evans (ed.), Viral Infections of Humans, Epidemiologyand Control, pp. 621- 27. Feron,V. J., and Kruysse, A. Effects of exposure to acroleinvapor in hamsters 652. New York: Plenum Medical Book Company, 1982.

CANCER RESEARCH VOL. 46 FEBRUARY 1986 961

Downloaded from cancerres.aacrjournals.org on October 2, 2021. © 1986 American Association for Cancer Research. Cantonese-style Salted Fish as a Cause of Nasopharyngeal Carcinoma: Report of a Case-Control Study in Hong Kong

Mimi C. Yu, John H. C. Ho, Shiu-Hung Lai, et al.

Cancer Res 1986;46:956-961.

Updated version Access the most recent version of this article at: http://cancerres.aacrjournals.org/content/46/2/956

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://cancerres.aacrjournals.org/content/46/2/956. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.