Age of diagnosis in familial Barrett’s associated neoplasia

Abstract

The identification of hereditary cancer genes for esophageal adenocarcinoma (EAC) and its precursor, Barrett’s esophagus (BE), may prove critical for the development of novel prevention and treatment strategies. Specifically, efforts for detecting BE and EAC susceptibility genes have focused on families with three or more affected members, since these individuals have an earlier age onset compared to non-familial individuals. Given that the use of BE may overestimate the likelihood of disease heritability, we evaluated the age of diagnosis in kindreds with a restricted definition including only confirmed high-grade dysplasia (HGD) or EAC. The Familial Barrett’s Esophagus Consortium database was used to identify individuals with HGD and EAC. These individuals were subsequently split into three kindred groups: non-familial—a single affected family member, duplex—two affected family members, and multiplex—three or more affected family members. Age of cancer diagnosis and other risk factors were compared between individuals in these groups. The study included 441 non-familial, 46 duplex, and 13 multiplex individuals. There was a statistically significant difference for age of diagnosis for individuals in the multiplex families compared to the non-familial and duplex families (56.0 versus 64.3, 63.5; p = 0.049). There was no significant difference between demographic factors and other cancer risk factors between family types. The results of this study support a genetic basis for familial Barrett’s associated neoplasia and evaluation of the genetic susceptibility to this disease should continue to focus on families with multiple (three or more) affected members.

Introduction

Esophageal adenocarcinoma (EAC) is a highly aggressive malignancy, yet the incidence of EAC has increased sharply in the past four decades, while the survival rate remains poor [1] Discovery of associated genetic factors may allow for the development of much needed novel therapeutic and diagnostic strategies [2]. Therefore, identification of families with a genetic predisposition to EAC is critical to enable studies of susceptibility genes.

EAC and its precursor, Barrett’s esophagus (BE) are complex diseases caused by the interplay of genetic susceptibility and environmental factors, including gastroesophageal reflux disease (GERD), age, sex, race, central obesity, smoking, and a family history of BE or EAC [3]. Our group, and others, have noted that BE and EAC aggregate in families [4, 5]. This familial clustering, termed Familial Barrett’s Esophagus (FBE), implicates genetic factors in BE and EAC susceptibility [6, 7]. Early work by our group looking at an FBE registry, maintained as part of an ongoing multi-center FBE study, did not show a difference in the age of diagnosis for familial cancers compared to non-familial cancers [8]. Subsequent work using the same FBE registry, demonstrated that families with three or more members affected by BE and/or EAC, termed “multiplex” families, are likely to have inherited susceptibility genes, while families with two or less affected members do not [9]. Leveraging these findings, a novel germline disease susceptibility variant in VSIG10L was discovered through genetic evaluation of a family with a rare, exceptional pedigree, with 10 family members affected with BE or EAC [10].

Our research program aims to identify families with germ-line susceptibility for BE and EAC, permitting focused future studies for cancer predisposition genes. For complex diseases, such as EAC, a younger age at diagnosis may be an indicator of possible genetic susceptibility to disease. Previous analyses of the FBE database included both reported and confirmed cases of EAC, and defined affected family members as those with non-dysplastic BE, dysplastic BE, or EAC [8, 9]. The use of non-dysplastic BE, dysplastic BE, and EAC as the “affected” trait may underestimate the heritability of EAC by leading to overestimation of the likelihood of disease heritability. In order to identify families likely to have germ-line susceptibility for BE and EAC, thus guiding future studies evaluating cancer predisposition genes, we performed an updated analysis to determine the age of diagnosis when restricting the definition of affectedness to only BE with high-grade dysplasia (HGD) or EAC, and include only pathologically confirmed cases.

Methods

Recruitment

The patients involved in this study have been recruited through the FBE Consortium starting in 1998 with the goal to identify susceptible genes for BE and its associated cancers. It is registered at clinicaltrials.gov (identifier NCT00288119) according to Institutional Review Board approved protocols. The institutions that have recruited patients with BE and EAC to this study include Case Western Reserve University (University Hospitals Cleveland Medical Center), Cleveland Clinic, Mayo Clinic, Johns Hopkins Medical Institute, University of Pennsylvania, Washington University at Saint Louis, University of North Carolina, and Fred Hutchinson Medical Center. Specifically, patients with newly diagnosed BE or EAC, those undergoing surveillance endoscopy for BE, and those undergoing endoscopy for palliation or staging for known EAC were recruited in endoscopy suites. After informed consent, patients were given a questionnaire that collected data on risk factors for BE and EAC (age, gender, race, smoking, GERD symptoms), self-reported obesity measures (waist circumference, hip circumference, weight, and height) to calculate body mass index (BMI) and waist-hip ratio (WHR) at baseline, 1, 5, and 10 years prior to diagnosis, proton pump inhibitor (PPI) use, and family history of BE and EAC (including JAC). Patients had the opportunity to confirm family history with relatives before submitting the questionnaire. Permission to contact family members was acquired to have the family members participate in the study by mail and to obtain medical records to confirm reported diagnoses. If the family member was deceased, the next of kin was contacted to obtain the age of diagnosis and other clinical variables.

Definitions

Each institution designated a gastrointestinal pathologist, who reviewed all the biopsies from family members who had agreed to participate in the FBE study. HGD was confirmed through a review of the pathology report. Cancer, which included both EAC and esophagogastric junctional adenocarcinoma (JAC), was defined as a tumor primarily involving the tubular esophagus for the former or the esophagogastric junction for the latter with histological evidence of an adenocarcinoma. For relatives, records were requested from outside hospitals to confirm the diagnosis. Only patients and family members who had confirmed HGD and EAC through pathology reports and had a reported age of cancer diagnosis were included in the study. Family members with non-dysplastic BE were not considered as affected.

These individuals were categorized into three family groups. The non-familial group was defined as one affected participant who did not have any family members with HGD, JAC or EAC. Duplex kindred were defined as two affected family members. Finally, multiplex kindred were defined as three or more affected family members.

Statistical analysis

The age of diagnosis, demographic factors (gender and self-identified race), risk factors (obesity, acid regurgitation, heartburn, and smoking), PPI use, and cancer stage were compared between the three family groups. The BMI (kg/m²) within 1 year prior to diagnosis was used to compare obesity. The non-familial group was used as a control to compare to the duplex and multiplex families. Fisher’s exact test for differences in proportions and Kruskal–Wallis test for differences in means were utilized for descriptive statistics. Mixed effects ordinal logistic regression was utilized to assess the association between age at diagnosis and family type, adjusting for familial correlation. P-values < 0.05 were considered statistically significant.

Results

The total FBE dataset consisted of 16,653 individuals. Restricting the search to individuals who had HGD, JAC, or EAC narrowed the data set to 902 individuals. Restricting the study to individuals whose age at diagnosis was recorded, further narrowed the dataset to 500 individuals (Flowchart 1). The non-familial type had 441 (88.2%) individuals, the duplex had 46 (9.2%), and the multiplex had 13 (2.6%) (Table 1). A larger proportion of males (432, 86.4%) were affected compared to females (68, 13.6%). The non-familial group included 380 males (86.2%) and 61 females (13.8%) and the duplex kindred was comprised of 40 males (87%) and 6 females (13%), while the multiplex kindred consisted of 12 males (92.3%) and 1 female (7.7%). No significant difference in sex distribution amongst the groups was identified (Table 1). Similarly, there was no statistical difference of race distribution amongst the family groups with the majority of the family types self-identifying as white. The non-familial group consisted of 425 white (96.4%) and 9 non-white (2%) individuals, the duplex kindred comprised of 45 white (96.4%) individuals, and the multiplex kindred involved 13 white individuals (100%) (Table 1). Symptoms and exposures among the three family groups, including BMI, smoking status, heartburn, acid regurgitation, PPI use, and cancer stage were not significantly different compared across the family groups (Table 1).

Flowchart 1.

Patient selection

Table 1.

Descriptive statistics, demographics and risk factors by family type

	Family type			p
	Nonfamilial	Duplex	Multiplex
Overall, n (%)	441 (88.2%)	46 (9.2%)	13 (2.6%)
Age of diagnosis, mean (SD)	64.3 (12.4)	63.5 (13)	56.0 (12.9)	0.049
BMI	29.1 (6.3)	28.7 (6.5)	30.6 (13.7)	0.849
Unknown/Missing	48 (10.9%)	17 (37%)	6 (46.2%)
Sex
Female	61 (13.8%)	6 (13%)	1 (7.7%)	0.999
Male	380 (86.2%)	40 (87%)	12 (92.3%)
Unknown/Missing	0 (0%)	0 (0%)	0 (0%)
Race
Nonwhite	9 (2%)	0 (0%)	0 (0%)	0.999
White	425 (96.4%)	45 (97.8%)	13 (100%)
Unknown/Missing	7 (1.6%)	1 (2.2%)	0 (0%)
Cigarette smoker
No	109 (24.7%)	14 (30.4%)	2 (15.4%)	0.064
Yes	299 (67.8%)	17 (37%)	4 (30.8%)
Unknown/Missing	33 (7.5%)	15 (32.6%)	7 (53.8%)
Heartburn
No	146 (33.1%)	15 (32.6%)	3 (23.1%)	0.222
Yes	265 (60.1%)	14 (30.4%)	4 (30.8%)
Unknown/Missing	30 (6.8%)	17 (37%)	6 (46.2%)
Acid regurgitation
No	166 (37.6%)	8 (17.4%)	4 (30.8%)	0.252
Yes	241 (54.6%)	21 (45.7%)	3 (23.1%)
Unknown/Missing	34 (7.7%)	17 (37%)	6 (46.2%)
PPI use
Yes	266 (60.3%)	18 (39.1%)	5 (38.5%)	0.4
No	134 (30.4%)	13 (28.3%)	1 (7.7%)
Unknown/Missing	41 (9.3%)	15 (32.6%)	7 (53.8%)
Stage
I	86 (19.5%)	6 (13.0%)	1 (7.7%)	0.11
II	64 (14.5%)	4 (8.7%)	0 (0%)
III	125 (28.3%)	9 (19.6%)	3 (23.1%)
IV	37 (8.4%)	9 (19.6%)	1 (7.7%)
Unknown/Missing	129 (29.3%)	18 (39.1%)	8 (61.5%)

Age of diagnosis

The median age at the time of diagnosis of HGD and EAC was lower for individuals in the multiplex families, compared to the non-familial and duplex families (56.0 versus 64.3, 63.5; p = 0.049) (Table 1). The mixed effects ordinal logistic regression model testing for the relationship between family type and age at diagnosis was not statistically significant (OR (95% CI) = 0.995 (0.882–1.121); p = 0.926). However, the model does suggest that an individual at any age is 0.005% less likely to be classified as multiplex or duplex compared to someone 1 year younger.

Discussion

In order to more efficiently identify families with an inherited genetic predisposition to EAC we evaluated the age of disease onset using a strict definition of FBE, with affected family members as having either HGD or EAC. Our expectation was that families with two members with HGD or cancer would be more likely to have a genetic predisposition and would therefore show a younger age of disease onset. However, using bivariate analysis we noted an earlier age at diagnosis only for multiplex families, and not for duplex families, compared to affected individuals without a family history of HGD or EAC.

Our study supposed that FBE, defining affected relatives as having BE or EAC, may be distinct from familial associations where affected relatives are restricted to only more advanced Barrett’s associated lesions, BE with HGD or EAC. However, since the age at diagnosis for duplex families was no different than non-familial cases, our results suggest that the cohort of duplex pedigrees represent a combination of families with two instances of sporadic disease along with families who share susceptibility traits, consistent with prior findings for studies of FBE [8, 9]. It is possible that even in multiplex families there are individuals with sporadic disease, which will make gene discovery more challenging. However, multiplex families should continue to form the basis for investigations of hereditary cancer genes associated with Barrett’s neoplasia.

While the bivariate analysis was remarkable for an earlier age at diagnosis for multiplex families, the mixed effects ordinal logistic regression was not statistically significant. There was a high level of missing information for the cancer risk factor variables within the duplex (~ 35%) and multiplex (~ 50%) family types, impacting the statistical power of this test. Due to the level of missingness of other cancer risk factors, the model could not be adjusted for these additional variables. However, when directly comparing risk factors, aside from family size, multiplex families did not differ significantly from non-familial or duplex families. This further supports a strong genetic susceptibility in patients with a high degree of familial HGD and EAC.

The limitations of the study include sample size, family size, and selection bias. The small number of multiplex families, limited by confirmed pathology, could have led to a type II error. Based on the results of the study, confirmed and reported cases should be included in future studies to ensure an ample sample size. The second limitation is the size of families as larger families are more likely to have multiple affected members simply by chance. Additionally, family member diagnoses of BE or EAC may influence an earlier endoscopy screening in family members with symptoms, leading to an early diagnosis of non-dysplastic BE or BE with dysplasia, even if there was no real genetic susceptibility. Thirdly, the database study is inherently limited due to selection bias as it relies heavily on the participation of family members with these conditions. Data collection on family members is thorough by first contacting the family member and then the next of kin to ensure the completeness of the database.

In summary, we show that HGD and EAC occur at a younger age in families with three or more affected members in the bivariate analysis. Therefore, our results support the targeting multiplex families, defined as those with three or more affected members, for studies of BE/EAC susceptibility genes. The definitions of FBE for these families can include BE and EAC, and it is reasonable to rely on reported cases for affected family members.