Twin and family studies reveal strong environmental and weaker genetic cues explaining heritability of eosinophilic esophagitis

Abstract

Background: Eosinophilic esophagitis (EoE) is a chronic antigen-driven allergic inflammatory disease, likely involving the interplay of genetic and environmental factors, yet their respective contributions to heritability are unknown.

Objective: To quantify the risk associated with genes and environment on familial clustering of EoE.

Methods: Family history was obtained from a hospital-based cohort of 914 EoE probands (n = 2192 first-degree "Nuclear-Family" relatives) and an international registry of monozygotic and dizygotic twins/triplets (n = 63 EoE "Twins" probands). Frequencies, recurrence risk ratios (RRRs), heritability, and twin concordance were estimated. Environmental exposures were preliminarily examined.

Results: Analysis of the Nuclear-Family-based cohort revealed that the rate of EoE, in first-degree relatives of a proband, was 1.8% (unadjusted) and 2.3% (sex-adjusted). RRRs ranged from 10 to 64, depending on the family relationship, and were higher in brothers (64.0; P = .04), fathers (42.9; P = .004), and males (50.7; P < .001) than in sisters, mothers, and females, respectively. The risk of EoE for other siblings was 2.4%. In the Nuclear-Family cohort, combined gene and common environment heritability was 72.0% ± 2.7% (P < .001). In the Twins cohort, genetic heritability was 14.5% ± 4.0% (P < .001), and common family environment contributed 81.0% ± 4% (P < .001) to phenotypic variance. Probandwise concordance in monozygotic co-twins was 57.9% ± 9.5% compared with 36.4% ± 9.3% in dizygotic co-twins (P = .11). Greater birth weight difference between twins (P = .01), breast-feeding (P = .15), and fall birth season (P = .02) were associated with twin discordance in disease status.

Conclusions: EoE RRRs are increased 10- to 64-fold compared with the general population. EoE in relatives is 1.8% to 2.4%, depending on relationship and sex. Nuclear-Family heritability appeared to be high (72.0%). However, the Twins cohort analysis revealed a powerful role for common environment (81.0%) compared with additive genetic heritability (14.5%).

Keywords: Eosinophilia; drug hypersensitivity; food allergy; gastrointestinal diseases; gene-environment interaction; heritability; immune system diseases; medical genetics; skin diseases; twins.

Figure I

Recruitment Algorithms and Case Identification for Nuclear-Family and Twin Cohorts A. Nuclear-Family Cohort. B. Twin Cohort. A. Nuclear-Family cohort from the Cincinnati Center for Eosinophilic Disorders; B. EoE Twins International Registry cohort. EGD, esophagogastroduodenescopy; EoE, eosinophilic esophagitis; Not EoE, unaffected by eosinophilic esophagitis; MZ, monozygotic; DZ, dizygotic.

Figure I

Recruitment Algorithms and Case Identification for Nuclear-Family and Twin Cohorts A. Nuclear-Family Cohort. B. Twin Cohort. A. Nuclear-Family cohort from the Cincinnati Center for Eosinophilic Disorders; B. EoE Twins International Registry cohort. EGD, esophagogastroduodenescopy; EoE, eosinophilic esophagitis; Not EoE, unaffected by eosinophilic esophagitis; MZ, monozygotic; DZ, dizygotic.

Figure II

Rates of EoE in Twin Cohort and Nuclear-Family Cohort Sibling Non-probands Frequency of EoE in dizygotic (DZ) non-proband co-twins (n=36), non-proband Nuclear-Family siblings of proband (n=782) compared to population prevalence by X²df=1. MZ, monozygotic.

Figure III

Summary Pedigrees Support a Complex Mode of EoE Inheritance. A. Nuclear Family Cohort. B. Twin Cohort (Monozygotic). C. Twin Cohort (Dizygotic) Diamond shape represents both brothers and sisters whose number range by “Number of probands’ siblings.” Frequency (%) is the percent of families with that summary pedigree as a percent of all families in panels A, B, and C. In the large Nuclear-Family cohort, families with unaffected parents and at least one additional brother or sister with EoE comprise 1.9%.

Figure III

Summary Pedigrees Support a Complex Mode of EoE Inheritance. A. Nuclear Family Cohort. B. Twin Cohort (Monozygotic). C. Twin Cohort (Dizygotic) Diamond shape represents both brothers and sisters whose number range by “Number of probands’ siblings.” Frequency (%) is the percent of families with that summary pedigree as a percent of all families in panels A, B, and C. In the large Nuclear-Family cohort, families with unaffected parents and at least one additional brother or sister with EoE comprise 1.9%.

Figure III

Summary Pedigrees Support a Complex Mode of EoE Inheritance. A. Nuclear Family Cohort. B. Twin Cohort (Monozygotic). C. Twin Cohort (Dizygotic) Diamond shape represents both brothers and sisters whose number range by “Number of probands’ siblings.” Frequency (%) is the percent of families with that summary pedigree as a percent of all families in panels A, B, and C. In the large Nuclear-Family cohort, families with unaffected parents and at least one additional brother or sister with EoE comprise 1.9%.

Figure IV

A: Twin Cohort ACE Model More Accurately Estimates Heritability by Separating Common Environment. B. Twin Cohort ACE Heritability Model Estimates Compared to Twin Cohort AE and Nuclear-Family AE Cohort Estimates A. “ACE” latent class path analysis estimates (point prevalence estimate at 5.5/10,000) represent a generalized model across all twins and all families. By convention, latent variables are represented as ovals and measured variables as squares; MZ, monozygotic; DZ, dizygotic. B. Twin cohort ACE path analysis (black) separates common family environment, estimating heritability at 14.5±4% (p<0.001) with superior model fit (p=0.56). As expected, using the same data and model but excluding common family environment (dark gray) inflates heritability to 99.5%. Similarly, Nuclear-Family cohort (light gray) inflates heritability estimate to 72±2.7% (p<0.001;liability threshold model); A, additive genetic variance (heritability); C, common, shared household, environmental variance; E, unique environment “error” variance.

Figure IV

A: Twin Cohort ACE Model More Accurately Estimates Heritability by Separating Common Environment. B. Twin Cohort ACE Heritability Model Estimates Compared to Twin Cohort AE and Nuclear-Family AE Cohort Estimates A. “ACE” latent class path analysis estimates (point prevalence estimate at 5.5/10,000) represent a generalized model across all twins and all families. By convention, latent variables are represented as ovals and measured variables as squares; MZ, monozygotic; DZ, dizygotic. B. Twin cohort ACE path analysis (black) separates common family environment, estimating heritability at 14.5±4% (p<0.001) with superior model fit (p=0.56). As expected, using the same data and model but excluding common family environment (dark gray) inflates heritability to 99.5%. Similarly, Nuclear-Family cohort (light gray) inflates heritability estimate to 72±2.7% (p<0.001;liability threshold model); A, additive genetic variance (heritability); C, common, shared household, environmental variance; E, unique environment “error” variance.