European Journal of Cancer 155 (2021) 116e126

Available online at www.sciencedirect.com

ScienceDirect

journal homepage: www.ejcancer.com

Original Research

Increased risk of Barrett’s oesophagus and related

neoplasia in individuals with a positive family history

Yonne Peters a,*, Lotte J. Huibertse a, Ruud W.M. Schrauwen b,

Adriaan C. Tan c, Rachel S. van der Post d, Peter D. Siersema a

a
Department of Gastroenterology and Hepatology, Radboud Institute for Health Sciences, Radboud University Medical
Center, Nijmegen, the Netherlands
b
Department of Gastroenterology and Hepatology, Bernhoven Hospital, Uden, the Netherlands
c
Department of Gastroenterology and Hepatology, Canisius Wilhelmina Hospital, Nijmegen, the Netherlands
d
Department of Pathology, Radboud University Medical Center, Nijmegen, the Netherlands

Received 21 March 2021; received in revised form 27 June 2021; accepted 12 July 2021
Available online 7 August 2021

KEYWORDS Abstract Background: Considering the poor prognosis of oesophageal adenocarcinoma

Oesophageal cancer; (EAC), it is important to identify individuals at increased risk of developing EAC who may
Risk factors; benefit from early detection and prevention strategies. We aimed to determine whether indi-
Barrett’s oesophagus; viduals with a positive family history of Barrett’s oesophagus (BE) and EAC are at an
Family history; increased risk of oesophageal neoplasia.
Detection; Methods: In a multi-centre caseecontrol study, BE patients with or without related oesopha-
PALGA geal neoplasia and randomly selected population controls filled out a questionnaire to collect
information on family history and other risk factors for BE and EAC. Positive family history
was defined as having 1 first-degree relative with BE or EAC whose diagnosis was histolog-
ically confirmed in the Dutch nationwide histopathology database.
Findings: We included 480 BE patients and 420 controls without BE who had a total of 6393
first-degree relatives. A pathologically confirmed positive family history was significantly high-
er in BE patients compared with controls (6.5% versus 0.9; p < 0.001). Positive family history
was independently associated with an increased risk of BE (OR 5.04; 95% CI 1.45e17.58;
p Z 0.01) after adjusting for known risk factors, such as gastroesophageal reflux disease
and body mass index, and family size.
Interpretation: We found that familial clustering of BE and EAC is present in 6.5% of Dutch
BE patients. Subjects with 1 first-degree relative with BE or EAC have a 5-fold increased risk

* Corresponding author: Radboud university medical center Nijmegen, Department of Gastroenterology and Hepatology, P.O. Box 9101, In-
ternal postal code 455, 6500 HB, Nijmegen, the Netherlands.
E-mail address: [email protected] (Y. Peters).

https://doi.org/10.1016/j.ejca.2021.07.007
0959-8049/ª 2021 The Author(s). Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/
licenses/by/4.0/).
 Y. Peters et al. / European Journal of Cancer 155 (2021) 116e126 117

of BE and EAC. These findings emphasize the importance of a detailed family history in pa-
tients with BE or EAC to identify individuals at increased risk who may benefit from early
detection strategies to prevent EAC-related mortality.
ª 2021 The Author(s). Published by Elsevier Ltd. This is an open access article under the CC
BY license (http://creativecommons.org/licenses/by/4.0/).

1. Introduction the association of positive family history and the risk of

BE and related neoplasia to develop a recommendation
Despite advances in multimodality treatment, oeso- for a personalized screening program in individuals with
phageal adenocarcinoma (EAC) is associated with a a family history of BE or EAC.
poor prognosis with a 5-year survival rate of 20% [1,2].
As most EACs develop over several years, its main 2. Methods
precursor lesions, Barrett’s oesophagus (BE) and
associated dysplasia, can be identified and treated 2.1. Study design and population
before progression to invasive EAC [3]. Current
guidelines do not advocate population-based screening In this caseecontrol study, all patients diagnosed with
for BE with upper endoscopy, as this would subject too BE with or without dysplasia or early EAC and patients
large numbers of individuals to a potentially unnec- undergoing surveillance of BE in three participating
essary invasive procedure with uncertain benefits, psy- hospitals, one tertiary referral and two secondary hos-
chological morbidity, and substantial costs [4e8]. pitals, between July 2015 and July 2020, were eligible for
However, various society guidelines suggest endoscopic recruitment. Inclusion criteria were: (i) age 18 years,
screening for patients with chronic gastroesophageal (ii) able to provide informed consent, (iii) available
reflux disease (GERD) symptoms combined with other contact information, and (iv) confirmed BE, defined as
risk factors for EAC. >1 cm BE at endoscopy with histopathological evidence
Only selecting individuals for screening based on the of intestinal metaplasia confirmed in at least one biopsy.
presence of GERD is likely to miss a large proportion of Patients with only gastric-type metaplasia or with only
EAC cases because approximately 40% of patients with an irregular Z-line were excluded. Deceased patients,
EAC have no clinically significant GERD symptoms [9]. patients who were unable to fill out the survey or who
Less than 10% of patients with EAC are known with a provided incomplete family history were also omitted.
prior diagnosis of BE [10,11]. This indicates that 90% of Population controls were randomly selected from the
EACs are not detected by current screening and sur- municipality population registries of the service areas of
veillance strategies and instead present with symptom- participating hospitals. Control participants with a self-
atic advanced EACs with associated high mortality reported history of BE or EAC were excluded, but
rates. Hence, personal risk assessment is needed to risk- otherwise, inclusion and exclusion criteria were identical
stratify individuals for appropriate screening strategies. for both cases and controls. The Central Research
EAC has a complex aetiology with both genetic and Ethics Committee of the Radboud University Nijmegen
lifestyle factors contributing to the individual risk of Medical Centre, the Netherlands and the local ethics
developing EAC. Known risk factors for BE and EAC committees of the participating centres approved the
are male gender, increasing age, GERD, and lifestyle study protocol. Written informed consent was obtained
factors such as smoking, poor diet, physical inactivity, from all participants.
and central obesity [3,12]. Family history could also be a
useful screening tool for evaluating cancer risk [13,14]. 2.2. Data collection
Case reports, twin studies, and clinical series have sug-
gested a familial component to BE and EAC [15e18]. All selected individuals (index patients) received a sur-
However, an association between family history and BE vey and a reminder three weeks later in case of non-
and EAC has not unambiguously been established, and response. Additional endoscopic and histological data
previous studies have shown conflicting results [18e20]. were collected from the patient electronic health systems
Since the risk of familial predisposition to developing for all eligible BE patients. The structured survey con-
BE and EAC is not completely clear, there is no sisted of three sections: information about risk factors
consensus on the necessity of screening individuals with for BE and EAC, GERD symptoms and proton pump
a family history of BE or EAC [4e8]. We, therefore, inhibitor (PPI) use, and family history. The risk factor
aimed to determine the prevalence of BE and EAC in section included details on weight, lifestyle, smoking
first-degree relatives of patients with BE with or without history, and alcohol consumption. GERD symptoms
related neoplasia. Furthermore, we aimed to determine were assessed using a validated symptom questionnaire
118 Y. Peters et al. / European Journal of Cancer 155 (2021) 116e126

[21]. The severity of heartburn and acid regurgitation Proportions with 95% confidence intervals (CIs) of
was recorded as mild, moderate, and severe. The format probands with a confirmed family history of BE or EAC
and content of the family history section were adapted (i.e., diagnosis in reportedly affected relative confirmed by
from instruments used by genetic counsellors that are histopathology review) were determined with the method
developed to obtain detailed family history and study of Wilson. Binary logistic regression was used to calculate
genetic aspects and from a previously validated family unadjusted and adjusted odds ratios (ORs) with 95% CIs
history screening questionnaire [22e25]. The instrument for the association between a family history of BE or
was modified for this study to obtain detailed histories EAC and the development of BE with or without related
related to GERD, BE, and EAC of all first-degree rel- neoplasia. The impact of potential confounders was
atives. To minimize recall bias, we only included first- evaluated by assessing their effect on the OR of the main
degree relatives in our study. Permission to contact the effects in the multivariate logistic models. The following
reportedly affected relatives and their contact informa- variables were considered potential confounders or effect
tion was obtained from the index patients by phone. modifiers based on the proposed association between BE
and family history, i.e., age, gender, ethnicity, education,
2.3. Family history and histopathological confirmation alcohol consumption, smoking status, body mass index
(BMI), GERD symptoms, family size, and age of siblings.
A positive family history was defined as having at least Each potential confounder was added separately to the
one first-degree relative (that is, parents, full siblings, or regression model. Variables that changed the OR for the
offspring) with BE or EAC whose diagnosis was histo- family history indicator by >10% and additional factors
logically confirmed in PALGA, the Dutch nationwide with a known or probable association with BE (p < 0.10)
registry of histo- and cytopathology [26]. Since its were considered as confounders and included in the
nationwide coverage in 1991, the reports of all pathol- multivariate model [27]. In the case of high collinearity
ogy laboratories in the Netherlands are centrally (correlation coefficient >0.8) between potential con-
archived in PALGA. We attempted to contact all rela- founders, the most significant confounder was selected.
tives who reported having a history of BE or oesopha- We evaluated the presence of effect modification by using
geal cancer. Written consent from reportedly affected cross-product terms in the logistic regression model and
relatives (or the index patient, if the affected relative was by evaluating stratum-specific ORs in the following sub-
deceased) was obtained to confirm their histological groups, i.e., number of first-degree relatives affected (1,
diagnosis in PALGA. Patients and relatives were con- 1, and 2 first-degree relatives) and age at diagnoses of
tacted by phone at least twice before concluding that the affected relatives. All analyses were performed with SPSS
pathology reports were unavailable. Version 25.0 (IBM Corp, Armonk, NY). Statistical sig-
Relatives with an EAC diagnosis or at least two pa- nificance was assumed to be present at p < 0.05.
thology reports reporting the presence of intestinal
metaplasia in the oesophagus (meaning the presence of 3. Results
BE) were included as affected cases. Reportedly affected
relatives without intestinal metaplasia or with oeso- 3.1. Participants
phageal squamous cell cancer or cancer that did not
involve the oesophagus were classified as false positive. A total of 650 patients diagnosed with BE were identi-
Similarly, if the diagnosis of BE or EAC could not be fied, of which 11 patients were excluded (Fig. 1). The
confirmed in PALGA (e.g., because no permission was survey was returned by 502 of the 639 remaining BE
given, patient identification was incorrect, or the diag- patients (response rate 78.6%). Of the 984 selected and
nosis was made before nationwide coverage of PALGA eligible population controls, 432 individuals returned
was realized), we considered the reported family history the survey (response rate 43.9%). One respondent was
of BE or EAC as negative. Sensitivity analyses were excluded due to a known diagnosis of BE. After
performed, including first-degree relatives whose his- excluding 31 other respondents who did not (entirely)
tology reports could not be confirmed in PALGA. complete the family history section, survey and family
history data were available for 902 subjects (480 cases
2.4. Statistical analysis and 422 controls).
Characteristics of included patients are summarized
Descriptive variables were expressed as means (SD) or in Table 1. Age, gender, and disease stage were com-
medians (interquartile range). Differences in character- parable between responders and non-responders. Pa-
istics between cases and controls and familial BE and tients with BE were generally older than controls
sporadic BE cases were tested with independent t-tests (p < 0.001) and were more likely to be male (p < 0.001).
for normally distributed data or ManneWhitney U tests GERD symptoms were significantly more common in
for nonnormally distributed data. Chi-squared tests or BE patients than in population controls (88.2% versus
Fisher’s exact tests were used to evaluate differences 36.0%; p < 0.001). The median BE length was
between categorical variables. C2M4 (IQR C0e4, M2e5), 353 patients (73.7%) were
 Y. Peters et al. / European Journal of Cancer 155 (2021) 116e126 119

Fig. 1. Flowchart of included BE patients with or without dysplasia or early adenocarcinoma and population controls.

diagnosed with a BE-segment length 3 cm, and 53 Supplementary Fig. S1. In confirmed familial BE, 35
patients (11.0%) were diagnosed with high-grade of the 254 first-degree (13.8%) relatives were known with
dysplasia (HGD) or EAC. The median age at the BE or EAC. Reflux symptoms were more frequently
time of BE diagnosis was 59 (IQR 52e65) years. reported in first-degree relatives of patients with familial
Barrett than in those with sporadic BE (29.5% versus
3.2. Family history of GERD, BE, and EAC 11.9%; p < 0.001). Clinical characteristics of BE patients
with and without a positive family history are shown in
A positive family history was reported by 67 BE patients Table 2. No differences were found regarding gender,
(14%) and 15 controls (3.6%). As two index cases were ethnicity, GERD symptoms, BMI, and lifestyle factors.
related, only the first respondent was included in the Age of GERD onset, age at BE and EAC diagnoses, and
analysis for familial BE. Of the 104 reportedly affected dysplasia grade were also not significantly different be-
relatives, permission to confirm the diagnosis in PALGA tween groups. Patients with familial BE had a signifi-
was provided by 80 relatives, who were related to 51 BE cantly shorter Barrett’s segment than patients with no
index patients and 11 controls, with 50 of them confirmed family history (median BE length: C1M3 versus C2M4;
to be registered in PALGA (Fig. 2). Family history was p Z 0.03). Median age at EAC diagnosis was 63 (IQR
determined as false-positive in seven BE patients (10.6%) 58e72) years in familial BE index patients (n Z 4) and
and three controls (20%). Despite efforts to obtain his- first-degree relatives (n Z 21) compared with 70 (IQR
topathology reports for the reportedly affected first- 60e75) years for sporadic index patients diagnosed with
degree relatives, family history could not be confirmed EAC (n Z 26) (p Z 0.33).
or was concluded to be not present in 28 BE index pa-
tients (42.4%) and eight controls (53.3%). After verifica- 3.4. Risk of BE and related neoplasia associated with a
tion, a positive family history of BE or EAC was family history of BE or EAC
confirmed in 6.5% (95% CI 4.6e9.0) of BE patients (31 of
479) and 0.9% (95% CI 0.4e2.4) of population controls (4 A family history of at least one first-degree relative with
of 422) (p < 0.001). Reflux symptoms were present in BE or EAC was significantly associated with the risk of
13.2% of first-degree relatives of BE patients and 7.4% of developing BE and related neoplasia (OR 7.23; 95% CI
population controls (p < 0.001). 2.53e20.66). GERD symptoms were the only
confounder that changed the effect of a positive family
3.3. Characteristics of familial BE history on the risk of BE and EAC by >10% to 4.57
(95% CI: 1.46e14.28). Multivariate logistic regression
Of the 31 index patients with familial BE, 27 probands analyses showed that male sex, older age, white
had one first-degree relative with BE or EAC, and four ethnicity, low level of education, high BMI, a history of
had 2 affected first-degree relatives. Pedigrees of these GERD symptoms, and a positive family history were
four families are shown in. independently associated with the risk of BE and EAC
120 Y. Peters et al. / European Journal of Cancer 155 (2021) 116e126

Table 1
Baseline characteristics of patients with Barret’s oesophagus (cases) and controls.
Characteristics Casesa (n Z 480) Controls P Value
(n Z 422)
Gender e Male 339 (70.6%) 217 (52.0%) <0.001
Age (years) 65.8  9.9 62.5  7.4 <0.001
Cultural background e White 473 (98.5%) 401 (96.4%) 0.04
Education level <0.001
High school or less 250 (52.2%) 139 (33.3%)
Vocational college 93 (19.4%) 60 (14.4%)
College/university 136 (28.4%) 218 (52.3%)
Centre N.A.
Tertiary referral centre 183 (38.1%)
Secondary hospital 297 (61.9%)
Body mass index (kg/m2) 26.4 (24.4e29.5) 24.7 (22.5e27.8) <0.001
Unavailable, n 0 3
Waist Hip Ratio 0.987  0.081 0.942  0.094 <0.001
Unavailable, n 52 29
Smoking status e Ever smoked 319 (66.5%) 253 (60.0%) 0.05
Alcohol consumption per week 0.19
None, or <1 drink 65 (14%) 59 (14.5%)
1e7 drinks 222 (47.7%) 221 (54.2%)
8e14 drinks 107 (23.0%) 78 (19.1%)
>14 drinks 71 (15.3%) 50 (12.3%)
GERD symptoms <0.001
Current 129 (27.1%) 60 (14.4%)
Previous 291 (61.1%) 90 (21.6%)
None 56 (11.8%) 267 (64.0%)
Heartburnb 174 (38.5%) 66 (16.2%) <0.001
Regurgitationb 170 (37.4%) 50 (12.3%) <0.001
GERD frequencyc <0.001
Every day 161 (39.7%) 31 (22.3%)
Once a week 159 (39.2%) 42 (30.2%)
<Once a week 86 (21.2%) 66 (47.5%)
GERD severityc <0.001
Mild 142 (34.5%) 90 (62.1%)
Moderate 160 (38.9% 39 (26.9%)
Severe 109 (26.5%) 16 (11.0%)
GERD age of onsetc 0.42
<20 years 39 (9.4%) 8 (5.5%)
20e30 years 63 (15.1%) 19 (13.0%)
30e40 years 58 (13.9%) 26 (17.8%)
40e50 years 90 (21.6%) 29 (19.9%)
>50 years 167 (40.0%) 64 (43.8%)
PPI use 445 (93.1%) 67 (16.3%) <0.001
Note. Values presented as n (%), mean  standard deviation, or median (interquartile range).
GERD, gastroesophageal reflux disease; PPI, proton-pump inhibitor.
a
Cases: patients with Barrett’s oesophagus with or without dysplasia or early adenocarcinoma.
b
Symptoms at least once a week.
c
Only patients with current or previous GERD symptoms were included.

(Table 3). Cases had more first-degree relatives, and the of BE or EAC in any first-degree relative was not
mean age of siblings was higher than controls associated with malignant progression risk (defined as
(Supplementary Table 1). As the age of index patients HGD or EAC in BE) (OR 1.60; 95% CI 0.59e4.4;
and age of siblings were correlated (correlation coeffi- p Z 0.36).
cient 0.86), final models were adjusted for the number Sensitivity analyses, also including relatives whose
and not for the age of first-degree relatives. diagnosis could not be confirmed in PALGA due to the
Table 4 shows the risk of BE, dysplastic BE, inability to obtain permission, incorrect patient identi-
and EAC in relation to family history of BE and EAC. fication information, or a diagnosis of BE or EAC
In the final adjusted model, a positive family history before nationwide coverage of PALGA, showed that the
in a first-degree relative was associated with a 5-fold association between a positive family history and the
increased risk of BE and related neoplasia (OR presence of BE or EAC was still significant (adjusted
5.04; 95% CI 1.45e17.58; p Z 0.01). A family history OR 4.89; 95% CI 2.06e11.59; p < 0.001).
 Y. Peters et al. / European Journal of Cancer 155 (2021) 116e126 121

Fig. 2. Confirmed BE and EAC diagnoses in first-degree relatives.

Age, education, smoking, BMI, and family size did general population after adjusting for other known risk
not significantly modify the association between BE and factors. However, a positive family history was not
a positive family history (p > 0.1 on interaction terms). associated with the risk of malignant progression in BE
Effect modification of gender, ethnicity and alcohol use patients.
could not be calculated due to limited subgroup sample Robust data on BE and EAC risk in individuals
sizes. As none of the population controls had 2 first- with a positive family history are lacking, and pub-
degree relatives with confirmed BE or EAC or a first- lished studies have reported conflicting results
degree relative younger than the age of 50, we could not [18e20,22,28e38]. The only study investigating a
stratify the ORs of BE in number and age of affected confirmed family history showed that individuals with
relatives. a positive family history were 12 times more likely to
have BE, EAC, or gastroesophageal junctional
4. Discussion adenocarcinoma than GERD patients [18]. This study
was limited by thesmall sample size, the inclusion of
This multicentre, population-based caseecontrol BE patients from tertiary hospitals, GERD patients
study, including 902 probands and 6393 first-degree undergoing endoscopy as the control group, and
relatives, showed that a positive family history of BE or combining BE, EAC and gastroesophageal junctional
EAC is an important risk factor for BE and related adenocarcinoma [39]. A meta-analysis, including four
neoplasia. We observed histopathologically confirmed studies related to screening first-degree relatives of BE
familial clustering of BE and EAC in 6.5% of BE pa- patients, reported a diagnostic yield for BE of almost
tients, which is in line with the findings of previous 25% [12]. Interestingly, one study reported newly
studies [22,28,29]. Individuals with a first-degree relative diagnosed BE in 40% of individuals with 2 relatives
with BE or EAC were found to have a five-fold higher with BE or EAC compared with only 5% in relatives of
risk of developing BE and related neoplasia than the sporadic BE patients [30].
122 Y. Peters et al. / European Journal of Cancer 155 (2021) 116e126

Table 2
Differences between patients with familial and sporadic Barrett’s oesophagus.
Characteristics Sporadic Familial P Value
Barrett’s oesophagus (n Z 447) Barrett’s oesophagus (n Z 31)
Gender e Male 316 (70.5%) 22 (71.0%) 0.96
Age (years) 65.9  9.9 64.6  9.8 0.48
Cultural background e White 441 (98.4%) 31 (100%) 1.00
Educational level 0.28
High school or less 237 (53.0%) 12 (38.7%)
Vocational college 86 (19.2%) 7 (22.6%)
College/university 124 (27.7%) 12 (38.7%)
Body mass index (kg/m2) 26.4 (24.4e29.6) 26.0 (24.2e28.0) 0.63
Waist Hip Ratio 0.987  0.082 0.984  0.057 0.85
Smoking status e Ever smoked 299 (66.7%) 19 (61.3%) 0.53
Alcohol consumption 400 (89.3%) 28 (90.3%) 1.00
GERD symptoms 0.16
Current 118 (26.6%) 11 (35.5%)
Previous 271 (61.0%) 19 (61.3%)
None 55 (12.4%) 1 (3.2%)
GERD frequencya 0.92
Every day 148 (39.5%) 13 (43.3%)
Once a week 148 (39.5%) 11 (36.7%)
<Once a week 79 (21.1%) 6 (20.0%)
GERD severitya 0.47
Mild 135 (35.4%) 7 (24.1%)
Moderate 146 (38.3%) 13 (44.8%)
Severe 100 (26.2%) 9 (31.0%)
GERD age of onseta 0.28
<20 years 36 (9.3%) 3 (10.0%)
20e30 years 55 (14.2%) 8 (26.7%)
30e40 years 55 (14.2%) 2 (6.7%)
40e50 years 82 (21.2%) 8 (26.7%)
>50 years 158 (40.9%) 9 (30.0%)
Most severe histology 0.22
Intestinal metaplasia 349 (77.9%) 25 (80.6%)
Low-grade dysplasia 51 (11.4%) 1 (3.2%)
High-grade dysplasia 22 (4.9%) 1 (3.2%)
Oesophageal adenocarcinoma 26 (5.8%) 4 (12.9%)
Age at BE diagnosis (years) 58.3  10.4 58.8  10.3 0.98
Age at neoplasia diagnosis (years) 66.1  8.5 71.5  9.2 0.12
BE length e Circumference (cm) 2 (1e4) 1 (0e3) 0.03
BE length e Maximal (cm) 4 (2e5) 3 (2e4) 0.03
Presence of hiatal hernia 401 (89.5%) 27 (87.1%) 0.56
Note. Values presented as n (%), mean  standard deviation, or median (interquartile range).
BE, Barrett’s oesophagus; GERD, gastroesophageal reflux disease.
a
Only patients with current or previous GERD symptoms were included.

Familial clustering of BE and EAC may be the significantly changed the effect of a positive family his-
consequence of shared environmental exposure, com- tory on the risk of BE and EAC. Furthermore, reflux
mon behavioural risk factors, or a genetic susceptibility symptoms were more common in first-degree relatives of
that increases the development of new lesions or affects familial BE cases than in population controls and spo-
the transition from BE to EAC. When combining index radic BE patients. These findings suggest a contribution
patients and affected relatives, we found a non- of both genetic and environmental factors with possibly
significantly younger age at EAC diagnosis in familial a central role for GERD [22].
cases compared with non-familial cases. Early age at Insight into the role of familial involvement in BE is
cancer onset in families with multiple affected members also important to increase the understanding of the
has been confirmed in previous studies and may be exact phenotype of hereditary BE and could be key to
indicative of inherited genetic mutations [22,28,31,40]. identify causal gene defects underlying this condition.
We largely excluded the possibility that shared envi- Studies have suggested an autosomal dominant mode of
ronmental factors on their own are responsible for the inheritance in clusters of BE and EAC patients [15,16].
observed association between a positive family and risk To adequately characterize the association between the
of BE and EAC by adjusting this association for risk of BE and EAC on the one hand and family history
modifiable risk factors. However, GERD symptoms on the other, genome-wide association studies, linkage
 Y. Peters et al. / European Journal of Cancer 155 (2021) 116e126 123

Table 3
Factors associated with Barrett’s oesophagus and related neoplasia.
Characteristics Univariable logistic regression model Final multivariable logistic regression model
OR (95% CI) P value OR (95% CI) P value
Male gender 2.21 (1.17e2.91) <0.001 2.72 (1.88e3.93) <0.001
Age (years) 1.04 (1.02e1.06) <0.001 1.05 (1.03e1.07) <0.001
White ethnicity 2.52 (1.02e6.25) 0.05 3.72 (1.27e10.90) 0.02
Educational level <0.001 <0.001
High school or less Reference Reference
Vocational college 0.87 (0.59e1.27) 0.46 0.70 (0.43e1.16) 0.17
College/university 0.35 (0.26e0.47) <0.001 0.31 (0.21e0.46) <0.001
Body mass index 1.10 (1.06e1.14) <0.001 1.05 (1.01e1.10) 0.02
Smoking status e Ever smoked 1.32 (1.00e1.72) 0.05
Alcohol consumption 1.06 (0.70e1.61) 0.79
History of GERD symptoms 13.30 (9.45e18.77) <0.001 16.79 (11.24e25.09) <0.001
First-degree relative with BE or EAC 7.23 (2.53e20.67) <0.001 5.04 (1.45e17.58) 0.01
Number of first-degree relatives 1.08 (1.03e1.14) 0.001 1.08 (1.01e1.15) 0.02

Table 4
Barrett’s oesophagus (BE) with and without neoplasia in first-degree relatives and the risk of non-dysplastic BE, dysplastic BE, and early
oesophageal adenocarcinoma.
Family history Risk of BE and Risk of NDBE Risk of HGD or Risk of early EAC
related neoplasia early EAC
OR (95% CI)a P OR (95% CI)a P OR (95% CI)a P OR (95% CI)a P
FDR with BE or EAC 5.04 (1.45e17.58) 0.01 5.54 (1.55e19.86) 0.01 6.64 (1.22e36.00) 0.03 8.19 (1.42e47.38) 0.02
FDR with NDBE e e e e
FDR with HGD or EAC 2.78 (0.75e10.30) 0.13 2.93 (0.77e11.22) 0.12 4.11 (0.68e24.88) 0.12 4.92 (0.71e34.00) 0.11
FDR with EAC 2.78 (0.75e10.30) 0.13 2.93 (0.77e11.22) 0.12 4.11 (0.68e24.88) 0.12 4.92 (0.71e34.00) 0.11
BE, Barrett’s oesophagus; EAC, oesophageal adenocarcinoma; FDR, first-degree relative; HGD, high-grade dysplasia; NDBE, non-dysplastic
Barrett’s oesophagus.
a
ORs were adjusted for: age, sex, ethnicity, education level, reflux symptoms, body mass index, number of first-degree relatives.

studies, and whole-exome sequencing of families with needs yet to be determined. For now, we advise gas-
aggregation of BE and EAC have been performed. troenterologists to obtain a detailed family history of
Several candidate causative genes, including single GERD, BE, and EAC in all BE and EAC patients to
nucleotide polymorphisms in CFTR genes, MSX1, and identify individuals most in need of undergoing upper
VSIG10L variants, have been identified in families with endoscopy or another minimally invasive screening
high penetrance of BE and EAC, but no clear mecha- modality [44e46]. We recommend endoscopic screening
nistic or phenotypic link to BE and EAC has been for individuals with at least two first-degree relatives
established yet [41e43]. with confirmed BE or EAC. Screening could also be
In the absence of genetic testing panels in current considered in first-degree relatives with 1 risk factor,
clinical practice, family history can be used to identify such as GERD or central obesity. Although family
individuals at increased risk for BE and EAC who may history is a non-modifiable risk factor, prevention could
benefit from targeted screening and prevention pro- reduce BE and EAC risk in individuals with a strong
grams. Current societal guidelines on screening in- family history by avoiding exposure to other known
dividuals with a family history of BE or EAC are largely modifiable risk factors and considering preventative
based on expert opinions and a small number of studies measures.
[18]. The American Society for Gastrointestinal Endos- Strengths of this study include a large sample size,
copy [6] recommends screening for BE in individuals inclusion of secondary and tertiary referral hospitals, the
with a family history of BE or EAC. In contrast, the inclusion of a population-based control group, and the
American Gastroenterology Association [4] and Euro- assessment of a large number of first-degree relatives.
pean Society for Gastrointestinal Endoscopy [8] have Strict criteria for BE diagnosis were applied, including
not included family history in their screening recom- the endoscopic presence of 1 cm columnar epithelium
mendations. Other societies recommend that a positive combined with the histopathologic presence of special-
family history should lower the threshold for screening ized intestinal metaplasia. Furthermore, reportedly
endoscopy [5,7]. affected relatives (or their representatives) were con-
How both our and previous findings should exactly tacted to histologically confirm their BE or EAC diag-
be applied to EAC screening recommendations nosis to minimize misclassification bias. This is also the
124 Y. Peters et al. / European Journal of Cancer 155 (2021) 116e126

first study assessing the association between a confirmed belief that a family history of BE and EAC is a risk
family history and BE with HGD and EAC, making the factor for developing BE and EAC. Family history
findings of clinical relevance. Additionally, we adjusted should, therefore, be included in (updated) guidelines on
for known risk factors for BE and EAC to assess screening for BE and EAC. The results emphasize the
whether shared environmental factors or inheritance of importance of eliciting detailed and regularly updated
common susceptibility genes conferred the increased family history information by physicians to recognize
risk. Finally, we accounted for family size and age dis- familial BE and potentially consider screening and pre-
tribution in the analyses. ventative measures for relatives at risk to prevent
Some limitations should be considered, which may morbidity and mortality from EAC.
have resulted in overestimating or underestimating the
true prevalence of a positive family history. First, the Author contributions
control group consisted of individuals randomly
selected from population registries, and only a minority Yonne Peters: study concept and design, data acquisi-
of controls had previously undergone upper endos- tion, quality control of data, data analysis and inter-
copy. Nonetheless, the prevalence of BE in the general pretation, statistical analysis, manuscript preparation,
population is thought to be only 1e2%, making the risk editing and review.
of classification error very low [47]. Furthermore, our Lotte Huibertse: data acquisition, quality control of
approach has the advantage of generalizability of the data, statistical analysis, manuscript review.
results to the general population. Second, only four Ruud W.M. Schrauwen: data acquisition, data
controls were found to have a positive family history, interpretation, critical revision of the manuscript.
resulting in a relatively large confidence interval of the Adriaan C. Tan: data acquisition, data interpretation
OR for the association between a positive family his- critical revision of the manuscript.
tory and the development of BE with or without related Rachel S. van der Post: study concept, data acquisi-
neoplasia. In order to assess robustness, we calculated tion, critical revision of the manuscript.
that the OR would still be significant (OR 2.0; 95% CI Peter D. Siersema: study concept and design, data
1.1e3.8) if 14 controls would have a positive family interpretation, critical revision of the manuscript, study
history. Third, the response rate of BE patients was supervision.
significantly higher compared with controls. Hence,
risk estimates could be higher than reported, as con- Funding
trols with a positive family history may have responded
more often than subjects with unaffected relatives. On This research received no specific grant from any
the other hand, population controls may be more often funding agency in the public, commercial or non-profit
unaware of a positive family than cases and as a result sectors.
more likely to underreport their family history.
Furthermore, relatives of BE patients may have un- Conflict of interest statement
dergone upper endoscopy more frequently compared
with controls due to increased awareness of BE and
The authors declare the following financial interests/
EAC in relatives. Fourth, although detailed analyses
personal relationships, which may be considered as po-
provided little evidence of confounding, we cannot
tential competing interests: Yonne Peters, Lotte Hui-
conclude that unmeasured socioeconomic or lifestyle
bertse, Ruud Schrauwen, Adriaan Tan, and Rachel S.
factors might have affected the results and that in-
van der Post do not report any conflicts of interest. Peter
teractions between measured confounders were not
D. Siersema received an unrestricted grant from Pentax
completely adjusted for in the logistic models. Finally,
(Japan), Norgine (UK), Motus GI (USA), and The
the caseecontrol design without follow-up and the
eNose Company (Netherlands).
small number of patients with a family history of at
least two first-degree relatives with BE or EAC may be
Appendix A. Supplementary data
considered a drawback. Consequently, multivariable
analyses using the number of relatives affected and
Supplementary data to this article can be found online
relatives’ age at diagnosis to make more tailored rec-
at https://doi.org/10.1016/j.ejca.2021.07.007.
ommendations on the optimal timing of screening
endoscopy were not possible. Longitudinal studies are
definitely needed to determine the familial risk of BE References
and EAC according to the number of affected relatives,
the age of cancer diagnosis and relatives’ age. [1] van Putten M, de Vos-Geelen J, Nieuwenhuijzen GAP, et al.
Long-term survival improvement in oesophageal cancer in the
In conclusion, we showed that familial BE is found in Netherlands. Eur J Cancer 2018;94:138e47.
6.5% of BE patients, with 14% of the first-degree rela- [2] Smyth EC, Lagergren J, Fitzgerald RC, et al. Oesophageal cancer.
tives being affected. Our findings support the common Nature Reviews Disease Primers 2017;3:17048.
 Y. Peters et al. / European Journal of Cancer 155 (2021) 116e126 125

[3] Peters Y, Al-Kaabi A, Shaheen NJ, et al. Barrett oesophagus. [22] Verbeek RE, Spittuler LF, Peute A, et al. Familial clustering of
Nature Reviews Disease Primers 2019;5(1):35. Barrett’s esophagus and esophageal adenocarcinoma in a Euro-
[4] Spechler SJ, Sharma P, Souza RF, et al. American Gastroenter- pean cohort. Clin Gastroenterol Hepatol 2014;12(10):
ological Association medical position statement on the manage- 1656e16563.e1. the official clinical practice journal of the
ment of Barrett’s esophagus. Gastroenterology 2011;140(3): American Gastroenterological Association.
1084e91. [23] Williams RR, Hunt SC, Barlow GK, et al. Health family trees: a
[5] Shaheen NJ, Falk GW, Iyer PG, et al. ACG clinical guideline: tool for finding and helping young family members of coronary
diagnosis and management of barrett’s esophagus. Am J Gas- and cancer prone pedigrees in Texas and Utah. Am J Public
troenterol 2016;111(1):30e50. quiz 1. Health 1988;78(10):1283e6.
[6] Qumseya B, Sultan S, Bain P, et al. ASGE guideline on screening [24] Emery JD, Reid G, Prevost AT, et al. Development and valida-
and surveillance of Barrett’s esophagus. Gastrointest Endosc tion of a family history screening questionnaire in Australian
2019;90(3):335e59. e2. primary care. Ann Fam Med 2014;12(3):241e9.
[7] Fitzgerald RC, di Pietro M, Ragunath K, et al. British Society of [25] Houwink EJF, Hortensius OR, van Boven K, et al. Genetics in
Gastroenterology guidelines on the diagnosis and management of primary care: validating a tool to pre-symptomatically assess
Barrett’s oesophagus. Gut 2014;63(1):7e42. common disease risk using an Australian questionnaire on family
[8] Weusten B, Bisschops R, Coron E, et al. Endoscopic management history. Clin Transl Med 2019;8(1):17.
of barrett’s esophagus: European society of gastrointestinal [26] Casparie M, Tiebosch AT, Burger G, et al. Pathology data-
endoscopy (ESGE) position statement. Endoscopy 2017;49(2): banking and biobanking in The Netherlands, a central role for
191e8. PALGA, the nationwide histopathology and cytopathology data
[9] Lagergren J, Bergstrom R, Lindgren A, et al. Symptomatic network and archive. Cell Oncol 2007;29(1):19e24.
gastroesophageal reflux as a risk factor for esophageal adeno- [27] Hernán MA, Hernández-Dı́az S, Werler MM, et al. Causal
carcinoma. N Engl J Med 1999;340(11):825e31 [see comments]. knowledge as a prerequisite for confounding evaluation: an
[10] Bhat SK, McManus DT, Coleman HG, et al. Oesophageal application to birth defects epidemiology. Am J Epidemiol 2002;
adenocarcinoma and prior diagnosis of Barrett’s oesophagus: a 155(2):176e84.
population-based study. Gut 2015;64(1):20e5. [28] Ash S, Vaccaro BJ, Dabney MK, et al. Comparison of endoscopic
[11] Verbeek RE, Leenders M, Ten Kate FJ, et al. Surveillance of and clinical characteristics of patients with familial and sporadic
Barrett’s esophagus and mortality from esophageal adenocarci- Barrett’s esophagus. Dig Dis Sci 2011;56(6):1702e6.
noma: a population-based cohort study. Am J Gastroenterol [29] Peters Y, Van Grinsven E, Siersema PD. Effect OF family history
2014;109(8):1215e22. ON the risk OF BARRETT’S esophagus and esophageal
[12] Qumseya BJ, Bukannan A, Gendy S, et al. Systematic review and adenocarcinoma: systematic review and meta-analysis. Gastro-
meta-analysis of prevalence and risk factors for Barrett’s esoph- enterology 2021;160(6):S636e.
agus. Gastrointest Endosc 2019;90(5):707e717.e1. [30] Chak A, Faulx A, Kinnard M, et al. Identification of Barrett’s
[13] Yoon PW, Scheuner MT, Peterson-Oehlke KL, et al. Can family esophagus in relatives by endoscopic screening. Am J Gastro-
history be used as a tool for public health and preventive medi- enterol 2004;99(11):2107e14.
cine? Genet Med 2002;4(4):304e10. [31] Chak A, Ochs-Balcom H, Falk G, et al. Familiality in Barrett’s
[14] Guttmacher AE, Collins FS, Carmona RH. The family history– esophagus, adenocarcinoma of the esophagus, and adenocarci-
more important than ever. N Engl J Med 2004;351(22):2333e6. noma of the gastroesophageal junction. Cancer Epidemiol Bio-
[15] Drovdlic CM, Goddard KAB, Chak A, et al. Demographic and markers Prev 2006;15(9):1668e73.
phenotypic features of 70 families segregating Barrett’s oesoph- [32] Dhillon PK, Farrow DC, Vaughan TL, et al. Family history of
agus and oesophageal adenocarcinoma. J Med Genet 2003;40(9): cancer and risk of esophageal and gastric cancers in the United
651. States. Int J Canc 2001;93(1):148e52.
[16] Sun X, Elston R, Barnholtz-Sloan J, et al. A segregation [33] Jiang X, Tseng CC, Bernstein L, et al. Family history of cancer
analysis of Barrett’s esophagus and associated adenocarci- and gastroesophageal disorders and risk of esophageal and gastric
nomas. Cancer Epidemiol Biomarkers Prevention 2010;19(3): adenocarcinomas: a case-control study. BMC Canc 2014;14:60.
666e74. a publication of the American Association for Cancer [34] Juhasz A, Mittal SK, Lee TH, et al. Prevalence of Barrett
Research, cosponsored by the American Society of Preventive esophagus in first-degree relatives of patients with esophageal
Oncology. adenocarcinoma. J Clin Gastroenterol 2011;45(10):867e71.
[17] Sun X, Elston RC, Barnholtz-Sloan JS, et al. Predicting barrett’s [35] Lagergren J, Ye W, Lindgren A, et al. Heredity and risk of cancer
esophagus in families: an esophagus translational research of the esophagus and gastric cardia. Cancer Epidemiol, Bio-
network (BETRNet) model fitting clinical data to a familial markers Prevention 2000;9(7):757e60. a publication of the
paradigm. Cancer Epidemiol Biomarkers Prevention 2016;25(5): American Association for Cancer Research, cosponsored by the
727e35. a publication of the American association for cancer American Society of Preventive Oncology.
research, cosponsored by the Am Society Preventive Oncol. [36] Mussetto A, Manno M, Fuccio L, et al. Screening for Barrett’s
[18] Chak A, Lee T, Kinnard MF, et al. Familial aggregation of oesophagus with oesophageal capsule endoscopy in first-degree
Barrett’s oesophagus, oesophageal adenocarcinoma, and oeso- relatives of patients affected by Barrett’s oesophagus: results of
phagogastric junctional adenocarcinoma in Caucasian adults. Gut a pilot study. Arab J Gastroenterol 2013;14(2):51e4.
2002;51(3):323e8. [37] Rubenstein JH, Tavakkoli A, Koeppe E, et al. Family history of
[19] Kharazmi E, Babaei M, Fallah M, et al. Importance of tumor colorectal or esophageal cancer in barrett’s esophagus and
location and histology in familial risk of upper gastrointestinal potentially explanatory genetic variants. Clin Transl Gastro-
cancers: a nationwide cohort study. Clin Epidemiol 2018;10: enterol 2020;11(4):e00151.
1169e79. [38] Tofani CJ, Gandhi K, Spataro J, et al. Esophageal adenocarci-
[20] Romero Y, Cameron AJ, Schaid DJ, et al. Barrett’s esophagus: noma in a first-degree relative increases risk for esophageal
prevalence in symptomatic relatives. Am J Gastroenterol 2002; adenocarcinoma in patients with Barrett’s esophagus. United
97(5):1127e32. European Gastroenterol J 2019;7(2):225e9.
[21] Jones R, Junghard O, Dent J, et al. Development of the GerdQ, a [39] Rüdiger Siewert J, Feith M, Werner M, et al. Adenocarcinoma of
tool for the diagnosis and management of gastro-oesophageal the esophagogastric junction: results of surgical therapy based on
reflux disease in primary care. Aliment Pharmacol Ther 2009; anatomical/topographic classification in 1,002 consecutive pa-
30(10):1030e8. tients. Ann Surg 2000;232(3):353e61.
126 Y. Peters et al. / European Journal of Cancer 155 (2021) 116e126

[40] Chak A, Chen Y, Vengoechea J, et al. Variation in age at cancer [44] Fitzgerald RC, di Pietro M, O’Donovan M, et al. Cytosponge-
diagnosis in familial versus nonfamilial Barrett’s esophagus. trefoil factor 3 versus usual care to identify Barrett’s oesoph-
Cancer Epidemiol Biomarkers Prevention 2012;21(2):376e83. a agus in a primary care setting: a multicentre, pragmatic, rand-
publication of the American Association for Cancer Research, omised controlled trial. Lancet 2020;396(10247):333e44.
cosponsored by the American Society Preventive Oncology. [45] Sami SS, Subramanian V, Ortiz-Fernandez-Sordo J, et al.
[41] van Nistelrooij AMJ, van Marion R, van Ijcken WFJ, et al. Performance characteristics of unsedated ultrathin video
Germline variant in MSX1 identified in a Dutch family with endoscopy in the assessment of the upper GI tract: systematic
clustering of Barrett’s esophagus and esophageal adenocarci- review and meta-analysis. Gastrointest Endosc 2015;82(5):
noma. Fam Cancer 2018;17(3):435e40. 782e92.
[42] Fecteau RE, Kong J, Kresak A, et al. Association between [46] Peters Y, Schrauwen RWM, Tan AC, et al. Detection of Barrett’s
germline mutation in VSIG10L and familial Barrett neoplasia. oesophagus through exhaled breath using an electronic nose de-
JAMA Oncology 2016;2(10):1333e9. vice. Gut 2020;69(7):1169e72. https://doi.org/10.1136/gutjnl-
[43] Gabre J, Chak A, Rustgi A. Familial Barrett’s esophagus and 2019-320273. gutjnl-2019-320273.
esophageal adenocarcinoma. Curr Treat Options Gastroenterol [47] Ronkainen J, Aro P, Storskrubb T, et al. Prevalence of Barrett’s
2020;18(4):616e22. esophagus in the general population: an endoscopic study.
Gastroenterology 2005;129(6):1825e31.