Cancer Epidemiology in Hispanic Populations: An Analysis of Funded Observational Research at the National Cancer Institute

More than 62 million people self-identified as Hispanic/Latino (H/L) in the 2020 United States census, comprising ∼19% of the population (1). Geographically across the United States, the proportion of H/L in the population varies substantially, with the highest number in New Mexico (49% of state population), Texas (40%), California (39%), Arizona (32%), and Florida (26%). Among H/L subgroups, Mexican Americans account for the largest group (62%), followed by Puerto Ricans (10%), Salvadorans (4%), Cubans (4%), Dominicans (4%), and Guatemalans (3%; ref. 2). The majority of H/Ls were born in the United States, with 33% reporting as foreign-born (i.e., born outside the United States and its territories and neither parent was a U.S. citizen; ref. 1). H/Ls are one of the fastest growing ethnic subgroups in the United States, with estimates projecting this subgroup to comprise 28% of the U.S. population by 2060. A continued dominance of U.S.-born H/Ls compared with foreign-born H/Ls is projected, although sociopolitical influences could affect migration patterns. For example, the Venezuelan foreign-born population living in the United States grew by 313% from 75,000 in 2000 to 309,000 in 2017 (1). The changing landscape of H/L populations has important implications for U.S. cancer-related incidence and outcomes (3, 4).

H/Ls have lower incidence rates of the four most common cancers (breast, colorectal, lung, and prostate) diagnosed in the United States compared with the non-H/L White (NHW) population (5). Despite these lower incidence rates, cancer is the leading cause of death among H/Ls, representing 21% of all adult deaths (6). One contributing factor to the higher cancer mortality is that H/L patients tend to be diagnosed at more advanced stages (6) than NHW patients, although differences in stage at diagnosis do not fully explain the differences in mortality. Further, H/Ls have higher incidence rates for liver, stomach, cervical, and gallbladder cancers, suggesting the presence of specific risk factors, exposure patterns, and/or genetics affecting risk of these cancers among H/Ls (5). The cancer-related burden varies across different H/L subgroups, yet this variation is often masked when H/L subgroup data are aggregated (4). Identifying the unique risk factors that contribute to differences in cancer risk and mortality patterns within the H/L population is critical to reducing the cancer burden.

The current analysis sought to provide an overview of NCI-recently funded research focusing on cancer epidemiology in H/L populations by identifying and describing key characteristics of relevant grants and highlighting research gaps and available resources. Information derived from this analysis coupled with the discussions from the NCI-sponsored virtual workshop, “Cancer Epidemiology Research in Hispanic Populations,” held on September 14 to 16, 2021, will help to inform future research opportunities (https://events.cancer.gov/egrp/cehp).

This analysis examined NCI-funded H/L-relevant cancer epidemiology research grants (e.g., research project grants and cooperative agreements) funded in fiscal years 2016 through 2021 for which NCI was either a primary or a secondary funder. Grants were identified using a key word search (Fig. 1) in an internal NIH tool to search applications and awards, Query/View/Report (last search date, September 2021).

To be eligible for inclusion into the final portfolio analysis dataset, a grant had to be an observational cancer epidemiology study that included or focused on a H/L population. These inclusion criteria were assessed using the Research, Condition, and Disease Category (RCDC) codes (https://report.nih.gov/funding/categorical-spending/rcdc) and a review of the Specific Aims of each grant. Relevance to H/L populations was determined on the basis of two criteria: (i) the specific aim(s) of the grant focused specifically on H/L populations or (ii) if not focused specifically on H/L populations, the grant contained at least one specific aim comparing racial/ethnic groups in which one of the comparator groups was H/L. Of the 450 eligible grants identified in the initial search, 211 had not been assigned a RCDC ‘epidemiology code’ and were thus excluded; an additional 40 grants were excluded due to the lack of inclusion and/or focus on H/L populations in the Specific Aims of the grant (Fig. 1). Thus, the final analytical dataset included 41 grants. To verify the reproducibility of the generated portfolio analysis, a separate but complementary search was conducted using the NIH RePORTER (https://reporter.nih.gov), a publicly accessible electronic tool that allows users to search a repository of NIH-funded research projects. This search yielded similar results of relevant grants.

Coding procedures were consistent with prior NIH-wide portfolio analyses (7, 8). After abstracting general grant characteristics [e.g., NIH Institute, funding opportunity announcement (FOA), funding mechanism], the 41 grants were reviewed and coded independently by a pair of reviewers [(T.K. Lam and L. Gallicchio) or (T.K. Lam and J.W. Elena)] using a codebook developed to facilitate consistent coding. Twelve percent of the grants were coded independently by all three reviewers (T.K. Lam, J.W. Elena, and L. Gallicchio) for quality control. Each grant was coded for the following variables: study design(s), area(s) on the cancer control continuum addressed in the grant, exposure(s), risk factors(s) and/or prognostic factor(s) examined, and cancer type(s). If reported in the grant, we abstracted additional information related to H/L populations (e.g., heritage and proposed sample size) and the types of biospecimens collected. Because biospecimen collection (e.g., blood, plasma, serum, tissue, DNA, urine, or fecal samples) varied widely by grants, we simplified the coding to “yes” or “no” when reported in this paper and did not specify the type(s) of biospecimen or assess how biospecimens were used in the study.

The data used in this study are publicly available in NIH RePORTER (https://reporter.nih.gov), a publicly accessible electronic tool that allows users to search a repository of NIH-funded research projects.

There were 41 cancer epidemiology grants identified that met the eligibility criteria outlined above, of which 32% focused solely on H/L populations and 68% included a specific aim that reported H/Ls specific-results, often as part of a comparison across racial/ethnic groups (Table 1). Approximately half (54%) of the grants were submitted in response to a topic-specific (i.e., solicited) FOA. NCI served as the primary funder for the majority of the grants (85%), with the remaining grants primarily funded by other NIH institutes (15%), including the National Institute of Minority Health and Health Disparities (NIMHD; 5%). It is important to note that grants funded entirely by other NIH institutes without NCI involvement were not included in these analyses. Over half of the grants were funded through the research project grant R01 mechanism (59%) and four (10%) were funded through a cooperative agreement (U) mechanism. Training grants (K mechanism) with research aims relevant to H/L populations comprised 10% of the total identified grants. Grants focused on areas across the cancer control continuum, with 18 grants (44%) focused on cancer risk/etiology, 20 (49%) focused on survivorship, and fewer grants focused on cancer detection (22%), treatment (22%), and diagnosis (5%).

Table 2 presents select characteristics of the 41 cancer epidemiology grants represented. More than half (61%) of the studies did not specify H/L participant heritage information beyond identification as H/L. Of those that did, the most reported heritages of the participants being studied were Mexican American (n = 15) and Puerto Rican (n = 4). Approximately half of the grants collected biospecimens, with most of those studies collecting blood samples. Most grants (54%) had projected sample sizes of less than 1,000 H/L participants. Twenty-four percent of the grants were larger studies with 1,000 to 10,000 participants, and 15% had studies with more than 10,000 participants.

Table 3 presents key characteristics of the studies included in the eligible grants, stratified by those focused on etiology (n = 18) or survivorship (n = 20); 2 grants included both etiology- and survivorship-focused projects in their proposed aims. Most grants categorized as etiology or survivorship used either a prospective or retrospective cohort study design (cancer etiology, 50%; survivorship, 62%). Cancer etiology grants used several study designs, including cohort (50%), case–control (15%), and nested case–control designs (7%), while survivorship grants primarily used the cohort design (75%). The median number of total participants for cohort studies was 4,925 (863 H/Ls), case–control studies had median participant counts of 650 for total population as well as H/Ls participants, and the median participant size for all other study designs combined [e.g., genome-wide association studies (GWAS), cross-sectional] was 1,208 for all participants (833 H/L). With regards to exposure(s) of interest, most grants investigated some type of information from biological specimens (e.g., genetics, biomarkers, epigenetics) or at least one lifestyle factor (e.g., smoking, obesity, diet) as an exposure of interest. However, all of the etiology grants included biological factors, whereas only half of the survivorship grants did. Half of the survivorship grants included at least one healthcare-related, clinical, and/or sociocultural and structural factor; conversely, each of these categories of factors was examined in less than 25% of the etiology grants. Few etiology and survivorship grants examined an environmental factor as an exposure of interest (10%). While 10% of grants examined multiple cancer types, the most commonly investigated cancer type for survivorship grants was breast (25%) whereas for etiology grants, colorectal (22%) and liver (22%) cancers were the top investigated cancer types.

This portfolio analysis characterized the 41 NCI-supported cancer epidemiology grants in H/L populations funded in fiscal years 2016 to 2021. The grants spanned the cancer control continuum from etiology to survivorship, with the notable exception of cancer prevention, an understudied and important area for this population. It should be noted that prevention grants may have not been captured in our analysis, as these are often intervention studies and thus not coded as “epidemiology” grants at NCI, which was an inclusion criterion of our study. More than half of the grants identified were submitted in response to a topic-specific FOA, which underscores the impact of initiatives designed to boost representation of diverse populations in research (e.g., NIH's UNITE initiative, https://www.nih.gov/ending-structural-racism/unite). While grants of any topic may be submitted to the parent NCI announcement (PA-20–185), topic-specific FOAs highlight opportunities in priority areas.

The 41 grants included in this portfolio analysis reflect current funding on cancer epidemiology research in H/L populations. The projects described in the analyses represent a variety of study designs, with cohort studies being the most common. Although research in H/L populations is growing, there are few longitudinal studies with detailed ethnic subgroup information and sizeable H/L populations with the statistical power to allow for further subgroup analysis (e.g., stratified by country of origin or type of cancer). One notable exception is the Multiethnic Cohort Study (MEC), which recruited H/L participants from the greater Los Angeles area and served as the source of data for approximately a third of the grants in this analysis. Leveraging the MEC resources represents an efficient use of existing data, yet additional studies with broader geographical representation would allow for the examination of exposures not captured in MEC and facilitate the replication of previous findings in new populations.

Most of the identified grants included a variety of risk factor information on lifestyle, sociocultural, and structural factors, healthcare-related factors, and clinical information, as well as information from biological specimens. As expected, studies of cancer survivors included more healthcare and clinical factors related to treatment decisions and cancer-related outcomes, while etiology studies focused more on lifestyle and other traditional risk factors. Although half of the survivorship studies reported sociocultural factors, only 11% of etiology studies did. This is critical information for cancer epidemiology research in H/L populations and thus represents a notable gap in the NCI portfolio for both etiology and survivorship studies. Socioeconomic status and other social determinants of health are established risk factors for cancer incidence and poor cancer survival outcomes (9, 10) and thus, should be captured in studies designed to reduce disparities and inform culturally appropriate interventions. In addition, few studies included environmental-related risk factors such as water and air pollution, infectious agents, and toxic chemicals. H/L populations, especially immigrants to the United States, are often exposed to environmental risk factors throughout their lives that are different or may be less prevalent (e.g., aflatoxin, infectious agents) than those typically studied in the United States, which may affect their risk of certain cancers. For example, Mexican Americans and Mexican immigrants have higher mortality rates than Mexicans for many cancers, including endometrium, colorectal, pancreas, non-Hodgkin lymphoma, kidney, lung, and among men, liver, and esophageal cancers (11). Some of these cancers are obesity-related, which may explain in part these noted disparities, but there are most likely additional factors that have not been identified and adequately studied yet. There are also observed differences in cancer patterns between H/L and NHW populations, with H/Ls having a higher risk of infection-related cancers (stomach, liver, and cervix) and gallbladder cancers; however, incidence varies substantially by nativity, Hispanic origin group, and duration of U.S. residence, with rates in some groups approaching or surpassing those of NHWs, particularly among U.S.-born Hispanic individuals (6, 11, 12). As patterns of risk factors vary between NHW and H/L populations, so does the distribution of cancer, which underscores the importance of supporting research in H/Ls across less common and understudied cancer sites.

Limited information with regard to H/L identity was noted, with very few grants reporting data on heritage, place of birth, language spoken in the home, and other measures of acculturation. Given the heterogeneity of the H/L population, collecting these factors is important to ensure diverse ethnic representation in H/L studies and allowing subgroup analysis. This provides greater understanding of their contribution to cancer and may inform prevention and intervention strategies across different H/L populations. Currently, epidemiologic data are typically aggregated across H/L populations as a monolithic group, which may mask the nuanced factors among H/L subgroups in relation to cancer. This approach may lead to misclassification of data on race and ethnicity whereby single-race population estimates, derived from the original multiple-race categories, may lead to uncertainties in population estimates and obscure the differential relationships between varying risk factors/behaviors and cancer within subgroups comprise. It is important to note that each grant does not necessarily include a unique study population, as multiple grants used the same source of data (e.g., the MEC has been used for multiple ancillary studies).

There are some limitations to the analysis reported in this manuscript. The criteria used to determine eligible grants in cancer epidemiology may have missed grants that did not use the specific search terms; however, the analysis included a robust collection of commonly used terms. It does not cover cancer-related grants being funded at NIH without NCI involvement (i.e., NCI is not a primary or secondary funder) and those funded by non-NIH sources. Nevertheless, a previous analysis of NIH-funded research from 2008 to 2015 related to H/L populations across six health topics, including cancer, sufficiently captured the cancer-related grants funded at NIH during that time (13). Because we restricted our analyses to observational studies, interventions and clinical trials were not included in the current analysis. We also acknowledge additional important work done along with the basic to translational science continuum outside of epidemiology and across the cancer continuum (e.g., prevention, diagnosis, and treatment). Finally, because our analysis was restricted to information found in the specific aims, cancer epidemiology studies embedded in large cooperative grants (e.g., the Comprehensive Partnerships to Advance Cancer Health Equity U54 Program PAR 18–361) may have been missed.

The future direction of cancer epidemiology research needs to reflect the changing dynamics of the U.S. demography with a youthful and growing H/L population that is majority U.S.-born (65%). Studies with diverse, representative populations and well-characterized ethnicity are needed to better understand strategies to reduce the cancer burden. Moving forward, research that carefully measures the relevant biological, environmental, behavioral, psychologic, sociocultural, and clinical risk factors will be critical to better understanding the nuanced patterns that are influencing cancer risk, health outcomes in survivors, and mortality in the heterogenous H/L population.

K.E. Akif reports that within the last 36 months an ESOP (employee benefit) from her previous employer (Westat); however, no longer owns/has this ESOP and has not had this ESOP for over 12 months. No disclosures were reported by the other authors.

J.W. Elena: Conceptualization, formal analysis, visualization, methodology, writing–original draft, project administration, writing–review and editing. L. Gallicchio: Conceptualization, methodology, writing–original draft. C.A. Pottinger: Formal analysis, writing–original draft, writing–review and editing. K.E. Akif: Writing–review and editing. R. Hanisch: Writing–review and editing. A.E. Kennedy: Conceptualization, methodology, writing–original draft, writing–review and editing. G.Y. Lai: Formal analysis, writing–review and editing. S. Mahabir: Writing–review and editing. D.N. Martin: Conceptualization, writing–review and editing. S. Srinivasan: Conceptualization, writing–review and editing. C.T. Yu: Writing–review and editing. T.K. Lam: Conceptualization, formal analysis, methodology, writing–original draft, project administration, writing–review and editing.

All work was conducted as part of the authors’ work-related duties at NIH.

The publication costs of this article were defrayed in part by the payment of publication fees. Therefore, and solely to indicate this fact, this article is hereby marked “advertisement” in accordance with 18 USC section 1734.