High Incidence of Gastric Cancer in El Salvador: A National Multisectorial Study during 2000 to 2014

Gastric cancer is the fourth leading cause of global cancer mortality and the leading infection-associated cancer. Globally, gastric cancer is one of the most common cancers, contributing to more than 1 million cases per year and 5.7% of all cancer diagnoses (1, 2). Gastric cancer has significant geographic variability (15–20 fold) with a high incidence in East Asia, Eastern Europe, and the mountainous regions of the Pacific littoral of Latin America (3–5). Immigrants from high incidence regions who emigrate to low incidence regions (North America, Western Europe) maintain the incidence rates and risk of their nations of origin (6–9). In the United States, gastric cancer represents a marked cancer disparity with two to three times the incidence rates among all non-White populations, including Latinos, and immigrant populations are a contributing factor (10–13).

The World Health Organization (WHO) reports that approximately 70% of the global cancer burden occurs in lower- and middle-income countries (LMIC; refs. 14–16). It is estimated that 30% to 50% of cancers in LMICs are preventable (14, 15). Risk factors for gastric cancer in Latin America are multifactorial and include a high prevalence of Helicobacter pylori chronic infection, host genotypes and responses, and dietary, behavioral, and environmental factors (17, 18).

The Central American Four (CA-4) is the geopolitical region of El Salvador, Honduras, Nicaragua, and Guatemala and is the largest LMIC region in the western hemisphere with a population of >44 million and with >6 million US immigrants. The CA-4 lacks population-based cancer registries (PBCR), although several are in process (19). The International Agency for Cancer Research (IARC) usually imputes data for the region. The GC burden in the CA-4 region is projected to increase by 80% by the year 2030 (1). Although clinicians and health professionals are aware that gastric cancer is a daily and significant challenge, historically, the lack of resources has precluded reliable data and PBCRs, in contrast to high-income countries/high- and middle-income countries (e.g., Costa Rica, Colombia, Chile; ref. 19).

El Salvador lacks reliable gastric cancer incidence data. Based on the limited information reported, the WHO identified gastric cancer as the second most common cancer in El Salvador, which has an insufficient healthcare workforce and infrastructure (7, 8). In 2015, El Salvador’s Ministry of Health (MINSAL) recognized gastric cancer as the leading cause of cancer hospital discharge diagnosis. IARC currently estimates the age-standardized incidence rate (ASIR) in El Salvador of 11.9 and 9.2 in men and women, respectively, based upon imputation methods (1, 20). The aim of this study is to estimate the ASIR of gastric cancer in El Salvador using all available endoscopy and pathology databases of the three sectors of the health system.

A unique collaboration was formed with the three sectors of the El Salvador hospitals and health system: public system (Ministry of Health, MINSAL), “Social Security” system for government employees (Instituto Salvadoreño de Seguridad Social, ISSS), and private gastroenterology groups through the national professional society (Sociedad de El Salvador de Gastroenterologia, AGEDES). A retrospective analysis of data from 16 different hospitals and centers in El Salvador was conducted to estimate the ASIR of gastric cancer. The hospitals and endoscopy facilities are located in three principal urban areas: San Salvador, Santa Ana, and San Miguel.

The Republic of El Salvador is located on the coast of the Pacific Ocean, occupying an area of 21,041 km². The estimated population for the year 2021 is 6.3 million inhabitants with an approximate demographic density of 323 inhabitants/km², which makes it the most densely populated country on the American continent. Its territory is organized into 14 departments, 262 districts, and 44 municipalities. San Salvador is the capital and the most populated city in the country; its metropolitan area includes 14 nearby districts and concentrates the political and economic activity of the republic. The cities of San Miguel and Santa Ana are other important centers of the country. Nearly 2.5 million Salvadorans live in the United States, the equivalent of one-third of the El Salvador population.

The population structure by sex and 5-year age groups for each calendar year was obtained from the Department of Census and Statistics in El Salvador (DIGESTYC). The estimates were based on the national population statistics and official population projections from 2005 to 2014 and the independent calculations from 2000 to 2004(https://www.bcr.gob.sv/documental/Inicio/apartado/10). Eighteen 5-year age groups were considered, using the categories (e.g., 15–19, 20–24, etc.).

Men and women of any age, residents in San Salvador, with a histologically confirmed diagnosis of incident malignant tumor of the stomach were registered in the study. Included patients had histologic and endoscopic (or surgical) data, per national standards. Given the retrospective nature of the study and general lack of electronic medical records, not all patients had available both endoscopy and pathology data (e.g., one public hospital lost several years of histology records due to a natural disaster). Each case represented a unique case, as the databases at each site were reviewed for duplication. Early gastric cancer was defined as an invasive gastric cancer that is confined to the mucosa or submucosa, irrespective of lymph node metastasis (T1, any N). Proximal gastric cancer versus distal gastric cancer, and the Borrman classification system were used to characterize the endoscopic findings of gastric cancer (18). The Lauren classification system was used for the subtypes of gastric cancer with a focus on intestinal and diffuse (18). Malignant gastric outlet obstruction was classified as complete, incomplete, and absent, and was used as a surrogate for advanced disease as computed tomography was not generally available during the majority of the study period (18, 21).

Endoscopy and pathology databases were reviewed at each endoscopy center to estimate the incidence of gastric cancer between 2000 and 2014. In the 5-year period from 2010 to 2014, all 16 facilities were operational and therefore provided the most robust data. The initial survey of national endoscopy volumes confirmed that these facilities represented >95% of national volumes and capacity determined by a prestudy survey of endoscopic volumes conducted by AEDES. For patients without pathology records, the endoscopy reports were independently adjudicated (LRC, DRM), as previously described (19). In this time period, manual data extraction was required given the lack of electronic health records and databases. Trained personnel used a standardized form for endoscopy and pathology, with unique identifiers. The ICD-10 coding system (RRID:SCR_010351) was used when data were available. Individuals at least 15 years of age were included, the age of adulthood in Central America. This methodology was previously used in a population-based study in rural Western Honduras, which demonstrated an ASIR three times greater than that reported by the IARC GLOBOCAN estimation (19).

The prevalence of HER2 and EBV-encoded RNA (EBER) in gastric cancer was assessed in an independent cohort of patients in the period from 2014 to 2016. High-quality, formalin-fixed, paraffin-embedded samples were used for the IHC assessments. Tissues were considered to be HER2-positive when assessment revealed either an IHC score of 2+ or 3+. The Roche Ventana IHC platform was used (RRID:SCR_021254), performed in El Salvador. The number of subgroups for HER2 and EBER was 275 and 97, respectively. Fluorescence in situ hybridization was not performed due to the lack of in-country expertise and budget constraints.

Crude incidence rates of gastric cancer were calculated for males and females. Two independent statistical analyses were performed to calculate the crude rates and ASIRs for males and females for each year in the study period using SAS 9.4 (RRID:SCR_008567) and SEER*Stat software (RRID:SCR_003293), respectively. Incidence rates are expressed as annual average rates for a 5-year observation period; the denominator (total population or specific stratum) provides an estimate of person-years of observation. Individuals of the opposite sex were excluded from the denominator of sex-specific cancer rates. Crude rates, standardized rates, and age-specific rates are expressed per 100,000 person-years. Standardization by age was done with the direct method using the world standard population proposed by Segi (RRID:SCR_005422). To describe the incidence the acronyms ASIR-W (I) will be used (22).

The trend of the gastric incidence rates through 2000 to 2014 was examined using the annual percentage change in the rates (APC) with 95% confidence intervals (95% CI). It was used the weighted least squares method, implemented by default in the SEER ^∗Stat program (RRID:SCR_003293). To identify trend changes over time, the JoinPoint (RRID:SCR_018129) regression model was used. The APC represents the average annual percentage increase or decrease in cancer rates during a specific period. In describing the change, the terms “increase” or “decrease” were used when the APC was significantly different from zero (two-sided P-values <0.05); otherwise, the term “stable or flat” was used (23).

This study and data collection were approved by the Institutional Review Board of Vanderbilt University (DRM) and the El Salvador Ministry of Health (LRC).

The data generated in this study are available upon request from the corresponding author.

A total of 10,039 unique incident cases of gastric cancer were identified during the period 2000 to 2014 (Table 1). A total of 54.5% (n = 5,484) were male and 45.5% (n = 4,566) were female, with a mean age of 65 ± SD 14.5. Seventy-nine percent (n = 7,913) of the patients were 55 years or older and 9.4% were 45 years or less. Of the 10,039 patients with confirmed gastric cancer, detailed data were available for endoscopy (60.1%, n = 6,031), pathology (27.1%; n = 2,725), and both endoscopy and pathology 12.8% (n = 1,283) databases.

The endoscopic evaluation documented 83.4% (n = 2,560) of cancers were classified as Borrman type III or IV (49.7% and 33.7% for Borrman III and Bormann IV, respectively). The rates of early gastric cancer and Bormann type I were 2.3% (n = 71) and 5.6% (n = 171), respectively. The size of the tumors was 3.1 ± 1.7 cm (mean ± SD). Among the 4,008 cases with detailed histology data gastric cancer, 82% (n = 3,288) were classified as adenocarcinoma. The Lauren histological subtypes of adenocarcinomas were intestinal (55.6%, n = 2,220), diffuse (19.9%, n = 797), mixed (0.2%, n = 10), and indeterminate (6.5%, n = 261; Table 1). The most common other tumor types, comprising 18% (n = 720) of non-adenocarcinoma gastric cancers, were lymphoma (n = 120, 16.6%), gastrointestinal stromal tumor (n = 77, 10.7%), squamous cell carcinoma (n = 11, 1.5%), neuroendocrine tumors (n = 11, 1.5%), and not otherwise specified (NOS) tumor types (n = 496, 68.9%). Nearly one-quarter of the patients (23.2%) had an element of pyloric obstruction, either complete (13.2%, n = 960) or incomplete (10.0%, n = 733). Pyloric obstruction served as a surrogate for advanced-stage disease, given the lack of availability of CT imaging.

The 15-year average crude incidence rates among males and females were 12.3 and 9.2 per 100,000 persons, respectively, annually between 2000 and 2014 (Table 2). After direct standardization, the average 15-year ASIR for males during the study period was 15.8 (95% CI, 14.2–17.5) and 10.4 (95% CI, 9.2–11.6) for females per 100,000 person-years (Fig. 1).

The ASIR showed an increasing trend during the three 5-year periods. In 2000 to 2004, the ASIR were 12.0 and 7.9 for males and females, respectively. In 2005 to 2009, the ASIR were 15.3 and 10.1 for males and females, respectively. The 5-year ASIR during 2010 to 2014, with all facilities operational, was 18.9 (95% CI, 14.4–20.7) and 12.2 (95% CI, 10.9–13.5) per 100,000 person-years for males and females, respectively. The highest ASIR in males was observed in 2012 (21.1) and in females in 2008 (14.0) per 100,000 person-years [see Supplementary Tables S1 and S2 for the observed events, population times, crude rates, standard errors, and ASIR with 95% confidence interval (CI) per year by sex]. The mean annual increase in gastric cancer incidence rates was 4.4% in males; and 4.3% in females. The APC was significantly different from zero at the α = 0.005, (Fig. 2).

Overall, more than half of the cases were diagnosed in the public hospitals (53.7%). Nearly one-third (31%) of the cases were diagnosed in private facilities and 15.3% in the Social Security system hospitals (ISSS). The mean number of cases increased annually, with an average of 492 cases per year in the first 5-year period (2000–2004) to 811 in the last period (2010–2014). This trend is due in part to the increased number of endoscopy facilities, with a reduction in patients expiring without diagnosis, as well as an aging and growing population. Public hospitals accounted for nearly two-thirds (62.2%) of the cases in the first 5-year period, decreasing to 47.1% in the final 5 years, potentially related to the increase in ISSS governmental endoscopy facilities (Table 3). Data availability per center type can be found in Supplementary Table S3.

The prevalence of HER2 and EBER GC was assessed in an independent cohort of patients in the period to 2014 to 2016 (Table 4). The HER2 subgroup included 275 patients: 148 (53.8%) males and 127 (46.2%) males. A total of 195 (60%) patients were more than 60 years old. The HER2 status was reported negative “0” in 182/275 (66.2%), negative “+1” in 47/275 (17.1%), “+2” in 18/275 (6.5%), and “+3” in 28/275 (10.2%). The EBER subgroup included 97 cases: 46 (47.4%) males and 51 (52.6%) males. Seventy-eight (80.4%) patients were aged 60 years. The EBER status was negative in 79 (81.4%) patients, positive in 9 (9.3%), and unknown (unsatisfactory specimen) in 9 (9.3%). In summary, 16.7% were HER2 2+/3+ positive and 10.2% were EBER positive.

Gastric cancer is a leading cancer in northern Central America and one of the few LMIC areas globally with a high incidence. In addition, the CA-4 region and El Salvador in particular account for a large US immigrant population. This retrospective study of gastric cancer incidence for the 15-year period of 2000 to 2014 confirms the high incidence in males and females. The ASIRs (M, F) were 18.9 (95% CI, 14.4–20.7) and 12.2 (95% CI, 10.9–13.5) per 100,000 person-years, respectively, in the final 5-year reporting period with all of the current endoscopic facilities operational. In El Salvador, despite the global downward trend, there has been a significant increase in cases, with an APC of 4.4% for males and 4.3% for females. This is attributed to the growing and aging population at-risk as well as the expanded number of facilities responsible for reporting cases. The majority of patients presented with advanced-stage disease, as summarized by a Borrmann type III or IV tumor (83.4%), partial or complete pyloric obstruction (23.3%), and mean tumor size >3 cm.

In the substudy, during 2014 to 2016, 16.7% were HER2 2+/3+ positive and 10.2% were EBER positive. This is an expected prevalence and is instructive for therapy planning (e.g., HER2⁺ treatment resources for breast and GC, in the context of WHO essential medicines list). This provides invaluable baseline data for future epidemiology studies, as historically, the lack of resources has precluded reliable data and PBCRs. The identification of HER2-positive gastric cancers has therapeutic implications, as these patients may benefit from emerging HER2-targeted therapies (18). The documentation of gastric cancer HER2 prevalence provides synergistic data with breast cancer to ensure that HER2-targeted therapies are available in the public sector through the Ministries of Health in this LMIC region. EBV-associated gastric cancer accounts for 10% of cases globally and has a distinct pathogenesis and opportunity for targeted therapies (e.g., PD-L1 expression; refs. 24–26).

The incidence rates of gastric cancer were comparable to those reported in Central America and South America. Direct comparisons require caution owing to variations in the study design and population. A study in Costa Rica reported ASIRs of 15.7 and 10.2 per 100,000 person-years in males and females, respectively (27). In neighboring Western Honduras, a rural area, ASIRs of 30.8 and 13.9, respectively, for males and females, respectively, are reported for the period 2000 to 2009 (19). In South America, rates are variable, with the highest rates in the Andean nations (27). Globally, the incidence of gastric cancer is decreasing owing to screening programs (e.g., East Asia) and decreased H. pylori infection prevalence and tobacco use, although stable or increased rates are projected in some areas of Latin America, as in the case of El Salvador herein (28). Notably, in neighboring western Honduras, the H. pylori and CagA seroprevalence rates are >80% (29).

The high gastric cancer incidence rates have implications for national cancer control plans in El Salvador and Central America as well as for US immigrants and cancer control in Latino populations. The equivalent of more than one-third of the El Salvador population now lives in the United States, nearly 2.5 million (30). This represents the third largest Latino community in the United States. The majority are foreign-born, having emigrated since 1980 due to the civil war followed by endemic gang violence. Immigrants from high-incidence regions who migrate to low-incidence regions (North America, Western Europe) have been shown to maintain the incidence rates and risk of their nations of origin. Thus, as guidelines and policies emerge for gastric cancer screening and surveillance programs in the United States for high-risk groups, El Salvador would be a priority (31).

There are inherent limitations to a retrospective study of gastric cancer, which utilized endoscopy and pathology records and databases to identify incident cases. A rigorous protocol was established with trained personnel for the manual review of records. The endoscopy/pathology facilities included >95% of national volumes, based on the AEDES capacity survey, and only a handful of small endoscopy private clinics (e.g., single endoscopist) were excluded. The number of facilities nearly doubled by the end of the study, and we emphasize the discussion around the data from 2010 to 2014 period, where all centers were operational, being the most robust. The study findings likely represent an underestimation of the incidence, particularly in the early time period, given the lack of diagnostic facilities and the reality that many individuals would expire without a diagnosis (e.g., rural areas). Nonetheless, this provides the best available data, in the absence of national systems for cancer reporting and statistics, and a lack of PBCRs, as is usual in LMICs. The partnership among the three sectors (MINSAL, ISSS, and AGEDES) is unique.

The high gastric cancer incidence rates in El Salvador underscore the burden of disease in Mesoamerica, as well as in immigrant populations from the region. This has immediate implications for cancer control planning in El Salvador, Central America, and the US Latino populations. Notably, the equivalent of one-third of the El Salvador population lives in the United States.

D.R. Morgan reports grants from NCI during the conduct of the study; other support from Panbela Therapeutics, Thorne, Inc., American Molecular Laboratory, and Freenome, Inc. outside the submitted work; in addition, D.R. Morgan has a patent for disposible gastroscope, patent held by University issued. No disclosures were reported by the other authors.

L. Ruiz de Campos: Conceptualization, supervision, validation, investigation, visualization, methodology, writing–original draft, writing–review and editing. M. Valdez de Cuellar: Conceptualization, supervision, validation, investigation, visualization, methodology, writing–review and editing. D.A. Norwood: Software, formal analysis, investigation, visualization, methodology, writing–original draft, writing–review and editing. T.Y. Carrasco: Data curation, software, investigation, visualization, methodology, writing–original draft, writing–review and editing. E.E. Montalvan-Sanchez: Investigation, methodology, writing–original draft, writing–review and editing. M.-V. Rodriguez-Funez: Data curation, supervision, visualization, methodology, writing–review and editing. T. Beasley: Data curation, software, formal analysis, validation, investigation, visualization, methodology, writing–original draft, writing–review and editing. R.L. Dominguez: Conceptualization, validation, visualization, methodology, writing–review and editing. L.E. Bravo: Formal analysis, validation, visualization, methodology, writing–review and editing. D.R. Morgan: Conceptualization, validation, investigation, visualization, methodology, writing–original draft, writing–review and editing.

We'd like to thank the Ministerio de Salud de El Salvador (MINSAL), Instituto Salvadoreño de Seguridad Social (ISSS), and La Asociación de Gastroenterología y Endoscopía Digestiva de El Salvador (AEGDES). We recognize Dr. Balmore Cruz (ISSS) for his important contributions to this study, who passed away in the early days of the pandemic. We recognize the First Cancer Bioinformatics Congress for the Central America Four region (2014) as a catalyst for this study and the Cancer Epidemiology Unit in the El Salvador Ministry of Health (MINSAL). US NCI (DRM, P01 CA028842) and the NCI Center for Global Health (DRM, P30CA068485, PAR-15-155, HHSN 261200800001E).