Tobacco Documents Online

I I 1 t I I I I I I I I DIET AS A CONFOUNDER OF THE ASSOCIATION BETWEEN AIR POLLUTION AND FEMALE LUNG CANCER: HONG KONG STUDIES ON EXPOSURES TO ENVIRONMENTAL TOBACCO SMOKE, INCENSE, AND COOKING FUMES AS EXAMPLES Linda C. Koo, J.H-C. Ho Cancer Research Laboratory Hong Kong Anticancer Society Nam Long Hospital, Hong Kong Introduction A comparative study of Chinese, Japanese, White, and Hawaiian women residing in various localities in the Pacific Basin indicated that the lung cancer incidence rate among nonsmoking Chinese in Hong Kong, Shanghai, or Hawaii ranged from 15.2 to 20.5/100,000, whereas that for women of other ethnic origins residing in the same localities only ranged from 7.3 to 7.5/100,000 (1). Some clues helping to explain this "mystery" of the unusually high incidence of lung cancer among nonsmoking Chinese females in such warm climate communities as Hong Kong, Singapore, or Hawaii (2) can be obtained from epidemiological statistics and studies. From the 1992 edition of Cancer Incidence in Five Continents (3) it can be noted that Hong Kong's world standardized female lung cancer incidence rate of 32.6/100,000 is among the highest in the world. This high incidence is a phenomenon of recent decades since mortality data from the Hong Kong Cancer Registry indicates that female lung cancer deaths increased from 7.7/100,000 in 1961 to 23.3/100,000 in 1990 (4). However, the lung cancer mortality rate of Chinese females from Guangdong province in China, the origins of most Hong Kong Chinese, was only 2.9/100,000 for the period 1973-75 (5). A further factor that has been identified from epidemiological studies in Hong Kong (6) and Singapore (7) is that within the Chinese ethnic/dialect groups, Cantonese females are about two to three times more likely to have lung cancer as those of Chiu Chow origins, even when they live in the same city. ~ These statistics indicate that Chinese women, especially those of Cantonese origin, share a common environmental exposure which they retain when they migrate overseas. The fact that , nonsmoking overseas Chinese women still have higher lung cancer rates than other ethnic groups living in the same area, would seem to rule out the influence of outdoor ambient air pollution as an important etiological explanation. In terms of indoor home sources of air pollution, the contribution of cooking and ' heating equipment would also seem to be minor since these factors are not unique to Chinese homes. Moreover, unlike the female lung cancer studies in cold climate areas in China such as Yunnan (8) and Manchuria (9), where the use of smoky coal stoves for heating can produce such concentrated fumes that , one can barely see one's outstretched hand when standing in such rooms,, the problems of Chinese women in Hong Kong and overseas communities is not one of air pollution from coal space heaters. The trends in mortality data from Hong Kong indicating a tripling increase over 30 years also suggest that the N I factor(s) are due to a recently introduced exposure(s) or changes in traditional exposures which are related o to contemporary Chinese cultural habits. The question we'd like to pose is: Is it mostly due to inhaled ~ or ingested substances? ~ ta O 00 I

Methods Over the last 14 years, we have been doing a series of epidemiological studies to narrow down possible factors affecting lung cancer in nonsmoking women in Hong Kong: *1981-3 retrospective case-control study of 200 female lung cancer patients and 200 district matched controls (10-12). *1985 cross-sectional study measuring personal exposures to nitrogen dioxide and prevalence of respiratory symptoms among 362 primary school children and their 319 nonsmoking mothers (13, 14). * 1988 measured the concentration of airborne carcinogens in the kitchen and living room of 33 working-class homes by 24 hour sampling in each room and analyzing the effects of emission and ventilation sources on the concentration of 7 polycyclic aromatic hydrocarbons (PAH) including benzo(a)pyrene (15). * 1993-4 cross-sectional telephone survey of 500 women on their dietary habits, exposure to environmental tobacco smoke, and prevalence of respiratory symptoms. Although the design and specific objective of each of the four Hong Kong epidemiological surveys was different,, we collected data on ETS in all four studies, and in the early three we gathered data on incense burning and exposure to cooking fumes. Data on dietary habits was only gathered in the 1981-3 case-control study and the 1993-4 telephone survey. Further details on the methodology in each study are provided in the references. Results Environmental Tobacco Smoke: Exnosure levels: Estimates of Hong Kong women's exposure to ETS varies with how it is defined. In terms of the person/place/time paradigm in epidemiology, it can be defined as simply the smoking habits of the current and/or ex-husband(s) or all family members, whether the home and/or workplace is considered, if current and/or past exposures are counted, and if only cigarettes or all forms of tobacco smoke are included (11). These complications in definition were addressed in our 1981-3 case-control study where we identified four possible exposure categories if we looked at the place of exposure (i.e- home, work, home+work, and none) (16), and another four categories if we looked at time of exposure (i.e. childhood, adulthood, both, none) (10). Generally, however, most researchers are defming it as the presence of a smoking spouse at home. Using this common definition, 60% (53 out of 88) of the nonsmoking female lung cancer patients and 49% (67 out of 137) of the district controls were exposed (10). In contrast to this older population of women with a mean age of 59, the 1985 study of 362 primary school children and their mothers (mean age 38) estimated that 36 % of the households had a smoking father/husband (13). -2- I I I I I I I I , I I I I 1 I

I I I I I I I I I I I I I 1 These data indicate that as smoking patterns vary by age group and time, one must be cautious in assuming that levels of exposure measured at one point in time of a subject's lifespan are representative of her lifetime dose. When smoking occurs at home, the amount of exposure and the overall effects of exposure may differ. In our 1985 study of personal levels of exposure to nitrogen dioxide (NO2) among primary school children and their mothers, increased smoking by the father/husband resulted in slightly increased N02 among the children but decreased NOz among their mothers (14). In another detailed analysis of life history correlates of ETS among the nonsmoking controls from the case-control study, it was found that the lifetime hours or years of exposure among wives with husbands who were heavy smokers, i.e. > 20 cigarettes/day, were not significantly higher than those with husbands who smoked less, i.e. 1-20 cigarettes/day (11). Yet this is a common assumption in studies on ETS and health effects, that there is a positive correlation between the number of cigarettes smoked by the husband and the amount of cumulative lifetime ETS exposure by the wife or other household members. The actual amount of ETS pollution that a wife with a smoking husband may inhale is also an issue that needs further investigation. From our NOZ personal monitoring study in 1985, where increased smoking at home was weakly associated with reduced levels of NOZ among the 319 mothers, we suggested that this was because when smoking occurred at home, the mothers would increase ventilation, thus resulting in reduced overall pollution from the cigarettes as well as other sources (14). This pattern was supported by our 1988 24-hour site monitoring study measuring the concentration of 7 polycyclic aromatic hydrocarbons (PAH) in the airborne particulates in 33 Hong Kong homes (15). As shown in Figure 1, increased cigarette smoking in the living room resulted in reduced concentrations of PAH in airborne dust. Raure 1 Effects of Clgarette Smoke on LMna Room Concentrations of PM None Inw Hph N- 19 6 e 0 14 b11 M~~py~ 0 S16 ibIDO MwvMwiw.Uwmf4V/~+•b.cfyqP.NR~.bwm0* • ~/tMWb~~aa.eara+~dMmm~wmwbrxo...dm.eama.~1.ww. ... rqr.eaa.aa a,om~.wmwiaw.wwd...eanea~erwa.nca eo:.e.: rcc wm is'~'iqO I I ' N O Co , s -4 00 , -3- i o I

With normal levels of smoking in homes, and given the subtropical climate, these data suggest that the Hong Kong inhabitants' response to smoking is to increase ventilation by opening windows and doors. The result is that the concentration of all indoor air pollutants, even those from other emission sources like cooking fires and incense, are diluted. This effect is especially apparent at low levels of smoking in the living room (1-2 cigarettes or 3-15 minutes of cigarette smoke per day). However, at higher levels of smoking, the effects of the extra ventilation activities as a response to perceived ETS can be overwhelmed so that cigarette smoke can be a major source of indoor air pollution where there is continuous or high levels of smoking, as in commercial environments like bars or discos. The behavioral response to ETS exposure, leading to overall dilution of air pollution in homes may also be more apparent in communities like Hong Kong where highly dense living conditions mean that people cannot move to another room to avoid cigarette smoke, so they do something about it. Other exposure study results also support the finding that the Hong Kong Chinese woman's contact with ETS is not unusually high when compared with women in other countries. In a 10-country, 13-site comparative study of urinary cotinine/creatinine ratios (cotinine is a metabolite of nicotine) the levels among Hong Kong Chinese women ranked 4th whereas those for Shanghai women were the lowest of all, ranking 13th (17). In fact Hong Kong's ratios would have been considerably reduced if the absolute cotinine levels had been compared instead of being adjusted by creatinine. From studies in Western populations, the usual range of creatinine excretion is 1.0 to 2.5 g/day (18) but among our Hong Kong subjects, 75 % had levels less than 1.0 g/day and the mean level among our 101 subjects was 0.77 g/day (SD=0.48). Our very low creatinine levels meant that cotinine adjustment by it would result in very high cotinine to creatinine ratios. We have no explanation for our abnormally low urinary creatinine levels, although it is known that it is increased by meat consumption (19). Lung cancer risk: As discussed above, assessing lifetime exposure to ETS is complicated because of the need to account for changing environments and lifestyles throughout the lifecycle of a subject. Defining exposure doses is also intricate because of different types of smoking methods, smokers, degree of ventilation, etc. This is especially pertinent in case-control studies on lung cancer, since the patients and controls are older subjects. In our previous analyses on the risk of ETS to lung cancer, we used such summary measurements of lifetime dose as total years, total hours, hours/day, and cigarettes/day (i.e. sum of the number of cigarettes/day smoked by each household member weighted by the years of exposure from that smoker). We also investigated whether a combination of such variables as hours/day plus years of exposure, or age at first exposure plus years of exposure led to dose-response relationships. In all these analyses, we could not establish a statistically significant association between lifetime ETS exposure and lung cancer risk among women who had never smoked (10). Stratification of the tumor data by histological type, lobar location of the primary tumor, and whether the tumor was proximally or peripherally situated suggested a weak possibility that peripheral tumors of the squamous or small cell type that were located in the middle or lower lobes might be affected by ETS. However, only 8 of our 88 nonsmoking cases had this combination of tumor characteristics, this number is too small to do statistical analyses (10). -4- , ' I I I I I I I I 1 I I I I I I I I

I I I I I I I I I I I I I I I I Overall, the possibility that ETS exposure explains the unusually high lung cancer incidence among nonsmoking Chinese women in Hong Kong is unlikely. Hong Kong Chinese women do not have extremely high ETS exposures compared to women living in other regions in the world (17). Our studies in Hong Kong have also shown that female lung cancer patients who have never smoked were significantly younger, by 7 years, than patients who had ever-smoked (20). In fact, all of our female patients under 40 years of age had never smoked (2). Since the concentration of noxious gases and particulates in mainstream smoke is 100 to 100,000 times higher than the inhaled concentration of such agents in room-diluted sidestream smoke, depending on the particular agent being measured (18), and the active smoker is also a passive smoker by inhaling the sidestream smoke of her own and others' cigarettes, it would be more logical that among lung cancer patients, passive smokers would be older than active smokers. In fact, the very young age of nonsmoking female lung cancer patients in Hong Kong suggests that the suspected agents are either introduced very early in life, have a stronger carcinogenic effect than active smoking, and/or result in lower exposure to protective agents. Diet: From the controls used in the 1981-3 case-control study, we identified 136 women who had ever-married and never-smoked to see whether there were other lifestyle variables being correlated with having a smoking husband. After analysis of 97 quantifiable variables that ranged from other exposures at home, personal habits and recreational activities, cooking and heating fuels, health histories and consumption of medications,, etc., dietary habits were the most significant variables correlated with ETS exposures. Wives with nonsmoking husbands consumed more cruciferous vegetables, carrots, beans/legumes, fermented bean products, milk, and home-cooked soup than wives with smoking husbands. On the other hand, wives with smoking husbands were significantly more likely to consume pickled vegetables, chili, and alcohol than wives with nonsmoking husbands, with their greater consumption of salted fish of borderline significance (p=0.09) (11). According to the U.S. National Research Council's 1982 report on Diet. Nutrition, and Cancer the former foods are generally protective of cancer and the latter potential inducers of cancer (21). More recently, from 1993-4, we conducted a telephone survey of women to study the relationship of dietary preferences with ETS exposures. From among 500 interviewed subjects, 232 were currently married, of which 67 had a currently smoking husband and 165 did not. This cohort of women, with a mean age of 44, is younger than the 1981-3 control subjects mentioned above. Each subject was asked how often she consumed a list of 9 food items and whether she liked or disliked eating a list of 14 food items on a 5-point scale. Wives with smoking husbands significantly preferred eating meat, chili, salted fish, cured meats, seafood sauces, and alcohol more than wives with nonsmoking husbands. On the other hand, the latter group significantly preferred eating soybean curd and milk, and were more frequent consumers of fresh fruit. From these two studies we can see some similar dietary habits correlated with ETS exposure although their ages are different and the studies were done about 10 years apart. Nonsmoking wives with smoking husbands tended to eat poorer diets by consuming more salted fish, cured meats, and alcohol, and were lower consumers of fresh fruit and milk. In a previous report on diet and lung cancer risk -5- I

I among nonsmokers in Hong Kong, cured meat and alcohol were associated with increased risk and milk and fresh fruit with decreased risk (12). Thus the pattern of dietary habits associated with ETS exposures in Hong Kong coincides with their independent risk for lung cancer in nonsmoking women. In an international comparative study of dietary correlates of ETS in Hong Kong, Sweden, United States, and Japan, there was a generally consistent pattern that among all these countries nonsmoking wives with smoking husbands were less likely to eat fresh fruit and vegetables than wives with nonsmoking husbands (22). Another study among women in Hawaii found intakes of betacarotene and cholesterol to be inversely associated with ETS exposures (23). Thus diet can be a significant confounder of the association between ETS and various diseases in a variety of industrialized urban cross-cultural settings. Incense: Exposure levels: From our 1985 survey measuring personal exposures to NOZ by the use of monitor badges, incense was identified as the most important emission source in the time that increased NO2 exposures among the mothers (14). From our 1988 study measuring the concentration of 7 PAH compounds in the airborne particulates in Hong Kong homes, incense was also found to be the major contributor to PAHs, like benzo(a)pyrene, in the living room, as shown in Figure 2 (15). These studies suggest that incense is a major source of gaseous and particulate air pollution in Hong Kong homes. Rpure 2 Effects of Incerse Smoke an uvYip Room Concentratbns of PAH M o 1-2 3.24 ,~bumed..3tlcft + Mt.~.4ara.aaw.~~Y~amimo.lNfo.~wYw. sn.csrnnria.iaw On any given day, about half the homes in Hong Kong will burn incense (24). Estimates on population exposure levels to incense increase with the age of the woman. From our 1985 survey, 48 % of the mothers (mean age 38) burned incense at home (14), whereas among the nonsmoking controls of the 1981-3 case-control study, 77% of this older cohort with mean age of 59 said that they burned it. -6- , I I I I I I I I I r I I I I I

I I I I I I I I I I I I I I Lune cancer risk: The high exposures to incense among Hong Kong women, is likely to be a major contributor to airborne carcinogens in the home. Figure 3 shows the lung cancer relative risk (RR) estimates from the 1981-3 case-control data. The subjects included 189 cases and 197 controls who had ever been married. Increasing years of incense exposure did not affect lung cancer RR in nonsmokers. However in smokers, increasing incense use resulted in significantly decreased RR. At the highest level of use, i.e. 40-70 years, the lung cancer risk among smokers was only 17% of those who did not burn incense at all. Although the data shown are adjusted by demographic variables and smoking, similar results were obtained in the unadjusted RR. Rgure 3 Incense Smoke and Wna Cancer Risk ,~_ • pw.e a nP. r,.. nw., arqna,aec~.y •• M~ u ~o.. qs LIYr.IMaeco, 9a,o•:IAIdOSCwMdBFWy These findings were unexpected. However, it is interesting to note that a case-control study of female lung cancer in Taiwan also found that incense burning was associated with significantly reduced risk for lung cancer (25). In the Taiwan study on female adenocarcinoma cases, the RR for burners of incense at the highest level, i.e. 14+ times per week, was about a fourth of that among the nonburners. Diet: Since incense burning in Hong Kong is done for the purpose of communicating with gods or ancestors, it is representative of a traditional lifestyle that may be correlated with other traditional behaviors. As diet was found to be an important factor affecting lung cancer risk in Hong Kong women (12) , we studied the relationship of diet with incense burning. When the ever-smoked female lung cancer patients were stratified by their incense burning habits (+/-), incense burners were found to consume more fresh fish, retinol, and dim sum (Cantonese pastries), and less alcohol and chili than those who did not burn incense. In our previous study of diet and lung cancer risk, the former items were associated with reduced risk and the latter with increased risk for lung cancer in nonsmokers (12). Thus the significantly reduced risk for lung cancer among smoking female lung cancer patients who burned incense may have been due to its correlation with traditional food items that are protective of cancer. -7- I

Cookine Fumes: Exoosure levels: The overwhelming majority of Chinese women have been exposed to cooking fames for some part of their adult life. From the 1985 study, 19 out of 319 mothers did not cook (6%), and among the 400 cases and controls in the 1981-3 study, 22 did not cook (6%). This leads us to discuss the types and sources of pollutants in cooking fumes (from the fire and/or from the cooked food), and whether significant amounts of fumes are inhaled by Chinese women. From our 1985 study on nitrogen dioxide exposures, such indices of cooking exposure such as the number of meals cooked at home and the frequency of frying had no effect on the children's or mothers' NOZ levels. On the other hand, the type of fuel used for cooking significantly affected the mother's NOZ levels. Mothers whose cooking stoves were fueled by liquid petroleum gas had significantly higher levels, and those who had piped gas had significantly lower NO2 levels (14). From our 1988 fixed site monitoring study of kitchen levels of PAH, we found that the total duration of time that cooking fires were lit (none of the homes had electric stoves), was associated with significant increases in benzo(a)pyrene and all 7 PAH compounds as shown in Figure 4. However, when we investigated the effects of different cooking methods on airborne PAH levels, frying and stir-frying generally led to reduced concentrations of PAH in dust. This is probably because women, sensing such oily fumes being generated, tried to disperse them by increasing ventilation (15). Electrically powered ventilation fans installed on kitchen windows were found to significantly reduce air pollution levels from both the personal NOZ measurements in 1985 and the fixed site monitoring in 1988. Figure 4 Tha Fffarte nf Stnva Firn< nn ICitrhrn PAH lavalc PAH ng/m dust t j ril.eBAP ' 1e.1 Y0. oY e.e ~ 15-30 31-60 61-90 91-360 Minutes Stove fire • M7Wt~i bf W41. wBf bfpr ql. er. aMnN qhTfn ~IMan. uf. nI Hnlw~ rmHlrtlne Hn. IaL.niity ef iMMi~ Wrn1nO e rna u1 Mr~~d kncfa eoerf. Scurc. x Ceo .t .1. n/N The results of the two sets of data, one monitoring personal exposures to air pollutants and the other a fixed site monitoring study, show the importance of doing both types of measurements to -8- t I I I I I I I I I I I I I I I

I I I I I I I I I I I I I I I I I I understand the dynamics of air pollution exposures in humans. People do not stay in one room, nor do they passively breathe in air pollution from emission sources over which they may have some control. When there is a noticeable source of air pollution, be it stir-frying or cigarette smoke, windows and doors will likely be opened to dispel such pollutants. When a passing vehicle spews out black soot on to the breathing zone of a pedestrian, she will frequently cover her nose and stop breathing. Thus the measurements of air pollution from fixed site monitoring studies may bear little relationship to the exposures of free ranging human beings. From our 1988 study of 33 female household heads, their mean duration of staying in the kitchen was 1.4 hours per day (15). This relatively short period of time may explain why cooking activities were not found to significantly affect personal NOZ levels among the 319 mothers in the 1985 study. It is normally assumed that there is a linear relationship between emission levels and personal inhalation of such pollutants. Our data from Hong Kong seem to indicate that this association can vary depending on people's reactions to the emission source. Unlike workplace or commercial indoor environments, inhabitants at home have more control over their air circulation and ventilation patterns. When an inhabitant perceives air pollution coming from a noticeable source, she can increase ventilation so that not only is the pollution from the perceived source reduced, but the accumulated pollution from other sources is also significantly reduced. This was found in our 1988 PAH study for such emission sources as gas powered water heaters, cigarette smoke, and stir-fry cooking whereby increasing emissions from these sources were associated with significantly reduced airborne PAH levels (15). By contrast, when boiling, steaming, or stewing activities took place, there was a dose-response increase in kitchen PAH levels. These variations in PAH levels by specific cooking activities are examples of the results of the importance of human perception. Unlike the large amount of oil and fumes generated from stir-frying; the latter cooking activities do not produce such noticeable air pollutants. Consequently women are less likely to actively increase ventilation in the kitchen. On the other hand some concentrated emission sources may be less affected by ventilation, as was found for incense, or be too overwhelming to be diluted by increased ventilation, as was found for the total time that the stove fire was lit (15). Lung cancer risk: In a case-control study of female lung cancer in Shanghai, China, Gao et al.(26) found that lung cancer RR increased when more meals were cooked by stir-frying, deep frying, and boiling. Shanghainese women who reported eye irritation and house smokiness when cooking also had increased RR. Although the authors postulated that exposure to oily cooking vapors may be etiologically related to lung cancer, it is interesting to note from their RR estimates that the most significant fmdings among the cooking methods was that from boiling, which does not produce oil vapors. Moreover, their study asked about current cooking habits, and the relevance of such current exposures to lung cancer etiology, which has a latency period of several decades, may be questionable. The possibility of reporting bias among cases is also a problem in all studies on respiratory diseases where exposure assessments are done by questionnaires. This is because of lay perceptions that respiratory diseases are caused/affected by inhalation of air pollutants. Thus cases may be more likely -9- I

I to report exposure to air pollution, or be more sensitive to its occurrence than controls. This may be especially true when cases are asked about recent exposures to air pollutants. In contrast to the Shanghai study on current exposures to cooking, our 1981-3 study asked about duration of cooking in years. Our results indicated that increasing years of cooking significantly reduced lung cancer risk among nonsmoking women (Figure 5) (27). Women who cooked for more than 25 years had their risk decreased by more than 60%, and the trend was highly significant (p<0.001). Hpure 5 tq. aeeaun a,e Y.pcalewaYt ane,p n.v.e.niaa.a Nmp I(enpwhm.n r +.n 1 o-w nau 41+ 1fBmf af CodOq ~ M~b~m~.~.hMn.vtlclnoFYNF) 9m.a~I~iu1SN Diet: ~ hend u~l .1. The relationship between duration of cooking and dietary patterns was more complicated. A comparison of those who cooked for shorter vs. longer durations among 88 nonsmoking cases indicated no significant differences in frequency of consumption of 17 food items when adjusted by age and years of education. However, among the controls, women who cooked more than 25 years were less frequent consumers of fresh fish (p=0.005) and foods containing retinol (0=0.006) and calcium (p=0.01), but were more frequent drinkers of alcohol (p = 0.04), than those who cooked for < 25 years. Again, these dietary patterns are consistent with lung cancer risk factors in Hong Kong(12). Discussion From epidemiological studies on risk factors for lung cancer among Chinese women, at least two different patterns are emerging. Among Chinese women living in colder climates in China, e.g. Manchuria and northern Ytnman, the influence of heavy doses of indoor air pollution from smoky coal fires is apparent (8,9), although diet cannot be excluded as an additional factor in those populations. For southern Chinese who are not exposed to air pollutants from space heating, and for overseas Chinese in developed urban communities, outdoor air pollution and fumes from heating and cooking appliances are -10- I I I I I I I I I I I I I I