How to Cite this Article: Calloway TJ, Martin LJ, Zhang X, Tandon A, Benson DW, Hinton RB. 2011. Risk factors for aortic valve disease in bicuspid aortic valve: A family-based study. Am J Med Genet Part A 155:1015–1020.
Bicuspid aortic valve (BAV, OMIM#109730) is the most common cardiovascular malformation (CVM), occurring in 1–2% of the general population, and is a risk factor for aortic valve disease (AVD) [Roberts, 1970; Ward, 2000; Rajamannan et al., 2003; Roberts et al., 2005]. AVD (stenosis and/or insufficiency) typically manifests later in life, affecting more than 2% of the population, and remains a significant surgical problem with approximately 100,000 valve replacement procedures performed each year in the US [Bonow et al., 2006; Nkomo et al., 2006; Otto, 2006; Rosamond et al., 2008; Lloyd-Jones et al., 2009]. The majority of AVD cases at any age have an underlying BAV, and longitudinal studies in young adults with BAV have shown that >20% ultimately develop AVD requiring intervention [Tzemos et al., 2008]. Two patterns of BAV morphology are commonly observed; ∼70% of cases have fusion of the right and left (RL) coronary cusps with the remainder consisting almost entirely of those with fusion of the right and non (RN) coronary cusps [Ward, 2000; Roberts and Ko, 2005]. There have been conflicting reports regarding the association between BAV morphology and AVD [Fernandes et al., 2004; Tzemos et al., 2008]. Fernandes et al.  identified an association between RN BAV and AVD in a pediatric population, while Tzemos et al.  found no association in an adult population . Despite clear evidence that BAV is a heritable condition with a recurrence risk of approximately 8%, it remains unclear whether BAV morphology is also genetically determined and whether it predicts the development of AVD [Cripe et al., 2004; Fernandez et al., 2009; Hinton et al., 2009].
MATERIALS AND METHODS
In this cross-sectional study, we sought to investigate risk factors for AVD in individuals with BAV using family-based information. BAV families were identified at Cincinnati Children's Hospital Medical Center (CCHMC) from 2003 to 2008 [Cripe et al., 2004; Hinton et al., 2009]. Families enriched for BAV were identified by probands with either BAV or hypoplastic left heart syndrome (HLHS), a severe form of aortic valve malformation. Our previous studies have shown that such families have similar recurrence risk for BAV [Cripe et al., 2004; Hinton et al., 2007]. Individuals with HLHS were excluded from analysis. Further, patients with genetic syndromes (e.g., Turner syndrome), infective endocarditis, or a history of rheumatic heart disease were excluded. This study was approved by the Institutional Review Board at CCHMC.
A comprehensive phenotypic analysis was performed using 2-D and Doppler echocardiography on all family members [Cripe et al., 2004]. BAV was defined as having two cusps instead of three in the parasternal short axis view, and valve morphology was identified as RL or RN [Cripe et al., 2004]. Outside echocardiogram reporting of BAV morphology (21% of BAV studies) was variable; in some cases, morphology was identified as indeterminate. Those with indeterminate morphology were excluded from AVD risk and BAV morphology heritability analyses. AVD was defined as at least mild stenosis (mean gradient ≥10 mmHg), or mild or greater insufficiency using established qualitative observations [Bonow et al., 2006; Otto, 2006]. Other CVMs when present were noted. History of aortic valve intervention (catheter-based or surgical) was determined from medical record review; in these cases, AVD status was defined pre-procedure. In some individuals, it was not possible to determine BAV status using echocardiography at our institution. In these cases, medical records, for example, echocardiogram report, cardiac catheterization report, and/or operative note during valve replacement surgery, from outside institutions were relied on for the diagnosis.
Family-based data were used to assess the effects of age, gender, and BAV morphology on AVD risk; isolated trivial stenosis or insufficiency was censored. To account for both intra- and inter-family variability, a Generalized Estimating Equation (GEE) model was used. Covariance structure was evaluated, and the unstructured covariance model was determined to fit the data best. All analyses were performed using Statistical Analysis Software version 9.2 (SAS Institute, Inc., Cary, NC). A P-value <0.05 was considered significant.
It has been suggested previously that BAV morphology has strong genetic determinates [Fernandez et al., 2009]. Thus, to examine the extent to which the relationship between BAV and AVD is confounded by shared underlying genetic factors, we also estimated the heritability (h2; proportion of phenotypic variation attributable to genetics) of AVD and BAV morphology using variance components analysis in Sequential Oligogenic Linkage Analysis Routines (SOLAR). This program was optimal to use in this study as it can handle diverse pedigree structures. To confirm segregation patterns, we examined BAV morphology in family members to determine if the concordance of morphology was greater than expected by chance. To determine the expected concordance rate we first identified all pairs of first-degree relatives who both had BAV and whose morphology was known. We then examined the RL frequency by proband status. We found that the probands had a lower frequency of RL (66.7) compared with nonprobands (75.5). Therefore, the expected concordance rate was calculated as the probability of concordance for RL (RL frequencyprobands × RL frequencynonprobands) plus the probability of concordance for RN (RN frequencyprobands × RN frequencynonprobands). We then used a χ2 goodness-of-fit test to compare the observed proportion concordant to the expected proportion concordant.
A total of 1,128 individuals from 226 families were evaluated (mean age ± SD, 27 ± 20 years; range, 0–88 years; 41% pediatric [≤18 years of age]; 591 [52%] males). There were 281 (25%) family members with BAV, and 167 (59%) of these had AVD (Table I; Fig. 1). Among BAV patients, 204 were male (73%, 2.7:1 male/female), and 73 had other CVMs (Table II). In addition, there were 34 family members with AVD and tricommissural aortic valve morphology (not BAV). AVD in family members with BAV consisted of 59 (35%) with stenosis, 63 (38%) with insufficiency, and 45 (27%) with both stenosis and insufficiency. Trivial disease was censored in 33 cases. RL morphology was present in 160 (57%), RN in 74 (26%), and indeterminate in 47 (17%) individuals. A total of 56 individuals with BAV underwent intervention of the aortic valve, 23/56 had indeterminate morphology (Table I).
Table I. Study Population
Number of patients (%)
RL morphology (n = 160)
RN morphology (n = 74)
Indeterminate morphology (n = 47)
Total (n = 281)
Individuals with more than two CVMs were only counted once.
For the heritability analysis, 940 individuals from 165 families in whom relationships were genetically verified were included (127 BAV; 38 HLHS) (Fig. 1). Although our previous studies identified a high heritability for BAV (h2 = 0.89 ± 0.06, P < 0.00001) [Cripe et al., 2004], the heritability of AVD was only 0.07 ± 0.17 (P = 0.33). Further, when restricting the analysis to individuals with BAV (n = 172), AVD still did not exhibit significant evidence of heritability (h2 = 0.23 ± 0.46; P = 0.32). For the heritability analysis of BAV morphology, we included 178 individuals whose familial relationships were validated genetically [Hinton et al., 2009]. Overall, the heritability of aortic valve morphology was 0.61 ± 0.21 (P = 0.11). When examining concordance of BAV morphology, we examined 29 first-degree relative pairs who had BAV of known morphology. We found 22 concordant pairs, resulting in a concordance rate of 76%. Based on the frequencies of RL and RN in probands and nonprobands, we expected 59% of individuals to be concordant by chance. Using a goodness-of-fit test, we determined that while this difference was not statistically significant, it approached significance (P = 0.058). Taken together these results suggest that there may be a significant but incomplete genetic basis underlying BAV morphology and AVD is determined largely by nongenetic factors.
AVD Risk Differs by BAV Morphology
Effects of BAV morphology, age, and gender on AVD risk were assessed. AVD was significantly associated with BAV morphology (P = 0.0027) and age (P = 0.0068) (Table I; Fig. 2). No significant effect of gender was detected (data not shown). There was a significant interaction between BAV morphology and age in predicting AVD risk (P = 0.055). The interaction term permits not only differences in AVD risk by BAV morphology but also allows RN and RL to have different age trajectories with respect to AVD risk. At birth, the odds of AVD in RL BAV were estimated to be 23% of that in RN (95% CI: 9–60%). However, the risk of AVD in individuals with RL BAV increases with age, so that by age 42, there is a similar risk of AVD in individuals with RN and RL BAV (Fig. 3). Two additional analyses examining AVD, one including all instances of trivial disease and one including only moderate or greater (clinically significant) AVD, were performed separately with similar results to the reported analysis (data not shown). Taken together, these findings demonstrate that BAV with RN morphology is more likely to manifest AVD in childhood, while BAV with RL morphology is more likely to manifest AVD in adulthood.
In this cross-sectional study, we identified BAV morphology as an age-dependent risk factor for AVD; pediatric patients with AVD are more likely to have RN BAV morphology while adult AVD patients are more likely to have RL BAV. These findings suggest that different BAV morphologies may represent different developmental abnormalities, such that RN morphology predisposes the aortic valve to early manifestation of disease and RL morphology results in latent AVD. Interestingly, similar findings have been demonstrated in Turner syndrome patients, who have been shown to have an increased risk for BAV with RL morphology and a low risk for AVD early in life [Sybert, 1998; Sachdev et al., 2008].
Outcomes of AVD in children have recently been described, but the risk factors for AVD in this age group are poorly understood [Mahle et al., 2010]. Indeed, a National Heart Lung and Blood Institute Working Group on AVD recently identified the need to identify “clinical risk factors for the distinct phases of initiation and progression of AVD” [Rajamannan et al., 2009]. From a clinical perspective, the age-dependent association of BAV morphology and AVD has important implications for the prognosis and clinical surveillance of BAV patients. Potential clinical benefits may include improved surveillance strategies in pediatric patients with RN BAV, and ultimately a better understanding of AVD natural history. Current recommendations for BAV patients include screening echocardiograms every 5 years [Bonow et al., 2006]. Given the results of the present study, surveillance in pediatric patients with RN morphology every 2 years may identify new AVD in a timelier manner. Conversely, individuals with RL BAV could be monitored less aggressively in early childhood as the risk of having AVD at this time is relatively low in our model.
Importantly, we also examined the possibility that BAV morphology type may be due to underlying genetic factors having pleiotropic effects on both BAV morphology and AVD. However, unlike BAV, which is strongly genetically determined (high heritability) [Cripe et al., 2004], BAV morphology does not exhibit segregation patterns in families that are consistent with strong genetic determination of morphology (moderate heritability), that is, families with multiple individuals affected with BAV do not all have the same BAV morphology. Further, when we estimated the heritability of AVD in our cohort, there was no evidence of underlying genetic effects (absent to low heritability). Taken together, these findings suggest that while BAV is determined largely by genetic effects and is a risk factor for AVD, epigenetic and environmental factors play a significant role in BAV morphology and later manifestations of AVD. For these reasons, studies of epigenetic mechanisms are warranted.
The moderate estimate of heritability for BAV morphology in the human families we studied is in contrast to findings in animal studies that suggest BAV morphology is entirely genetic [Fernandez et al., 2009]. Nongenetic factors that distinguish mice and hamsters from humans include a significantly higher heart rate resulting in a shorter diastolic component to the cardiac cycle and consequently different hemodynamics, potentially including blood flow patterns during cardiac development, such that different BAV morphology patterns may predispose to different disease states. Understanding the cellular and molecular underpinnings of AVD, including the roles of cells and matrix proteins, may further elucidate this relationship [Aikawa et al., 2006; Hinton et al., 2006]. Taken together, these findings are consistent with human genetics studies that identify aortic valve malformation as a complex trait, but demonstrate that associated disease states, that is AVD, are determined largely by nongenetic factors [Martin et al., 2007; Fernandez et al., 2009].
Despite the unique strengths of family-based analyses, this study has notable limitations. The majority of BAVs were identified in pediatric patients presenting with AVD resulting in an ascertainment bias. There was an unusually high frequency of BAV with indeterminate morphology (17%) due to dependence in some cases on medical records from nonlocal family members (74%) and inclusion of patients with a history of aortic valve replacement (47%). We believe that the decreasing probability of AVD risk with older age (Fig. 3) may be due in part to attrition secondary to an increasing proportion of patients undergoing intervention. For the AVD risk analysis, there was an inadequate sample size to stratify patients with both BAV and AVD (n = 167) by disease type. Finally, we do not have detailed information pertaining to potentially pertinent adult comorbidities, for example, hyperlipidemia, hypertension, and diabetes mellitus, and this should be considered in future studies. Nonetheless, family-based data provide a unique cross-sectional perspective on a latent disease phenotype that is difficult to study using a longitudinal study design.
In summary, BAV morphology may predict the timing of AVD onset, that is, individuals with RN BAV are more likely to develop AVD as children, while individuals with RL BAV are more likely to develop AVD as adults. Further, while BAV has a strong genetic component, BAV morphology appears to be only moderately influenced by genetic factors; moreover, inheritance of AVD is not consistent with underlying Mendelian segregation. Thus, these studies suggest that the relationship between AVD and BAV and its morphology is not due to shared underlying genetic influences. Clearly, future studies should examine other factors which may help explain the increased risk of AVD in BAV patients.
The authors would like to thank the families who participated in the study. This work was supported by grants from the National Institutes of Health HL069712 (D.W.B.), HL074728 (L.J.M., D.W.B.), and HL085122 (R.B.H.).