BPG is committed to discovery and dissemination of knowledge
Retrospective Cohort Study Open Access
Copyright: ©Author(s) 2026. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution-NonCommercial (CC BY-NC 4.0) license. No commercial re-use. See permissions. Published by Baishideng Publishing Group Inc.
World J Gastrointest Oncol. Nov 15, 2026; 18(11): 123432
Published online Nov 15, 2026. doi: 10.4251/wjgo.123432
Foregut malignancy after discordant fecal immunochemical test: A cohort study at a United States safety-net hospital
Andrew John Cruz, Sandeep Gurram, Isabel Obias, Allen Luo, Gevork Salmastyan, Aravind Rajagopalan, Stefanie Bui, Nikhilesh Kumar, Erick Cruz Grave, Eric Yang, Amarachi Erondu, Theodor Griggs, Department of Medicine, University of Southern California/Los Angeles General Medical Center, Los Angeles, CA 90033, United States
Roxanna Sharifi, Chester Hui, Keck School of Medicine, Los Angeles, CA 90033, United States
ORCID number: Andrew John Cruz (0009-0006-2035-3024).
Author contributions: Cruz AJ designed the study, collected and analyzed the data, performed the statistical analysis, and drafted the manuscript; Gurram S, Obias I, Luo A, Salmastyan G, Rajagopalan A, Bui S, Sharifi R, Kumar N, Hui C, and Cruz Grave E performed chart review and data abstraction; Yang E performed chart review, data abstraction, and detailed malignancy case review; Erondu A contributed to data collection and critical revision of the manuscript; Griggs T conceived and supervised the study, provided mentorship, and critically revised the manuscript for important intellectual content; all authors have read and approved the final manuscript.
AI contribution statement: The authors did not use any AI tools in the preparation of this manuscript. All data analysis, writing, and revision were performed entirely by the authors.
Institutional review board statement: This study was approved by University of Southern California Institutional Review Board (No. HS-25-00684).
Informed consent statement: Informed consent was waived by the IRB given the retrospective, de-identified nature of the study.
Conflict-of-interest statement: All authors declare no conflict of interest in publishing the manuscript.
STROBE statement: The authors have read the STROBE Statement – checklist of items, and the manuscript was prepared and revised according to the STROBE Statement – checklist of items.
Data sharing statement: The dataset is available from the corresponding author upon reasonable request.
Corresponding author: Andrew John Cruz, MD, Department of Medicine, University of Southern California/Los Angeles General Medical Center, 2020 Zonal Avenue, Los Angeles, CA 90033, United States. andrewjohn.cruz@med.usc.edu
Received: May 19, 2026
Revised: July 2, 2026
Accepted: July 28, 2026
Published online: November 15, 2026
Processing time: 170 Days and 23.9 Hours

Abstract
BACKGROUND

Whether fecal immunochemical test-positive (FIT+) patients whose colonoscopy does not identify a bleeding source should undergo esophagogastroduodenoscopy (EGD) is unclear. Prior data report foregut cancer rates below 1% in population screening cohorts, but yield by colonoscopy concordance has not been directly compared, particularly in United States safety-net Hispanic populations.

AIM

To compare the yield of EGD between FIT+ patients with discordant (non-explanatory) and concordant (explanatory) colonoscopy at a safety-net hospital.

METHODS

We performed a retrospective cohort study of FIT+ adults aged ≥ 45 years who underwent colonoscopy (2019-2025) and EGD within one year at Los Angeles General Medical Center, a safety-net hospital. Colonoscopy was classified as concordant if it identified a lesion plausibly explaining a FIT+ result, including colorectal neoplasia, vascular lesions, or colitis. Foregut malignancy and other EGD findings were compared between groups. Firth penalized logistic regression (exploratory) adjusted for age, sex, anemia, foregut symptoms, nonsteroidal anti-inflammatory drug/aspirin use, and cirrhosis.

RESULTS

Of 379 FIT+ patients (172 discordant, 207 concordant), 11 foregut malignancies occurred in the discordant group (6.4% vs 0%; P < 0.001; Firth adjusted odds ratio = 26.14, 95%CI: 2.16-315.82). Premalignant lesions were more common in concordant patients (7.7% vs 2.3%, P = 0.02); other EGD findings did not differ significantly between groups. Foregut symptoms were equally prevalent in both groups (44.2% vs 38.2%, P = 0.25), indicating that symptom-driven referral does not account for the observed difference. Among the 11 malignancy patients, 0 were ever-smokers and All patients underwent EGD for standard clinical indications; the observed association reflects a clinically selected cohort rather than protocolized EGD for all discordant patients.

CONCLUSION

In a clinically selected safety-net cohort, all foregut malignancies occurred exclusively among discordant FIT+ patients despite similar symptom profiles between groups, suggesting that discordant colonoscopy may help stratify foregut malignancy risk in populations with elevated gastric cancer burden.

Key Words: Fecal immunochemical test; Esophagogastroduodenoscopy; Foregut malignancy; Gastric cancer; Safety-net hospital; Colonoscopy concordance

Core Tip: At a United States safety-net hospital, all 11 foregut malignancies in fecal immunochemical test-positive patients occurred in those whose colonoscopy did not explain the positive result (6.4% vs 0%), including patients without symptoms or anemia. This association persisted after multivariable adjustment (adjusted odds ratio = 26.14) and across multiple sensitivity analyses. Foregut symptoms were equally prevalent in both groups, and all patients underwent esophagogastroduodenoscopy for standard clinical indications rather than discordant fecal immunochemical test status alone. None of the malignancy patients were smokers or heavy drinkers. These hypothesis-generating findings warrant confirmation in multicenter prospective studies.



INTRODUCTION

Colorectal cancer screening with the fecal immunochemical test (FIT) has become the predominant stool-based strategy in the United States and worldwide. FIT detects human globin using monoclonal antibodies, and because globin from the upper gastrointestinal (GI) tract is largely degraded by intestinal proteases before reaching the colon, the test was designed to be specific for colorectal bleeding[1]. In practice, 40%-70% of colonoscopies after a FIT-positive (FIT+) do not identify advanced neoplasia[2,3].

Whether to interrogate the foregut following a negative colonoscopy for a FIT+ result remains a clinical conundrum. The American Gastroenterological Association’s 2000 position statement on occult GI bleeding (GIB) recommended against further evaluation of asymptomatic patients with a positive fecal occult blood test (FOBT+) and negative colonoscopy unless anemia develops[4]. Further, a 2010 systematic review concluded that foregut cancer rates following FOBT+ and negative colonoscopy did not justify routine esophagogastroduodenoscopy (EGD)[5]. These recommendations were based largely on guaiac FOBT data, but more recent FIT-specific studies have reinforced the same message; a Dutch population-based screening cohort of over 16000 FIT participants found only 0.14% were diagnosed with foregut cancer within three years[6], and a 2023 meta-analysis estimated the pooled prevalence of foregut cancer at 0.8% among FOBT+ and FIT+ patients[7]. Together, these data have not supported the practice of routinely pursuing EGD after a non-explanatory colonoscopy.

These prior studies are difficult to apply directly to United States safety-net populations. Most were conducted in population-based screening programs in Europe and Asia, where demographics, Helicobacter pylori (H. pylori) prevalence, and foregut malignancy incidence differ substantially from those of safety-net hospitals in the United States[6,8]. Hispanic individuals in the United States carry approximately twice the gastric cancer incidence of non-Hispanic white individuals and present with more advanced disease at younger ages[9,10]. Prior studies also predate widespread adoption of FIT and rely on guaiac-based FOBT tests that are biologically and analytically distinct[1]. Nearly all studies identified report overall EGD yield without comparing outcomes to a concordant control group (patients whose colonoscopy did identify an explanatory lesion). Without this comparator, it is difficult to determine whether EGD findings in the discordant group reflect true enrichment for foregut pathology or simply the background prevalence of foregut findings.

Hemorrhoids are a common but contested classification. Clinicians frequently attribute FIT+ results to hemorrhoids when no other colonic lesion is found. However, a 2018 systematic review and meta-analysis found no significant association between hemorrhoids and false-FIT+ (pooled relative risk = 0.99, 95%CI: 0.93-1.97)[11], and the largest United States study to date found no difference in FIT specificity by hemorrhoid status[12]. FIT detects intact globin mixed into stool during colonic transit, whereas hemorrhoidal bleeding typically deposits blood on the stool surface after formation, limiting detection. If hemorrhoids do not plausibly explain FIT positivity, patients whose only colonoscopy finding is hemorrhoids should be considered discordant, a classification with implications for the decision to pursue EGD.

We conducted a retrospective cohort study at Los Angeles General Medical Center, a large United States safety-net hospital to compare the yield of EGD between FIT+ patients whose colonoscopy was discordant (non-explanatory) and those whose colonoscopy was concordant (explanatory).

MATERIALS AND METHODS
Study design and setting

This was a retrospective cohort study at Los Angeles General Medical Center, a 600-bed county hospital and the largest safety-net facility in Los Angeles County, serving a predominantly Hispanic and publicly insured population. The study was approved by the Institutional Review Board (No. HS-25-00684); informed consent was waived given the retrospective de-identified design.

Study population

We identified adults aged 45 years or older who underwent colonoscopy for FIT+ results between January 1, 2019 and September 30, 2025. Potentially eligible colonoscopies were identified by searching the institutional endoscopy database (Cerner GGastro, Oracle Cerner, Kansas City, MO, United States) for indicated terms including FIT, fecal, immunochemical, testing, anemia, melena, and hematochezia. A confirmed FIT+ result was then verified on chart review of the electronic medical record before cohort assignment. Patients whose colonoscopy indication did not correspond to a confirmed FIT+ were excluded. Patients with guaiac-based FOBT rather than FIT were excluded because guaiac FOBT detects heme peroxidase activity from any source and has different specificity for lower GIB. Patients were also required to have undergone EGD within one year of the index colonoscopy. The age threshold of 45 was selected to align with current United States Preventive Services Task Force and American College of Gastroenterology colorectal cancer screening recommendations[13,14]. Additional exclusion criteria were incomplete endoscopic records, prior GI malignancy, and known inflammatory bowel disease, defined as a documented diagnosis of Crohn’s disease or ulcerative colitis. Of 1610 records screened, 379 FIT+ patients met all criteria (Figure 1). The most common reason for exclusion was absence of a confirmed FIT+ result prior to index colonoscopy (n = 1152). Other exclusions included incomplete endoscopic records (n = 45), guaiac FOBT rather than FIT (n = 14), no EGD within one year (n = 13), aborted or poor-quality examination (n = 4), prior GI malignancy (n = 2), and known inflammatory bowel disease (n = 1). This search strategy may have enriched the cohort for patients undergoing GIB evaluation beyond routine colorectal cancer screening, which should be considered when interpreting the reported EGD yield and the generalizability of these findings to broader FIT+ screening populations.

Figure 1
Figure 1 Cohort derivation flow diagram. Of 1610 records screened, 379 fecal immunochemical test (FIT)-positive patients met inclusion criteria and were classified as discordant (n = 172) or concordant (n = 207) based on colonoscopy findings. The most common reason for exclusion was absence of a confirmed positive FIT on chart review (n = 1152). Discordant patients included those with no explanatory finding or hemorrhoids with additional non-concordant findings (n = 115) and those with hemorrhoids only (n = 57), classified as discordant based on meta-analytic evidence that hemorrhoids do not significantly contribute to false-positive FIT results. Among concordant patients, colorectal neoplasia accounted for 183 (88.4%) and an alternative explanatory lesion for 24 (11.6%). FIT+: Fecal immunochemical test-positive; FOBT: Fecal occult blood test; EGD: Esophagogastroduodenoscopy; GI: Gastrointestinal; IBD: Inflammatory bowel disease; aOR: Adjusted odds ratio.
Colonoscopy concordance classification

Each colonoscopy was classified as concordant or discordant based on whether the findings plausibly explained the FIT+ result. Concordant colonoscopy required identification of colorectal neoplasia [any adenomatous polyp, sessile serrated lesion, traditional serrated adenoma, or colorectal malignancy including carcinoma, neuroendocrine tumor (NET), lymphoma, or GI stromal tumor] or an alternative explanatory colonic lesion (angioectasia, portal hypertensive colopathy, colonic or rectal varices, or colitis). Hemorrhoids were classified as discordant rather than concordant based on meta-analytic evidence that they do not significantly contribute to false-FIT+ results[11,12,15]. Uncomplicated diverticulosis alone was not considered explanatory. Hyperplastic polyps were recorded but were not considered explanatory given their non-neoplastic histology[13,14]. All remaining patients were classified as discordant.

Outcomes

All macroscopic and histopathologic EGD findings were recorded. The primary outcome was foregut malignancy, defined as any biopsy-proven malignant neoplasm identified on EGD, including but not limited to adenocarcinoma, squamous cell carcinoma, neuroendocrine neoplasms (well-differentiated NETs and poorly differentiated neuroendocrine carcinomas), lymphomas, and GI stromal tumors. Premalignant foregut lesions were defined as gastric intestinal metaplasia (GIM), Barrett’s esophagus (BE), and esophageal or gastric epithelial dysplasia confirmed on biopsy. Additional EGD findings recorded included esophagitis, gastritis, peptic ulcers, gastric antral vascular ectasia, Cameron lesions, gastric polyps, and H. pylori status on biopsy.

Covariates

Baseline data included age, sex, ethnicity, comorbidities (hypertension, diabetes, cirrhosis, chronic kidney disease, coronary artery disease), medication exposures [nonsteroidal anti-inflammatory drugs (NSAIDs)/aspirin, antiplatelet agents, anticoagulants, iron supplementation, proton pump inhibitors], foregut symptoms (dyspepsia, defined as upper abdominal pain or discomfort, heartburn, acid regurgitation, nausea, vomiting, postprandial fullness, early satiety, and epigastric pain including burning, bloating, or belching), and laboratory values (hemoglobin, mean corpuscular volume, iron studies). Anemia was defined as hemoglobin below 120 g/L in women or 130 g/L in men. Baseline clinical and laboratory data were complete for all 379 included patients no imputation was performed. Smoking status, alcohol use, and insurance type were collected through additional chart review for all 379 patients. Cirrhosis etiology was abstracted for all cirrhotic patients.

Statistical analysis

Categorical variables were compared using χ2 or Fisher’s exact tests. Continuous variables were assessed for normality using the Shapiro-Wilk test; non-normally distributed variables are reported as median (interquartile range) and compared using the Mann-Whitney U test with a P value < 0.05 considered significant. Multivariable logistic regression estimated the adjusted association between discordant status and foregut malignancy, with covariates selected a priori: (1) Age; (2) Sex; (3) Anemia; (4) Foregut symptoms; (5) NSAID/aspirin use; and (6) Cirrhosis. Because all foregut malignancies occurred in the discordant group (complete separation), Firth’s penalized maximum likelihood was used to obtain finite parameter estimates[16,17]. One prespecified sensitivity analysis reclassified hemorrhoid-only patients as concordant. Hemorrhoid-only patients were defined as those whose only colonoscopy finding was hemorrhoids, without diverticulosis, colonic polyps, varices, or terminal ileum polyps. Analyses were performed using Python 3.11 with SciPy 1.11 for hypothesis tests and the firthlogist package for Firth’s penalized logistic regression.

RESULTS
Cohort and baseline characteristics

Of 1610 records screened, 379 FIT+ patients met inclusion criteria: (1) 172 discordant; and (2) 207 concordant (Figure 1). Among discordant patients, 57 (33.1%) had hemorrhoids as their only colonoscopy finding and 115 (66.9%) had no explanatory finding or hemorrhoids with additional non-concordant findings such as diverticulosis. Baseline characteristics are shown in Table 1. Groups were similar in age [61 (55-67) years vs 61 (56-66) years, P = 0.78], foregut symptoms (44.2% vs 38.2%, P = 0.25), and anemia (34.9% vs 37.2%, P = 0.67) for discordant and concordant groups, respectively. Concordant patients were more often male (48.3% vs 36.6%, P = 0.03) and had higher rates of cirrhosis (28.0% vs 10.5%, P < 0.001). Among concordant patients, colorectal neoplasia accounted for 183 (88.4%) and an alternative explanatory lesion for 24 (11.6%) (Table 1). The proportion of Other/unknown race did not differ between groups (10.5% vs 12.1%; P = 0.75).

Table 1 Baseline characteristics of discordant and concordant groups, n (%) or median (interquartile range).
Characteristic
Discordant (n = 172)
Concordant (n = 207)
P value
Age61 (55-67)61 (56-66)0.78
Male sex63 (36.6)100 (48.3)0.03
Race/ethnicity
    Hispanic129 (75.0)147 (71.0)0.42
    White non-Hispanic4 (2.3)15 (7.2)0.03
    Asian9 (5.2)11 (5.3)1.00
    Black12 (7.0)9 (4.3)0.37
    Other/unknown18 (10.5)25 (12.1)0.75
Foregut symptoms76 (44.2)79 (38.2)0.25
Anemia at FIT60 (34.9)77 (37.2)0.67
Hemoglobin (g/L)128 (112-140)131 (114-145)0.18
MCV88.6 (83.3-91.0)89.3 (83.8-93.1)0.10
Cirrhosis18 (10.5)58 (28.0)< 0.001
NSAID/aspirin use61 (35.5)57 (27.5)0.12
Smoking status
    Never145 (84.3)164 (79.2)
    Former21 (12.2)28 (13.5)
    Current6 (3.5)15 (7.2)
Ever-smoker27 (15.7)43 (20.8)0.23
Alcohol use
    None121 (70.3)118 (57.0)
    Low-risk drinking22 (12.8)18 (8.7)
    Heavy drinking/AUD9 (5.2)35 (16.9)< 0.001
Any alcohol use31 (18.0)53 (25.6)0.08
    Not documented20 (11.6)36 (17.4)
Insurance type0.045
    Medi-Cal112 (65.1)111 (53.6)
    Medicare18 (10.5)20 (9.7)
    Uninsured10 (5.8)12 (5.8)
    Private2 (1.2)0 (0.0)
    Other10 (5.8)28 (13.5)
    Not documented20 (11.6)36 (17.4)
Gastric biopsy obtained162 (94.2)193 (93.2)0.83
Same-day bidirectional79 (45.9)101 (48.8)0.61

Ever-smoker rates were similar between groups (15.7% discordant vs 20.8% concordant; P = 0.23). Heavy alcohol use or alcohol use disorder (AUD) was significantly more common in the concordant group (16.9% vs 5.2%; P < 0.001), consistent with the higher cirrhosis prevalence. The cohort was predominantly publicly insured: 65.1% of discordant and 53.6% of concordant patients had Medi-Cal as their primary insurance. Among the 76 cirrhotic patients, the most common etiologies were alcohol (34.2%), metabolic dysfunction-associated steatotic liver disease (17.1%), mixed etiologies (10.5%), hepatitis C (9.2%), autoimmune (3.9%), and unknown/cryptogenic (5.3%).

Foregut malignancy

All 11 biopsy-proven foregut malignancies occurred in the discordant group (6.4% vs 0%; Fisher’s exact P < 0.001; absolute risk difference 6.4%, 95%CI: 2.6%-9.6%, Table 2). On Firth’s penalized logistic regression adjusting for age, sex, anemia, foregut symptoms, NSAID/aspirin use, and cirrhosis, discordant status remained independently associated with foregut malignancy [adjusted odds ratio (aOR) = 26.14, 95%CI: 2.16-315.82, P < 0.001; Table 3]. Within the same model, older age was associated with foregut malignancy (aOR = 1.05 per year, P = 0.036).

Table 2 Esophagogastroduodenoscopy findings by colonoscopy concordance group (n = 379), n (%).
EGD finding
Discordant (n = 172)
Concordant (n = 207)
P value
Foregut malignancy111 (6.4)0 (0.0)< 0.001
Premalignant lesion24 (2.3)16 (7.7)0.02
Esophagitis18 (10.5)19 (9.2)0.73
Gastritis40 (23.3)32 (15.5)0.07
Peptic ulcer10 (5.8)9 (4.3)0.64
GAVE0 (0.0)4 (1.9)0.13
Gastric polyp14 (8.1)19 (9.2)0.86
H. pylori positive18 (10.5)23 (11.1)0.87
No abnormal findings94 (54.7)119 (57.5)0.60
Table 3 Firth penalized logistic regression: Predictors of foregut malignancy (exploratory analysis, n = 379).
Variable
aOR
95%CI
P value
Discordant status26.142.16-315.82< 0.001a
Age (per year)1.051.01-1.100.036a
Male sex1.450.46-4.570.554
Anemia1.710.49-6.010.439
Foregut symptoms3.520.96-12.900.067
NSAID/aspirin use1.000.30-3.340.995
Cirrhosis1.360.23-8.170.758

Among the 11 foregut malignancies, 7 were gastric adenocarcinoma, 2 were gastric NET (one World Health Organization grade 1, Ki-67 2%; one World Health Organization grade 2, Ki-67 12%), one was esophageal adenocarcinoma, and one was gastroesophageal junction adenocarcinoma. One malignancy occurred in a patient whose only colonoscopy finding was hemorrhoids, and 2 occurred in patients without foregut symptoms or anemia. Among the 8 patients in whom cancer was not the primary clinical suspicion, EGD indications included gastroesophageal reflux disease (4 patients), GIM or H. pylori surveillance (n = 2), anemia workup (n = 1), and dysphagia (n = 1). While cancer was not suspected, all underwent EGD for standard clinical indications rather than discordant FIT status alone (Table 4).

Table 4 Patient-level characteristics of the 11 foregut malignancy cases (demographics, timing, and clinical presentation).
Patient
Age
Sex
Interval (days)
Same-day
Foregut Sx
Anemia
EGD indication
Suspected before EGD
Cancer site
Smoking
Alcohol
166F0YesYesNoGERDNoGastricNeverNone
254M0YesYesYesGIM/H. pylori surveillanceNoGastricNeverSocial
356M36NoYesNoGERDNoGastricNeverNone
469F54NoNoYesGERDNoGastric (NET)NeverNone
562F75NoYesN/AIDA workup, esophageal thickening on CTYesGastric (NET)NeverNone
670M96NoNoYesAnemia workupNoGastricNeverNone
771F102NoNoNoAbnormal CT-gastric wall thickeningYesGEJNeverNone
869F121NoYesNoAbnormal CT-gastric outlet obstructionYesGastricNeverNone
955M237NoYesNoGERDNoGastricNeverNone
1070M280NoYesYesDysphagiaNoEsophagealNeverNone
1169F350NoNoNoGIM surveillanceNoGastricNeverNone
Other EGD findings

Premalignant foregut lesions were identified in 4 discordant patients (2.3%) and 16 concordant patients (7.7%; P = 0.02). All 16 cases were GIM (n = 15) or BE (n = 1); no esophageal or gastric dysplasia was identified. Rates of gastritis, esophagitis, peptic ulcers, gastric antral vascular ectasia, and H. pylori did not differ significantly between groups (Table 2). More than half of patients in each group had no abnormal EGD findings (54.7% vs 57.5%, P = 0.60).

Same-day bidirectional endoscopy

Nearly half the cohort (180/379, 47.5%) underwent same-day EGD and colonoscopy, with similar rates between groups (45.9% vs 48.8%, P = 0.61). In these patients, EGD was planned before colonoscopy findings were known. Among same-day patients, foregut malignancy was found in 2 discordant and 0 concordant. Among patients who did not receive colonoscopy and EGD on the same day, malignancy rates were 9 of 93 discordant (9.7%) and 0 of 106 concordant (0%). Gastric biopsies were obtained in 355 of 379 patients (93.7%), with similar rates between groups [162/172 (94.2%) discordant vs 193/207 (93.2%) concordant; P = 0.83]. Biopsies were performed at endoscopist discretion without a standardized protocol.

Sensitivity analysis

In a sensitivity analysis reclassifying hemorrhoid-only patients as concordant (115 discordant, 264 concordant), foregut malignancy was identified in 10 discordant (8.7%) and 1 concordant (0.4%; Fisher exact P < 0.001; Firth’s aOR = 21.93, 95%CI: 3.98-120.80, P < 0.001). In an additional sensitivity analysis restricting concordance to advanced adenoma (defined as polyp size 10 mm or greater, villous histology, high-grade dysplasia, serrated lesion, or carcinoma), colorectal cancer, colitis, or vascular lesions, 89 patients with non-advanced tubular adenoma only were reclassified as discordant. All 11 malignancies remained in the expanded discordant group [11/261 (4.2%) vs 0/118 (0%); Fisher exact P = 0.020]. In a sensitivity analysis excluding all cirrhotic patients (n = 303), foregut malignancy remained concentrated in the discordant group [10/154 (6.5%) vs 0/149 (0%); Fisher exact P = 0.002]. A summary of all sensitivity analyses is provided in Supplementary Table 1.

DISCUSSION

This study sought to compare the yield of EGD between FIT+ patients with concordant and discordant colonoscopy findings at Los Angeles General Medical Center. The concentration of all 11 biopsy-proven foregut malignancies in the discordant group is an unexpected finding that warrants further investigation. After adjustment for age, sex, anemia, foregut symptoms, NSAID/aspirin use, and cirrhosis, the discordant status remained independently associated with foregut malignancy (aOR = 26.14). This association persisted in a sensitivity analysis reclassifying hemorrhoid-only patients as concordant (aOR = 21.93). Two of the 11 patients with foregut malignancy had no foregut symptoms or anemia, the very criteria most clinicians would use to decide whether to pursue EGD. Had EGD been limited to symptomatic or anemic patients, these two cancers would have been missed. This suggests that conventional alarm features alone may not identify all patients who would benefit from foregut evaluation in this population. The median colonoscopy-to-EGD interval among the 11 malignancy patients was 96 days (range 0-350). Two malignancies were detected on same-day bidirectional endoscopy; 8 of 11 (72.7%) were detected within 121 days of colonoscopy. Only 3 of 11 patients had cross-sectional imaging findings suggesting malignancy before EGD. The remaining 8 underwent EGD for standard clinical indications: (1) Gastroesophageal reflux disease (4 patients); (2) GIM or H. pylori surveillance (n = 2); (3) Anemia workup (n = 1); and (4) Dysphagia (n = 1). While cancer was not the primary clinical suspicion in these 8 patients, all underwent EGD for conventional indications rather than discordant FIT status alone. Critically, however, these standard indications do not explain the concentration of malignancies in the discordant group: Foregut symptoms were equally prevalent in both groups (44.2% vs 38.2%, P = 0.25) and anemia rates were similar (34.9% vs 37.2%, P = 0.67), yet all 11 malignancies occurred in the discordant group. Among the 11 malignancy patients, 0 of 11 were ever-smokers and 0 of 11 had heavy alcohol use or AUD. Patient-level characteristics of the 11 foregut malignancy cases are detailed in Tables 4 and 5.

Table 5 Patient-level characteristics of the 11 foregut malignancy cases (imaging, staging, and outcomes).
Patient
Prior imaging
Imaging finding
Stage (TNM)
Metastatic
Treatment
Vital status
Prior UGI surg
1NoneN/ANot stagedNoNoneAliveNo
2NoneN/ANot stagedNoNoneAliveNo
3NoneN/ANot stagedNoNoneAliveNo
4NoneN/AT1N0M0NoEndoscopic resectionAliveNo
5CTEsophageal thickeningT2N0M0NoSurgeryAliveNo
6NoneN/AT1N0M0NoNoneAliveNo
7CTGastric wall thickening, lymphadenopathyT2N0M0NoSurgery + chemotherapyAliveNo
8CTGastric outlet obstructionT3N0M0NoSurgery + chemotherapyDeceasedNo
9NoneN/AT3N1M0NoChemotherapy + radiationDeceasedNo
10NoneN/AT1N0M0NoNoneAliveNo
11NoneN/ANot stagedNoNoneAliveNo

The 6.4% foregut malignancy rate observed in the discordant group is markedly higher than the rates of less than 1% reported in prior population-based screening cohorts[5-8] which may suggest an enrichment for patients at elevated risk for foregut malignancy undergoing colonoscopy for FIT+ results. This difference may reflect several factors including referral bias by internal county practitioners, and a higher baseline foregut malignancy risk in our Hispanic patient population here at Los Angeles General Medical Center compared to European cohorts evaluated in prior studies[5-8]. Hispanic individuals in the United States have approximately twice the gastric cancer incidence of non-Hispanic white individuals and a higher prevalence of H. pylori infection, a known gastric carcinogen[9,10,18,19]. These findings raise the question of whether the threshold for foregut evaluation should differ in Hispanic and safety-net populations with a presumably higher baseline gastric cancer risk. Further, to our knowledge, this is the first study to directly compare EGD yield between discordant and concordant patient populations. The inclusion of a concordant comparator group strengthens the analysis by providing a reference population against which to measure the yield of EGD in discordant patients. Although Hispanic ethnicity provides important context for the elevated overall foregut malignancy rate compared to European cohorts, it does not explain the concentration of malignancies in the discordant group, as Hispanic ethnicity was equally distributed between groups (75.0% vs 71.0%).

These findings should be interpreted in light of a recent analysis of the Dutch national FIT-based screening program by de Klaver et al[20], which found that while FIT+ screenees without advanced neoplasia had a significantly increased risk of cancers proximal to the colon, the 3-year cumulative incidence remained less than 1%. Several key differences distinguish that population-based study from ours. First, de Klaver et al[20] used cancer registry linkage for passive detection, whereas our study used EGD for active detection. Second, the Dutch population has substantially lower baseline gastric cancer incidence than the predominantly Hispanic population served by Los Angeles General Medical Center. Third and most importantly, our cohort represents a clinically selected population in which EGD was performed at clinician discretion for standard indications, not a population-based screening sample. The 6.4% foregut malignancy rate therefore reflects a clinically selected cohort and should not be interpreted as the expected yield of protocolized EGD for all discordant FIT+ patients. However, de Klaver et al[20] themselves concluded that future studies should evaluate which additional risk factors might justify EGD; our findings in a high-risk Hispanic safety-net population directly address this call.

The mechanism underlying the observed enrichment for foregut malignancy in the discordant group remains uncertain. If discordant colonoscopy identifies patients whose FIT+ result originated from a foregut source rather than a colonic one, it would explain both the non-explanatory colonoscopy and the high foregut malignancy yield. One explanation for how this could occur is that some foregut lesions bleed enough for FIT to detect residual globin despite intestinal transit, although this study was not designed to test that mechanism. Conditions that accelerate intestinal transit, such as diarrhea or rapid gut motility, could further reduce globin degradation and increase the likelihood of FIT detecting foregut bleeding. Future prospective studies correlating fecal hemoglobin levels with foregut lesion characteristics would help clarify this question. The higher rate of premalignant foregut lesions in concordant patients (7.7% vs 2.3%, P = 0.02) may reflect shared risk factors between gastric and colonic epithelial pathology. In a large United States case-control study, GIM was independently associated with colorectal tubular adenomas (odds ratio = 1.65, 95%CI: 1.59-1.71), possibly mediated through H. pylori infection and diminished gastric acid barrier function[21]. This is consistent with our observation that patients with colonic neoplasia, who dominate the concordant group, also had higher rates of GIM. Notably, these premalignant lesions are not themselves a plausible source of occult bleeding; rather, their co-occurrence with colonic neoplasia likely reflects shared mucosal risk factors rather than a causal relationship with FIT positivity. The concordant neoplasia rate of 88.4% exceeds typical adenoma detection rates at screening colonoscopy and likely reflects FIT+ selection enriching for colonic neoplasia. The finding that discordant patients were more often female is consistent with known sex differences in adenoma prevalence[2]. Because women develop fewer adenomas, they are less likely to have a colonic lesion explaining a FIT+ result, which may account for their overrepresentation in the discordant group. An additional consideration in premenopausal women is whether menstrual blood loss may have contributed to FIT+ results, although this cannot be determined from these data and represents a potential area for further investigation.

Two baseline differences between groups merit comment. First, concordant patients were more often male, which is consistent with the higher adenoma prevalence observed in men across screening populations[2], this is thought to be because men are more likely to harbor colorectal neoplasia, they are more likely to have a colonoscopy finding that explains a FIT+ result, and consequently cluster in the concordant group. Second, concordant patients also had substantially higher rates of cirrhosis (28.0% vs 10.5%), which is expected because cirrhosis produces portal hypertensive colopathy, colonic varices, and other vascular lesions that constituteconcordant findings by definition. Neither imbalance undermines the primary analysis, as both sex and cirrhosis were included as covariates in this Firth’s regression model. Additionally, heavy alcohol use and AUD were significantly more common in the concordant group (16.9% vs 5.2%, P < 0.001), consistent with the higher cirrhosis prevalence. Among the 11 malignancy patients, 0 were ever-smokers and 0 had heavy alcohol use, strongly arguing against tobacco or alcohol as confounders for the primary finding. Cirrhosis etiology was predominantly alcohol (34.2%), followed by metabolic dysfunction-associated steatotic liver disease (17.1%), mixed etiologies (10.5%), and hepatitis C (9.2%). In a sensitivity analysis excluding all cirrhotic patients, the association between discordant status and foregut malignancy persisted (6.5% vs 0%, P = 0.002), confirming that cirrhosis does not drive the primary finding.

As a retrospective observational study, our findings are subject to inherent limitations. EGDs were performed at clinician discretion rather than by protocol, so the discordant group represents patients whom clinicians chose to evaluate further, that is to say, not all FIT+ patients with non-explanatory colonoscopy received an EGD. Several findings temper this concern: (1) 47.5% of the cohort underwent same-day bidirectional endoscopy in which EGD was planned before colonoscopy results were available, and same-day rates were similar between groups (P = 0.61); (2) 2 of 11 discordant patients with foregut malignancy had no symptoms or anemia, which are the usual triggers for EGD referral; and (3) The findings were consistent under an alternative hemorrhoid classification (aOR = 21.93, P < 0.001). Furthermore, foregut symptoms were equally prevalent in both groups (44.2% vs 38.2%, P = 0.25), indicating that symptom-driven referral does not account for the observed difference in foregut malignancy between groups. Additional limitations include the single-center design and the small number of malignancy events (11 total). The complete separation of malignancies into the discordant group required Firth’s penalized regression, and the wide confidence interval (2.16-315.82) reflects the limited precision inherent to rare-event analyses and the low power of this study. Despite these limitations, the consistent direction of the finding across unadjusted and adjusted analyses, and further under an alternative hemorrhoid classification, suggests a real signal that merits confirmation in larger future studies. Importantly, the 6.4% foregut malignancy rate reflects a clinically selected cohort in which patients underwent EGD for standard clinical indications, and cannot be generalized to all discordant FIT+ patients. Whether discordant FIT status independently identifies patients who would benefit from EGD beyond standard clinical indications can only be determined by prospective studies with protocolized EGD for all discordant patients regardless of symptoms. The FIT quantitative value was not uniformly available, and unmeasured confounders including smoking, alcohol use, birthplace, liver disease severity, and weight loss were not always available and may have influenced both EGD referral and malignancy risk. Finally, gastric biopsies were obtained at endoscopist discretion rather than by protocol, so findings such as GIM, BE, and H. pylori status may have been missed more often in one group than the other depending on how frequently each endoscopist chose to biopsy. However, biopsy rates were similar between groups (94.2% vs 93.2%, P = 0.83), suggesting that differential biopsy practices are unlikely to explain between-group differences in histologic findings. The overall GIM detection rate of 5.0% may underestimate true prevalence compared to studies using protocolized gastric mapping. As a single-center study at a large urban safety-net hospital serving a predominantly Hispanic population, these findings may not generalize to other demographic groups, community settings, or organized population-based screening programs where FIT+ patients may differ in baseline foregut malignancy risk. Additional unmeasured confounders include body mass index, gastroesophageal reflux disease or Barrett esophagus history, weight loss, birthplace or immigration region, prior H. pylori treatment history, and dietary factors, which were not systematically available in our retrospective dataset. Capsule endoscopy was performed in 11 patients (2.9%); computed tomography enterography and device-assisted enteroscopy were not systematically captured, and small-bowel bleeding sources cannot be excluded.

CONCLUSION

In this clinically selected cohort, foregut malignancy clustered exclusively among discordant FIT+ patients despite similar clinical profiles between groups, suggesting that discordant colonoscopy status may help stratify foregut malignancy risk in safety-net and Hispanic populations with elevated baseline gastric cancer burden. These findings are hypothesis-generating and suggest that discordant FIT+ colonoscopy may identify a subgroup of patients warranting further prospective study, particularly in safety-net populations with elevated baseline gastric cancer risk. Multicenter prospective studies with protocolized EGD for all FIT+ patients with non-explanatory colonoscopy, standardized gastric biopsy protocols, and systematic collection of smoking, alcohol, birthplace, and socioeconomic data would address the limitations of this retrospective single-center analysis.

ACKNOWLEDGEMENTS

The authors thank the patients of Los Angeles General Medical Center and the Department of Medicine house staff who contributed to data collection.

References
1.  Allison JE, Fraser CG, Halloran SP, Young GP. Population screening for colorectal cancer means getting FIT: the past, present, and future of colorectal cancer screening using the fecal immunochemical test for hemoglobin (FIT). Gut Liver. 2014;8:117-130.  [RCA]  [PubMed]  [DOI]  [Full Text]  [Full Text (PDF)]  [Cited by in Crossref: 147]  [Cited by in RCA: 140]  [Article Influence: 11.7]  [Reference Citation Analysis (4)]
2.  Lee JK, Liles EG, Bent S, Levin TR, Corley DA. Accuracy of fecal immunochemical tests for colorectal cancer: systematic review and meta-analysis. Ann Intern Med. 2014;160:171.  [RCA]  [PubMed]  [DOI]  [Full Text]  [Cited by in Crossref: 583]  [Cited by in RCA: 553]  [Article Influence: 46.1]  [Reference Citation Analysis (11)]
3.  Robertson DJ, Lee JK, Boland CR, Dominitz JA, Giardiello FM, Johnson DA, Kaltenbach T, Lieberman D, Levin TR, Rex DK. Recommendations on Fecal Immunochemical Testing to Screen for Colorectal Neoplasia: A Consensus Statement by the US Multi-Society Task Force on Colorectal Cancer. Gastroenterology. 2017;152:1217-1237.e3.  [RCA]  [PubMed]  [DOI]  [Full Text]  [Cited by in Crossref: 203]  [Cited by in RCA: 321]  [Article Influence: 35.7]  [Reference Citation Analysis (8)]
4.  Zuckerman GR, Prakash C, Askin MP, Lewis BS. AGA technical review on the evaluation and management of occult and obscure gastrointestinal bleeding. Gastroenterology. 2000;118:201-221.  [RCA]  [PubMed]  [DOI]  [Full Text]  [Cited by in Crossref: 339]  [Cited by in RCA: 318]  [Article Influence: 12.2]  [Reference Citation Analysis (4)]
5.  Allard J, Cosby R, Del Giudice ME, Irvine EJ, Morgan D, Tinmouth J. Gastroscopy following a positive fecal occult blood test and negative colonoscopy: systematic review and guideline. Can J Gastroenterol. 2010;24:113-120.  [RCA]  [PubMed]  [DOI]  [Full Text]  [Full Text (PDF)]  [Cited by in Crossref: 25]  [Cited by in RCA: 29]  [Article Influence: 1.8]  [Reference Citation Analysis (0)]
6.  van der Vlugt M, Grobbee EJ, Bossuyt PM, Bos ACRK, Kuipers EJ, Lansdorp-Vogelaar I, Spaander MCW, Dekker E. Risk of Oral and Upper Gastrointestinal Cancers in Persons With Positive Results From a Fecal Immunochemical Test in a Colorectal Cancer Screening Program. Clin Gastroenterol Hepatol. 2018;16:1237-1243.e2.  [RCA]  [PubMed]  [DOI]  [Full Text]  [Cited by in Crossref: 14]  [Cited by in RCA: 29]  [Article Influence: 3.6]  [Reference Citation Analysis (0)]
7.  Shah A, Eqbal A, Moy N, Koloski N, Messmann H, Kendall BJ, Sharma P, Dulleck U, Jones MP, Holtmann GJ. Upper GI endoscopy in subjects with positive fecal occult blood test undergoing colonoscopy: systematic review and meta-analysis. Gastrointest Endosc. 2023;97:1005-1015.e30.  [RCA]  [PubMed]  [DOI]  [Full Text]  [Cited by in Crossref: 7]  [Cited by in RCA: 15]  [Article Influence: 5.0]  [Reference Citation Analysis (0)]
8.  Bini EJ, Rajapaksa RC, Valdes MT, Weinshel EH. Is upper gastrointestinal endoscopy indicated in asymptomatic patients with a positive fecal occult blood test and negative colonoscopy? Am J Med. 1999;106:613-618.  [RCA]  [PubMed]  [DOI]  [Full Text]  [Cited by in Crossref: 20]  [Cited by in RCA: 27]  [Article Influence: 1.0]  [Reference Citation Analysis (0)]
9.  Karimi P, Islami F, Anandasabapathy S, Freedman ND, Kamangar F. Gastric cancer: descriptive epidemiology, risk factors, screening, and prevention. Cancer Epidemiol Biomarkers Prev. 2014;23:700-713.  [RCA]  [PubMed]  [DOI]  [Full Text]  [Cited by in Crossref: 1420]  [Cited by in RCA: 1418]  [Article Influence: 118.2]  [Reference Citation Analysis (16)]
10.  Vitiello GA, Hani L, Wang A, Porembka MR, Alterio R, Ju M, Turgeon MK, Lee RM, Russell MC, Kronenfeld J, Goel N, Datta J, Maker AV, Fernandez M, Richter H, Correa-Gallego C, Berman RS, Lee AY. Clinical Presentation Patterns and Survival Outcomes of Hispanic Patients With Gastric Cancer. J Surg Res. 2021;268:606-615.  [RCA]  [PubMed]  [DOI]  [Full Text]  [Cited by in Crossref: 1]  [Cited by in RCA: 6]  [Article Influence: 1.2]  [Reference Citation Analysis (0)]
11.  de Klerk CM, Vendrig LM, Bossuyt PM, Dekker E. Participant-Related Risk Factors for False-Positive and False-Negative Fecal Immunochemical Tests in Colorectal Cancer Screening: Systematic Review and Meta-Analysis. Am J Gastroenterol. 2018;113:1778-1787.  [RCA]  [PubMed]  [DOI]  [Full Text]  [Cited by in Crossref: 32]  [Cited by in RCA: 58]  [Article Influence: 7.3]  [Reference Citation Analysis (1)]
12.  Ebner DW, Rushlow D, Mou J, Porter K, Finney Rutten LJ, Limburg P, Sancar F, Imperiale TF. Stool-Based Colorectal Cancer Screening Test Performance Characteristics in Those With and Without Hemorrhoids. Mayo Clin Proc Innov Qual Outcomes. 2023;7:320-326.  [RCA]  [PubMed]  [DOI]  [Full Text]  [Full Text (PDF)]  [Cited by in RCA: 1]  [Reference Citation Analysis (0)]
13.  US Preventive Services Task Force, Davidson KW, Barry MJ, Mangione CM, Cabana M, Caughey AB, Davis EM, Donahue KE, Doubeni CA, Krist AH, Kubik M, Li L, Ogedegbe G, Owens DK, Pbert L, Silverstein M, Stevermer J, Tseng CW, Wong JB. Screening for Colorectal Cancer: US Preventive Services Task Force Recommendation Statement. JAMA. 2021;325:1965-1977.  [RCA]  [PubMed]  [DOI]  [Full Text]  [Cited by in Crossref: 1663]  [Cited by in RCA: 1619]  [Article Influence: 323.8]  [Reference Citation Analysis (8)]
14.  Shaukat A, Kahi CJ, Burke CA, Rabeneck L, Sauer BG, Rex DK. ACG Clinical Guidelines: Colorectal Cancer Screening 2021. Am J Gastroenterol. 2021;116:458-479.  [RCA]  [PubMed]  [DOI]  [Full Text]  [Cited by in Crossref: 675]  [Cited by in RCA: 622]  [Article Influence: 124.4]  [Reference Citation Analysis (13)]
15.  Nakama H, Kamijo N, Fujimori K, Horiuchi A, Abdul Fattah S, Zhang B. Immunochemical fecal occult blood test is not suitable for diagnosis of hemorrhoids. Am J Med. 1997;102:551-554.  [RCA]  [PubMed]  [DOI]  [Full Text]  [Cited by in Crossref: 14]  [Cited by in RCA: 12]  [Article Influence: 0.4]  [Reference Citation Analysis (0)]
16.  Firth D. Bias reduction of maximum likelihood estimates. Biometrika. 1993;80:27-38.  [PubMed]  [DOI]  [Full Text]
17.  Heinze G, Schemper M. A solution to the problem of separation in logistic regression. Stat Med. 2002;21:2409-2419.  [RCA]  [PubMed]  [DOI]  [Full Text]  [Cited by in Crossref: 1170]  [Cited by in RCA: 1500]  [Article Influence: 62.5]  [Reference Citation Analysis (5)]
18.  Hooi JKY, Lai WY, Ng WK, Suen MMY, Underwood FE, Tanyingoh D, Malfertheiner P, Graham DY, Wong VWS, Wu JCY, Chan FKL, Sung JJY, Kaplan GG, Ng SC. Global Prevalence of Helicobacter pylori Infection: Systematic Review and Meta-Analysis. Gastroenterology. 2017;153:420-429.  [RCA]  [PubMed]  [DOI]  [Full Text]  [Cited by in Crossref: 2676]  [Cited by in RCA: 2357]  [Article Influence: 261.9]  [Reference Citation Analysis (21)]
19.  Kumar S, Metz DC, Ellenberg S, Kaplan DE, Goldberg DS. Risk Factors and Incidence of Gastric Cancer After Detection of Helicobacter pylori Infection: A Large Cohort Study. Gastroenterology. 2020;158:527-536.e7.  [RCA]  [PubMed]  [DOI]  [Full Text]  [Full Text (PDF)]  [Cited by in Crossref: 234]  [Cited by in RCA: 217]  [Article Influence: 36.2]  [Reference Citation Analysis (22)]
20.  de Klaver W, van der Vlugt M, Spaander MCW, Bossuyt PM, Dekker E. Risk of Cancers Proximal to the Colon in Fecal Immunochemical Test Positive Screenees in a Colorectal Cancer Screening Program. Gastroenterology. 2024;167:788-797.e2.  [RCA]  [PubMed]  [DOI]  [Full Text]  [Cited by in Crossref: 12]  [Cited by in RCA: 13]  [Article Influence: 6.5]  [Reference Citation Analysis (0)]
21.  Sonnenberg A, Turner KO, Genta RM. Associations between gastric histopathology and the occurrence of colonic polyps. Colorectal Dis. 2020;22:814-817.  [RCA]  [PubMed]  [DOI]  [Full Text]  [Cited by in Crossref: 6]  [Cited by in RCA: 16]  [Article Influence: 2.7]  [Reference Citation Analysis (0)]
Footnotes

Peer review: Externally peer reviewed.

Peer-review model: Single blind

Corresponding Author's Membership in Professional Societies: American College of Gastroenterology, No. 75384.

Specialty type: Oncology

Country of origin: United States

Peer-review report’s classification

Scientific quality: Grade B, Grade B, Grade C, Grade C

Novelty: Grade B, Grade C, Grade C, Grade C

Creativity or innovation: Grade B, Grade B, Grade B, Grade B

Scientific significance: Grade B, Grade C, Grade C, Grade C

P-Reviewer: Abdulrasak M, MD, MRCP, PhD, Sweden; Men CJ, PhD, China S-Editor: Luo ML L-Editor: A P-Editor: Zhao YQ

Write to the Help Desk