|Home | About | Journals | Submit | Contact Us | Français|
Malignancies of the lymphoid cells, including non-Hodgkin lymphomas (NHLs), Hodgkin lymphoma (HL) and multiple myeloma (MM), occur at much lower rates in Asians than other racial/ethnic groups in the United States (US). It remains unclear whether these deficits are explained by genetic or environmental factors. To better understand environmental contributions, we examined incidence patterns of lymphoid malignancies among populations characterized by ethnicity, birthplace, and residential neighborhood socioeconomic status (SES) and ethnic enclave status.
We obtained data regarding all Asian patients diagnosed with lymphoid malignancies between 1988 and 2004 from the California Cancer Registry and neighborhood characteristics from US Census data.
While incidence rates of most lymphoid malignancies were lower among Asian than white populations, only follicular lymphoma (FL), chronic lymphocytic leukemia/small lymphocytic lymphoma (CLL/SLL), and nodular sclerosis (NS) HL rates were statistically significantly lower among foreign-born than US-born Asians, with incidence rate ratios ranging from 0.34 to 0.87. Rates of CLL/SLL and NS HL were also lower among Asian women living in ethnic enclaves or lower-SES neighborhoods than those living elsewhere. Conclusions: These observations support strong roles of environmental factors in the causation of FL, CLL/SLL, and NS HL.
Studying specific lymphoid malignancies in US Asians may provide valuable insight towards understanding their environmental causes.
Malignancies of the lymphoid cells, including non-Hodgkin lymphomas (NHLs), Hodgkin lymphoma (HL) and multiple myeloma (MM), are highly heterogeneous with respect to clinical features and patterns of occurrence (1). Despite this variation, United States (US) populations of Asian origin consistently have much lower incidence rates of lymphoid malignancies than populations of Caucasian or African origin (2). In our recent assessment of lymphoid malignancies in the US, overall incidence rates were substantially lower among Filipinos (67% of non-Hispanic white rate), South Asians (64%), Vietnamese (62%), Japanese (53%), Chinese (47%), and Koreans (33%) than among non-Hispanic whites (2). These striking differences may contain important clues as to genetic or environmental risk factors for these diseases, especially if incidence rates change with migration from low-risk (e.g., Asia) to higher-risk areas (e.g., US).
Unfortunately, cancer incidence rates according to measures of migration or acculturation cannot be readily calculated. Although patient birthplace is collected by cancer registries, it is missing in a biased manner for a substantial proportion of patients (3–6), and the birthplace-specific annual population counts needed for rate denominators are not readily available from governmental agencies. Measures of immigrant acculturation (e.g, years since migration, language used at home) are not collected at all by cancer registries. To surmount these challenges, we have developed a resource (7) incorporating cancer registry data from California and population data needed to calculate cancer incidence rates among US Asians by birthplace and two residential neighborhood measures of acculturation: socioeconomic status (SES) and ethnic enclave status. Using this resource, we examined variation in incidence rates of lymphoid malignancies among US Asians by these factors.
From the California Cancer Registry (CCR), which comprises three of the National Cancer Institute’s SEER program registries(8), we obtained information on all California residents diagnosed with a primary invasive lymphoid malignancy (International Classification of Disease for Oncology, 3rd Edition, (ICD-O-3) morphology codes 9590–9591, 9650–9655, 9661–9734, 9761, 9764, 9823, 9827–9837, 9940, 9948, and 9970) during the period 1 January 1988 through 31 December 2004. Using InterLymph Consortium guidelines (9, 10), we further classified lymphoid malignancies by histologic subtype into diffuse large B-cell lymphoma [DLBCL] (ICD-O-3 morphology codes 9680 and 9684, excluding those with code 9684 and a T-cell, NK-cell or null cell immunophenotype); follicular lymphoma [FL] (codes 9690, 9691, 9695, and 9698); chronic lymphocytic lymphoma/small lymphocytic leukemia [CLL/SLL] (codes 9823 and 9670); multiple myeloma [MM] (codes 9731–9734); classical HL (codes 9650–9655, 9661–9667) and its two most common subtypes, nodular sclerosis [NS] HL (codes 9663–9665, 9667) and mixed cellularity [MS] HL (code 9652); and T- or NK-cell NHL [TCL] (codes 9700–9719, 9729, 9827, 9831, 9834, 9837, and 9948 plus codes 9590, 9591, 9675, 9684, 9727, 9820, 9832, 9835, or 9970 and a T-cell or NK cell immunophenotype). Overall NHL was categorized as including codes 9590–9591, 9670–9729, 9761, 9764, 9820, 9823, 9827, 9831–9837, 9940, 9948, and 9970.
We included in these analyses 8,638 lymphoid malignancies (6,712 NHL, 526 NHL, and 1,410 MM) occurring in patients from the six Asian ethnic populations that together comprised 92% of all Asian and Pacific Islander patients with lymphoid malignancies in the CCR in the study period. Of these, 2,385 (28%) cases were Chinese, 1,246 (14%) were Japanese, 2,913 (34%) were Filipino, 506 (6%) were Korean, 701 (8%) were South Asian (including Asian Indians, Pakistanis, Sri Lankans and Bangladeshis), and 887 (10%) were Vietnamese. Approximately 5% of patients were originally coded in the registry data as “Asian, not otherwise specified (NOS)”; for 55% of these patients, we were able to determine a more specific Asian ethnic designation based on birthplace and names (first, maiden, last) by applying the North American Association of Central Cancer Registries (NAACCR) Asian/Pacific Islander Identification Algorithm (11). We also included as a reference group 110,789 non-Hispanic white patients diagnosed with lymphoid malignancies (85,465 NHL, 8,967 HL, and 16,357 MM) during the same period.
Because patients in the cancer registry with unknown birthplace data are more likely to be US-born than those with available data (3–6), we developed a method based on patients’ Social Security numbers (SSN) to more accurately classify patient immigrant status, as described previously (7). Briefly, we used: 1) registry-based birthplace data available for 81% of cases (73% from hospital medical records and 8% from death certificates); and 2) for the 19% of cases with unknown birthplace, statistical imputation of immigrant status using the patient’s SSN. Comparing the age of SSN issue with self-reported birthplace in previously interviewed cancer patients (N=1,836) and based on maximization of the area under the receiver-operating characteristic curve and confirmation with logistic regression modeling, we considered cases who received an SSN before age 25 years US-born, and those who had received an SSN at or after age 25 years as foreign-born. This age cut-point resulted in 84% sensitivity and 80% specificity for assigning foreign-born status across the Asian populations. Fewer than 3% of cases with missing or invalid SSNs were assigned an immigrant status based on the ethnicity-sex-age birthplace distribution of the overall sample.
Using patient residential address and small area (census tract) information from the US Census, we classified neighborhood SES and ethnic enclave status for all Asian patients diagnosed between 1 January 1998 and 31 December 2002. We considered all Asians together as a single group because detailed ethnicity-specific population estimates are not available for census tracts, and chose the time period in question (i.e., within two years of the 2000 US Census) because census tract estimates are only available for decennial census years. Census tracts were geocoded from patient residential address at time of diagnosis. The 3% of eligible cases whose address could not be precisely geocoded to a census tract were randomly assigned to a census tract within their county of residence. We assigned neighborhood SES using a previously described index (12) that incorporates 2000 Census data on education, income, occupation, and housing costs. We categorized this measure by quintiles based on the distribution of the composite SES index across the state of California, then re-categorized into two groups because of small sample sizes in the quintiles: lower SES (quintiles 1, 2, and 3) or higher SES (quintiles 4 and 5). Because the CCR does not collect any individual-level information on patient SES, we could not assess neighborhood-level effects separately from those at the individual-level.
We defined a neighborhood ethnic enclave as a geographical unit that is relatively more concentrated in terms of its population and language (in this study specific to Asians) than other geographical units in California. To characterize residence in an ethnic enclave, we applied principal components analysis (PCA) (13) to selected census variables at the block group level, which was in turn averaged to the census tract level. The census variables included in the ethnic enclave index were: percent of Asian-language-speaking households that are linguistically isolated, percent of all Asian-language speakers who speak limited English, percent recent immigrants, and percent Asian. This index explained 63% of the variability in the data. Neighborhood ethnic enclave was classified into quintiles based on the distribution of the composite ethnic enclave index across the state of California, then re-categorized into two groups because of small sample sizes in the quintiles: lower (quintiles 1, 2, and 3) or higher (quintiles 4 and 5) enclave status.
From the 1990 and 2000 Census Summary File 3 (SF-3), we obtained population counts by sex, race/ethnicity, immigrant status, and five-year age group for the state of California. We also used data from the 20% Integrated Public-Use Microdata Sample of the Census to estimate age- and birthplace-specific population counts for the six Asian groups (14–17) by smoothing with a spline-based function(18). For intercensal years, we estimated the percent foreign-born using cohort component interpolation and extrapolation methods(19), adjusting estimates to the populations by age and year provided by the California Department of Finance for years 1988–1989 and by the US Census for years 1990–2004 due to data availability.
We used SEER*Stat software (20) to compute age-adjusted incidence rates (standardized to the 2000 US standard million population) and 95% confidence intervals (CIs). To comply with CCR regulations, we do not present case counts or rates based on fewer than 15 cases. For HL, we also calculated age-adjusted rates for persons ages 15–44 (N=159 males, 153 females), and 45 and above (N=110 males, 70 females), at diagnosis) because of strong previous evidence of etiologic differences between these groups (21). We calculated incidence rate ratios (IRRs) to compare incidence rates. Because of small case numbers, Asian ethnic groups were combined for analyses of NHL subtypes, MM, and HL, and analyses of HL rates jointly by SES and enclave could not be undertaken by age group. We could not perform joint analyses by birthplace and neighborhood SES or ethnic enclave status due to the lack of census-tract-level population data by birthplace. All analyses had the approval of the institutional review board of the Cancer Prevention Institute of California.
Among the 8,638 Asians diagnosed with a lymphoid malignancy in California in the years 1988 through 2004, the majority (75%) were foreign-born, although this proportion was much lower for Japanese (32%) (Table 1). Among histologic subtypes of NHL, the most common among Asians was DLBCL (N=2,345, 35% of all NHL), followed by TCL (N=721, 11%), FL (N=661, 10%), and CLL/SLL (N=560, 8%); by comparison, non-Hispanic whites had higher proportions of FL, CLL/SLL and a lower proportion of TCL. 1,410 Asians were diagnosed with MM and 516 with HL, including 322 (62%) with NS HL and 96 (19%) with MC HL.
For NHL overall, age-adjusted incidence rates for most Asian ethnic groups were substantially lower than those for non-Hispanic whites (Table 2). For example, the IRR among foreign-born Filipino males vs. white males was 0.70 (95% CI 0.65–0.75) and the corresponding IRR for females was 0.76 (95% CI 0.70–0.83); the IRR among foreign-born Chinese males vs. white males was 0.49 (95% CI 0.44–0.54) and the corresponding IRR among females was 0.50 (95% CI 0.45–0.55). The only groups for which rates were not significantly different from those of non-Hispanic whites were the relatively small populations of foreign-born Japanese men and US-born South-Asian and Vietnamese men and women.
For most specific lymphoid malignancy subtypes, incidence rates for both US- and foreign-born Asians were lower than those for non-Hispanic whites. The most marked deficits were observed for CLL/SLL (among males, IRR for foreign-born Asians vs. whites=0.22, 95% CI 0.18–0.25; among females, IRR=0.24, 95% CI 0.18–0.30) and NS HL (among males, IRR for foreign-born Asians vs. whites=0.25, 95% CI 0.13–0.37; among females, IRR=0.19, 95% CI 0.13–0.26). However, for TCL and DLBCL (in most ethnic groups), rates were comparable to those for non-Hispanic whites.
For overall NHL, foreign-born Chinese, South Asian, and Vietnamese men and women had consistently lower incidence rates than their US-born counterparts (Table 2). By contrast, foreign-born Japanese men had incidence rates 71% higher than their US-born counterparts, whereas no birthplace difference was observed among Japanese women or Korean men or women.
For specific subtypes, Table 3 shows that FL incidence was consistently lower among foreign-born than US-born Asian men and women. For DLBCL, for which numbers of cases were adequate for examining rates for Chinese and Japanese, patterns were similar to those observed for overall NHL, with lower incidence rates for foreign-born vs. US-born Chinese men and women, in contrast to higher rates among foreign-born vs. US-born Japanese men. Among Japanese and other Asian (Filipina, South Asian, and Vietnamese) women, there were no significant differences in the incidence rate of DLBCL by birthplace. Incidence rates of TCL did not vary by birthplace among Asian men or women. For MM, rates were marginally higher (37%) among foreign-born than US-born Asian women, but comparable by birthplace in men.
For overall HL, rates among foreign-born Asians were approximately half those among US-born Asians. IRR patterns were similar for NS HL, but no nativity differences occurred in rates of MC HL. In data stratified by age, the protective effect of foreign birthplace was limited to young adults of both genders for HL overall; for NS HL, it was apparent for both younger and older females (IRR’s of 0.34 (95% CI 0.23–0.50) for ages 15–44, and 0.32 (95% CI 0.14–0.80) for ages 45 and above).
Although we had limited statistical power for to assess rate changes over time stratified by birthplace, rates did not vary significantly between the periods 1988–1996 and 1997–2004 (data not shown). For overall NHL, rates increased significantly among US-born Chinese and Filipino men and foreign-born Korean men, but not among women of the same groups. IRRs comparing foreign- vs. US-born Asians were generally similar between 1988–1996 and 1997–2004, although for overall HL and NS HL, they were statistically significant only in the latter period (data not shown).
Among Asian men, ethnic enclave status did not impact the incidence rates of overall NHL, DLBCL, FL, CLL/SLL, TCL, MM, overall HL, NS HL, or MC HL (Table 4). By contrast, among Asian women, overall NHL, CLL/SLL, overall HL and MC HL were significantly less common in neighborhoods with higher ethnic enclave status. For HL, these patterns did not differ by age group for either gender. Asian men living in higher-SES neighborhoods had significantly elevated incidence rates of FL and NS HL (an effect limited to young adult men), but not of other lymphoma subtypes. Asian women in higher-SES neighborhoods had significantly higher incidence rates of FL, TCL and overall HL (apparent only for women over age 45 at diagnosis); marginally higher rates of NS HL and MC HL; but lower rates of CLL/SLL.
When we analyzed rates by neighborhood ethnic enclave status and SES jointly, we found that Asian men living in areas with both lower ethnic enclave status and higher SES had significantly higher incidence rates of overall NHL and FL, as well as a marginally higher incidence rate of DLBCL, than Asian men living in areas with both higher enclave status and lower SES (Table 4). For overall HL and NS HL, rates were marginally higher for Asian men living in neighborhoods of higher than lower SES irrespective of their ethnic enclave status. Compared to Asian women living in neighborhoods with both higher ethnic enclave status and lower SES, Asian women living in neighborhoods with lower ethnic enclave status and higher SES also had significantly elevated incidence rates of overall NHL, FL, overall HL and NS HL; elevations were particularly marked for the latter two (IRR=4.1, 95% CI 2.15–7.74, and 2.5, 95% CI 1.17–5.14, respectively). In addition, Asian women who resided in neighborhoods with both higher ethnic enclave status and higher SES had elevated incidence rates of FL, TCL, and overall HL but a lower rate of CLL/SLL. MM incidence rates did not vary by neighborhood enclave status and SES among Asian men or women.
A role for environmental exposures in cancer etiology can be inferred from changes in cancer incidence after migration from low- to high-risk areas. In a large population-based series of US Asians with lymphoid malignancies, we found that rates were substantially lower in foreign-born than US-born patients for certain lymphoma subtypes, specifically CLL/SLL, FL, and NS HL. Rates of CLL/SLL and NS HL were also significantly lower among Asian women living in ethnic enclaves or lower-SES neighborhoods, compared with rates of Asian women living in lower-enclave or higher-SES neighborhoods, respectively. For HL, the risks associated with higher-SES and lower-enclave neighborhoods were stronger in females than in males. For MM, incidence rates did not differ according to birthplace, ethnic enclave status, or neighborhood SES. We also confirmed that the incidence rates of most subtypes were substantially lower than rates in non-Hispanic white populations; for TCL and DLBCL—the two subtypes for which absolute incidence rates were most similar between Asians and non-Hispanic whites—we did not observe consistent differences in incidence according to birthplace or neighborhood characteristics.
There is little published information regarding the incidence patterns of lymphoid malignancy subtypes among Asians according to detailed ethnicity and birthplace. Our recent analysis based on SEER data documented lower incidence of lymphoid malignancies among six Asian ethnic groups compared with whites(2), but lacked the data to consider differences by birthplace. In SEER data, an assessment of NHL cases diagnosed in the period 1973–86 and classified according to the Working Formulation scheme also found reduced risk of FL in foreign-born compared with US-born Chinese and Japanese (but not Filipinos), with incidence rates 60–80% lower than rates in their US-born counterparts(22). However, in that analysis, the authors assumed that the SEER cases without birthplace information had randomly missing data. As we have shown that those with missing data are more likely to be US-born(5), this earlier analysis may have underestimated rate differences by birthplace, which may explain the difference in findings for CLL/SLL and HL. To our knowledge, ours is the first study to address lymphoid malignancy incidence patterns among US Asians according to neighborhood characteristics, although we did report previously that rates of young-adult HL were lower among Asian women (but not men) living in the lowest terciles of neighborhood SES in California(23). Our findings of lowered rates of CLL/SLL and NS HL among Asian women living in impoverished or ethnic enclave communities as compared to more affluent and presumably more acculturated communities further support the notion that the causation of these particular lymphoid malignancy subtypes involves environmental exposures more common in westernized environments.
Differences in cancer incidence rates between Asians who immigrate to the US (and their descendents) and those who remain in Asia have long been considered strong evidence of environmental influences on carcinogenesis, although it is possible that there are also genetic differences among persons who are healthy enough to emigrate. For breast cancer, incidence rates among Chinese and Filipina women born in the US are nearly twice those of women living in Asia, and these differences are increasingly thought to relate to reproductive and dietary changes associated with westernized lifestyle(7, 24). For NS HL, exposures of interest include correlates of the childhood social environment(25) (e.g., family size, household crowding) and measures of microbial burden or other immunologically relevant environmental exposures (e.g., age at diagnosis with mononucleosis)(26), particularly in early life (21, 25, 27–33). Childhood environment has not been consistently associated with risk of FL or CLL/SLL(34–36), although a recent pooled analysis including over 13,500 NHL cases did report for FL significantly positive associations with both birth order and sibship size(36). However, risk of both FL and CLL/SLL has been inversely associated with atopic disease(37), which could be associated in turn with early-life microbial exposures. It is uncertain if chronic infection with hepatitis viruses, linked to doubled risks of NHL(38, 39) and endemic in Asia(40, 41), are relevant to the observed rate patterns. Although US-born Asians have lower rates of chronic infection with hepatitis B and C viruses than their foreign-born counterparts in the US and Asia(40, 42), the associations of viral hepatitis with risk of specific NHL subtypes (e.g., DLBCL) do not correspond the observed incidence rate differences by birthplace in our study(43).
The stronger effects of birthplace and neighborhood characteristics for CLL/SLL and HL observed in females than males could, in part, reflect socially determined differences in exposure opportunities (such as those involving children) and biologically determined differences in immune response to exposures(23). For HL, the gender difference in the effect of nativity may result from protection in low-acculturation women afforded by both early exposure to infection and higher parity or lactation, as hormonal exposures through pregnancy and breast-feeding may interact with childhood exposures to affect risk of HL(44).
For HL, the varying impact of birthplace by age group is consistent with prior evidence of differing pathogenesis of HL by age (21). It also is relevant to the differential effect of birthplace on incidence of the NS and MC subtypes, given that young adult HL is predominantly of the NS subtype. Further, subtype differences in birthplace associations may reflect etiologic differences in immune control and age at infection of Epstein-Barr virus (EBV), as EBV is more commonly found in tumors of the MC than NS type (45, 46) and of Asians than whites (47).
Dietary patterns and energy balance/obesity, which also vary by birthplace among US Asians (48), may also be associated with risk of certain lymphoid malignancies (49–52), and therefore represent important areas for future study in Asian immigrant populations. For MM, our observation of substantially lowered rates among Asians as compared to non-Hispanic whites, but no difference according to birthplace or neighborhood characteristics, suggests a more important role for genetic susceptibility and less of an influence of environmental exposures that change with acculturation. In support of this hypothesis, MM risk has been associated with polymorphisms in genes thought to influence innate immunity and immunoregulatory processes (53, 54).
By using over 16 years of SEER data from California, we were able to capitalize on the relatively large size of the Asian population in this state and to draw conclusions based on the representativeness of these high-quality, population-based data. We consider the ethnic and birthplace classifications used here to have low probabilities of misclassification or bias. Specific Asian ethnic group was classified directly from registry records or, for those without specific registry information on ethnicity, from applying a validated ethnicity classification algorithm. With this approach, a small proportion (<3%) of patients was excluded from these analyses because of missing ethnic classification. Furthermore, cancer registry classification of specific Asian ethnicity shows good agreement with self-reported information (55). For cases for whom birthplace information was reported to the registry (the vast majority), we have also demonstrated that this classification shows excellent agreement in comparison with self-reported birthplace (4, 5); for the remaining cases, we applied a validated birthplace classification algorithm with good sensitivity and specificity.
Despite these important strengths, our results also may be subject to some limitations. First, we had limited statistical power to analyze certain subgroups, such as specific Asian ethnic groups and uncommon lymphoid malignancies. Second, the heterogeneity in the complex pathologic methods required to diagnose and classify lymphoma cases may have resulted in misclassification of some cases by subtype. Our prior comparisons of cancer registry ICD-O-3 classifications to those obtained from uniform re-review of pathologic specimens suggest a high degree of reliability for the diagnosis of overall NHL and HL(56) and for particular subtype classifications including FL (89%), SLL (79%), DLBCL (90%), and NS HL (95%), but more moderate reliability for rarer subtypes(57, 58). In addition, cancer registry data lack detail regarding certain histopathologic characterizations (e.g., t(14;18) translocations for FL, and EBV tumor-cell status for HL), as well as information regarding parental race/ethnicity, individual-level education and other measures of SES, medical history, age at immigration, duration of immigration and other risk factors that could be relevant to our observed incidence rate differentials. Lastly, these data cannot speak to the independent or joint influence of genetic factors in modifying risk of FL and CLL/SLL across populations. Recent genomewide association studies found genetic variants that influence risk for FL(59) and SLL/CLL(60) and the absolute difference between rates in US-born Asians and whites does not rule out a role for genetic predisposition to FL and CLL/SLL. Regardless, our results suggest that environmental exposures have greater influence than genes on the variation in incidence rates by ethnicity and nativity.
The markedly lowered rates of lymphoid malignancies among Asians relative to other racial/ethnic groups in the US and among foreign-born Asians relative to US-born Asians have suggested some kind of protection from lymphomagenic processes, but it has been unclear whether this protection relates to genetic or environmental differences. Our data suggest a clear pattern of increased risk of FL, CLL/SLL and HL in Asians according to US birthplace and neighborhood acculturation indicators, and thereby point to a strong influence of environmental factors that change with immigration and acculturation to a westernized lifestyle. Future studies of FL, CLL/SLL and HL designed to collect a wide array of environmental exposure information (and implicated genetic variants of risk) are warranted among Asian immigrant populations in the US and other westernized countries, as they may identify heretofore unrecognized and modifiable causes of these malignancies.
Financial support: This study was supported by the National Cancer Institute’s Surveillance, Epidemiology and End Results (SEER) program under contract HHSN261201000036C awarded to the Cancer Prevention Institute of California (CPIC).
Other information: The collection of cancer incidence data used in this study was supported by the California Department of Public Health as part of the statewide cancer reporting program mandated by California Health and Safety Code Section 103885; the SEER Program under contract N01-PC-35136 awarded to CPIC (formerly the Northern California Cancer Center), contract N01-PC-35139 awarded to the University of Southern California, and contract N01-PC-54404 awarded to the Public Health Institute; the Centers for Disease Control and Prevention’s National Program of Cancer Registries, under agreement 1U58DP00807-01 awarded to the Public Health Institute; and R01-ES015552 from the NIEHS and R01-CA121052 from NCI. The ideas and opinions expressed herein are those of the authors and endorsement by the State of California, Department of Public Health the National Cancer Institute, and the Centers for Disease Control and Prevention or their Contractors and Subcontractors is not intended nor should be inferred. The authors thank Sarah Shema and Tim Miller for their contributions to this study.
No authors report any financial conflict of interest.