Gastric cancer in the pediatric population, a multicenter cross-sectional analysis of presentation and coexisting comorbidities

Abstract

Objectives

Solid tumors of the stomach in children are rare, adenocarcinoma being most frequently reported. Risk factors and clinical presentation are poorly understood. We undertook a nationwide database analysis to evaluate pediatric CA stomach presentation, comorbidities, and metastatic pattern.

Methods

The Cerner Health Facts Database^® (CHFD) was queried for pediatric age range (1–21) patients, 2010–2017 inclusive. The pediatric gastric cancer cohort was defined by ICD9 and ICD 10 CM diagnoses attributable to primary (non-GIST, non-hematologic) solid tumors of the stomach limited to diagnosis priority < 5 and validated by filtering for supportive diagnoses. Demographic characteristics, comorbidities, before and throughout the medical record were analyzed and compared to the base population.

Results

The cohort included 333 patients from a base population of 9.6 million children. The M:F ratio was 1.15:1, mean age at diagnosis was 11.8 years. Stomach cancer was most prevalent in non-Hispanic whites, less in Asians and African Americans. Symptoms included abdominal pain, vomiting, anemia, diarrhea and weight loss. Reflux symptoms, esophagitis, gastritis, including H. pylori and duodenitis were reported in 10.2%. Obesity, obesity-related comorbidities, tobacco exposure and family history of colonic polyps, gastrointestinal and breast cancer were all more prevalent (P < 0.0001) in the cohort.

Discussion

We defined patient demographic characteristics, anatomic distribution in a large cohort of children with solid tumors of the stomach. Reported symptoms in our cohort are similar to those observed in adults. Associated comorbidities which may reflect risk factors include obesity, tobacco exposure and family history of intestinal polyps and malignancy.

Background

Malignant solid tumors of the stomach encompass several histologic subtypes. Adenocarcinoma is the predominant subgroup and is the second leading cause of cancer-related mortality and the fourth-highest incident cancer globally. There are distinct differences in incidence rates based on geographic regions, and it is less common in the United States. Common associated factors include smoking, obesity, and H. pylori infection in the adult population. Other factors include cancer-predisposing hereditary syndromes, peptic ulcer disease, partial gastrectomy, radiation, and drug exposure.

Age is a critical determinant of the risk of stomach cancer; adenocarcinoma is rare (0. 1% of cases) in the pediatric age range (Tessler et al. 2019). The annual incidence of pediatric foregut and small intestinal solid tumors has been estimated at 0.027 cases per million. Risk factors in children are poorly understood, largely resulting from difficulty accruing a significant cohort to study. To date, limited presentation and outcomes data has been derived from single institution or national cancer database reports.

Drawing upon the case reports and small patient series in pediatric patients with malignant, non-GIST, gastric solid tumors (Table 1) (Strobel et al. 1978; Conley et al. 1988; McGill et al. 1993; Katz et al. 1997; Bees et al. 1997; Wolach et al. 1998; Cacciaguerra et al. 1998; Sasaki et al. 1999; Bläker et al. 2000; Michálek et al. 2000; Dokucu et al. 2002; Harting et al. 2004; Caudill et al. 2004; Hara et al. 2005; Ukiyama et al. 2005; Khurshed et al. 2007; Curtis et al. 2008; Luzzatto et al. 1989; Lagmay et al. 2009; Raphael et al. 2011; Subbiah et al. 2011; Rizzo et al. 2011; Patiroglu et al. 2013; Tuna Kirsaclioglu et al. 2014; Hunter 2015; Gumuscu et al. 2016; Zheng et al. 2016; Sabree et al. 2018; Medina Carbonell et al. 2020; Manohar et al. 2021), tumors are more prevalent in females (F:M, 1.3:1) and present at a mean age of 12.4 (SD 4.9) years. This suggests an age spread of cases including presenting in the first decade of life. When reported, racial background reflects a higher prevalence in Asians including Middle Eastern children, consistent with the pattern observed in adults. In children, the predominant histologic subtype is adenocarcinoma, specifically poorly differentiated signet cell adenocarcinoma. When specified the localization of the tumor included antro-pyloric (9); lesser (8); and greater (7); curve and in cardia (3); posterior wall (3) and body (1).

Table 1.

Literature Review, Pediatric Gastric (Solid Tumor) Malignancies

Authors	Pub.year	Age (yrs)	Gender	Race/ethnicity	Histologic subtype	Location (in stomach)	Risk factors	Metastatic pattern	LN metastatic involvement	Outcome
Strobel et al.	1978	19	F	Hisp.	PD-SRCC	Antrum	NR	Peritoneal seeding		D
Conley et al.	1988	11	F	NR	PGC	Lesser curve	(CVID), gastritis FH CA stomach	No metastasis		D
Luzzatto et al.	1989	11	M	NR	Leiomyoblastoma	Greater curve	–	No metastasis		CR
McGill et al.	1993	16	F	SE Asian	PGC	Cardia	NR	LN	Intraabdominal, intrathoracic	CR
Katz et al.	1997	14.5	M	NR	PGC	Antrum	Pernicious anemia/chronic gastritis (atrophic)	LN	NR	CR
Bees et al.	1997	10	F	NR	Gastric Malignant Schwannoma	Lesser Curve	NR	No metastasis		NR
Wolach et al.	1998	14	M	NR	Adenoca	Antrum	Pernicious anemia Cardiomyopathy	LN	NR	CR
Cacciaguerra et al.	1998	16	F	NR	PD-SRCC	Lesser curve	NR	Ovarian, peritoneal seeding, LN		D
Sasaki et al.	1999	11	F	Asian	Adenoca	Cardia	FH malignancy	No metastasis		CR
Bläker et al.	2000	11	M	NR	PD-SRCC	Antrum	NR	Lung and LN	NR	D
Michálek et al.	2000	9	M	NR	Adenoca	Greater curve	FH CA stomach	Peritoneal seeding and distant mets	NR	D
Dokucu et al.	2002	2.5	F	NR	WGC	Lesser curve	NR	Liver and LN	Supraclavicular	D
Harting et al.	2004	8	F	Middle Eastern	PD-SRCC	Greater curve	FH malignancy H. Pylori gastritis	Splenic and liver involvement		R
Caudill et al.	2004	4	F	NR	Neuroblastoma	Post. Wall	NR	NR		D
Hara et al.	2005	17	M	NR	PD-SRCC	NR	NR	Distant mets		D
Ukiyama et al.	2005	2	M	NR	Yolk Sac tumor	Greater Curve	History of resected gastric teratoma	No metastasis		CR
Khurshed et al.	2007	20	M	SE Asian	PD-SRCC	NR
		20	F	SE Asian	PD-SRCC
		20	F	SE Asian	Adenoca
Curtis et al.	2008	2	M	NR	Adenoca	Lesser curve	NR	LN	NR	D
		15	F	NR	Adenoca	Greater Curve		No metastasis		R
		5	F	NR	Embryonal rhabdomyosarcoma	Posterior wall		NR		D
Lagmay et al.	2009	10	F	Cauc	Clear cell sarcoma	Body	NR	Liver and LN	Intrabdominal / retroperitoneal	CR
Raphael et al.	2011	13	M	NR	PD-SRCC	Cardia	FH CA stomach H. Pylori gastritis	Peritoneal seeding		D
Subbiah et al.	2011	16.3	F	Hisp.	Adenoca	NR		Peritoneal seeding		NR
		8.6	F	Middle Eastern	Adenoca			No metastasis		NR
		17.3	M	Hisp.	Adenoca			Liver		NR
		16.8	M	African Am	Adenoca			Liver and peritoneal seeding		NR
		16.9	F	Hisp.	Adenoca			Liver, lung, intraabdominal, vertebral		NR
		12	M	NR	IMT	Antrum/pylorus	NR	No metastasis		R
Rizzo et al.	2011	14	F	NR	PD-SRCC	Lesser curve	FH malignancy	Bone/bone marrow		D
Lu et al.	2012	12	F	NR	PD-SRCC	Greater curve	NR	No metastasis		CR
Patiroglu et al.	2013	18	F	Cauc	PD-SRCC	Antrum	Ataxia-telangiectasia H. Pylori gastritis Hashimoto's thyroiditis	Peritoneal seeding		D
Kirsaclioglu et al.	2014	14	F	NR	Carcinoid tumor	Greater curve	Chronic atrophic gastritis	No metastasis		CR
Riera Llodrá et al.	2015	12	M	NR	PD-SRCC	Diffuse	H. Pylori gastritis	Peritoneal seeding		D
Hunter et al.	2015	13	M	Hisp.	Adenoca	Lesser curvature	NR	Liver, lung		D
Lin et al.	2015	16	M	NR	PD-SRCC	NR	Smoking H. Pylori gastritis FH gastric cancer	Peritoneal seeding		D
Gumuscu et al.	2016	13	F	African Am	NET	Lesser curve	NR	LN	NR	CR
Zheng et al.	2016	11	M	NR	PD-SRCC	Antrum	NR	Liver and LN	Head and neck	D
		3	F	NR	Mixed Germ cell Tumor	Antrum		LN	Intrabdominal / retroperitoneal	CR
		12	M	NR	Epitheliomesenchymal biphasic tumor	Antrum		No metastasis		CR
Sabree et al.	2018	15	M	Cauc	PD-SRCC	Pylorus	NR	Liver, LN	Intrabdominal/retroperitoneal	D
Medina et al.	2020	14	F	Hisp.	PD-SRCC	Post. wall	NR	LN + pancreas	Hear and neck	D
Manohar et al.	2021	13	M	NR	Synovial sarcoma	NR				NR

Literature review summary: case reports/small case series pediatric malignant gastric solid tumors (excluding GIST)

NR not reported, FH family history, Hisp Hispanic, Cauc. Caucasian, WGC Well differentiated gastric carcinoma, PGC Poorly differentiated gastric carcinoma, PD-SRCC (Poorly differentiated) Signet ring cell carcinoma, NET Neuroendocrine tumor, CR Complete Remission, R Recurrence, D Deceased

In the largest cohort to date, using the National Cancer Database, Tessler and colleagues (Tessler et al. 2019) reported on 129 pediatric patients with gastric adenocarcinoma and noted presentation with more advanced disease including stage 4 and aggressive (Signet ring) histology in children compared with adults but with no overall survival difference between the two groups.

A Japanese survey of 80 children with stomach cancer showed equal gender distribution and was more prevalent (90%) in patients older than 10 years. A family history of cancer was noted in 20% of patients. In that study, only 3 children were tested for H. pylori and 2 were positive. The authors speculated that the observed decrease in incidence in stomach cancer parallels a decrease in H. pylori prevalence (Okuda et al. 2019).

The evidence behind the role of H pylori in pediatric gastric cancer is less clear. In a study in 750 Turkish children undergoing endoscopy, 52% were H pylori positive and of whom 2% were diagnosed with intestinal metaplasia in gastric mucosal biopsy compared to none in the H. pylori negative group (Cam 2014).

A genetic predisposition for gastric cancer identified through familial clustering is present in up to 10% of patients with stomach cancer. Identified genetic syndromes account for 1–3% of stomach cancer overall. Primarily gastric cancer-predisposing conditions include hereditary diffuse gastric cancer (HDGC); gastric adenocarcinoma and proximal polyposis syndrome (GAPPS) and familial intestinal gastric cancer (FIGC). Other hereditary cancer syndromes conferring risk of gastric cancer in young adults, older than 20 years of age, include hereditary nonpolyposis colon cancer (HNPCC), Peutz-Jeghers Syndrome, juvenile polyposis, familial adenomatous polyposis, MYH-associated polyposis (Setia et al. 2015) and familial breast cancer (Setia et al. 2015; Oliveira et al. 2015).

Gastric solid tumors in children present with a broad spectrum of symptoms including abdominal pain, fever, anemia, and gastrointestinal bleeding. In addition, Okuda and colleagues reported vomiting, diarrhea, neck and abdominal mass, and weight loss specifically in children with adenocarcinoma of the stomach (Okuda et al. 2019). Dysphagia may be a presenting symptom in fundal adenocarcinoma.

The published literature on pediatric stomach cancer reflects an increased likelihood of metastatic spread at the time of presentation. The pattern in pediatric patients includes peritoneal seeding, and hematogenous spread primarily to the liver, lung, and vertebrae (Subbiah et al. 2011).

To date there are no robust studies in pediatric stomach cancer detailing the pattern of presentation or a detailed analysis of associated comorbidities. The impact of genetic and medical comorbidities as risk factors therefore remains largely obscure. The increased sophistication of administrative health databases (ADH) in the last decade has allowed for greater access to information in a greater pool of patients that harbor rare diagnoses including stomach cancer. In this study we have outlined the epidemiologic characteristics, presentation, and associated comorbidities in a large, validated cohort of pediatric patients diagnosed with stomach cancer.

Methods

Cerner Health Facts Database® (CHFD) is populated by the daily extraction of discrete electronic health record (EHR) data from participating organizations. These organizations have provided data rights to Cerner and allow the integration of de-identified information into a data warehouse. CHFD stores electronic health information in “fact” tables, which correspond to various types of information recorded during a visit. Each visit by a patient to a medical facility is called an “encounter,” which can be of different types, such as emergency, inpatient, or outpatient encounter. Each encounter can create various kinds of information; many encounters will document one or more diagnoses or one or more medical procedures that may have been performed during a visit. CHFD consists of structured data fields and only documents medication information (dosing, frequency) for inpatient visits. CHFD data is de-identified to HIPAA standards; text documents and images are not included. Children’s Mercy is a contributor to CHFD and has received a copy of the full database to support research. The data is installed in Microsoft (Redmond, WA) Azure, and queries are performed with R Studio version 1.3.1093 with R version 4.0.3. This work was performed with the 2018 version of the CHFD with data from 2000 through 2017. Data from 664 facilities associated with 100 nonaffiliated health systems are included in this release. This version of the CHFD data includes 68.7 million patients, 506.9 million encounters, 4.7 billion lab results, 729 million medication orders, 989 million diagnoses, and 6.9 billion clinical events.

The Children’s Mercy Institutional Review Board has designated research with CHFD data as “non-human subjects research.” We included all patients aged 1–21 years from 2010–2017. The base population was then filtered for (12) ICD9 and (13) ICD 10 CM diagnoses attributable to primary malignancy involving the stomach excluding both Gastrointestinal Stromal Tumors (GIST) and lymphoma (Appendix Table 8), Diagnosis priority < 5, and documentation of a supportive coded event (validation filter) from a predefined list of provider encounter types, medications or procedures were applied to derive our pediatric gastric cancer cohort (Fig. 1). The cohort was analyzed using phenotype-code mapping (Phecodes) (Wu et al. 2019; Denny et al. 2013) and comorbidity-prevalence ratios (Martinez et al. 2017). Lifetime reported comorbidities recorded were segregated and analyzed in 5 segregated time intervals relative to the gastric cancer index visit: > 1 year before, 1 year before, during, 1 year after, and > 1 year after index visit (Appendix Table 9). When analyzing for symptoms and related diagnoses at presentation, the cohort encounters were broadened to include all diagnoses from 12 months before the qualifying diagnostic code for gastric cancer. When analyzing for potential associated or predisposing comorbidities, the comorbidities recorded prior to the index encounter were included except for family history of malignancy and polyps of the colon wherein the search parameters included lifetime diagnoses.

Table 8.

ICD9 and ICD 10 CM diagnoses identifying primary malignancy involving the stomach (25 codes analyzed)

ICD	DIAGNOSIS_DESCRIPTION	n	n encounter	n patient	Median priority
ICD9 151.9	MALIGNANT NEOPLASM OF STOMACH, UNSPECIFIED	403	250	163	1
ICD10-CM C16.9	Malignant neoplasm of stomach, unspecified	64	51	47	1
ICD9 151.0	MALIGNANT NEOPLASM OF CARDIA	52	43	38	1
ICD9 V10.04	HISTORY OF MALIGNANT NEOPLASM OF STOMACH	51	42	38	3
ICD9 151.8	MALIGNANT NEOPLASM OF OTHER SPECIFIED SITES OF STOMACH	23	15	15	1
ICD10-CM Z85.028	history of other malignant neoplasm of stomach	15	13	13	4
ICD9 151.2	MALIGNANT NEOPLASM OF PYLORIC ANTRUM	15	10	10	1
ICD10-CM C16.0	Malignant neoplasm of cardia	12	10	9	1
ICD10-CM C16.2	Malignant neoplasm of body of stomach	8	8	8	1
ICD9 151.4	MALIGNANT NEOPLASM OF BODY OF STOMACH	21	17	8	1
ICD9 151.3	MALIGNANT NEOPLASM OF FUNDUS OF STOMACH	7	5	5	1
ICD10-CM C16.8	Malignant neoplasm of overlapping sites of stomach	6	4	4	2.5
ICD9 151.1	MALIGNANT NEOPLASM OF PYLORUS	4	3	3	1
ICD10-CM C16.5	Malignant neoplasm of lesser curvature of stomach, unspecified	3	2	2	1
ICD9 151.5	MALIGNANT NEOPLASM OF LESSER CURVATURE OF STOMACH, UNSPECIFIED	3	2	2	1
ICD10-CM C16.1	Malignant neoplasm of fundus of stomach	1	1	1	4
ICD10-CM C16.3	Malignant neoplasm of pyloric antrum	1	1	1	1

Fig. 1.

Pediatric gastric cancer population derivation from base population in Health facts Data Warehouse

Table 9.

Distribution/Category of pediatric stomach cancer cohort related diagnoses

interval	n diagnoses
> 1 year prior	3394
< 1 year prior	6916
At index Ca Stomach Dx	1824
Through 1 year after index Dx	7198
> 1 year after index Dx	5.096

Summary statistics are presented as frequencies and proportions. Fisher Exact test for probability, Relative Risk, and Odds Ratios were used for statistical inferences on significance between groups (MedCalc Software Ltd^®). Statistical significance was determined at the alpha level of 0.05.

Results

Our cohort included 333 patients from 27 health systems and was derived from a pediatric age base population of 9.6 million patients, 38.9 million encounters from 84 health systems (Fig. 1).

This represents a population prevalence of 0.03 per 1000 pediatric patients or 1 in 29,000 children. The M:F ratio for stomach cancer diagnosis was 1.15:1 with a mean age at diagnosis of 11.8 (range 1–21, SD 6.9, SE 0.3) years. The racial distribution was different from the base population (Table 2). Stomach cancer was significantly more prevalent in Caucasians and less prevalent in Hispanic patients (P < 0.025). It was less common in African American and Asian patients although this did not achieve statistical significance.

Table 2.

Racial/ethnic breakdown of pediatric gastric cancer cohort compared with base pediatric population

Race/ethnic description	Cohort		Base population
		%		%
Caucasian	210	63.06	5,308,065	55.15%	P = 0.0037
African American	54	16.22	1,771,414	18.40%	P = 0.3046
Asian	5	1.50	174,457	1.81%	P = 0.6713
Hispanic	1	0.30	329,329	3.42%	P = 0.0017
Biracial	1	0.30	66,958	0.70%	P = 0.3813
unknown	62	18.62	1,974,462	20.51%	P = 0.3930

Italicized data points indicate P < 0.025

We explored the pediatric subset for diagnosis clusters (phenotype–Tags) reported during the year prior to (first) diagnosis of stomach cancer. We limited the analysis for > 10 patients (3.3% of the subset) within any given predefined phenotype Tag. The most prevalent related diagnoses are summarized in Table 3. The most common symptoms included abdominal pain (16.8%); vomiting (13.5%); anemia (10.8%); diarrhea (9.6%); and weight loss (9%). Interestingly, venous thromboembolic diagnoses (3.9%) were markedly more prevalent than the base population (PR 29.3). Infectious diagnoses; non–pneumonia (16.5%); and pneumonia (3%); were reported but only slightly over base population prevalence (1.2, 1.3 respectively), whereas ascites (3%); anemia (10.8%); and thrombocytopenia (3.6%) were significantly more prevalent compared to the base population. Fever was a statistically significant symptom associated with gastric cancer diagnosis (P < 0.025). Pooled reflux, with or without, esophagitis, gastritis, including H. pylori gastritis and duodenitis was reported in 10.2% of patients and was 4 times more prevalent than in the base population. Globally, PUD diagnoses were more prevalent in the CA stomach cohort, most notably gastric ulcer, (PR ICD9; 16.2/ICD10; 32.3), helicobacter/other/unspecified gastritis (PR ICD9; 15.7, 10.7, 5.6) and esophageal reflux (PR 5.7) reflux and unspecified esophagitis (PR 6.1 and 3.5) (Table 6).

Table 3.

Symptoms (a) and diagnoses (b) recorded prior to gastric cancer diagnosis

	n encounter	n patient	nICD	Prevalence	Prevalence ratio
(a) Symptom
Abdominal pain	124	56	9	16.82%	1.507
Vomiting	84	45	7	13.51%	1.347
Anemia	88	36	9	10.81%	5.060
Diarrhea	46	32	2	9.61%	2.349
Weight loss/malnutrition	63	30	10	9.01%	6.673
Fever	59	30	4	9.01%	0.684
Constipation	37	25	3	7.51%	1.808
GI bleeding	16	12	5	3.60%	6.412
(b) Diagnosis
Infection (not pneumonia)	105	55	8	16.52%	1.249
Anemia	88	36	9	10.81%	5.060
GERD ± esophagitis, gastritis, duodenitis	69	34	10	10.21%	4.017
Venous thrombo-embolism	20	13	4	3.90%	29.332
Thrombocytopenia	15	12	2	3.60%	19.529
Ascites	11	10	2	3.00%	45.474
Pneumonia	14	10	3	3.00%	1.374

Italicized data points indicate P < 0.025

Filtered for cohort n > 10

nEncounters frequency of recorded event, nPatients number of patients with Dx recorded, nICD number of ICD codes within phenotype cluster, Prevalence Ratio cohort prevalence/base population Prevalence

Table 6.

Prior comorbidities, chronic diagnoses recorded > 1 year before index Ca Stomach visit (base population prevalence > 0.01)

Dx tag description	N ICD	% base population prevalence	Cohort encounters	Cohort patients	Cohort prevalence	% Cohort prevalence	Prevalence ratio
Hypertension	2	1.39	86	27	0.0811	8.1	5.8300	P < 0.0001
Psychiatric comorbidity	12	5.91	46	26	0.0781	7.8	1.3217	P = 0.1423
Hyperlipidemia	1	0.61	29	17	0.0511	5.1	8.3239	P < 0.0001
Type 2 DM	3	0.83	31	15	0.0450	4.5	5.4514	P < 0.0001
Esophageal reflux	1	1.31	20	12	0.0360	3.6	2.7521	P = 0.0004
Constipation	4	4.15	28	12	0.0360	3.6	0.8679	P = 0.6108
Diarrhea	2	4.09	13	10	0.0300	3.0	0.7341	P = 0.3143
Hyperlipidemia/hypercholesterolemia	1	0.19	11	9	0.0270	2.7	13.8660	P < 0.0001
Obesity	3	2.25	11	8	0.0240	2.4	1.0657	P = 0.8614
Tobacco	2	1.86	9	8	0.0240	2.4	1.2913	P = 0.4688
OSA	1	0.99	5	4	0.0120	1.2	1.2132	P = 0.7013

Anatomic localization of the tumor was reported in 108 out of 318 informative cases. The reported distribution (Table 4) was gastric cardia (52%), body (18%), antrum (16%), fundus (9%), lesser curve (5%) and overlapping sites.

Table 4.

Anatomic distribution pediatric Ca stomach

Anatomic distribution	n	%	% when specified
Cardia	47	14.78	52.81
Fundus	8	2.52	8.99
Body	16	5.03	17.98
Lesser curve	4	1.26	4.49
Pylorus-antrum	14	4.40	15.73
Overlapping sites	4	1.26
Other specified site	15	4.72
Unspecified	210	66.04

Metastatic spread, including lymphatic spread, was reported in 40 patients (12%) (Table 5) and the pattern of spread in decreasing frequency included lung, peritoneal and retroperitoneal, liver, bone and bone marrow, and intra-abdominal lymphatic spread.

Table 5.

Metastatic spread pattern in 1st year following Ca stomach diagnosis

	n	%
Metastatic site
Lung	7	17.5
Retroperitoneum and peritoneum	6	15
Liver	5	12.5
Bone/bone marrow	3	7.5
Brain and spinal cord	3	7.5
Other digestive organs and spleen	3	7.5
Pleura	1	2.5
Other specified	1	2.5
Large intestine	1	2.5
Lymphatic spread
Intra-abdominal	3	7.5
Unspecified (lymphatic)	3	7.5
Extremities; axillary and inguinal, upper and lower limbs	2	5
Intra-thoracic	1	2.5
Head, face and neck	1	2.5

Chronic comorbidities recorded before the index stomach cancer diagnosis were computed as subset prevalence as well as subset to base population prevalence ratio (PR) and in this population (Table 6).

Comorbidities noted to be more prevalent in the cohort over the pediatric base population included obesity as well as diagnoses that are more prevalent in overweight children, including essential hypertension, type 2 diabetes, hyperlipidemia, hypercholesterolemia, and obstructive sleep apnea (OSA). Esophageal reflux was more prevalent in the cohort compared with the pediatric base population (PR 2.7). Specific peptic ulcer disease-related diagnoses (gastritis) were infrequent but relatively more prevalent than in the base population (Appendix Table 10).

Table 10.

Peptic ulcer disease tagged diagnoses recorded > 1 year before index Ca Stomach visit

ICD	DIAGNOSIS_DESCRIPTION	Base population prevalence	n cohort encounter	n cohort patient	Prevalence	% Prevalence	Prevalence ratio
ICD9 535.50	UNSPECIFIED GASTRITIS AND GASTRODUODENITIS, WITHOUT MENTION OF HEMORRHAGE	0.0042	3	3	0.0090	0.9	2.1261	P = 0.1909
ICD9 535.40	OTHER SPECIFIED GASTRITIS, WITHOUT MENTION OF HEMORRHAGE	0.0008	2	2	0.0060	0.6	7.1826	P = 0.0052

Tobacco use was more prevalent pre-diagnosis and was reported in 17 patients in the year before diagnosis of stomach cancer. This was significantly more prevalent when compared with the base population (PR 2.7; RR 2.7 P < 0.0001). Pre-diagnosis H. pylori was diagnosed in 2 patients, this increased to 7 patients when including all encounters with a subset to base prevalence ratio (PR) of 39.5. (P < 0.0001).

Our coding resources (ICD9, ICD10-CM) do not accurately identify hereditary cancer syndromes, however, a documented family history of colonic polyps, gastrointestinal and breast cancer were all significantly more prevalent in the cohort compared with the pediatric base population (P < 0.025) (Table 7).

Table 7.

Family history of colonic polyps, malignancy in pediatric gastric cancer cohort compared with base population. Italicized data points indicate p<0.025

Family history	n encounters	Prevalence ratio (ICD9)		Prevalence ratio (ICD10-CM)
Colonic polyps	5	101.59	P < 0.0001	162.83	P < 0.0001
Malignant neoplasm of the digestive tract	22	108.83	P < 0.0001	32.88	P < 0.0001
Malignant neoplasm of the breast	13	67.58	P < 0.0001	12.48	P = 0.0116

Italicized data points indicate P < 0.025

Conclusions

This study is the largest analysis to date of a group of children diagnosed with gastric cancer. We describe the demographic, presenting symptoms, anatomic localization, metastatic spread, and associated diagnoses identifying potential risk factors. We have focused on gastric solid tumors excluding GIST and lymphoma to limit our study to a more homogenous spectrum of disorders.

The breakdown in racial characteristics in the cohort is interesting. In contrast to the our literature review, Asians were not over-represented in this U.S. sample with only 1.5% of the affected individuals compared to 1.8% of the base population (P = NS). Our inferences in this respect are limited by the size of the cohort but do not appear to relate to under-representation of the Western region of the US with the higher distribution of Asians (Appendix Table 11). The reduced prevalence in Asians is consistent with the earlier observation that the incidence of stomach cancer among Asian immigrants is lower than among native Asians suggesting that lifestyle factors are a significant determinant of stomach cancer risk (Kim et al. 2015). In our pediatric subset, Caucasians were significantly over-represented (P = 0.003) whereas African American and Hispanic children were underrepresented (P = NS, P = 0.0017 respectively). These observations can be explained given our observed anatomic distribution of stomach cancer in children. In adults, cardia originating stomach cancer is more prevalent in non-Hispanic whites than in Hispanics (Shah et al. 2020), our observations suggest that this may be a factor in childhood stomach cancer.

Table 11.

HealthFacts^® regional distribution in comparison with US census data population distribution

US 2021 census data			HealthFacts® 2018 regional distribution		Ratio health facts/US census
Region	Population	Percentage	Population	Percentage
Northeast	57,159,838	17.20%	12,790,039	18.62%	1.08
Midwest	68,841,444	20.70%	22,878,296	33.30%	1.61
West	78,667,134	23.70%	20,931,249	30.47%	1.29
South	127,225,329	38.30%	12,096,746	17.61%	0.46

The observed age distribution suggests several children with gastric malignancy presented under the age of 10. Earlier studies reporting on gastric adenocarcinoma suggest a preponderance of adolescent and young adult (10–21 year old) patients. However several series, with more liberalized inclusion of other histopathologic subtypes report patients as young as 8 months of age. These reports include rare entities such as rhabdomyosarcoma, gastrointestinal stromal tumors (GIST), neuroendocrine tumors, and lymphoma. We specifically excluded lymphoma and GIST tumors when deriving our pediatric subset (and in our literature review) however the inclusion of other rare tumors may potentially factor in our observed mean age at diagnosis.

The observed symptoms at presentation in our cohort of patients mirrors the adult patient experience; abdominal pain, nausea and vomiting, gastrointestinal hemorrhage manifesting overtly or as iron deficiency, weight loss, and ascites were prominent.

In our pediatric subset, 12 affected subjects (3.9%) had venous and thrombo-occlusive phenomena reported at the time of diagnosis. This has been reported in gastric adenocarcinoma including in pediatric patients (Hunter 2015). Our observation suggests that increased vigilance for gastrointestinal including gastric malignancy may be warranted in children with unexplained thromboembolic phenomena.

We can only speculate on possible risk factors for pediatric gastric cancer from our pediatric subset. Obesity has been recognized as a risk factor for gastric cancer in adults, especially in males and non-Asians (Rawla and Barsouk 2019). The underlying mechanism of this association is unclear and may reflect obesity as a surrogate for a diet high in meat and low in fruits and vegetables. In our cohort,obesity and comorbidities that are strongly associated with obesity were markedly more prevalent. These include hypertension, hyperlipidemia, and type 2 diabetes mellitus. In turn, obesity itself is significantly under-reported in clinical claims databases (Ammann et al. 2018) presumably including our own, rendering of our observed difference in prevalence likely more significant.

A genetic predisposition to gastric cancer is well established in the adult literature. An increased risk of gastric cancer in children with hereditary polyposis syndromes is broadly cited but largely unsubstantiated. Our observations are necessarily limited insofar as the diagnostic codes related to hereditary predisposition to cancer are not granular enough to provide detailed, syndrome-specific insight on risk. However, we observed a striking and significantly increased likelihood of patients reported with a positive family history of colon polyps, intestinal cancer, and breast cancer suggesting a strong genetic influence in our pediatric subset and to some extent conflicting with our observations on obesity and race.

Esophagitis is a recognized risk factor, in adults, specifically for gastric cardia cancers (Cavaleiro-Pinto et al. 2011) which was the predominant localization noted in our pediatric subset. Esophageal reflux was more prevalent in our pediatric subset with cancer than the base population (RR 2.7, P = 0.0004) suggesting a similar pathophysiologic paradigm in a subset of affected individuals.

H pylori gastritis was more prevalent in our pediatric stomach cancer patients. H pylori-associated gastritis is a well-established risk factor in gastric cancer in diverse populations including tenuous evidence in smaller pediatric reports. The mechanism of carcinogenesis is thought to be multifactorial and includes direct and inflammation-related DNA damage mediated in part by virulence factors that activate cell signaling pathways controlling cell proliferation (Alipour 2021).

The exposure to antacid therapy in our population could not be studied. Long-term antacid therapy, both proton pump inhibitor (PPI) as well as histamine-2 receptor antagonists (H2RAs), are associated with gastric neoplasia (Ahn et al. 2013). Antacids may induce a hypergastrinemic state or the association may be spurious and confounded by the indication.

In a national-registry-based study on metastatic spread patterns in predominantly adult patients with gastric cancer, 26% and 13% of patients had metastasis to single and multiple sites respectively. The most common sites of metastasis were the liver (48%), peritoneum (32%), lung (15%) and lymph nodes or poorly defined (11%) A relatively higher incidence of lymph node metastasis in younger patients has been reported (Ahn et al. 2013). There are no prior published reports on metastatic spread in robust pediatric cohorts. Peritoneal spread, liver, followed by lung metastases are noted in smaller series (Riihimäki et al. 2016; Esaki et al. 1990). In our study metastases were noted in 12% of patients within 1 year of the index diagnosis. The most frequently reported metastatic spread patterns were lung, direct peritoneal seeding, liver, bone and bone marrow, brain and vertebral column, and other gastrointestinal sites. Lymphatic spread to intrabdominal lymph nodes (7.5%) and extremities (5%) were reported.

Our study has several limitations. Our methodology relies on the accuracy of ICD coded and recorded diagnoses which are subject to several potential sources of error (Esaki et al. 1990). It is difficult to determine precisely the impact of these factors on the accuracy of the pediatric subset definition. Clearly, a central determinant of accuracy, defined as agreement between clinical chart and administrative data (ICD-9-CM and ICD-10-CM) is the character or category of the diagnosis, wherein malignancy, as in our study, consistently ranks high positive and negative predictive values (O’Malley et al. 2005). The specific literature on the subject suggests that order of diagnosis is a critical determinant (Quan et al. 2008) (Zafirah et al. 2018). In our study we have restricted the pediatric subset definition to the first 5 diagnoses recorded for a given encounter and even then, our median priority was 1. In addition to a purely ICD diagnosis of interest defined pediatric subset, we have additionally filtered our subset for secondary diagnoses that are consistent with the clinical scenario surrounding the diagnosis of gastric malignancy. A greater degree of validation, namely through chart review, given the size of our pediatric subset would be methodologically impracticable. Another limitation of our study also relates to the diagnostic codes for gastric cancer being nonspecific in terms of histology so that all histologic subtypes of primary stomach cancer are pooled. The specific codes used exclude both GIST (leiomyosarcoma), and lymphoma but may include rare histologic subtypes not attributable to other diagnostic codes.

Conclusion

Gastric cancer is a rare diagnosis in pediatric patients. In this large sample from the United States, the prevalence of non-GIST solid tumor malignancy in individuals through 21 years of age is 1 in 33,000. It is more common in male, non-Hispanic white patients, and can present early in life. Presentation is similar to adults although earlier studies suggest more advanced disease at the diagnosis. Cardia localization of stomach cancer is the most common in children and may relate to an association with obesity and esophageal reflux. A family history of colon polyps, intestinal and breast malignancy as well as a history of H. pylori gastritis is more prevalent in children with stomach cancer. Further studies are needed to better define the role of lifestyle and genetic risk factors for stomach cancer in children.

Appendix

Tables 8, 9, 10 and11.

Author contributions

All authors contributed to the study conception and design. Material preparation, data collection and analysis were performed by EFG, UO, NS and TMA. The first draft of the manuscript was written by TMA, UO, SDSP and MAT. All authors commented on previous versions of the manuscript. All authors read and approved the final manuscript.

Funding

This work was supported by internal research funding from the Gastroenterology Division, Children’s Mercy Hospital Kansas City.

Data availability

The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request.

Declarations

Conflict of interest

The authors have no relevant financial or non-financial interests to disclose.

Ethics approval

This is an observational study. The Children’s Mercy Hospital Kansas City Research Ethics Committee has determined this as non-human subject research and confirmed that no ethical approval is required.