Cancer incidence in Asian migrants to New South Wales, Australia.

Cancer incidence during 1972-90 in Asian migrants to New South Wales, Australia, is described. Overall cancer incidence was lower than in the Australia born in most migrant groups, and this reached significance in migrants born in China/Taiwan, the Philippines, Vietnam and India/Sri Lanka, and in male migrants born in Indonesia. For the majority of cancers, rates were more similar to those in the Australia born than to those in the countries of birth. For cancers of the breast, colorectum and prostate, rates were relatively low in the countries of birth, but migrants generally exhibited rates nearer those of the Australia born. For cancers of the liver and cervix and, in India/Sri Lanka-born migrants, of the oral cavity, incidence was relatively high in the countries of birth but tended to be lower, nearer Australia-born rates, in the migrants. For these cancers, environmental factors related to the migrant's adopted country, and migrant selection, appeared to have a major effect on the risk of cancer. For certain other cancers, incidence was more similar to that in the countries of birth. Nasopharyngeal cancer, and lung cancer in females, had high rates in both the countries of birth and in migrants to Australia. Nasopharyngeal cancer rates were highest in China/Taiwan and Hong Kong-born migrants, and were also significantly high in migrants from Malaysia/Singapore, Vietnam and the Philippines. Rates of lung cancer were significantly high in women born in China/Taiwan, and the excess was greater for adenocarcinoma than for squamous cell carcinoma. Melanoma had low rates in both the migrants and in the countries of birth. For these cancers, it was probable that genetic factors, or environmental factors acting prior to migration, were important in causation.

Many studies of cancer risk in migrants have been performed in Australia. one-fifth of whose population has been born elsewhere (Castles. 1989). Although there have been some people of Asian origin in Australia since the gold rushes of the 1850s. until the relaxation of the White Australia Polics in the late 1960s migrants to Australia were predominantly of European origin. In recent years. however. Asians have accounted for over 400/0 of Australia's immigrant intake (Borowski and Shu. 1992). and bv 1986 the Asia born constituted 17% of the immigrant population in Australia (Castles. 1989).
Previous studies of cancer risk in migrants to Australia have not included Asian migrants (e.g. McMichael and Giles. 1988;McMichael et al.. 1989) or have examined rates in persons of Asian origin by region rather than by individual country of birth (Armstrong et al.. 1983: McCredie et al.. 1990). In New South Wales (NSW) cancer mortality has been reported in the China Taiwan-born. but covered too few deaths to examine cancer sites in detail (Zhang et al.. 1984). and cancer incidence has been described in the China Taiwan born by individual site ). The accumulated data on cancer incidence bv country of birth (COB) in NSW (McCredie et al.. 1993) allowed a description of cancer incidence in Asians in Australia by individual COB for the first time.
Populations and methods NSW is the most populous state in Australia. with 5 898 731 residents at the 1991 census. The NSW Central Cancer Registry receives statutory notifications from hospitals and radiotherapy departments. as well as pathology reports and death certificates. for all cases of invasive cancer which occur in NSW (McCredie et al.. 1991) September 1994 recorded routinelv. The countries examined in this report comprised all those Asian countries, excluding those in the Middle East. for which at least 100 cases of cancer occurred during 1972 -90. Malaysia and Singapore. and China and Taiwan. have not always been coded separately by the Registrv. and so cannot be examined individualls. India and Sri Lanka were grouped together after an examination of the data revealed no major differences in disease patterns. For incident cases diagnosed between 1972 and 1990 and notified to the Registry. data were tabulated according to 5 year age group (0-4. 5-9. 75-9. 80 and over). sex. COB and cancer site (ICD 9). All cases originally coded to ICD 8 were bridge coded to ICD 9 codes (Coates and ). For lung cancer. cases were also tabulated by morphology (SNOMED;Cote. 1982). COB was unknown for 2.7% of all cases, excluding those with melanoma, and for 24.6% of cases with melanoma. The effect of duration of residence could not be analysed as the data were incomplete. Year of cancer incidence was not a good proxy for duration of residence. as migration continued from Asia throughout the period of the study. Data on Asian ethnicits are not collected bv the registry.
Populations by age. sex and COB were derived from data supplied by the Australian Bureau of Statistics (ABS) for the 1971. 1976. 1981censuses (ABS. unpublished tables: ABS. 1993: McCredie et al.. 1993. Indirectly agestandardised incidence ratio percentages (SIRs) were calculated using rates in Australian-born residents of NSW as standard. Confidence intervals (CIs) were calculated assuming that the observed cases followed a binomial distribution. The level of significance was set at 0.01 because of the many comparisons that were made. Average annual cancer incidence rates. directly standardised to the 'world' population (Doll. 1976). were calculated for each COB to allow comparisons with published incidence rates in the countries of ongin (Sarjadi. 1990: Parkin et al.. 1992: Pham et al.. 1993.
Unpublished data on religion and occupational status of the immigrant groups were obtained through the Bureau for Immigration Research. Most of these data related to the 1991 census as earlier inforrnation was not available.

Results
Sociodemographic background of the immigrant groups Over 130 000 persons born in the Asian countries considered here were resident in NSW at the 1986 census. The age and sex distributions varied markedly by COB (Table I) (Tables II and III). In none of the COB groupings was overall cancer incidence higher than in the Australia born. The incidence was significantly low in all groups except those born in Hong Kong and Malaysia Singapore. and in women born in Indonesia.
Rates of oral cancer were low in males from most COB. significantlv so for those born in China Taiwan. SIRs for nasophary ngeal cancer were markedly raised in most Asian migrants. approaching 5000 in the Hong Kong born. Rates of other pharyngeal cancers were not significantly different from those in the Australia born (not shown). Oesophageal cancer incidence was significantly raised in Hong Kong-born males, as was stomach cancer incidence in the China Taiwan born of both sexes. SIRs for cancers of the colon and rectum were similar and are presented here combined. Colorectal cancer incidence was significantlv low in the China'Taiwan and India Sri Lanka born. in males born in Vietnam. and in females born in the Phillipines. Liver cancer incidence was raised in most immigrant groups. and was greatest in the Vietnam born. in whom the SIR was over 2600 in males and 1000 in females.
Rates of cancer of the larynx were low in most Asians. but did not reach significance compared with the Australia born. Lung cancer incidence rates were low in all male immigrant groups. but this was significant only in those born in Malaysia Singapore and India Sri Lanka. In females, there was a different pattern. with significantly high rates in the China Taiwan born, and non-significantly high rates in those born in Hong Kong. Malaysia/Singapore, Vietnam and India Sri Lanka. This was predominantly due to raised rates of adenocarcinoma of the lung. Incidence of adenocarcinoma was significantly high in China Taiwan-born females (SIR = 245. 99% CI 139-398). but not in males (SIR = 124. 99% . In China Taiwan-born females it was the most common form of lung cancer. Rates of squamous cell carcinoma were non-significantly raised in China Taiwanborn women .
Breast cancer rates were significantly low in females born in Vietnam and in China Taiwan. Cervical cancer SIRs were significantly high in the Vietnam born and were significantly low in the India born. Incidence of cancers of the prostate, testis. and bladder was low in most groups, reaching significance in only a few instances.
Melanoma incidence was significantly lower in all migrant groups than in the Australia born, with SIRs ranging from 0 to 31. Thyroid cancer incidence was raised in most migrant groups. and was highest in those born in the Phillipines, although this reached significance only in females. Rates   countries of birth were rates for any other cancer sites significantly different to those of the Australia born.

Discussion
Exploring cancer patterns in Australia's Asian migrants is of importance not only' in the planning of health care for these communities, but also for the study of possible aetiological factors. We have identified patterns in cancer incidence which are similar to those of Chinese migrant populations in the US (King and Locke. 1980) and Singapore (Lee et al.. 1988). and have also examined cancer incidence in migrants from other Asian nations for which there are few incidence data. Although were were unable to analyse cancer rates by duration since migration. some inferences could be made regarding trends by comparing rates between the countries of origin, the migrants in NSW and the Australia born in NSW (Tables IV and V). In general, cancers with rates in migrants intermediate between Australia's and those of the COB are most likely to be related to environment, whereas cancers with rates in migrants similar to those of the COB are more likely to be related to hereditary factors or early environment. However, in interpreting the current data, it is necessary to consider other influences on cancer rates in the migrants, and in their countries of birth.
Migrants may be unrepresentative of the population of their country of birth. For example, migrants from Vietnam were predominantly refugees (Borowski and Shut 1992), their rates of unemployment in Australia were high (Bureau of Immigration Research, 1991) and their occupational status was low. In contrast, occupational status was high in migrants from Hong Kong, India,Sri Lanka and Malaysia, 'Singapore. Migrants may come predominantly from areas which have rates of cancer different from national rates. In China. cancer rates vary markedly by region (Chen et al.. 1990). Most Australian China-born migrants originate from Guangdong (Zhang et al.. 1984). but no current population-based data on cancer incidence are available from this province (Parkin et al.. 1992). The ethnic background of certain migrant groups differed from that of their country of birth.    (Moore. 1988: Pinnawala. 1988).
Duration of residence has also varied. In 1986. the longest mean penod of residence was in Indian and Sri Lankan migrants (15.5 and 12.3 years). and the shortest was in Filipino and Vietnamese migrants (5.0 and 5.6 years) (Castles. 1989).
In comparing cancer rates in the migrants with those in their countries of birth. the accuracy of the latter must also be considered. Most country of birth rates were extracted from Cancer Incidence in Five Continents. Published indices of data quality were generally highest for the cancer registries of New South Wales and Singapore. and were somewhat lower for the other registries (Parkin et al.. 1992). Rates for Indonesia and Vietnam were extracted from other published sources (Sarjadi. 1990: Pham et al.. 1993. and the quality of these data is uncertain. Cancers wtith rates more similar to ,4ustralia born than country of birth rates Although oral cancer is one of the most common cancers in India. chiefly related to chewing tobacco (WHO. 1984), rates were not raised in migrants from India Sri Lanka. It was unclear whether the decreased rates were because these migrants, who are of high socioeconomic status (SES) and predominantly of mixed Anglo-Indian ethnicity. had never chewed tobacco, or whether they stopped chewing tobacco on coming to Australia. Mortality rates from oral cancer in Indian migrants to England and Wales are increased above rates of those born in England but these migrants are of lower SES than Indian migrants to Australia (Berra and Swerdlow, in preparation).
Stomach cancer is the most common cancer in most of East Asia . However, in NSW it was significantly increased only in the China Taiwan born. and in none of the migrant groups was it the most common.
Rates of colorectal cancer were higher in migrants than in their country of origin, except in women born in the Philippines. However, in most migrants SIRs were significantly low in one or both sexes. The exceptions were the Hong Kong, Malaysia Singapore. and Indonesia born. Colorectal cancer risk in the Chinese in Singapore, China and the US has been associated with an increased food energy intake from fat (Whittemore et al.. 1990). and an increased meat vegetable consumption ratio (Lee et al.. 1989), and rates have been found to increase rapidly with transition to the American diet in Chinese migrants to the US (Yu et al., 1991). Rates in Singapore were lower in the ethnic Chinese born in China than in those born in Singapore (Lee et al.. 1988). In Singapore dunrng 1983-87 incidence rates of colorectal cancer in Indians were lower than rates in the Chinese (Parkin et al.. 1992), reflecting the pattern seen in this study. A possible explanation is that Indians may be more likely to be vegetarian. At the 1991 Australian census. 18.8% of the India born, and 30.8% of the Sri Lanka born, compared with 3.6% of the Hong Kong born, classified themselves as Buddhist or Hindu (Bureau for Immigration Research, unpublished data), religions which encourage avoidance of meat.
Liver cancer tended to be much less common than in the countries of origin, except in the Vietnam and Indonesia born. but rates were generally still above those of the Australia born. The raised rates were consistent with the distribution of hepatitis B. the principal cause of this cancer in Asia (Anthony. 1984). Hepatitis B infection in Asians is usually acquired vertically or during early childhood (Anthony. 1984). If early infection were the sole risk factor, one would expect rates in migrants to be similar to country of birth rates. However. Tables IV and V show that. in those migrant groups with high country of origin rates. incidence rates tended to be lower in Australia. This is consistent with the action of co-factors. acting later in life. in the aetiology of liver cancer. Research in China suggests that one such factor may be aflatoxin ingestion (Yeh et al.. 1989: Ross et al.. 1992). In males. low lung cancer rates in those born in Malaysia Singapore and India Sri Lanka were surprising given the almost equivalent rates of smoking in Asiaand Australiaborn males found by the National Health Survey (Castles. 1992). Possible explanations for this pattern include differential smoking patterns among migrants from within the Asian region. differences in duration of smoking from the Australia-born population. and the high SES of these immigrant groups. No information on duration of smoking by country of birth was available.
In the immigrant groups of high SES. breast cancer rates were similar to those in the Australia born. In China Taiwanand Vietnam-born women rates were low. but higher than in their countries of birth. In Singapore. rates of breast cancer were lower in ethnic Chinese women born in China than in those born in Singapore (Lee et al.. 1988). Breast cancer is more common in women of higher SES (Petrakis et al.. 1982) and in Chinese of high educational status in Singapore (Lee et al.. 1993). Early age at first full-term pregnancy has been found to be protective against breast cancer in both Caucasians (Petrakis et al.. 1982) and Chinese (Lee et al.. 1993). The highest fertility rates in Australian immigrant women were in Filipino (3.2) and in China Taiwan-(2.5) and Vietnam-born (2.2) women (Castles. 1989). the three groups with the lowest rates of breast cancer. In addition, women born in Hong Kong. Malaysia Singapore and India Snr Lanka. who were more likely to delay child bearing until after the age of 25 (Castles. 1989). had higher rates of breast cancer. That obesity was more common in Australia-born than in Asia-born women (Castles. 1992) may also be related to the variations in breast cancer incidence.
Low SES is strongly associated with risk of cervical cancer (Christopherson and Nealon. 1981) and appeared to be a predictor of risk in this study. Rates were low in the high-SES India Sri Lanka born. despite high country-of origin rates. Conversely, rates in Vietnam-born women, who were of low SES, were high, despite apparent low country of origin rates. However, recently published Vietnamese rates are from the north of Vietnam (Pham et al.. 1993). which may have lower rates of cervical cancer than the south. where cervical cancer has been previously reported as constituting over 50Go of all cancer in women (Parkin, 1986). Increased proportional incidence of cervical cancer has been described in Vietnamese migrants to Los Angeles County (Ross et al.. 1991). Evidence points towards a sexually transmitted virus. the human papillomavirus (HPV). as the cause of the majority of cases of cervical cancer (Bosch et al.. 1992). While we had no data on HPV infection, high levels of reactivity to tests for syphilis have been described in Vietnamese refugees in Australia (Bek and Levy. 1992).
Rates of prostate cancer were low in most immigrant groups. but were much higher than in the countries of origin. Rates were lowest in the China Taiwan born. who also have low mortality from prostate cancer in the US (King and Locke. 1980) and in Singapore (Lee et al.. 1988). However. rates in the Hong Kong born, who were of high SES. were not significantly low. The difference in rates between the China,Taiwan and Hong Kong born is compatible with environmental factors in the causation of this cancer. but could also potentially be explained by socio-economic differentials in usage of medical services. A high-fat diet has been implicated as a risk factor for prostate cancer (Greenwald. 1982 groups. This is consistent with previous findings of low rates of testicular cancer in Taiwan (King and Locke. 1980). in Indian migrants to England and Wales (Berra and Swerdlow. in preparation) and in other Asian populations (Parkin et al.. 1992). The non-significantly low rates of bladder cancer found in all Asian immigrant groups were in contrast to the high rates of bladder cancer previously described in male British. European and Middle Eastern migrants to NSW (McCredie et al.. 1990). It has been postulated that occupational exposure to hazardous chemicals in low-status jobs in Australia could explain these high rates. The absence of raised rates in Asian migrants may reflect the fact that they were the most recent migrants. and the potentially harmful exposures were no longer present. or that the carcinogenic effects of any Australian exposures were not yet apparent. In addition. the fact that many Asian migrants were of high SES would make them less likely to be occupationally exposed to hazardous chemicals.
Cancers with rates more similar to country of birth than Australia-born rates Rates of nasopharyngeal cancer (NPC) were close to those in the countries of origin. Raised rates of NPC have been described in Chinese migrants in the US (King and Locke, 1980) and NSW (Zhang et al.. 1984;McCredie and Coates. 1989). and in Vietnamese migrants to the US (Ross et al.. 1991) and England and Wales (Swerdlow, 1991). In southern China. rates of NPC are thought to be higher in persons of low socioeconomic status (SES) (Yu et al., 1986). but we found high rates in the Hong Kong born. who were of high SES. Population-based studies in southern China have found that consumption of Cantonese-style salted fish as a weaning food is a strong risk factor for NPC (Yu et al.. 1986(Yu et al.. . 1988). It might be expected then that rates would stay high in all Chinese who migrate after infancy. It has long been recognised that Epstein-Barr virus infection is associated with this cancer (de-The. 1993), and others have postulated genetic susceptibility as a strong risk factor (Ho et al.. 1982;Lu et al., 1990). Our findings, of high rates in both the high-SES Hong Kong born and the lower SES China Taiwan born. could be explained either by genetic risk factors or by a risk factor acting early in life that was not differentially distributed by SES. The finding of raised SIRs in other immigrant groups was consistent with either genetic or cultural intermingling of the ethnic Chinese in South-East Asia.
Melanoma rates were consistent with the expected protective effect of skin pigmentation. Melanoma constituted about 8% of all registered cancers in the Australia born during the period of this review in NSW but accounted for less than 2% in most of the immigrant groups.
The high rates of lung cancer found in female migrants from China Taiwan. and in migrant groups with a high proportion of ethnic Chinese. were at odds with the low prevalence of smoking found in Asia-born immigrant women in Australia (Castles. 1992). However, high rates of lung cancer. and a high proportion of adenocarcinoma, which is less strongly associated with smoking than squamous cell carcinoma (Lam et al.. 1987;Morabia and Wynder, 1991). have been previously described in Chinese women in Singapore. Hawaii. Hong Kong, the US (MacLennan et al.. 1977;Gao et al.. 1988;Koo and Ho. 1990) and NSW (McCredie et al.. 1990). In China. it has been estimated that only 25-35% of lung cancer in females is attributable to tobacco smoking (Wu-Williams et al., 1990;Liu et al., 1992), other possible risk factors including a deficiency of vitamin A (McLennan et al.. 1977). passive smoking (Lam et al.. 1987) and indoor air pollution (Liu et al.. 1993).
For the majority of cancers. environment factors including change to an Australian environment as well as socioeconomic status of the migrant group, appeared to be the major influences on cancer incidence. Only for the most visible difference between the races, skin colour. was there evidence of a genetic trait which dominated cancer risk (melanoma). Rates of nasopharyngeal cancer. and of lung cancer in females. were also similar to country of birth rates, consistent with either early environmental or genetic risk factors.