Abstract
Purpose :
Heritability (h2) estimates of primary open angle glaucoma (POAG) may be underestimated due to insidious onset, low prevalence, and patient-related incognizance. POAG related endophenotypes, intraocular pressure (IOP), central corneal thickness (CCT), retinal nerve fibre layer (RNFL) thickness, and cup-disc ratio (CDR) are heritable and are useful to physicians and in genetic epidemiological studies. This systematic review and meta-analyses examine h2 estimates of POAG and its endophenotypes, and underlying factors that contribute to its heterogeneity.
Methods :
A systematic retrieval of literature was performed, using the population, exposure, comparator, outcome (PECO) strategy up to Sept. 2017 within PubMed, ScienceDirect, EMBASE, and Web of Science by two independent reviewers. We included original studies that 1) had h2 as the outcome or allowed h2 to be re-estimated from intraclass correlations/linear regression coefficients, and 2) had POAG or related endophenotypes as the traits of interest. An inverse-variance random-effects meta-analyses of h2 estimates for these traits were performed. Sub-group, moderation analyses, and assessments of heterogeneity were conducted when traits had more than 10 studies.
Results :
The total number of eligible articles was 54; 27 twin, 24 family, and 3 genome-wide association studies (GWAS). The number of h2 estimates for POAG, IOP, CCT, RNFL, and CDR were 5, 38, 15, 3, and 8, respectively. The corresponding mean (95% confidence interval) pooled h2 (%) estimates were 52 (30-70), 49 (42-55), 81 (76-86), 74 (53-86), and 58 (48-66), respectively. Heritability estimates were on average higher in twin studies than family studies (e.g., h2 for CCT was 88 (87-89)% for twin, and 64 (62-66)% for family studies), and in studies using variance component methods as opposed to correlations (e.g. h2 for CDR was 74 (72-76)% for variance component, and 55 (51-58)% for correlation). Substantial heterogeneity was found within endophenotype analyses.
Conclusions :
Heterogeneity in h2 estimates arise not only from study design, like twin vs family, but also analysis method, where variance component analyses estimated higher trait h2 than correlations. These pooled heritabilities provide the most accurate estimates to date of the total genetic variance in POAG and its endophenotypes that can ultimately be explained by gene finding studies.
This is an abstract that was submitted for the 2018 ARVO Annual Meeting, held in Honolulu, Hawaii, April 29 - May 3, 2018.