Family screening for bicuspid aortic valve and aortic dilatation: a meta-analysis

Abstract

Aims

International guidelines recommend screening of first-degree relatives (FDR) of people with bicuspid aortic valves (BAVs). However, the prevalence of BAV and of aortic dilatation amongst family members is uncertain.

Methods and results

A systematic review and meta-analysis of original reports of screening for BAV. Databases including MEDLINE, Embase, and Cochrane CENTRAL were searched from inception to December 2021 using relevant search terms. Data were sought on the screened prevalence of BAV and aortic dilatation. The protocol was specified prior to the searches being performed, and standard meta-analytic techniques were used. Twenty-three observational studies met inclusion criteria (n = 2297 index cases; n = 6054 screened relatives). The prevalence of BAV amongst relatives was 7.3% [95% confidence interval (CI) 6.1%–8.6%] overall and per family was 23.6% (95% CI 18.1%–29.5%). The prevalence of aortic dilatation amongst relatives was 9.4% (95% CI 5.7%–13.9%). Whilst the prevalence of aortic dilatation was particularly high in relatives with BAV (29.2%; 95% CI 15.3%–45.1%), aortic dilatation alongside tricuspid aortic valves was a more frequent finding, as there were many more family members with tricuspid valves than BAV. The prevalence estimate amongst relatives with tricuspid valves (7.0%; 95% CI 3.2%–12.0%) was higher than reported in the general population.

Conclusion

Screening family members of people with BAV can identify a cohort substantially enriched for the presence of bicuspid valve, aortic enlargement, or both. The implications for screening programmes are discussed, including in particular the substantial current uncertainties regarding the clinical implications of aortic findings.

Structured Graphical Abstract

Structured Graphical Abstract.

Left: prevalence estimates from meta-analysis for identification of bicuspid aortic valves (BAV) in screened relatives and identification of an affected family with at least one individual with newly found BAV per families screened, in addition to prevalence estimates for aortic dilatation of screened relatives overall and specifically amongst individuals also identified to have BAV or tricuspid aortic valves. Right: a visual representation of the estimated proportions of each finding from pooled screening programmes, by individuals (above) and by families screened (below).

See the editorial comment for this article ‘Family screening for bicuspid aortic valve: indicated, but easy to implement?’, by A. Evangelista and H.I. Michelena, https://doi.org/10.1093/eurheartj/ehad057.

What is already known on this topic

Bicuspid aortic valve (BAV) is the commonest congenital heart defect. It is associated with aortic dilatation and can lead to substantial morbidity and need for cardiac surgery. It is known to be heritable in some families, and international guidelines recommend that family members undergo screening. However, the prevalence of (i) BAV and (ii) aortic dilatation in family members is uncertain, and there are no systematic nationally mandated family screening programmes. Furthermore, current studies have not been large enough to determine, with certainty, whether aortic enlargement is confined to relatives with bicuspid valves or is found in family members irrespective of valve morphology.

What this study adds

We performed a systematic review and meta-analysis of published studies of screened family members of index cases of BAV. This study adds three key findings. First, it provides the most reliable estimates available for the prevalence of BAV per individual screened, 7.3% (95% CI 6.1%–8.6%), and per family screened, 23.6% (95% CI 18.1%–29.5%). Second, it showed that aortic dilatation was quite common amongst screened relatives (9.4%; 95% CI 5.7%–13.9%) and affected nearly a third of relatives with BAV (29.2%; 95% CI 15.3%–45.1%). Third, aortic dilatation was not confined to those with BAV and occurred in 7.0% (95% CI 3.2%–12.0%) of those with normal valves, more than would be expected in the general population. The implications for the design and implementation of screening programmes are discussed.

Introduction

Bicuspid aortic valve (BAV) is the commonest congenital heart defect, with a prevalence of about 0.5%.^1,2 The natural history of BAV is often complicated by aortic valve stenosis, regurgitation, or endocarditis, and it is commonly associated with coarctation, dilatation, or occasionally dissection of the thoracic aorta.³ Contemporary cohort studies report that, whilst mortality rates appear similar to the general population over 10⁴ and 20³ years of follow-up, and aortic dissection is uncommon,⁵ cardiac or aortic surgical intervention is frequently necessary, occurring in about 25% of patients over 20 years.³

Bicuspid aortic valve has a genetic basis in some patients. It is strongly associated with the Turner syndrome⁶ and other genetic conditions,⁷ and BAV with or without aortopathy can be heritable as an autosomal dominant trait (sometimes with incomplete penetrance) in some families,⁸ and in a subset of these families (<5%), rare variants at particular loci have been identified which segregate with disease and can explain the genetic basis of the disease in that family.⁹ The extent to which remaining cases are genetically mediated and the nature of the genetic basis in familial cases is not fully understood. Estimates of the heritability of BAV have varied widely with h² estimates ranging between 0.07 (p > 0.05) and 0.47 (p < 0.01) in contemporary studies.^10,11 Estimates of the heritability of BAV-related aortic dilatation have been moderate: (h² 0.30 p > 0.05¹¹ and 0.32 p < 0.001¹²).

Screening of first-degree relatives (FDR) is widely recommended by international guidelines.^13–17 European [European Society of Cardiology (ESC)],¹⁶ US [American Heart Association (AHA)/American College of Cardiology (ACC)/ Society of Thoracic Surgeons (STS)),^13,14 and Canadian College of Cardiology (CCS)¹⁵ guidelines all recommend screening for FDR. European Society of Cardiology guidelines particularly recommend screening of FDR that are (i) male, (ii) athletic, and (iii) hypertensive.¹⁷ Despite these recommendations, there are no nationally implemented systematic screening programmes, and screening occurs sporadically and perhaps inequitably.

The decision to implement a screening programme such as those recommended by guidelines (but rarely undertaken in practice) is based on a careful assessment of the intended benefits, unintended harms, and expected costs of the programme relative to other claims on those resources. This depends critically, though not exclusively, on the estimated prevalence of the condition in the population screened. A number of studies have sought to estimate the prevalence of BAV in family members of probands or have provided published data which allow an estimate to be made. These studies have generally been small, consisting of just a few hundred participants at most, without any very large studies (e.g. n > 1000). Collation of these studies is likely to provide a more reliable estimate of the overall prevalence of both BAV and its associated aortopathy amongst screened family members than the results of any individual study. This information, when combined with a systematic review of the literature on screening for BAV, could be used to identify gaps in knowledge and inform healthcare policy in determining whether and how to implement international guideline recommendations to screen family members. We therefore undertook a meta-analysis and systematic review of published studies, to estimate the prevalence of both BAV and aortopathy in family members, in addition to a systematic review and appraisal of the literature on screening of family members for BAV.

Methods

We performed a systematic review and meta-analysis of studies enrolling family members of probands with BAV. The aims and analyses to be performed were pre-specified, and the study was registered on the Open Science Framework (OSF) (https://osf.io/ewb47/)¹⁸ in advance of the searches being performed. The findings are reported in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement. The systematic review and meta-analysis did not involve accessing or otherwise processing patient-identifiable information and hence did not require ethical approval.

Search strategy

We systematically searched MEDLINE, Embase, and the Cochrane Central Register of Controlled Trials (Central) from inception to 5 December 2021 using a combination of relevant index terms, MeSH terms, and keywords (see Supplementary material online, Table S1). Forward and backward citation searches were then conducted on the relevant papers. No limits were set for language. Two reviewers (JJHB and RF) independently screened titles retrieved for relevance and then assessed full texts for eligibility. Disagreements were resolved by consensus. We identified and excluded duplicates and collated multiple reports of the same study so that each study, rather than each report, was the unit of interest in the review (duplicates are shown in Supplementary material online, Table S2).

Eligibility criteria and population

Studies were eligible for inclusion if they enrolled family members of adults or children with BAV for screening and collected data relevant to the pre-specified outcomes of interest. No constraints were set on the age of screened family members or the timing of screening. We excluded studies only screening for umbrella conditions that include BAV, such as left ventricular outflow tract obstruction and other left-sided cardiac abnormalities, without specifically identifying BAV as a subgroup. Case reports, or very small case series, which we defined as describing five cases or fewer, were excluded. We excluded studies that exclusively enrolled participants with (i) syndromic forms of BAV (e.g. Turner syndrome) as family members in such genetically determined conditions are likely to be enriched for disease or (ii) complex forms of left ventricular outflow malformation, such as hypoplastic left heart syndrome.¹⁹

Outcomes of interest

We collected data on the prevalence of BAV in all family members screened, in FDR only (sensitivity analysis), and at the family level, counting a family as affected if at least one additional family member was affected. We also collected data on the prevalence of aortic dilatation in all relatives screened and in FDR only (sensitivity analysis), using the definition of dilatation used in each study. We analysed aortic dilatation in all screened relatives, amongst those found to have concomitant BAV and amongst those with apparently normal (tricuspid) aortic valves. Outcomes relating to the systematic review of the literature supporting screening for BAV appraised against the UK National Screening Committee (UK NSC) framework can be found in the supplement.

Data extraction

Data were extracted using pre-formatted excel spreadsheets (Office 365, Microsoft Corporation) on study design (type, setting, method of screening, timing, and screening invitation), population (type of relative, participant numbers, and demographics), and prevalence of BAV and aortic dilatation. Information on study characteristics, screening methodology, participant information, and the reported prevalence of BAV are shown in Table 1.

Table 1.

Study details and participant baseline characteristics

Study	Population	Participants	Number of participants total (index/relatives)	Age of participants index/relatives	Sex of participants index/relatives	Relative screened	Method of screening
			(N)	(years)	(% male)
Cozijnsen 202120	The Netherlands	Isolated BAV	375 (118/257)	60.0/48.0	69.5%/34.6%	FDR	TTE/surgery
Miller 202121	The USA	97.2% BAV/BAV + TAA	639 (251/388)	3.68/6.91	73.7%/49.2%	Siblings	TTE
Tessler 202122	France and Israel	BAV	408 (107/301)	48.1/40.2	72.5%/46.6%	FDR	TTE
Massardier 202023	Canada	BAV without CTD	695 (213/482)	11.0/−	68.5%/50.6%	FDR	TTE
Galian-Gay 201911	Spain	BAV without CTD	980 (256/724)	46.4/41.7	71.0%/49.2%	FDR	TTE
Dayan 201924	International	BAV	218 (78/140)	−/39.0	−/52.1%	FDR	TTE, exam
Bissell 201625	The UK	BAV	199 (57/142)	–/–	–/–	FDR	CMR/TTE
Dayan 201426	Uruguay	BAV	21 (0/21)	–/–	–/–	FDR	TTE
Hales 201427	The USA	Isolated BAV	388 (181/207)	<18/7.0	–/–	Siblings	TTE
Demir 201328	Turkey	60% isolated BAV + other LSCA	272 (66/206)	8.66/30.8	81.4%/48.3%	Sequential sampling of FDR	TTE
Carro 201329	Spain	BAV	378 (90/288)	49.6/−	75.0%/66.7%	FDR	TTE
Robledo-Carmona 201330	Spain	Isolated BAV	448 (100/348)	46.8/−	65.6%/45.0%	FDR	TTE/surgery
Panayotova 201131	The UK	BAV	76 (24/52)	–/–	–/–	FDR	TTE
Mahle 201132	The USA	BAV	276 (134/142)	<18/12.3	–/–	Siblings	TTE
Kerstjens-Frederikse 201133	The Netherlands	25% BAV + LVOTO without NCA	591 (142/449)	<18/8.5	67.0%/53.2%	FDR	TTE, ECG
Calloway 201110	The USA	BAV	1354 (226/1128)	<18/27.0	−/52.4%	Sequential sampling of FDR	TTE
Biner 200934	The USA	BAV	102 (49/53)	46.5/41.0	66.7%/49.0%	FDR	TTE
Loscalzo 200735	The USA	69% BAV + TAA	132 (8/124)	–/–	88.9%/54.5%	Full pedigrees	TTE/Hx
Martin 200736	The USA	BAV	353 (38/315)	–/–	65.8%/−	Sequential sampling of FDR	TTE
Lewin 200437	The USA	40% BAV + AVS/CoA	149 (45/104)	–/–	71.0%/−	FDR	TTE
Cripe 200438	The USA	BAV	309 (50/259)	<18/0–78	66.0%/45.2%	Sequential sampling of FDR	TTE/FHx
Huntington 199739	Canada	BAV	220 (30/190)	45.0/−	–/–	FDR	TTE
Emanuel 197840	The UK	BAV	229 (41/188)	36.3/−	73.8%/−	FDR	TTE, CXR, exam

AVS, aortic valve stenosis; BAV, bicuspid aortic valve; CoA, coarctation of the aorta; CTD, connective tissue disease; CXR, chest radiograph, exam, clinical examination; FHx, family history; FDR, first-degree relative; Hx, clinical history including family history; LSCA, left-sided cardiac abnormality; LVOTO, left ventricular outflow tract obstruction; NCA, non-cardiac abnormalities; SDR, second-degree relative; TAA, thoracic aortic aneurysm.

*BAV/aortic stenosis.

Synthesis of results, statistics, and design

Prevalence and risk ratios were calculated using DerSimonian and Laird random-effect model meta-analyses, with heterogeneity estimated from the inverse-variance fixed-effect model in Stata (17.0, StataCorp LLC, College Station, TX, USA).^41,42 Heterogeneity was quantified using the I² measure and the p-value from the chi-square test. Strict thresholds for interpreting I² are not recommended, but >50% and heterogeneity p < 0.10 may represent substantial to considerable heterogeneity.⁴³ Small-study bias was assessed for using funnel plots and the Egger’s regression-based test where included studies ≥10.⁴⁴ We calculated number needed to screen as 1/(prevalence of BAV in screened relatives—prevalence of BAV identified before screening), assuming all relatives underwent screening. Descriptive statistics are reported as means ± standard deviation (SD). For all statistics relating to probands, averages were calculated as the mean for each study, weighted by the number of probands in the study as a proportion of the total number of probands in that analysis; for statistics relating to relatives, averages were means weighted by the number of relatives participating in each study, as a proportion of the number of relatives in that analysis. Statistics were undertaken using IBM SPSS (version 25). Subgroup comparison was made using the Cochran’s Q statistic provided by Stata 17.0.

Subgroup and sensitivity analysis

Subgroup analysis was performed by studies that included paediatric and adult probands, paediatric studies were separated on the basis that they exclusively recruited paediatric probands. Sensitivity analyses were performed with only (i) inclusion of studies that studied FDR (excluding studies on siblings only^21,27,32) and (ii) inclusion of studies with ‘low’ and ‘moderate’ risk of bias (see Supplementary material online, Table S3). Further analyses were performed for estimates of BAV prevalence based on study-related factors that may alter prevalence by excluding studies with (iii) ‘enriched’ cohorts (described as ‘enriched’ by authors)^10,36,38 and (iv) studies that recruited probands undergoing aortic valve surgery³¹ (Table 2).

Table 2.

Sensitivity analysis and subgroup analysis

Aspect interrogated	Outcome	Prevalence %
		(95% CI)
Subgroup analysis
Individual BAV prevalence	Paediatric proband	7.6% (5.9%–9.5%)11,20,22,24–26,29–31,34,35,39,40
	Adult proband	7.0% (5.3%–8.8%)10,21,23,27,28,32,33,36–38
Family BAV prevalence	Paediatric proband	22.2% (15.9%–29.3%)11,20,22,29,30,35,39,40
	Adult proband	25.7% (15.1%–37.9%)23,28,33,36,38
Aortic dilatation in BAV and TAV	Paediatric proband	5.2% (2.5%–8.8%)10,21,23,28,36
	Adult proband	12.4% (6.8%–19.4%)11,20,22,24,29–31,34,35,39,45
Aortic dilatation in BAV	Paediatric proband	21.6% (6.0%–42.6%)10,21,23,36
	Adult proband	34.6% (20.4%–50.2%)11,20,29–31,35,45
Aortic dilatation in TAV	Paediatric proband	3.6% (0.3%–9.9%)21,23,36
	Adult proband	8.9% (4.0%–15.4%)11,20,22,24,30,35,39,45
Sensitivity analysis
Individual BAV prevalence	‘Low’ or ‘moderate’ risk of bias for screening	7.0% (5.6%–8.4%)10,11,20–24,27–30,33,34,37,39
	FDR only	7.1% (5.8%–8.5%)10,11,20,22–26,28–31,33–40
	Removal of ‘enriched’ cohorts	7.1% (5.7%–8.5%)11,20–35,37,39,40
	Removal of proband aortic valve surgery cohorts	7.5% (6.2%–8.8%)10,11,20–25,27–30,32–40 *4.3% (0.4%–10.8%)26,31
Family BAV prevalence	‘low’ or ‘moderate’ risk of bias for screening	19.1% (15.0%–23.6%)11,22,23,28–30,33,38–40,46
	FDR only	23.6% (18.1%–29.5%)11,20,22,23,28–30,33,35,36,38–40
	Removal of ‘enriched’ cohorts Removal of proband aortic valve surgery cohorts	22.7% (17.3%–28.7%)11,20,22,23,28–30,33,35,36,39,40
Aortic dilatation in BAV and TAV	‘Low’ or ‘moderate’ risk of bias for screening	9.4% (5.7%–13.9%)10,11,20–24,28–30,34,39
	FDR only	9.3% (5.3%–14.0%)10,11,20,22,23,26,28–31,34–36,39,45
Aortic dilatation in BAV	‘Low’ or ‘moderate’ risk of bias for screening	29.0% (12.2%–49.0%)10,11,20,21,23,29,30
	FDR only	26.3% (13.0%–41.9%)10,11,20,23,26,29–31,35,45
Aortic dilatation in TAV	‘Low’ or ‘moderate’ risk of bias for screening	7.6% (3.3%–13.5%)11,20–24,30,39
	FDR only	5.3% (2.0%–10.0%)11,20,22–24,30,39,45

Individual bicuspid aortic valve (BAV) prevalence refers to the prevalence of BAV amongst individual relatives found to have BAV through screening. Family BAV prevalence is the prevalence of at least one relative per family having a BAV. Aortic dilatation in BAV or tricuspid aortic valve (TAV) refers to the prevalence of aortic dilatation amongst screened individual relatives found to have either a BAV or TAV. ‘Removal of proband aortic valve surgery cohorts’ refers to removal of studies specifically recruiting probands undergoing aortic valve surgery. ‘Paediatric’ studies specifically included paediatric probands, whereas not all ‘adult’ studies specified inclusion of only adult probands and may have included some paediatric probands. (−) unavailable. * inclusion of only studies that recruited probands undergoing aortic valve surgery.

Risk of bias assessment

The risk of bias was evaluated used validated methods for the estimation of risk of bias in prevalence studies by Hoy et al.⁴⁷ An overall score of ≤6, 7–8, and >8 equates to ‘low’, ‘moderate’, and ‘high’ risk of bias, respectively.

Results

Twenty-three studies (including 2297 index cases and 6054 screened relatives) were eligible for inclusion in the analyses, based on the inclusion and exclusion criteria (Figure 1). Index cases were 32 ± 21 years of age, and 71 ± 3.8% were male. Screened relatives were 29 ± 13 years of age, and 50.6 ± 5.1% were male.

Figure 1.

Flow diagram of Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA).

Most studies were prospective (21/23 studies),^{10,11,20–26,28–31,33–40} and recruited participants from single-centre, university-affiliated teaching hospitals (18/23 studies)^{10,21,23,25,26,28–40} or were multicentre (3/23 studies).^11,22,24 Included studies were published between 1978 and 2021, and 18 were full text.^{10,11,20–24,27,28,30,33–40} Eight studies recruited exclusively paediatric index cases,^{10,21,23,27,28,32,37,38} Five studies included both adults and children,^{11,20,29,30,34} and 10 did not specify.^{22,24–26,31,33,35,36,39,40} Galian-Gay et al.¹¹ used a core laboratory to improve echocardiographic reproducibility.

Prevalence of bicuspid aortic valve

Twenty-three studies, including 6054 screened family members, included a quantitative assessment of BAV prevalence in relatives, from which the pooled prevalence of BAV amongst all screened relatives of probands with BAV was 7.3% (95% CI 6.1%–8.6%, I² 64% p < 0.0001, 442/6054 relatives) (Figure 2A).^{10,11,20–40}

Figure 2.

(A) Pooled prevalence of bicuspid aortic valve cases amongst screened relatives. (B) Pooled prevalence of screened families within which at least one additional case of bicuspid aortic valve was found in addition to the index case. BAV, bicuspid aortic valve.

The mean age of screened relatives with BAV was 30.0 ± 13.7 years,^11,20,27,29 and 66.3 ± 8.3% were male.^{10,11,20,22,27,29,30,35,37–39} There was no significant difference in BAV detection rate between study date and type (multicentre vs. single centre). From five included studies, the prevalence of BAV incidentally detected before screening was 0.47% (95% CI 0%–1.9%, I² 84% p < 0.0001); from this, a number needed to screen to identify one family member affected by BAV of 20 (95% CI 5–120) can be estimated.^{11,20,23,28,31} Subgroup and sensitivity analyses are summarised in Table 2. There was no evidence for small-study effects in the assessment of screened relatives (z 0.52, p > 0.05) (see Supplementary material online, Figure S1A). Subgroup comparison by age of proband was not significant.

Thirteen studies, including 1100 screened families, reported the number of screened families within which at least one screened family member had been newly identified as having BAV (in addition to the index patient with BAV). Pooled analysis showed a rate of 23.6% [95% CI 18.1%–29.5%, I² 77%, p < 0.0001, 227/1100 families, 3.3 ± 1.2 (range 1.2–6.2) relatives per family] (Figure 2B).^{11,20,22,23,28–30,33,35,36,38–40} Subgroup and sensitivity analysis is found in Table 2. The funnel plot of studies used to calculate the prevalence of BAV amongst screened families was asymmetric, particularly at higher standard errors (z = 2.43, p = 0.015) (see Supplementary material online, Figure S1B), indicating the possibility of small-study effects. Subgroup comparison by age of proband was not significant.

Prevalence of aortic dilatation

Sixteen studies, including 3446 screened family members, reported the prevalence of aortic dilatation amongst family members. The prevalence of aortic dilatation amongst all screened family members was 9.4% (95% CI 5.7%–13.9%, 310/3446, I² 93%, p < 0.0001) (Figure 3A).^{10,11,20–24,29–31,34–36,39,45} Screened family members found to have BAV had a prevalence of aortic dilatation of 29.2% (95% CI 15.3%–45.1%, I² 87% p < 0.0001, 11 studies, n = 91/485) (Figure 3B)^{10,11,20,21,23,28–31,35,36,45} compared to family members found to have tricuspid aortic valves, in whom the prevalence of aortic dilatation was 7.0% (95% CI 3.2%–12.0%, I² 87% p < 0.0001, 11 studies, n = 200/2749) (Figure 3C)^{11,20–24,30,35,36,39,45} at average ages of 33.2 ± 10.0 years and 39.0 ± 10.6 years, respectively. Family members found to have BAV were about six times more likely to have aortic dilatation than family members with tricuspid valves (risk ratio 6.1, 95% CI 3.4%–10.8%, I² 74% p < 0.001).^{11,20,21,23,30,35,36,45} One included study reported the age of screened relatives with aortic dilatation and tricuspid valves to be 57 years.²⁰ There were insufficient reports of the prevalence of hypertension alongside relatives screened to have aortic dilatation to produce a representative estimate. Subgroup and sensitivity analysis is found in Table 2. Subgroup comparison of aortic dilatation by studies that included paediatric vs. adult (±paediatric) probands was not significant, except for the comparison amongst relatives screened to have BAV (chi-square 4.04, p = 0.045) suggesting that studies recruiting adult probands may have a significantly higher prevalence of screened aortic dilatation than paediatric studies. There was no evidence of small-study bias in the analysis of aortic dilatation for screened relatives overall (z = 1.77, p = 0.08) and relatives with BAV (z = 1.71, p = 0.09) or tricuspid aortic valves (z = 1.54, p = 0.12) (see Supplementary material online, Figure S1C, S1D and S1E). Panayotova et al.³¹ found one case of ascending aortic aneurysm in FDR screened to have BAV that required prompt intervention.

Figure 3.

(A) Pooled prevalence of aortic dilatation amongst screened relatives. (B) Pooled prevalence of aortic dilatation amongst screened relatives found to have a bicuspid aortic valve. (C) Pooled prevalence of aortic dilatation amongst screened relatives found to have a tricuspid aortic valve. Brackets in aortic location measured represent locations measured but not used in data presented. A, annulus; AA, ascending aorta; BAV, bicuspid aortic valve; CI, confidence interval; Root, aortic root; STJ, sinotubular junction; SV, sinus of Valsalva; TAV, tricuspid aortic valve.

Evidence base for screening

There were no quantitative data on long-term improvement in morbidity or mortality from screening of BAV. In addition to very few qualitative assessments of public opinion for screening, full results of the systematic review of the literature supporting screening for BAV appraised against the UK National Screening Committee (UK NSC) framework can be found in the supplement.

Risk of bias assessment

Ten (43%), six (26%), and seven (30%) studies were considered ‘low-’, ‘moderate-’, and ‘high-’ risk of bias for the purpose of screening, respectively (see Supplementary material online, Table S3). The issue that affected 79% of included studies was the risk of non-response bias that in the vast majority of cases was due to a low uptake of screening or lack of information on the topic. Use of random selection or a census approach and study target population raised the risk of bias in 54% and 38% of cases, respectively. This was often due to a lack of participant characteristic information and the possibility of selection bias in the recruitment of cohorts. A case definition of BAV was commented on in 75% of cases.

Discussion

This systematic review and meta-analysis of 23 screening studies (including data from 6054 screened relatives) provide an estimate of the prevalence of BAV amongst screened family members of probands with BAV of 7.3% (95% CI 6.1%–8.6%). This estimate, whilst lower than sometimes reported,^48–50 is an order of magnitude greater than the prevalence reported for live births (0.5%), screened athletes (0.5%), and a large database of echocardiograms performed for clinical reasons (0.6%),^1,2 indicating that relatives of patients with BAV are significantly enriched for BAV.

To our knowledge, this is the largest study to date and therefore the most reliable estimate presently available. Our confidence in the estimate is increased because the finding was consistent across various sensitivity analyses, with similar estimates obtained when the analysis was confined to FDR family members, when studies at high risk of bias were excluded, and when subgroup analyses by proband age (paediatric compared to adult) were performed.

As each family on average has several family members, the likelihood of finding at least one family member with a BAV is greater than the likelihood of finding a BAV in a single family member. We found the prevalence of BAV per screened family to be ∼23.6% (95% CI 18.1%–29.5%). This estimate, which implies that about four or five families would need to be screened to detect each new BAV, is subject to greater uncertainty than the main analysis based on individual family members because (i) it is based on the analysis of only 1100 families in 13 trials; (ii) it is probably quite sensitive to family size; (iii) BAVs are not randomly distributed, but rather, cluster in families; and (iv) the prevalence estimate itself is probably an over-estimate, as there was some evidence of small-study effects (i.e. potentially publication bias) as many smaller studies did not publish their results in this format.

The prevalence of aortic dilatation amongst screened relatives in 16 studies (comprising 3446 family members) was 9.4% (95% CI 5.7%–13.9%). Family members with BAV were about six times more likely to have aortic dilatation [prevalence 29.2% (95% CI 15.3%–45.1%)] than those found to have tricuspid valves [prevalence 7.0% (95% CI 3.2%–12.0%)], risk ratio 6.1 (95% CI 3.4%–10.8%).The prevalence of aortic enlargement amongst FDR with BAV was similar to that reported in the literature for BAV probands in general.³

Importantly, our study confirmed a relatively high prevalence of aortic dilatation amongst family members, even where the valve was classified as tricuspid [prevalence 7.0% (95% CI 3.2%–12.0%)], whereas only about 2.5% of the population would be expected to have a z-score > 1.96, assuming a normal distribution of aortic root sizes.⁵¹ Thus, whilst BAV aortopathy was an occasional finding, occurring in 91/3234 (2.8%), aortic dilatation without associated BAV was a more frequent finding (200/3234), owing in part to the large number of screened family members who were classified as having normal valves. We sought evidence for heterogeneity by age of proband recruited and found no strong evidence for such heterogeneity. However, as we did not have access to individual patient data and as there were potentially paediatric probands in some of the non-paediatric studies, we interpret this finding with caution. There was insufficient data available to make a valid assessment of the age of included screened relatives with normal valves and aortic dilatation. Our analyses in relation to aortic enlargement are less definitive than those related to BAV prevalence because (i) they are based on fewer family members in only 16 studies; (ii) there were variations between studies in the precise locations at which aortic measures were made; (iv) the definition of normality varied from study to study, and it was not possible to harmonise definitions; and (v) echocardiography itself is subject to some limitations in assessing the ascending aorta. A minority of included studies were abstracts (6/23), and their inclusion probably reduces the risk of publication bias in our results, but it is possible that they could be preliminary results of future publications. As we sought to include all published screening studies, these original studies had a variety of aims that may explain the high degree of heterogeneity we found, although our results are likely to be relatively generalisable having included numerous populations from 10 separate countries. As some included studies did not specifically seek an unbiased prevalence of BAV or aortic dilatation amongst screened relatives, they may have preferentially sought to enrich their cohorts by including relatives with cases over those without, potentially introducing selection bias. Our sensitivity analysis considers studies that sought to ‘enrich’ their cohorts, but these studies were unclear about how they did this and the degree to which this may have introduced selection bias.

We found no studies that reported clinical outcomes for prospectively enrolled family members with BAV. Michelena et al.³ followed up 212 patients with asymptomatic BAV up to 20 years. It is not reported what proportion of the cohort were detected by screening, but this cohort provides information about the natural history of BAV. Twenty-four per cent required aortic valve surgery, and 5% required ascending aortic surgery. Forty-two per cent experienced cardiovascular medical or surgical events during follow-up.³ If we assume that screen-detected BAVs have similar outcomes to community BAV patients, then ∼3% of a screened population based on the prevalence we have found might experience an event over 20 years. These figures allow more refined modelling of the benefits of screening FDR to inform the assessment of screening programmes.

Comparison with other studies

This is the first review to combine and quantitatively analyse the prevalence of BAV and aortic dilatation in screened family members. Previous studies on screening have mostly recommended screening of relatives of individuals with BAV based on its prevalence,^{23,28,29,31,33,34,37,52} its heritability,⁵² the risk of complications including aortopathy,^21,34,35 and cost-effectiveness.^27,53 Galian-Gay et al. questioned the routine adoption of screening based on their study’s FDR prevalence findings.^11,54

Previous studies have identified aortic dilatation amongst relatives classified to have tricuspid valves on transthoracic echocardiography (TTE) but were found to have partial raphe using computed tomography (CT).^55,56 This highlights the complexity of defining bicuspid valve phenotype by imaging alone and suggests that there is a complex relationship between valve morphology and aortopathy in this syndrome which may explain the mechanism for aortic dilatation in some cases.

Implications for policy and practice

We provide an estimate for the prevalence of BAV in family members of affected probands, which is likely to be more reliable than the results of any individual study. We also provide a pooled estimate for screened families within which at least one new BAV was identified in addition to the index case and for aortic dilatation amongst screened relatives, although these estimates are subject to greater uncertainty than the main result. Further studies estimating the prevalence of BAV in FDR and families are unlikely to add substantially greater precision to our estimates unless very large. Whilst the prevalence of BAV (and to some extent aortopathy) in family members is now well established within narrow confidence limits, whether to embark on a programme of screening requires more than a knowledge of the prevalence of the disease in the target population. Many of the elements critical to determining whether screening will do more good than harm remain uncertain. In particular, the natural history of screen-detected BAV and of screen-detected aortopathy is not well understood, the potential harms arising from screening programs, and perceptions of family members are not defined. The effectiveness of particular screening programmes has not been rigorously evaluated, and robust outcome data are not available. There is little reliable data on (i) cost-effectiveness of such programmes, (ii) what resource implications such a programme might entail, (iii) what the alternative uses to those resources might be, (iv) whether such programmes might be considered good value for money, (v) whether they can be delivered equitably to all members of society likely to benefit (if any), and (vi) what barriers there may be to such a distribution and how they may be overcome.

Clinical practice guidelines recommend targeting screening, but the benefit from screening is deferred. Screening leads to benefit by finding an asymptomatic condition at an early stage which, if left untreated, would cause symptoms or complications years later. Thus, screening interventions have a ‘time lag to benefit’. This period starts from the date of screening, when the patient is exposed to the risks of screening (e.g. pain, worry, and social implications of the diagnosis), to the point in time when the benefits are observed. Diagnosis of BAV by TTE is not perfect, and false positives and negatives both occur in ∼10% of cases.⁵⁷ For patients with a long life expectancy such as children, the time lag to benefit may be many years, and screening exposes this group to immediate risks. These aspects relevant to the timing and frequency of screening should be explored in future studies. There is insufficient evidence on the age at which screening should be offered. Further evidence is needed on the age at which complications arise in family members.

Conclusion

The prevalence of BAV in screened family members was 7.3% (95% CI 6.1%–8.5%), representing a >10-fold enrichment compared to the general population but lower than early studies suggested. Moreover, further identification of BAV occurs in about a quarter of screened families. The overall prevalence of aortic dilatation in family members was about 9% overall. Whilst aortic dilatation was markedly more frequent in family members with bicuspid valves (29%), aortic dilatation was not confined to these family members with BAV, and the prevalence of aortic dilatation in family members with tricuspid valves (7%) was also higher than in the general population. Most family members had normal valves and normal aortas, and in most families, no recurrent cases of BAV were found.

The prevalence of BAV in family members is now known with high precision, and further studies are unlikely to add substantially greater precision unless very large (consisting of many thousands of patients). Consequently, subsequent research should now focus on closing the substantial gap between the guideline recommendations that recommend screening and the evidence base which is required to justify a systematic (and hence equitable) screening programme. In particular, (i) reliable estimation of the harms of screening (e.g. anxiety provoked by incidental findings), (ii) subsequent consideration of the best imaging modalities (e.g. cardiac CT or cardiac magnetic resonance) in relatives with pathological or ambiguous findings, (iii) subsequent consideration of cardiac CT in relatives found to have tricuspid valves but aortic dilatation, and (iv) patient and family member’s attitudes to screening and the acceptability of proposed models of screening the specific should be explored, and (v) the natural history of screen-detected disease (especially mild aortic enlargement without BAV)—a common finding—needs to be clarified with outcome data.

Author contributions

J.J.H.B.: guarantor; methodology; data collection; investigation; formal analysis; visualisation; writing—manuscript original draft; review and editing; and project administration. R.F.: data collection, investigation, and writing—review and editing. A.P.: visualisation and writing—review and editing. R.K.: conceptualization, visualisation, and writing—review and editing.

Transparency declaration

The authors, including the manuscript’s guarantor (J.J.H.B.), affirm that the manuscript is an honest, accurate, and transparent account of the review reported.

Supplementary data

Supplementary data is available at European Heart Journal online.

Data availability

Tabular data obtained from included studies and upon which analyses have been made are included within the associated supplement.

Funding

No funding was obtained for the production of this manuscript.