Cancer Mortality in Latino Populations by Birthplace and Generation: A Complex Analysis

In this issue of the Journal, Chen and colleague’s (1) important research addresses cancer mortality patterns among the Latino population, a sizable yet often overlooked, growing demographic in the United States. Using data from one of the few prospective cohort studies that include a significant number of Latino individuals, the multi-ethnic cohort study (MEC), they examine differences by birthplace and generation status specifically for Mexican Latino populations taking into account uniquely available data on cancer risk factors such as smoking, parity, body mass index (BMI), alcohol, and physical exercise.

In tandem with the exponential growth of US racial and ethnic diversity following the Immigration and Naturalization Act of 1965, researchers have been interested in examining the changing patterns of cancer risk among immigrants and successive generations. Many have proposed that acculturation, the adoption by immigrant groups of practices, values, and lifestyles of the host population (2,3), drives changes in cancer risk (4-8). However, because of data limitations, little direct scrutiny of this theorized effect with Mexican Latino populations has been possible until now.

Comparative studies of migration and cancer among Latino populations are complex and can generally be studied at 3 distinct junctures. First, there are differences between foreign-born immigrant groups and their countries of origin. Second, there are differences between foreign-born immigrant groups and US-born groups of the same ancestry. Third, there may be differences between successive generations of US-born groups, no longer immigrants, but sharing the same ancestry, such as Mexican Americans. All 3 transitions are different and impacted by different factors such as geography, birthplace, environment, and lifestyles. However, the analysis of these differences on a population basis is quite difficult: data on birthplace and generation are either incomplete or absent from cancer registry data. Consequently, incidence studies on this topic have been few and far between (5,9,10).

Alternatively, mortality data are highly complete, at least on birthplace, and have enabled some comprehensive population-based studies that examine patterns for all major cancer sites. Results from these show that at the first juncture, immigrant groups in the United States show higher cancer mortality than their counterparts in their country of origin for most of the common cancers including lung, colorectal, and breast; however, these increases are offset by substantial reductions in stomach, cervical, and prostate cancers (11). At the second juncture, when compared with their foreign-born counterparts living in the same state or region, US-born Latino populations show substantially higher mortality rates for the majority of cancers, especially colorectal, kidney, and liver among US-born males and breast and lung cancers among US-born females (11,12). However, decreases in stomach, cervical, and prostate cancers are no longer observed.

Chen et al. (1) analyzed differences at the second and third juncture. When examining the role of birthplace (second juncture), their results largely confirm previous findings. Perhaps more intriguing was that adjusting for common cancer risk factors, available in MEC but lacking in mortality data, failed to provide meaningful insights into observed differences between these groups. Rates remained largely the same pre- and post-adjustment, with the exception of adjustment for BMI with kidney cancer. Thus, common hypothetical explanations for the lower cancer mortality among immigrants, usually focused on differences in risk factor prevalence (obesity for colorectal cancer, parity for breast cancer, etc.) could not be confirmed. Innovatively, utilizing cohort variables not typically present in population data, Chen et al. tackle the comparison at the third juncture, focusing exclusively on generation status among US-born Mexican-ancestry groups. However, no statistically significant differences between successive generations for any specific cancer site were found, somewhat contrary to prevailing acculturation theories. Moreover, the hypothesized approximation of Latino rates to the rates of the majority non-Hispanic White population as generation increases remains dubious except for breast (and that finding was not statistically significant). Concordantly, John and colleague’s (5) 2005 study of postmenopausal breast cancer incidence across generations of Latina women found similar results: statistically significant differences by birthplace not explained by risk factor data and no statistically significant differences in frequency when comparing second- and third-generation Latina women. In fact, looking at not only breast but all cancer sites combined, rather than an observed similarity to non-Hispanic Whites as acculturation would suggest, US-born Mexicans’ specific patterns, including excessive rates of liver and kidney mortality (13-15), much more closely resemble patterns seen among American Indian groups, suggesting not only possible genetic commonalities but also similar pathways via the social determinants of health.

Of course, MEC data are not perfect; risk factors are measured once at baseline, and the number of deaths accrued per cancer is limited. Moreover, mortality rates encompass not only determinants of cancer incidence, including genetics and cancer risk factors such as smoking and obesity, but also cancer survival, like access to and uptake of cancer screenings and cancer treatments, which are known concerns among Mexican immigrants. Nonetheless, what is evident from this and others’ studies is that US-born Latino populations have statistically significantly higher cancer mortality than their foreign-born counterparts (11,12,16,17). In particular, because US-born Mexicans comprise the vast majority of the 22 million US-born Latino population (18), they clearly constitute an important and unique demographic. Aggregating US-born with foreign-born Latino populations into 1 group in cancer studies provides misleading results; Latino populations should be studied by distinct birthplace groups whenever possible. Although by itself, birthplace seems a meaningless characteristic, on a population basis, birthplace serves as a proxy for many factors, including immigration status, tied to access to health care, as well as environmental influences and lifestyle factors during the formative years. Although the birthplace variable is by no means perfect, as the foreign-born Latino population includes the sizable so-called 1.5 generation (19) brought into the United States as children (20%-25% of all foreign-born Latino population present in the United States as of 2020 are part of the 1.5 generation) (18) and their patterns will more closely resemble the US-born, it is an objective variable, more so than race and/or ethnicity, and it is obtainable. As such, it can help us better understand the epidemiology of cancer among the diverse populations of the United States.

Still, the novel usage of covariates by Chen et al. (1) did produce additional intriguing results. Lung cancer mortality estimates changed after adjustment for smoking years when comparing US-born Latino to non-Hispanic White populations, likely reflecting the high population attributable fraction of smoking in lung cancer. As would be expected, stratification by smoking status showed this effect only among smokers; among never smokers, lung cancer rates remained the same by both birthplace and subsequent generations. This demonstrates the added value obtainable by stratification in a data-rich study such as this. Accordingly, the existing MEC data present additional opportunities, such as studying breast cancer stratified by parity, colorectal cancer stratified by diet and/or BMI, and endometrial and ovarian cancers stratified by hormonal data. Special attention should be paid to liver cancer, currently the leading cause of cancer death among US-born Mexican males (13), and whose patterns among Latino groups follow marked etiological differences by sex, age, and country of origin (20). Although data on hepatitis B and C were not available in this MEC analysis, sex-specific strata by 2 important causes of liver cancer, nonalcoholic fatty liver disease (strongly linked to obesity) (21), and alcohol consumption, could further clarify the complex patterns of this malignancy among the Latino population.

Overall, Chen and colleagues’ study (1) highlights how little we know about cancer in the Latino population. It suggests we should be revisiting our theories on the effects of acculturation and finding more opportunities to assess them, possibly expanding from cohorts, limited in size, to larger and more generalizable population-based studies using cancer registries. Large prospective cohort studies are ideal to study specific epidemiological associations given the quality and availability of data on specific exposures; however, they are inevitably subjected to selection factors, which compromise the generalizability of estimates by race and/or ethnicity. For example, in this analysis, among Latina women aged 50-74 years, mortality rates are higher for lung than breast cancer, a reversal of what is seen in population-based studies (22). Cohort-style and population-based studies are part of a continuum with an increasing degree of generalizability, and researchers should seek to maximize the potential of both. Registries need to be enriched by investing in linkages, including with electronic medical records, to supplement risk factor data on a population basis. Combined, these efforts would facilitate finally approaching a true understanding of how and if geography, birthplace, and generation status impact cancer patterns. Only when we have a more complete understanding of the epidemiology of cancer in Latino populations can we tailor our cancer prevention and control efforts and reduce its impact.

Funding

This work was supported by the Bankhead Coley Research Program of the State of Florida, Grant Number 20B16. Research reported in this publication was also supported by the National Cancer Institute of the National Institutes of Health under Award Number P30CA240139.

Notes

Role of the funders: The funders had no role in the writing of this editorial or the decision to submit it for publication.

Disclosures: The author declares no conflicts of interest.

Disclaimer: The content is solely the responsibility of the author and does not necessarily represent the official views of the National Institutes of Health.

Author contributions: Writing, original draft—PSP. Writing, review and editing—PSP.

Acknowledgement: I wish to extend my special thanks to Dr. Karen Callahan.

Data Availability

No new data were generated or analyzed in support of this research.