Most studies exploring ethnic/racial disparities in nonsmall cell lung cancer (NSCLC) compare black patients with whites. Currently, the effect of Hispanic ethnicity on the overall survival of NSCLC is poorly understood. Therefore, the authors carried out a large-scale, population-based analysis using the Surveillance, Epidemiology, and End Results (SEER) data base to determine the impact of Hispanic ethnicity the survival of patients with NSCLC.
The authors identified 172,398 adult patients with pathologically confirmed NSCLC from the SEER data base who were diagnosed between 1988 and 2007. A multivariate Cox proportional hazards regression analysis was used to determine the impact of race/ethnicity on overall survival. Pair-wise comparisons were used to determine whether Hispanic ethnicity influenced NSCLC histology or stage at diagnosis.
Compared with non-Hispanic white patients, Hispanic white patients had a statistically significant better overall survival (hazard ratio [HR], 0.85; 95% confidence interval [CI], 0.83-0.87), and black patients had worse survival (HR, 1.091; 95% CI, 1.072-1.109). Within the bronchioalveolar carcinoma (BAC) subtype, Hispanic-white patients tend to be over represented (8.1% Hispanic whites vs 5.5% non-Hispanic whites vs 3.7% blacks; P < .001).
Lung cancer represents the leading cause of cancer-related death world wide; in the United States, 222,520 individuals are diagnosed, and 157,300 die annually from the disease.1 The disease burden of lung cancer is not shared equally in the US population, and there are distinct variations in incidence and overall survival among various racial/ethnic groups. For example, black patients with nonsmall cell lung cancer (NSCLC) are differentially afflicted; black patients have an increased incidence rate, a decreased overall survival rate, and lower rates of surgical resection.2, 3 Most previous studies that investigated racial/ethnic disparities in NSCLC examined differential outcomes in black patients and white patients, but few have focused specifically on Hispanics. According to the US census, Hispanics currently comprise 11% of the US population and are projected to grow to 23% of the population by 2050.
It has been demonstrated previously that Hispanic whites (HWs) have a lower incidence of lung cancer compared with both blacks and non-Hispanic whites (NHWs), despite a socioeconomic status that would predict otherwise.1, 4 However, the impact of Hispanic ethnicity on the survival of patients with NSCLC is less clear: Some studies have suggested increased mortality (compared with NHWs), and others have suggested improved survival.1, 5, 6 Moreover, although others have identified differences in survival, many have failed to adjust for confounding variables, and few have identified plausible mechanisms.
Therefore, we conducted the current large-scale, population-based analysis using the Surveillance, Epidemiology, and End Results (SEER) database to examine the effect of Hispanic ethnicity on the survival of patients with NSCLC. In addition, we sought to determine whether any observed survival disparities could be attributed to differences in histology or stage at diagnosis.
MATERIALS AND METHODS
Data were obtained retrospectively from the National Cancer Institute's SEER Program using the 17-Registry 1973 to 2007 data set (November 2009 submission, released in April 2010). The SEER Program is the only comprehensive source of population-based information in the United States that includes stage of cancer at the time of diagnosis, first course of treatment, and data regarding patient survival. We assembled a cohort of 172,398 patients aged ≥21 years with pathologically confirmed NSCLC diagnosed between 1988 and 2007. Additional covariates included patient age at diagnosis, sex, stage at diagnosis, and histology (adenocarcinoma [AdenoCA], large cell carcinoma [LCC], squamous cell carcinoma [SqCC], bronchioalveolar carcinoma [BAC], adenocarcinoma with mixed subtype [AMST], and carcinoma not otherwise specified [NOS]). To investigate Hispanic ethnicity, patients were classified as NHW, HW, or black. Country of birth was categorized further to specify whether HW patients were born in the United States, South America, Central America, or the Caribbean (Cuba/Dominican Republic).
Our primary endpoint was overall survival. Proportional hazards models were used to examine the adjusted association of race/ethnicity and other potential covariates with overall survival. Survival was calculated from the date of diagnosis to the date of death. Adjusted models also included age, sex, stage at diagnosis, and histology. Pair-wise comparisons were used to analyze the prevalence of race/ethnicity with respect to age, histology, and stage at diagnosis by comparing NHW patients with black patients and HW patients. A second pair-wise comparison was performed comparing US-born HWs with different cohorts of non-US-born HWs. Normalized histologic frequency (NHF) for each race/ethnicity is defined as the frequency of presenting with a particular NSCLC histology (ie, AdenoCA, LCC, etc) normalized to NHWs. Therefore, NHWs served as the reference group and had an NHF value of 1.00 for each histologic subtype. For example, the racial/ethnic NHF for BAC is diagrammed NHFHW = (BAC%HW/BAC%NHW), NHFBlack = (BAC%Black/BAC%NHW), and NHFNHW = (BAC%NHW/BAC%NHW). All data were analyzed using the SAS statistical software package (version 9.1; SAS Institute, Cary, NC).
We selected 172,398 adult patients who had pathologically confirmed NSCLC from the SEER database. NHWs accounted for 79.1% of all patients, blacks accounted for 10.3%, and HWs accounted for 10.9%. The median age at diagnosis was 67 years, and 42% of patients were women. The most common histological subtype was AdenoCA (35.4%), followed by SqCC (29.2%), NOS (17.3%), LCC (6.8%), BAC (5.6%), and AMST (5.6%). Compared with NHW patients, black patients and HW patients were diagnosed at a younger age and with a greater incidence of advanced disease. Significant trends in histology also existed, with distinct histologic incidences observed between black patients and HW patients compared with NHW patients. The complete analysis is provided in Table 1.
Table 1. Patient Characteristics (N = 172,398)
Abbreviations: AMST, adenocarcinoma with mixed subtype; BAC, bronchioalveolar carcinoma; HW, Hispanic white; NHW, non-Hispanic white; NOS, not otherwise specified.
Next, we investigated whether any variation existed among HW patients to determine whether differences in NSCLC variants existed between non-US-born HWs and US-born HW. Compared with US-born HW, patients with NSCLC who were born in Cuba/DR were more likely to be men, whereas those born in Central America were more likely to be women. The complete analysis is presented in Table 2. To simplify our survival analysis, we investigated HWs as either US-born or non-US-born (Mexico, Cuba/DR, Central America, and South America).
Table 2. Analysis of Hispanic White Patients Only (N = 18,206)
Abbreviations: AMST, adenocarcinoma with mixed subtype; BAC, bronchioalveolar carcinoma; DR, Dominican Republic; HW, Hispanic white; NHW, non-Hispanic white; NOS, not otherwise specified; US, United States.
Our initial analysis involved a Cox proportional hazards model of all patients within the cohort. Increasing age, men, advanced stage, and LCC histology were associated with worse survival. BAC histology was associated with a statistically significant improved survival. Black patients had worse survival (hazard ratio [HR], 1.091; 95% confidence interval [CI], 1.072-1.109) than NHW patients (HR, 1.00; reference category). HW patients, both US-born (HR, 0.848; 95% CI, 0.832-0.864) and non-US-born (HR, 0.872; 95% CI, 0.856-0.889), had significantly better survival than either NHW patients or black patients. The complete survival analysis is presented in Table 3.
Table 3. Multivariate Analysis of Overall Survival for All Patients and for Individual Subsets Based on Stage
Table 3 indicates that no survival differences were observed between US-born and non-US-born HW when our survival analysis was grouped into patients with stage I through IV disease (see Table 3, All Patients). Because stage IV disease is associated with an extraordinarily high risk of mortality for all patients, it is possible that including these patients in our analysis with other patients (stages I, II, and IIIA/IIIB) potentially may mask differences in survival that otherwise may not be apparent. Therefore, we performed a subset analysis of patients according to each individual stage in an effort to more precisely define the impact of Hispanic ethnicity on NSCLC survival. In our subset analysis, variables that were previously associated with increased mortality persisted (increasing age, men, NOS subtype, LCC subtype). Hispanic ethnicity continued to be associated with improved survival, although the significance was slightly less with advanced stage. It is noteworthy that no statistical difference in survival existed between US-born and non-US-born Hispanics. This finding suggests that immigration patterns do not explain the observed survival advantage for Hispanic patients with NSCLC (see Discussion, below).
The interpretation of how racial/ethnic histologic subtype disparities may impact survival can be difficult. Figure 1 provides a diagram of the association between histologic subtype, the HR associated with each subtype, and the frequency of histologic presentation by race/ethnicity (see Materials and Methods, above). We represent the mortality associated with each NSCLC subtype with a bar graph (using the HRs from Table 3). Below this, the racial/ethnic distribution of each NSCLC subtype is diagramed (using the percentages from Tables 1 and 2). The racial/ethnic distribution is represented as normalized values (NHF), in which NHWs served as the reference category. In summary, HWs tend to have an increased incidence of the NSCLC histologic subtypes that are associated with a significantly lower mortality risk.
In this study, we used the SEER database to examine a large cohort of patients with NSCLC, with a focus on Hispanic ethnicity, to determine whether racial/ethnic disparities exist in overall survival. Another objective of our analysis was to determine whether any differences in survival could be attributed to disease-associated variables. We observed that HWs with NSCLC have a decreased risk for overall mortality compared with NHWs and blacks. Regarding clinical variables that could account for the differences in survival, we observed that HWs were over-represented within the histology subtypes that were associated with better survival and under-represented within the histology subtypes that were associated with poor survival.
Prior studies examining the survival of HW patients with lung cancer have been limited. Data from the American Cancer Society suggest both lower incidence and lower mortality rates for HWs with NSCLC; however, these raw frequencies were not adjusted for confounders, such as NSCLC stage at presentation.1 Other studies examining Hispanic ethnicity and NSCLC survival produced mixed results. One group concluded that HWs had poorer survival compared with NHWs, but those authors limited their analysis to only patients with stage I disease.6 Another group similarly reported an increased risk of death for HWs; however, that study did not differentiate between small cell and nonsmall cell lung cancers.5
Although our study was performed in a large cohort and did control for potential confounders, such as stage at diagnosis, treatment, and histology, several limitations related to the use of the SEER database should be noted. Although the database has the benefit of providing large numbers, it does not provide information about comorbidities, smoking status, occupational exposures, socioeconomic factors, or access to health care resources. In addition, the SEER database provides no information about chemotherapy received and has only limited information regarding surgical or radiotherapeutic interventions.
It has been reported that HW patients have improved survival compared with NHW patients and black patients for several disease processes (cardiovascular disease, breast cancer, and prostate cancer). This finding was previously termed the “Hispanic paradox,”7 because HWs in the United States tend to have fewer resources and less access to care than NHWs and also tend to have a poverty rate similar to that of blacks.4 Although only 8.1% of NHWs live below the poverty line, 22.6% of HWs and 24.8% of blacks are indigent.4 Similarly, only 12% of NHWs aged <65 years lack health care coverage, but 35.1% of HWs and 18.6% of blacks aged <65 years are uninsured.4 Lower socioeconomic status and limited access to health care have been identified as important contributors to the increased risk of death observed among blacks patients with NSCLC.2 In fact, studies that adjust for access to health care have reported similar survival rates for blacks and NHWs.2, 8-10
Several explanations for the “Hispanic paradox” have been proposed and may apply to our findings, including increased family support, lower smoking rates, an increased propensity for healthy individuals to migrate to the United States, and a tendency for sick immigrants to return to their country of birth.7, 11 HWs in the United States have lower rates of cigarette smoking than NHWs and blacks, which likely contributes to the lower incidence of lung cancer in this population.1, 4, 12, 13 In addition, smokers are at increased risk for SqCC and LCC, which are the tumor types associated with a poor prognosis. Conversely, nonsmokers are more likely to develop subtypes that carry a favorable prognosis, like AdenoCA and BAC, subtypes that were identified more frequently in HW patients than in NHW patients and black patients.14-16 Although smoking trends likely contributed to the survival advantage observed in our Hispanic cohort, studies that compared HW and NHW smokers reported a decreased incidence of lung cancer among HW smokers, suggesting that other (perhaps genetic) factors also may be important.17
Among the explanations for the Hispanic paradox, 2 potentially artifactual explanations do not explain our observations. The first potential effect is known as the “healthy immigrant” effect, which proposes that there is a selective pressure on the foreign Hispanic population for healthier and fitter individuals to immigrate to the United States.11 The other potential explanation is the so-called “salmon bias,” which suggests that ill immigrants tend to return to their country of birth to live out their lives.11 Consequently, their mortality status would not by captured by SEER. It was because of these 2 possibilities that we specifically analyzed both US-born and non-US-born Hispanic patients with NSCLC. If immigration patterns were the reason for the observed Hispanic survival benefit, then immigrant Hispanics (non-US-born) would have a greater survival advantage than nonimmigrant Hispanics (US-born). According to our multivariate analysis, US-born Hispanics and non-US-born Hispanics had similar survival trends. Thus, immigration patterns likely do not explain the observed Hispanic survival advantage.
A significant finding in our study is the increased prevalence of BAC among the HW population. The results of our analysis suggest that different molecular phenotypes of NSCLC may be the product of an interaction between genetic and environmental factors that could be related to ethnicity. In previous studies, mutations in the epidermal growth factor receptor gene EGFR have been associated with AdenoCA and with an improved prognosis, and such mutations reportedly occur more frequently in nonsmokers, women, and, notably, Asians.18 Other biomarkers also have been associated with racial and/or ethnic differences in oncogenesis and treatment outcome. For example, mutations in the breast cancer susceptibility gene (BRCA) have been associated with breast cancer among women of Ashkenazi Jewish heritage19; and recent studies suggest a similar relationship for pancreatic cancer.20 By examining the racial/ethnic disparities within NSCLC, the potential exists to develop prevention/treatment strategies that will significantly alter the natural history of the disease.
In summary, we observed that HW patients with NSCLC had improved survival compared with NHW patients and black patients. Because HW patients had a higher prevalence of tumor subtypes with a favorable prognosis, further studies are warranted to explore the role of both environmental factors, such as cigarette smoke exposure, and genetic factors that may portend particularly favorable molecular phenotypes of NSCLC.