|Home | About | Journals | Submit | Contact Us | Français|
The higher incidence of acute lymphoblastic leukemia (ALL) among Hispanic children relative to that in other racial/ethnic groups is well-known. We evaluated incidence patterns of ALL in adults.
We analyzed the incidence patterns of ALL (ICD-03 codes 9835–9837) among all patients diagnosed from 1988–2004 in California using data from the California Cancer Registry to determine whether adult Hispanics also had higher incidence rates of ALL compared to non-Hispanic Whites (Whites). Age-adjusted incidence rates (AAIR), incidence rate ratios (IRR) and 5-year survival rates were obtained using SEER*Stat. AAIRs of other leukemia subtypes and IRRs relative to non-Hispanic whites were also examined as references of ALL.
AAIRs of ALL in Hispanic males and females ages 20–54 years were higher compared to those in White males and females (IRR=1.99,95% CI=1.74–2.28 and IRR=1.91,95% CI=1.60–2.25 respectively). A higher AAIR of ALL was also observed among older (55+ years) Hispanic females (IRR=1.84, 95% CI=1.52–2.21), but not males (IRR= 1.07, 95% CI= 0.84–1.34). Among Hispanics, low socioeconomic status (SES) was associated with a higher AAIR compared to high/middle SES (IRR= 1.33, 95% CI=1.04–1.70). The respective five-year survival rates among ALL patients were 38% and 30% for Whites and Hispanics ages 20–54 years, and 8% and 12% for patients 55 years of age or older. Compared to other racial/ethnic groups, Hispanics did not have an increased IRR of the other major leukemia subtypes.
Hispanics experience a higher incidence of ALL throughout life, but not other subtypes.
Acute lymphoblastic leukemia (ALL) is a neoplasm that arises from clonal expansion of lymphoblasts, the majority of which arise from B-lymphoblasts and a small subset from T-lymphoblasts (1). Although childhood ALL has been well-studied epidemiologically, the incidence pattern of ALL in adults has not, probably because of the relatively low incidence. Population-based studies of ALL in children have consistently reported higher incidence among Hispanics compared to other ethnic groups in the United States (2, 3, 4, 5). We performed the current study to examine whether adult Hispanics also experience higher incidence rates of ALL utilizing data from the California Cancer Registry. We also examined relative survival among ALL patients by racial/ethnic group.
The source of data for the study was the California Cancer Registry (CCR), the legally mandated system of regional cancer registries reporting to the California Department of Health Services, which consists of three population-based cancer registries designated as part of the National Cancer Institute Surveillance, Epidemiology and End-Results Program (SEER) (6). Data from these California registries contains at the minimum 98% of all incident cancers diagnosed in the state for any given year.
Cases were defined as California residents diagnosed with ALL (International Classification of Diseases for Oncology [ICDO-3] morphology codes 9835–9837, according to the accepted WHO/REAL classification system), during the years 1988–2004 (1). Diagnoses were verified by pathology report for 99% of the cases. AAIR’s and IRR’s of ALL by race/ethnicity were compared to other leukemia subtypes diagnosed during the same period, and classified as: B-cell chronic lymphocytic leukemia (CLL) (ICD-03 morphology codes 9823, 9670), acute myelogenous leukemia (AML) (9840, 9861, 9866, 9867, 9870–9874, 9891, 9895– 9897, 9910, 9920, 9930, 9931) and chronic myelogenous leukemia (CML) (9863, 9875, 9876). Adults were defined as subjects who were 20 years of age and older at diagnosis.
Demographic information collected by the registries includes age at diagnosis, sex, block group residence at diagnosis (to assign socioeconomic status [SES]), race/ethnicity and birthplace. Patients were categorized into the following racial/ethnic groups: non-Hispanic White (White), non-Hispanic African-American (African American), non-Hispanic Asian/Pacific Islander/American Indian/Alaskan Native (Asian/PI/AI/AN) and Hispanic. Assignment of Hispanic race/ethnicity was based on an algorithm that included information from the hospital record, birthplace, and surname (from the 1980 U.S. Census). Information on birthplace was collected from approximately 74% of all Hispanic cancer patients during the time period under study.
SES is based on census block group characteristics of the case residence at the time of diagnosis, incorporating information on median education, proportion unemployed, occupation, proportion with incomes below the twice the poverty level, median household income, median rent and median home value obtained from the 1990 and 2000 U.S. census and incorporated into a single socioeconomic indicator as described previously by Yost, et al. (7). Each block group in California was assigned a score and the scores ranked into quintiles irrespective of race or ethnicity. Each patient received an SES classification based on the quintile of the block group of his/her residence at the time of diagnosis. For this analysis, the time period was limited to the five-year time period surrounding the 2000 census, 1998–2002, since the census block group populations were not available for intracensal years (between census counts). To estimate our population-at-risk denominator for analyses by SES, the 2000 population counts were multiplied by 5 (8). SES quintiles were collapsed into two groups: high/middle (quintile 3–5), and low (quintile 1–2).
SEER*Stat software1 was used to calculate average annual age-adjusted incidence rates (AAIR) and 5-year relative survival rates by sex, race/ethnicity, and age (<19 years old, 20–54 years old, and 55 years or older) for cases diagnosed between 1988–2004; AAIRs are reported as cases per 100,000 people, and age-adjusted (within the broad age ranges) to the 2000 US population standard. Incidence rate ratios (IRR) were obtained based on the AAIRs using the same software, with Whites as the reference group. Tests of statistical significance are 2-sided, with a p-value of 0.05 considered statistically significant. For some analyses, all diagnoses in persons over 19 years old were considered together as “adult” cases. Incidence rates and IRRs by SES were also obtained using SEER*Stat software1.
Because population-based denominator data is not available by birthplace, we assessed the risk of ALL among adult Hispanics born outside of the United Sates (US) versus those born in the United States (reference group) in case-case comparisons with two “control” groups: all other Hispanic leukemia patients and all other Hispanic cancer patients, excluding ALL patients. Both birthplace comparisons were limited to Hispanic patients diagnosed at 20 years of age or older. Proportional odds ratios (PORs) for risk by birthplace were then calculated using a multivariable unconditional logistic regression model which included sex (male/female), age at diagnosis (continuous) and SES (high, middle, low) (9). Logistic regression analyses were calculated using SAS version 9.1.3 (10).
A total of 5,203 males and 3,972 female residents of California were diagnosed with ALL during the period 1988–2004 (Table 1). Incidence rates were highest during childhood among all sex- and racial/ethnic specific groups, decreased during young adulthood, and rose again slightly at older ages, demonstrating a bi-modal pattern (Tables 1a,,bb).
AAIRs for ALL were highest in male and female Hispanics within each age category compared to those in other racial/ethnic groups (Tables 1a,,b).b). The greatest difference in incidence rates relative to Whites was observed among Hispanics aged 20–54 years. In the oldest age group of Hispanics, the incidence rate in older Hispanic males was not much higher than that in older White males; Hispanic female incidence remained quite high relative to that of White females. Over all racial/ethnic groups combined, males had a 1.3-fold or greater AAIR of ALL compared to females (data not shown), with some variation by age and racial/ethnic group.
Among adult Hispanics, the AAIR of ALL was modestly higher in those of low SES relative to those of high/middle SES. No such difference by SES was seen among White ALL patients (Table 2).
Adult Hispanics with ALL were more likely to be born outside the U.S. relative to within the U.S. when compared to other Hispanic leukemia patients (POR=1.31, 95% CI= 1.12–1.55), and to all other (non-leukemia) Hispanic cancer patients (POR=1.57, 95% 1.36–1.81) diagnosed in California during the time period 1988–2004. Hispanic adults did not have higher incidence rates of any other major leukemia subtypes compared to other racial/ethnic groups, thus the excess incidence is specific only for ALL (Table 3).
Five-year relative survival was highest in the youngest age group, and dropped precipitously with older age at diagnosis, a pattern evident across all ethnicities (Figure 1). Survival rates among Hispanics were slightly lower than those of Whites among children (77% in Hispanics vs. 84% in Whites), and young adults 20–54 years old (30% in Hispanics vs. 34% in Whites). Among the oldest age group, Hispanic ALL patients had slightly higher 5-year relative survival rates compared to Whites (12% in Hispanics vs. 8% in Whites). Female Hispanics had slightly higher survival rates as children and teenagers, but slightly lower survival rates as adults compared to male Hispanics (data not shown). None of these differences were statistically significant.
To ensure that our results were not biased due to the inclusion of data prior to the adoption of the new classification of ALL, we restricted the analyses to the time period 1994–2004, and found essentially identical results (data not shown).
Hispanics have the highest incidence rates of ALL, but not other types of leukemia, throughout life. There is some evidence to support an association between low SES and increased ALL incidence rates among Hispanics. Hispanic ALL patients diagnosed up to 54 years of age had a 5-year survival rate that was marginally lower than that of Whites, but among those diagnosed at 55 years of age or older, the survival rate was marginally higher.
Although the higher incidence rates of ALL among Hispanic children is well-known (2–5), only one published report and one book have examined the pattern in adults (11,12). An examination of incidence patterns for 85 different types of cancer in Los Angeles County from 1972–1998 also demonstrated an excess of ALL among Hispanics throughout life (11). In another study, Matasar et al (12) reported that the excess incidence rates for Hispanics leveled off at older ages. We saw a similar pattern among Hispanic males only; the incidence rate among the oldest female Hispanics was still the highest of any ethnic group.
Previous studies have examined the relationship between SES and leukemia incidence rates with variable results but the relationship has rarely been examined in adults. In an extensive review of the literature, Poole and colleagues (13), summarized 47 studies on the relationship between SES and childhood leukemia incidence and concluded that there was no clear-cut evidence of an association. Because the negative association of SES with AAIR of ALL was observed among Hispanics but not among non-Hispanic whites, it is likely that different etiologic pathways are responsible for disease in different age and ethnic groups.
Our finding of lower survival of ALL in Hispanic children compared to White children is consistent with a previous report based on SEER data from other geographic areas in U.S. (14). There are no other reports of variation in survival rates by ethnicity for adults with ALL. The slightly higher survival rates of Hispanics compared to Whites in the oldest group of patients is surprising and cannot be easily explained; however the overall survival rates and the numbers of patients in this age group are so low that the differences are not statistically significant.
Limitations of our study include reliance on ecological data at the population level rather than individual data for some exposures such as SES, which could result in non-differential misclassification. There may be some misclassification of Hispanic ethnicity (15), since information on ethnicity was obtained from the medical records as part of the cancer reporting process. Furthermore, since California has a large number of undocumented Hispanics (estimated at 2,575,000 in 2005) (16), which the census may not count accurately, the incidence rates may be inflated due to an artificially low denominator. However, since the higher incidence rate among Hispanics was observed only for ALL and not other types of leukemia, this is not a likely explanation and the effect may be real. In addition, we observed a similar high IRR of ALL among adult Hispanics when the analysis was restricted to Los Angeles County and when it was expanded to include 16 U.S SEER registries (data not shown).
Since data on the population at risk by birthplace are unavailable, we used all other Hispanic cancer patients and all other Hispanic leukemia patients as comparison groups. This type of case-case analysis can produce biased results depending on the distribution of the exposure in the comparison group. However, the consistency of the association across the two comparisons suggests a true birthplace effect. Finally, we also confirmed the bimodal incidence peak for ALL reported by others (1213). A similar bimodal incidence peak was not seen in the UK (17). The persistent higher incidence rate among Hispanics throughout life suggests that the disease may have etiological factors in common when diagnosed at different ages.
There is evidence of both genetic and environmental risk factors in the etiology of childhood ALL, including en utero folate deficiency, parental exposure to radiation and pesticides, infections and certain genotypes involved in oxidative stress and folate metabolism (18). Latency from prenatal exposures up to 10 years has been suggested (18) but the period is unlikely to extend through adulthood. Chemotherapy is associated with an increased relative risk of AML, but not ALL, in adulthood (19). Little has been published on risk factors for adult ALL but one group of investigators have reported evidence of a protective association between variants of the folate metabolizing gene methylenetetrahydrofolate reductase (MTHRF) and adult ALL (20, 21), however Hispanics were not included in these studies based primarily on European-origin populations. A recent NHANES survey reported a higher prevalence of the protective C677T MTHRF genotype but a lower prevalence of the protective A1298C genotype in Mexican-Americans compared to non-Hispanic Whites (22), which offers a possible line of inquiry. One of the purposes of descriptive epidemiological studies is to identify associations between demographic factors and disease which can be used for hypothesis generation. This study raises an important observation worthy of further follow-up.
Presented as part of an oral presentation in the Epidemiology Group Minisymposium: “Higher incidence of acute lymphoblastic leukemia among adult Hispanics in the United States”, at the 98th annual meeting of the American Association of Research on Cancer (AACR), Los Angeles, California, April 16th, 2007.
This project has been funded in whole or in part with Federal funds from the National Cancer Institute, National Institutes of Health, Department of Health and Human Services, under Contract No. N01-PC-35139. The collection of cancer incidence data used in this publication was supported by the California Department of Health Services as part of the statewide cancer reporting program mandated by California Health and Safety Code Section 103885. The ideas and opinions expressed herein are those of the author, and no endorsement by the State of California, Department of Health Services is intended or should be inferred. This publication was made possible by grant number 1U58DP000807-01from the Centers for Disease Control and Prevention. Its contents are solely the responsibility of the authors and do not necessarily represent the official views of the federal government.
Authorship: Contribution: Sheeja T. Pullarkat contributed to the study design, analyzed data and wrote the manuscript. Kathleen Danley performed statistical analysis and developed the tables. Leslie Bernstein contributed to the study design, analyzed data, and critically reviewed the manuscript. Russell K. Brynes provided the concept, analyzed the data, and critically reviewed the manuscript. Wendy Cozen contributed to the study design, analyzed data and co-wrote the manuscript. Russell K Brynes and Wendy Cozen contributed equally to this study.
Conflict of interest disclosure: The authors declare no competing financial interests.
1Surveillance Research Program, NCI, http://seer.cancer.gov/seerstat