Beta‐blokery w zapobieganiu rozwarstwienia aorty u osób z zespołem Marfana
Abstract
Background
Marfan syndrome is a hereditary disorder affecting the connective tissue and is caused by a mutation of the fibrillin‐1 (FBN1) gene. It affects multiple systems of the body, most notably the cardiovascular, ocular, skeletal, dural and pulmonary systems. Aortic root dilatation is the most frequent cardiovascular manifestation and its complications, including aortic regurgitation, dissection and rupture are the main cause of morbidity and mortality.
Objectives
To assess the long‐term efficacy and safety of beta‐blocker therapy as compared to placebo, no treatment or surveillance only in people with Marfan syndrome.
Search methods
We searched the following databases on 28 June 2017; CENTRAL, MEDLINE, Embase, Science Citation Index Expanded and the Conference Proceeding Citation Index ‐ Science in the Web of Science Core Collection. We also searched the Online Metabolic and Molecular Bases of Inherited Disease (OMMBID), ClinicalTrials.gov and the World Health Organization (WHO) International Clinical Trials Registry Platform (ICTRP) on 30 June 2017. We did not impose any restriction on language of publication.
Selection criteria
All randomised controlled trials (RCTs) of at least one year in duration assessing the effects of beta‐blocker monotherapy compared with placebo, no treatment or surveillance only, in people of all ages with a confirmed diagnosis of Marfan syndrome were eligible for inclusion.
Data collection and analysis
Two review authors independently screened titles and abstracts for inclusion, extracted data and assessed trial quality. Trial authors were contacted to obtain missing data. Dichotomous outcomes will be reported as relative risk and continuous outcomes as mean differences with 95% confidence intervals. We assessed the quality of evidence using the Grading of Recommendations Assessment, Development and Evaluation (GRADE) approach.
Main results
One open‐label, randomised, single‐centre trial including 70 participants with Marfan syndrome (aged 12 to 50 years old) met the inclusion criteria. Participants were randomly assigned to propranolol (N = 32) or no treatment (N = 38) for an average duration of 9.3 years in the control group and 10.7 years in the treatment group. The initial dose of propranolol was 10 mg four times daily and the optimal dose was reached when the heart rate remained below 100 beats per minute during exercise or the systolic time interval increased by 30%. The mean (± standard error (SE)) optimal dose of propranolol was 212 ± 68 mg given in four divided doses daily.
Beta‐blocker therapy did not reduce the incidence of all‐cause mortality (RR 0.24, 95% CI 0.01 to 4.75; participants = 70; low‐quality evidence). Mortality attributed to Marfan syndrome was not reported. Non‐fatal serious adverse events were also not reported. However, study authors report on pre‐defined, non‐fatal clinical endpoints, which include aortic dissection, aortic regurgitation, cardiovascular surgery and congestive heart failure. Their analysis showed no difference between the treatment and control groups in these outcomes (RR 0.79, 95% CI 0.37 to 1.69; participants = 70; low‐quality evidence).
Beta‐blocker therapy did not reduce the incidence of aortic dissection (RR 0.59, 95% CI 0.12 to 3.03), aortic regurgitation (RR 1.19, 95% CI 0.18 to 7.96), congestive heart failure (RR 1.19, 95% CI 0.18 to 7.96) or cardiovascular surgery, (RR 0.59, 95% CI 0.12 to 3.03); participants = 70; low‐quality evidence.
The study reports a reduced rate of aortic dilatation measured by M‐mode echocardiography in the treatment group (aortic ratio mean slope: 0.084 (control) vs 0.023 (treatment), P < 0.001). The change in systolic and diastolic blood pressure, total adverse events and withdrawal due to adverse events were not reported in the treatment or control group at study end point.
We judged this study to be at high risk of selection (allocation concealment) bias, performance bias, detection bias, attrition bias and selective reporting bias. The overall quality of evidence was low. We do not know whether a statistically significant reduced rate of aortic dilatation translates into clinical benefit in terms of aortic dissection or mortality.
Authors' conclusions
Based on only one, low‐quality RCT comparing long‐term propranolol to no treatment in people with Marfan syndrome, we could draw no definitive conclusions for clinical practice. High‐quality, randomised trials are needed to evaluate the long‐term efficacy of beta‐blocker treatment in people with Marfan syndrome. Future trials should report on all clinically relevant end points and adverse events to evaluate benefit versus harm of therapy.
PICO
Streszczenie prostym językiem
Beta‐blokery w leczeniu zespołu Marfana
Pytanie
Czy korzyści z leczenia beta‐blokerami w zespole Marfana przeważają nad szkodami w porównaniu z placebo lub brakiem leczenia?
Wprowadzenie
Zespół Marfana to dziedziczna choroba genetyczna obejmująca wiele narządów. Jedną z najczęstszych i najważniejszych cech tej choroby jest poszerzenie aorty (największa tętnica transportująca krew z serca). Może to prowadzić do problemów zagrażających życiu, np. rozwarstwienia aorty, które jest przerwaniem ciągłości błony wewnętrznej, co skutkuje przedostaniem się krwi pomiędzy warstwy ściany aorty i jej gromadzenia oraz w konsekwencji rozerwania aorty.
Jako leki pierwszego rzutu w zespole Marfana zalecane są beta‐blokery (leki beta‐andrenolityczne; przyp. tłum.) ‐ grupa leków stosowana w celu obniżenia ciśnienia tętniczego. Dokładny mechanizm ich działania w przypadku zespołu Marfana nie jest znany.
Okres wyszukiwań
Dane naukowe są aktualne do czerwca 2017 r.
Charakterystyka badania
Do przeglądu włączyliśmy jedno badanie obejmujące 70 osób z zespołem Marfana w wieku od 12 do 50 lat, których przydzielono do dwóch grup. W jednej grupie podawano beta‐bloker o nazwie propranolol, a w drugiej nie zastosowano żadnego leczenia. Badanie trwało średnio 9,3 lat w grupie kontrolnej (bez leczenia) i 10,7 lat w grupie leczonej.
Źródła finansowania badania
Badanie finansowano z grantów Narodowego Instytutu Zdrowia (NIH), Agencji Żywności i Leków (FDA) oraz National Marfan Foundation.
Główne wyniki i wnioski
Przyjmowanie propranololu w porównaniu z brakiem leczenia nie obniżyło śmiertelności ani zachorowalności, obejmując rozwarstwienie aorty, niedomykalność zastawki aortalnej (przeciekanie krwi przez zastawkę aorty, skutkujące wstecznym przepływem krwi do serca), niewydolność serca (brak zdolności serca do pompowania odpowiedniej ilości krwi do narządów) oraz operację serca. Niemniej jednak leczenie zmniejszyło tempo poszerzania się aorty. W publikacji przedstawiającej wyniki badania nie uwzględniono wszystkich szkodliwych następstw leczenia. Według naszej oceny to badanie kliniczne obciążone jest dużym ryzykiem błędu, a uzyskane dane są niskiej jakości. Dane pochodzące z niego nie są wystarczające, by mogły stanowić cenną informację dla osób z zespołem Marfana, ich rodzin i opiekunów.
Authors' conclusions
Summary of findings
| Beta‐blockers compared with placebo, no treatment or surveillance only for preventing aortic dissection in Marfan syndrome | ||||||
| Patient or population: people with confirmed diagnosis of Marfan syndrome Settings: outpatient Intervention: beta‐blockers Comparison: placebo or no treatment or surveillance only | ||||||
| Outcomes Follow‐up: mean 9.3 years in control group and 10.7 years in treatment group | Anticipated absolute effects* (95% CI) | Relative effect | No of participants | Quality of the evidence | Comments | |
| Risk with placebo, no treatment or surveillance only | Risk with beta blockers | |||||
| All‐cause mortality | 53 per 1000 | 13 per 1000 (1 to 250) | RR 0.24 (0.01 to 4.75) | 70 | ⊕⊕⊝⊝ | |
| Clinically important non‐fatal end point as defined byShores 1994(aortic dissection, aortic regurgitation, congestive heart failure, or cardiovascular surgery) | 316 per 1000 | 249 per 1000 (117 to 534) | RR 0.79 (0.37 to 1.69) | 70 | ⊕⊕⊝⊝ | |
| Acute aortic dissection | 105 per 1000 | 62 per 1000 (13 to 319) | RR 0.59 (0.12 to 3.03) | 70 | ⊕⊕⊝⊝ | |
| Aortic regurgitation | 53 per 1000 | 63 per 1000 (9 to 419) | RR 1.19 (0.18 to 7.96) | 70 | ⊕⊕⊝⊝ | |
| Congestive heart failure | 53 per 1000 | 63 per 1000 (9 to 419) | RR 1.19 (0.18 to 7.96) | 70 | ⊕⊕⊝⊝ | |
| Cardiovascular surgery | 105 per 1000 | 62 per 1000 (13 to 319) | RR 0.59 (0.12 to 3.03) | 70 | ⊕⊕⊝⊝ | |
| Aortic root diameter > 6 cm | 26 per 1000 | 31 per 1000 (2 to 480) | RR 1.19 (0.08 to 18.24) | 70 | ⊕⊕⊝⊝ | |
| *The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). | ||||||
| GRADE Working Group grades of evidence High quality: we are very confident that the true effect lies close to that of the estimate of the effect. | ||||||
| 1Downgraded one step for risk of bias (high risk of selection, performance, detection, attrition and selective outcome reporting bias). | ||||||
Background
Description of the condition
Marfan syndrome is a hereditary disorder affecting the connective tissue and is caused by a mutation of the fibrillin‐1 (FBN1) gene. It was first discovered in 1896 by Professor Antoine‐Bernard Marfan, a French paediatrician (Franken 2012; Judge 2005; Koracevic 2012). The condition is estimated to have an incidence of 1 in 2000 to 1 in 5000 individuals (Danyi 2012; Franken 2012; Von Kodolitsch 2007; Wright 2012).
Marfan syndrome affects multiple systems of the body, most notably the cardiovascular, ocular, skeletal, dural and pulmonary systems (Castellano 2012; Pyeritz 2009). Aortic root dilatation is the most frequent cardiovascular manifestation (Von Kodolitsch 2004), and its complications, including aortic regurgitation, dissection and rupture are the main cause of morbidity and mortality in people with this disease (Canadas 2010; Franken 2012). One long‐term survival study of Marfan syndrome in the early 1970s showed the mean age of death to be 32 years for people who did not receive aortic surgery (Murdoch 1972).
Although other cardiovascular features such as valvular disease, endocarditis, cardiomyopathy and arrhythmias exist in these people with Marfan syndrome (Pyeritz 2000; Von Kodolitsch 2004), dissection of the aorta has been, and remains, the most common cause of death (Pyeritz 2008; Silverman 1995). Aortic wall weakness, aortic dilatation and arterial hypertension are the major mechanisms of dissection and rupture (Von Kodolitsch 2004). Research has shown that a defect in the FBN1 glycoprotein, a major constituent of the extracellular matrix microfibrils (Bunton 2001), leads to poor structural and functional integrity of the normal vessel wall by several potential mechanisms (Danyi 2011; Judge 2005; Von Kodolitsch 2007). The changes in the walls of the elastic arteries occur primarily in the medial layer and are associated with less distensibility and increased stiffness leading to consequent weakening and dilatation, beginning in the sinuses of the aortic root and extending to the proximal ascending aorta (Pyeritz 2000; Pyeritz 2009).
Aortic root pathology has hence become the most important target for improving survival in people with Marfan syndrome (Vaidyanathan 2008). As a result of improved diagnosis, careful monitoring, lifestyle guidance, medical and especially surgical management of this disease, the life expectancy of people with Marfan syndrome has increased by at least 30 years (Pyeritz 2009; Silverman 1995). Although the benefits of prophylactic aortic surgery have been clearly demonstrated, the value of reducing aortic dilatation medically is unclear in the clinical setting to reduce morbidity and mortality.
Description of the intervention
The primary aim of pharmacological therapy for Marfan syndrome is to slow down the rate of aortic dilatation with the goal to delay or prevent complications and surgical interventions (Franken 2012). The beta‐blockade strategy began in 1971 when Halpern and colleagues suggested that the reduction of the rate of increase in aortic pressure over time was an important additional factor to lowering blood pressure alone in decreasing haemodynamic stress on the proximal aorta (Halpern 1971; Keane 2008; McKusick 2004). Beta‐blockers became, and have remained, the standard preventive treatment since the mid‐1990s when one randomised controlled trial (RCT), Shores 1994, concluded that prophylactic beta‐adrenergic blockade with propranolol slowed the rate of aortic dilatation and reduced the development of aortic complications in people with Marfan syndrome. Subsequent studies have shown varying results on the efficacy of beta‐blockade therapy (Legget 1996; Rios 1999; Tierney 2007). Despite the lack of conclusive evidence, it has been recommended that people with Marfan syndrome and aortic aneurysm should be prescribed beta‐blockers to reduce the rate of dilatation unless contraindicated (Boodhwani 2014; Hiratzka 2010; Pyeritz 2012; Wright 2012). Current beta‐blockers in use include propranolol, atenolol and metoprolol (Wright 2012). Atenolol is currently the drug of choice as it has a longer half‐life and is more cardioselective than propranolol, with fewer side effects (Keane 2008). The drug dosage is adjusted according to heart rate, aiming to maintain a rate of 60 to 70 beats per minute at rest and fewer than 100 beats per minute after sub‐maximal exercise (Canadas 2010; Loeys 2010; Wright 2012). Beta‐blockers have a significant side‐effect profile and documented adverse effects include bronchospasm, exercise intolerance, fatigue, depression, and first‐ and third‐degree heart block (Gao 2011; Shores 1994).
Other anti‐hypertensive medications such as calcium channel blockers, angiotensin‐converting enzyme (ACE) inhibitors and, more recently, angiotensin receptor blockers (ARB) have been studied with mixed results (Bhatt 2015; Lacro 2014; Milleron 2015; Rossi‐Foulkes 1999; Topouchian 1998; Williams 2012; Yetman 2005). Several large RCTs assessing the effect of losartan (an angiotensin II type I receptor blocker) have recently been completed or are underway (Forteza 2016; Gambarin 2009; Lacro 2014; Mullen 2013; Moberg 2012). The use of an ACE inhibitor or an ARB is suggested for people who are unable to tolerate beta‐blockers or as an add‐on therapy if beta‐blockade monotherapy is unsuccessful at controlling blood pressure (Keane 2008; Pyeritz 2014; Singh 2016; Wright 2012). The Marfan Treatment Trialists' Collaboration has proposed a large, prospective, collaborative meta‐analysis of all the RCTs evaluating ARB treatment in Marfan syndrome (Pitcher 2015).
How the intervention might work
Although the exact mechanism of beta‐blockers in Marfan syndrome is unknown, the potential benefits have been attributed to their negative inotropic and chronotropic effects, resulting in a reduction in aortic wall stress (Canadas 2010). This is important in decreasing the progression of aortic dilatation as the degree of increase in aortic root diameter is a major indicator for the risk of dissection (Judge 2005). The anti‐arrhythmic and anti‐fibrillatory effects of beta‐blockade are also believed to be advantageous as other cardiovascular manifestations such as mitral valve prolapse and left ventricular dilatation are common in Marfan syndrome, pre‐disposing people to arrhythmias (Koracevic 2012).
The cardiac cycle is pulsatile in nature, with aortic expansion during systole and recoil during diastole. This pulsatile flow can be characterised by a change in pressure over time and contributes to the progression of aortic dissection compared with non‐pulsatile flow (Castellano 2012; Liao 2010). The aorta buffers the level of fluctuation between the extreme pressures of systole and diastole, allowing a nearly continuous blood flow from the central to the peripheral vascular system (Belz 1995; Koracevic 2012). This protective function, known as the Windkessel effect, relies on the elasticity of the aorta (Belz 1995), and, as Marfan syndrome is associated with abnormal aortic elastic properties (Hirata 1991), people are therefore compromised. Studies have shown that beta‐blocker therapy may directly affect the aortic wall by increasing its distensibility, and decreasing aortic stiffness and pulse‐wave velocity (Groenink 1998; Ladouceur 2007; Rios 1999). With the favourable effects of beta‐blockers on change in pressure over time, this pharmacological intervention became important in the medical management in aortic dissection (Castellano 2012; Danyi 2012; Liao 2010).
Why it is important to do this review
Beta‐blockers have remained the standard treatment for the prevention of aortic complications since medical intervention was introduced in the 1970s. Studies investigating the efficacy and therapeutic benefit of beta‐blockade have produced heterogeneous and conflicting results, leading to much debate on their life‐long use for people with Marfan syndrome (Ladouceur 2007; Tierney 2007). One meta‐analysis, Gersony 2007, included six studies, of which only one was an RCT, and concluded that there was no evidence that beta‐blocker therapy had clinical benefit in people with Marfan syndrome. Conversely, Gao 2011 concluded from their meta‐analysis that routine prescription of beta‐blockers may offer substantial benefit on clinical end points for children and adolescents with Marfan syndrome.
One Cochrane Review on the medical treatment for small abdominal aortic aneurysms (AAA) reported that there was no significant difference in AAA expansion or cardiovascular end points between beta‐blocker treatment and placebo. Furthermore, a significantly increased number of people discontinued beta‐blocker therapy for AAA management due to adverse effects and there was no significant difference in the combined overall mortality of the three included propranolol trials (Rughani 2012). Although thoracic aortic aneurysms (TAA) are more common in Marfan syndrome and extrapolating data from AAA to TAA studies should be conducted with caution (Castellano 2012), these findings do raise cause for concern.
Lifelong treatment, as beta‐blockade therapy in Marfan syndrome currently remains, is not a decision or commitment that should be taken lightly. Even with the anticipation of other pharmacological treatments currently under trial, it is important to establish the efficacy of the baseline treatment that has been in place for decades. This would hopefully decrease the discrepancy gap between clinical practice and current evidence, and provide the physician with more options to optimise and individualise a management plan for his/her patient. Since aortic pathology is present in the vast majority of people with Marfan syndrome, and is the most life‐threatening manifestation of this disease (Canadas 2010; Franken 2012; Gao 2011; Judge 2005), this review will focus on the effects of beta‐blockers for preventing aortic dissection in Marfan syndrome. Therefore, we will examine the most up‐to‐date literature to assess the long‐term efficacy of this treatment in people of all ages with Marfan syndrome and determine if current practice with life‐long beta‐blocker therapy is evidence‐based.
Objectives
To assess the long‐term efficacy and safety of beta‐blocker therapy as compared to placebo, no treatment or surveillance only in people with Marfan syndrome.
Methods
Criteria for considering studies for this review
Types of studies
All randomised controlled trials (RCTs) of at least one year in duration assessing the effects of beta‐blocker monotherapy in people with Marfan syndrome were eligible for inclusion.
Types of participants
We included participants of all ages with a confirmed diagnosis of Marfan syndrome and excluded people with previous aortic root surgery, planned aortic surgery within one year of study enrolment, previous aortic dissection and co‐existing diagnoses of connective tissue disease. We also excluded pregnancy in people with Marfan syndrome.
Types of interventions
We assessed beta‐blocker monotherapy compared with placebo, no treatment or surveillance only.
Types of outcome measures
Primary outcomes
-
All‐cause mortality (including mortality attributed to Marfan syndrome)
-
All non‐fatal serious adverse events (including all‐cause hospitalisations, cardiovascular events such as aortic dissection or rupture, and cardiovascular surgery)
Secondary outcomes
-
Measurements of the aortic root diameter taken by echocardiogram or other imaging modalities
-
Systolic and diastolic blood pressure
-
Total adverse events
-
Withdrawals due to adverse events
Search methods for identification of studies
Electronic searches
We identified trials through systematic searches of the following bibliographic databases on 28 June 2017:
-
Cochrane Central Register of Controlled Trials (CENTRAL; 2017, Issue 5) in the Cochrane Library
-
Epub Ahead of Print, In‐Process & Other Non‐Indexed Citations, MEDLINE Daily and MEDLINE (Ovid, 1946 to 28 June 2017)
-
Embase (Ovid, 1980 to Week 26, 2017)
-
Science Citation Index Expanded and the Conference Proceeding Citation Index ‐ Science, Web of Science Core Collection (Thomson Reuters, 1970 to 28 June 2017)
We adapted the search strategy for MEDLINE (Ovid) (Appendix 1) for use in the other databases. The Cochrane sensitivity‐maximising RCT filter was applied to MEDLINE (Ovid) (Lefebvre 2011) and an adaptation of it to Web of Science. For Embase, we applied RCT filter terms as recommended in the Cochrane Handbook for Systematic Reviews of Intrventions (Lefebvre 2011). Additionally, we applied an adverse events filter to MEDLINE and Embase (Loke 2011).
We also searched the Online Metabolic and Molecular Bases of Inherited Disease (OMMBID) and sought to identify ongoing, recently completed or unpublished trials by searching clinical trials registers; ClinicalTrials.gov (www.ClinicalTrials.gov) and the World Health Organization (WHO) International Clinical Trials Registry Platform (ICTRP) Search Portal (apps.who.int/trialsearch/), using the terms 'Marfan syndrome' or 'Marfan syndrome' (last updated search on 30 June 2017).
We did not impose any restriction on language of publication.
Searching other resources
We examined reference lists of eligible studies and reviewed articles manually for additional references. We contacted the authors of the included study for missing data via email.
Data collection and analysis
Selection of studies
Two review authors (HK and KL) independently reviewed the titles and abstracts of all potential studies identified by the search strategy outlined above. We initially screened studies and coded them as 'Yes', 'Maybe' or 'No'. We resolved discrepancies by consensus or discussion with the third review author (VM).
For inclusion, a trial had to meet the following criteria:
-
the study was a RCT;
-
the study population had a diagnosis of Marfan syndrome;
-
the intervention was beta‐blocker mono‐therapy compared with placebo, no treatment, or surveillance only;
-
the study reported one or more primary or secondary outcome measures.
We obtained full‐text versions of all potentially eligible studies and made a decision for inclusion or exclusion. Where there were any differences in opinion regarding the suitability of a study, we discussed these with the third review author (VM). We included detailed trial reports where methodology could be assessed.
Data extraction and management
Two review authors (HK and VM) independently extracted relevant data from the retrieved articles using a data extraction form (Appendix 2). We resolved any disagreement in the data extraction and management by discussion.
Assessment of risk of bias in included studies
Two review authors (HK and KL) independently assessed the risk of bias for the included study. We performed this assessment of methodological quality according to the criteria described in Chapter 8 of the Cochrane Handbook for Systematic Reviews of interventions (Higgins 2011), and included the following:
-
random sequence generation;
-
allocation concealment;
-
blinding of participants and personnel;
-
blinding of outcome assessment;
-
incomplete outcome data;
-
selective outcome reporting;
-
other bias.
Each domain was allocated as either 'low, high or unclear risk'. We resolved any disagreements by discussion with the third author (VM).
Measures of treatment effect
We summarised dichotomous data (all‐cause mortality, non‐fatal serious adverse events, total adverse events and withdrawal due to adverse events) using risk ratios (RR) with 95% confidence intervals (CI) where relevant and when data were available. We planned to use mean differences (MD) with 95% CI to summarise continuous outcomes (measurement of aortic diameter, systolic and diastolic blood pressure).
Unit of analysis issues
For dichotomous outcomes, the unit of analysis was the number of individuals assigned to beta‐blocker treatment and the number of individuals assigned to the control group. For continuous outcomes, we planned the unit of analysis was to be the mean, standard deviation, and the number of individuals in the treatment and control groups (Deeks 2011).
Dealing with missing data
One trial, Shores 1994 met the inclusion criteria. We used data available in the publication and also contacted the study authors via email for additional information regarding missing data. The study protocol was not available. This study is not registered with ClinicalTrials.gov.
Assessment of heterogeneity
As only one trial met the inclusion criteria, we could not perform heterogeneity assessments. If sufficient numbers of studies had been available, we would have performed a Chi2 test to determine the presence of statistical heterogeneity at a significance level of P value less than 0.10. If more studies become available in the future, we will assess the quantity of heterogeneity using the I2 statistic (Higgins 2003) with the following parameters:
-
I2 = 0% to 40% heterogeneity may not be important;
-
I2 = 30% to 60% may represent moderate heterogeneity;
-
I2 = 50% to 90% may represent substantial heterogeneity;
-
I2 = 75% to 90% considerable heterogeneity.
Assessment of reporting biases
If we identified at least 10 studies, we planned to assess reporting bias by reviewing the funnel plots and performing a linear regression test. As only a single study met the inclusion criteria we could not assess reporting bias.
Data synthesis
We used the most recent version of the Cochrane Review Manager 5 software (RevMan 5) for data synthesis and analysis (RevMan 2014). We utilised a fixed‐effect model and all data are accompanied by the 95% CI.
Subgroup analysis and investigation of heterogeneity
As only one study met the inclusion criteria, we did not perform subgroup analysis. If sufficient numbers of trials were available, we had planned to perform the following subgroup analyses:
-
severity of disease at baseline: mild, moderate or severe;
-
sex: female versus male;
-
age groups: children/adolescents versus adults.
Sensitivity analysis
We did not perform sensitivity analysis as only one trial met the inclusion criteria. If sufficient numbers of trials were available, we had planned to assess the following features of methodological quality of included studies: sequence generation, allocation concealment, blinding of participants and assessors, and incomplete outcome data.
Results
Description of studies
Results of the search
We screened a total of 1819 references from databases, trials registers and handsearching after de‐duplication. We retrieved 15 full‐text articles for further assessment. Only one study fulfilled the eligibility criteria for inclusion (Shores 1994). See study flow diagram (Moher 2009) (Figure 1).
Study flow diagram
Included studies
The one included study by Shores 1994, titled 'Progression of aortic dilatation and the benefit of long‐term β‐adrenergic blockade in Marfan syndrome' is an open‐label, randomised trial that included adolescent and adult participants with Marfan syndrome to receive either beta‐blocker (propranolol) or no treatment.
This single‐centre study recruited participants who met the international diagnostic criteria for Marfan syndrome (Beighton 1988) within one year of trial initiation. Participants less than 12 or more than 50 years old, or receiving ongoing treatment with propranolol were excluded. Other exclusion criteria included: aortic dissection, aortic regurgitation on auscultation, moderate or severe mitral regurgitation, previous cardiovascular surgery, dyspnea during moderate exercise, orthopnoea, peripheral oedema, left ventricular ejection fraction less than 50%, atrioventricular conduction delay of any degree, and disorders where propranolol was contraindicated.
Of 117 people considered for the study, 93 were eligible and 70 provided informed consent. Participants were randomised after giving consent by assigning the next available number on a list derived from a table of random numbers. 38 participants with even numbers received no treatment (control group) and 32 participants with odd numbers received propranolol (treatment group). Participants who were eligible for the study but who chose not to participate did not differ appreciably in any of the characteristics from participants in the control and treatment groups combined.
The two study groups were matched for sex (male:female, control 19:18, treatment 20:12), mean age (control vs treatment; 14.5 years vs 15.4 years), and proportion of participants less than 18 years old. Baseline cardiovascular characteristics including aortic root diameter, presence of mitral‐valve prolapse or regurgitation, blood pressure, heart rate and systolic time interval were recorded. In the treatment group, male participants had a significantly lower resting heart rate compared to the female participants. Additionally, the treatment group had a significantly greater aortic diameter at baseline compared to the control group, but did not differ significantly in aortic ratio.
Neither participant nor investigator was blinded to the study. The initial dose of propranolol was 10 mg four times daily in the treatment group. This was adjusted according to individual response of heart rate to exercise and systolic interval, assessed after two to four weeks of initiation. Optimal dose was reached when the heart rate remained below 100 beats per minute during exercise or the systolic time interval increased by 30%. The mean (± standard error (SE)) dose of propranolol was 212 ± 68 mg given in four divided doses daily. The time‐point at which this optimal dose was achieved is not provided.
Participants were reviewed every 6 to 12 months with history and physical examination, electrocardiography and echocardiography. Serum drug concentration and phonocardiography to measure the left ventricular systolic time interval were also evaluated in the treatment group to assess for optimal dosing.
Participants remained in the study until one of the following endpoints was reached: voluntary withdrawal, death, aortic dissection, aortic regurgitation detected by auscultation, cardiovascular surgery, congestive heart failure or an intractable adverse reaction to propranolol. The published study states that all comparisons of the two study groups included all participants, according to the intention‐to‐treat (ITT) principle. However, the study authors, through personal communication, confirmed that no clinically relevant endpoints were collected during the follow‐up period of 9.3 years in the control group and 10.7 years in the treatment group, therefore they did not perform an ITT analysis at the end of follow‐up.
Further details of the included study are presented in Characteristics of included studies.
Excluded studies
We obtained the full‐text versions for fifteen studies in total. Reasons for exclusion of five studies are presented in Characteristics of excluded studies. The other nine studies were not RCTs and did not meet the inclusion criteria.
Risk of bias in included studies
We have summarised the risk of bias of the included study (Shores 1994) in Figure 2 and also provided a detailed explanation in the 'Risk of bias' table. Based on the information available in the published study and through personal communication with the study authors, we considered the trial to have a low risk of random sequence generation (selection) bias and a high risk of allocation concealment (selection) bias. There was high risk of performance bias, detection bias, attrition bias and reporting bias, and an unclear risk of other bias.
Risk of bias summary: review authors' judgements about each risk of bias item for each included study
Allocation
After obtaining consent, participants were randomised by assigning the next available number from a list derived from a table of random numbers at the beginning of the study. Participants with an even number received no treatment (control group), and participants with an odd number received propranolol (treatment group). We therefore judged the study at low risk of bias for random sequence generation, however, the study author, through personal communication, stated that investigators were aware of treatment allocation and therefore we judged the study at high risk of bias for allocation concealment.
Blinding
This was an open‐label study with both participants and investigators aware of treatment allocation and therefore at high risk of performance bias. Echocardiographic data were interpreted throughout the study by the same investigator who was unaware of the participant's identity, age, study group, or the sequence of multiple tracings, and therefore considered low risk of bias. However, for all other outcomes, we judged a high risk of detection bias as outcome assessors were not blinded.
Incomplete outcome data
Withdrawal due to clinical end‐points was described, but the total withdrawals from the study was not. Furthermore, once participants reached a clinical end‐point, they were withdrawn and further follow‐up was not documented (confirmed by the study author). The published study states that all comparisons included all participants according to the ITT principle. However, the study author, through personal communication, confirmed that no clinically relevant endpoints were collected during the follow‐up period of 9.3 years in the control group and 10.7 years in the treatment group, therefore they did not perform an ITT analysis at the end of follow‐up. We therefore judged the study at high risk of attrition bias. The study author stated that there were no missing data and the last observation was carried forward.
Selective reporting
We judged the study at high risk of reporting bias. The study protocol was not available and this was confirmed by the author through personal communication. Not all endpoints described in the methods are reported in the published study including: voluntary withdrawal, cardiovascular surgery, congestive heart failure indicated by new‐onset dyspnoea, orthopnoea, peripheral oedema, or fatigability associated with a left ventricular ejection fraction of less than 40%, and intractable adverse reaction to propranolol. Data regarding two of these outcomes: cardiovascular surgery and congestive heart failure, were provided by the study author through personal communication. Additionally, in the published study, it appears that "aortic root > 6 cm" was added as a clinical end‐point in addition to those stated in the methods section, however the study author states that this was a pre‐determined clinical end‐point. Furthermore, adverse effects data were not collected systematically between groups as the study was not blinded or placebo‐controlled. The study stated that a propranolol dose reduction from 80 mg to 40 mg was required for one participant who developed third‐degree atrioventricular block. Personal communication with the study author reports that no adverse events occurred in the control group.
Other potential sources of bias
This study was supported by grants from the National Institute of Health, the Food and Drug Administration and the National Marfan Foundation. However other bias is unclear as findings from this study have not been replicated as only a single study met the inclusion criteria. The study author, through personal communication, confirmed that no power calculation was performed to determine the sample size.
Effects of interventions
Primary outcomes
1. All‐cause mortality (including mortality attributed to Marfan syndrome)
Two deaths (2.8%) were observed in the study, both in the control group. The two participants, a 14 year‐old boy and an 18 year‐old woman, had mitral‐valve prolapse and a history of paroxysmal tachyarrhythmia (one of whom had Wolff‐Parkinson‐White syndrome). No aortic dissection or obvious cause of death was identified in postmortem examinations for either participant. The time points at which these participants died were not provided. No mortality attributed to Marfan syndrome was reported. There was no difference between the two study groups in all‐cause mortality (RR 0.24, 95% CI 0.01 to 4.75; participants = 70; studies = 1; low‐quality evidence). See Analysis 1.1.
2. All non‐fatal serious adverse events (including all‐cause hospitalisations, cardiovascular events such as aortic dissection or rupture, and cardiovascular surgery)
All non‐fatal serious adverse events as defined above were not reported in the included study.
Important non‐fatal clinical end points
Other morbidity outcomes as defined and reported by the Shores 1994 study included aortic dissection, aortic regurgitation, congestive heart failure and cardiovascular surgery.
Information regarding congestive heart failure and cardiovascular surgery were provided by the study author through personal communication. Cardiovascular surgery was indicated in participants when aortic root diameter reached 6 cm.
Twelve participants in the control group and eight participants in the treatment group reached these non‐fatal clinical endpoints. However, two participants in the treatment group had never taken their propranolol. Analysis showed no difference between the two study groups (RR 0.79, 95% CI 0.37 to 1.69; participants = 70; studies = 1; low‐quality evidence). See Analysis 1.2.
Acute aortic dissection was reported in four participants in the control group and two participants in the treatment group. Analysis showed no difference between the two study groups (RR 0.59, 95% CI 0.12 to 3.03; participants = 70; studies = 1; low‐quality evidence). See Analysis 1.3.
Aortic regurgitation was reported in two participants in the control group and two participants in the treatment group. Analysis showed no difference between the two study groups (RR 1.19, 95% CI 0.18 to 7.96; participants = 70; studies = 1; low‐quality evidence) See Analysis 1.4.
Congestive heart failure was reported in two participants in the control group and two participants in the treatment group. Analysis showed no difference between the two study groups (RR 1.19, 95% CI 0.18 to 7.96; participants = 70; studies = 1; low‐quality evidence) See Analysis 1.5.
Cardiovascular surgery was reported in four participants in the control group and two participants in the treatment group. Analysis showed no difference between the two study groups (RR 0.59, 95% CI 0.12 to 3.03; participants = 70; studies = 1; low‐quality evidence) See Analysis 1.6.
Secondary outcomes
1. Measurements of the aortic root diameter taken by echocardiogram or other imaging modalities
Echocardiography was used to measure aortic diameter. M‐mode echocardiography was utilised until 1982 when cross‐sectional examinations were performed. The study states that for conformity, only M‐mode tracings by the leading‐edge method were analysed to determine aortic diameter. Maximal diameter, usually at the level of the sinuses, but occasionally at the sino‐tubular junction was measured in five consecutive cycles and averaged. We note that there is discrepancy between the manuscript text where it states that mean values were presented with standard errors, and the table of patient characteristics, where the values are stated as means with standard deviations.
The two study groups differed significantly in their initial aortic diameter (control vs treatment: 30.2 mm vs 34.6 mm). Initial mean aortic ratio did not differ significantly (1.3 vs 1.4) and therefore we treated it as a co‐variable in the analysis. The aortic ratio was calculated by dividing the measured aortic diameter by the diameter predicted from the participant's height, weight and age. Of the two participants reaching the clinical endpoint of aortic root greater than 6 cm, the initial aortic ratio of one participant in the control group was 1.4 vs 2.1 in the other participant in the treatment group. Table 1 provides information on the initial aortic diameter and ratio reported at baseline.
| Initial aortic diameter | Control male (N = 19) | Control female (N = 19) | Treatment male (N = 20) | Treatment female (N = 12) |
| Measured (mm) ± SD | 31.1 ± 6.9 | 29.4 ± 6.8 | 36.7 ± 9.3 | 31.2 ± 5.3 |
| Expected (mm) ± SD | 24.6 ± 3.9 | 23.3 ± 3.2 | 25.5 ± 4.1 | 23.3 ± 4.2 |
| Ratio | 1.27 ± 0.19 | 1.27 ± 0.26 | 1.43 ±0.26 | 1.37 ± 0.2 |
SD: standard deviation
The study reports that the rate of aortic ratio increase was significantly lower in the treatment group compared to the control group (mean slope of the aortic ratio plotted against time for treatment vs control: 0.023 vs 0.083 per year), (t = 6.73, P < 0.001; z = 6.64 by Mann Whitney nonparametric rank‐sum test, P < 0.001). The study authors also report little, if any relation between the rate of change in aortic ratio and the initial aortic diameter. Additionally, analysis of covariance showed that adjustment for the initial aortic diameter had a negligible effect on the significance of the difference in the rates of enlargement. We were unable to measure the treatment effect on aortic ratio as these data were unavailable.
The study states that participants from both study groups who reached an endpoint had higher average initial aortic ratios than the total study population. Again, data were unavailable to perform statistical analysis.
The study does report on participants with aortic root diameter greater than 6 cm. One participant in the control group and one participant in the treatment group reached this end point. Although not stated in the published study, the study author, through personal communication, confirmed that this was a pre‐determined clinical end point. Analysis showed no difference between the two study groups (RR 1.19, 95% CI 0.08 to 18.24; participants = 70; studies = 1; low‐quality evidence) See Analysis 1.7.
2. Systolic and diastolic blood pressure
Table 2 provides information on the blood pressure measured at baseline and during optimal dose.
| BP measured as mmHg ±SD | Control male (N = 19) | Control female (N = 19) | Treatment male (N = 20) | Treatment female (N = 12) |
| SBP at baseline | 118 ± 14 | 110 ± 13 | 115 ± 13 | 115 ± 14 |
| DBP at baseline | 72 ± 11 | 70 ± 10 | 73 ± 10 | 69 ± 13 |
| SBP during optimal dose | Not reported | Not reported | 108 ± 15 | 108 ± 8 |
| DBP during optimal dose | Not reported | Not reported | 66 ± 11 | 63 ± 7 |
| End of treatment | Not reported | Not reported | Not reported | Not reported |
BP: blood pressure; DBP: diastolic blood pressure; SBP: systolic blood pressure
Systolic and diastolic blood pressure values were reported at baseline but not at study endpoint therefore we could not perform analysis of magnitude of blood pressure reduction between beta‐blocker and no‐treatment groups.
Blood pressure data in the treatment group were reported when the optimal doses of propranolol were achieved.
-
In the male treatment group, the study reported a significant decrease in blood pressure values during optimal treatment dose as compared to baseline (systolic: P = 0.006, diastolic: P = 0.045).
-
In the female treatment group, the study reported a decrease in blood pressure values during optimal treatment dose as compared to baseline (systolic: P = 0.06, diastolic: P = 0.051) however, the P values were not significant.
3. Total adverse events
As the study was not blinded or placebo‐controlled, the control group was not queried systematically about adverse symptoms. The investigators reported that there was no atrioventricular conduction delay in the control group. The study author, through personal email communication, provided additional information regarding adverse events in the control group (see Table 3, which provides information on adverse effects of long‐term treatment for 30 participants complying with beta‐blocker therapy).
| Adverse effect | Beta‐blocker group (N = 30) | Control |
| Third‐degree heart block | 1 | Not reported in Shores 1994, however, the study author reported that no adverse events occurred in the control group (personal communication) |
| First‐degree heart block | 3 | |
| Lethargy | 8 | |
| Depression | 1 | |
| Insomina | 4 | |
| Dream disturbance | 3 | |
| Mild bronchospasm | 1 | |
| Accentuated effects of alcohol | 1 | |
| Total with one or more effects | 10 |
4. Withdrawals due to adverse events
Shores 1994 does not discuss any loss to follow‐up or withdrawals due to adverse events. The study author, through personal communication, confirmed that information on adverse events after participants withdrew from the study was not collected or documented.
Discussion
Summary of main results
The single study included in this review (Shores 1994) is an open‐label, randomised, single‐centre trial comparing beta‐blocker monotherapy with no treatment in 70 participants aged 12 to 50 years old with Marfan syndrome. Participants were randomly assigned to propranolol (N = 32) or control (N = 38) for an average duration of 9.3 years in the control group and 10.7 years in the treatment group.
Propranolol was initiated at 10 mg four times daily, then individually up‐titrated to an optimal dose when the heart rate remained below 100 beats per minute during exercise or the systolic time interval increased by 30%. The mean (± SE) optimal dose of propranolol was 212 ± 68 mg given in four divided doses daily.
There was no significant difference between the two study groups in all‐cause mortality. No mortality attributed to Marfan syndrome was reported. All non‐fatal serious adverse events were not reported. However, the study authors reported pre‐defined non‐fatal clinical endpoints that included aortic dissection, aortic regurgitation, congestive heart failure, and cardiovascular surgery. Their analysis showed no difference between the treatment and control groups in these outcomes. Additionally, beta‐blocker therapy did not reduce the incidence of acute aortic dissection, aortic regurgitation, congestive heart failure or cardiovascular surgery.
The trial authors reported a significantly reduced rate of aortic dilatation measured by M‐mode echocardiography in the treatment group (aortic ratio mean slope: 0.084 (control) vs 0.023 (treatment), P < 0.001). All other secondary outcomes (systolic and diastolic blood pressure, total adverse events and withdrawal due to adverse events) were not adequately reported in the treatment or control group at the study end‐point and therefore could not be analysed.
We judged this trial to be at low risk of random sequence generation but high risk of selection (allocation concealment) bias, performance bias, detection bias, attrition bias and selective reporting bias.
Overall completeness and applicability of evidence
Overall completeness
Only one RCT satisfied the inclusion criteria for this review. The sample size calculation was not reported. The power of the study to detect a difference in the aortic root dimensions (the primary outcome measure) was not reported.
Primary outcomes
Although all‐cause mortality from both study groups was reported, it was not the trial's primary outcome and considered as one of numerous clinical end‐points. It is therefore difficult to interpret all‐cause mortality, as participants withdrawn for other reasons were not followed up to the study's completion. We commend the average study duration (9.3 and 10.7 years in the control and treatment groups respectively), however, the rationale for this length is unclear as no protocol was available. Each participant differed in the length of follow‐up according to whether pre‐defined end‐points were reached, but individual participant data for this were not provided. The study authors report on the clinical end‐points reached, however no data were presented on other reasons for withdrawal.
Other primary outcomes pre‐defined by the review were not fully reported in the published article as previously described. Personal communication with the study authors by email resulted in obtaining important information regarding other non‐fatal clinical outcomes such as congestive heart failure and participants undergoing cardiovascular surgery.
Secondary outcomes
Aortic root dimensions, considered the primary outcome by Shores 1994, were measured by M‐mode echocardiography throughout the study duration to maintain conformity. This single modality is however a key limitation, as the current gold standards employ cardiovascular computed tomography (CT) or magnetic resonance imaging (MRI) to confirm the echocardiographic measurements of the aortic diameter (Hiratzka 2010, Wright 2012). Although all echocardiographic data were interpreted by the same investigator, ruling out inter‐observer variability in the study, this measurement tool is still prone to intra‐observer variability and error.
It is also important to note that the initial aortic diameter was significantly different between the study groups and the Shores 1994 investigators accounted for this by measuring aortic ratio as a parameter and assessing the rate of change in aortic ratio as a function of the initial aortic diameter. However, they did not provide timelines for the two participants (one from each study group) who reached the clinical end‐point of aortic root greater than 6 cm despite the differing initial aortic ratios (1.4 in control, 2.1 in treatment).
The rationale for the study was the effect of beta‐blockade on arterial haemodynamic function. Although the study provided blood pressure, heart rate and systolic time‐interval measurements at baseline for both study groups, these parameters were not provided for the study end‐point and only reported for the treatment group when optimal dose was achieved. Correlation with haemodynamic function with aortic root dimensions would have provided valuable information.
Other limitations of the study are the open‐label design and lack of placebo in the control group. The adverse effects of long‐term beta‐blocker therapy were not reported in the published study for either group, however personal communication with the study authors confirmed that there were no adverse effects in the control group. The timeline to when the adverse effects reported for the treatment group have not been provided. Although the study states that serum propranolol levels were taken, these data were not available in the published manuscript.
Applicability of evidence
The authors of this review feel that the applicability of the evidence is limited and can be broadly summarised in two categories:
Intervention
Current guidelines recommend beta‐1 selective agents such as atenolol or metoprolol and it is therefore difficult to determine the appropriate dosing schedules for these drugs as this trial used propranolol. It is recommended that dosing is adjusted to maintain a heart rate after submaximal exercise to less than 100 beats per minute in adults and less than 110 beats per minute in children (Wright 2012).
When this study began, propranolol, a non‐selective beta‐adrenergic blocker was the only preparation available for clinical use. Propranolol was the only beta‐blocker studied as compared to control. We did not identify any RCTs comparing atenolol or metoprolol to control in people with Marfan syndrome.
Population
This evidence cannot be applied to people with Marfan syndrome with aortic dissection, aortic regurgitation, moderate or severe mitral regurgitation, previous cardiovascular surgery, those with dyspnoea due to moderate exercise, orthopnoea or peripheral oedema, left ventricular ejection fraction of less than 50%, atrioventricular conduction delay of any degree, and those with disorders in which propranolol is contraindicated (diabetes mellitus or recurrent bronchospasm requiring medical treatment).
Further, the current best recognised tool to diagnose Marfan syndrome is the Revised 2010 Ghent nosology criteria (Loeys 2010), first proposed in 1996 (De Paepe 1996) to overcome the tendency to over‐diagnose people with Marfan syndrome (Wright 2016). As participants in Shores 1994 met the diagnostic criteria published by Beighton 1988, now largely considered a joint hypermobility score, it is possible that people without Marfan syndrome may have been included in the study.
Male and female participants in both study groups had greater measured initial aortic diameter than expected values, therefore the results from the study cannot be applied to people without aortic dilatation at the initiation of beta‐blocker therapy. Furthermore, as children less than 12 years old were excluded, the outcomes can only be generalised to the adolescent and adult population although the study authors suggest earlier treatment provides greater benefit.
Quality of the evidence
Although the single included study is a RCT, which is considered the highest‐quality study design, we judged the evidence to be of low quality using the GRADE approach for all outcomes. The single included study was at high risk of selection (allocation concealment), performance, detection, attrition and selective reporting bias. Additionally, we further downgraded the evidence because we judged imprecision to be serious, due to very wide confidence intervals for all outcome measures.
With the exception of mortality and aortic diameter, most other primary as well as secondary clinically relevant outcomes identified in this review were not adequately reported in the published manuscript. In this study when participants reached a pre‐defined end point, they were withdrawn from the trial and their follow‐up data were not reported. Adverse events data were poorly documented and therefore we could not evaluate the benefits versus harm of long‐term beta‐blocker therapy.
Since we judged the single trial to be of low quality, our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect. This study provides inadequate evidence to inform people with Marfan syndrome, and their families and care‐providers.
Potential biases in the review process
We performed a comprehensive literature search to identify all studies meeting the inclusion criteria from several databases. We did not limit our search to a particular language and we followed strictly the methodology described in the published protocol.
Only one randomised trial met the inclusion criteria, therefore study findings have not been replicated. It was not possible to collect all relevant information for the pre‐defined outcomes outlined in this review as they were not reported in the published study. However, we successfully contacted the study authors and obtained additional information.
Agreements and disagreements with other studies or reviews
The present study is the only RCT that compares beta‐blocker therapy with no treatment in Marfan syndrome. Although RCTs are considered the highest level of evidence, we recognise that data are available from non‐randomised studies on beta‐blocker monotherapy in Marfan syndrome and provide valuable information.
Several retrospective and prospective studies evaluating the efficacy of beta‐blocker therapy have demonstrated conflicting results and we have summarised these in Additional Table 4.
| Study | Study design | Study population | Results | Conclusions |
|
|
|
|
| |
|
|
|
|
| |
|
|
|
|
| |
|
|
|
|
| |
|
|
|
|
| |
|
|
|
|
| |
|
|
|
|
| |
|
|
|
|
|
ACEI: angiotensin‐converting enzyme inhibitor; AGV: aortic growth velocity; BP: blood pressure; CT: computerised tomography; MRI: magnetic resonance imaging
Beta‐blocker compared to no treatment
The non‐randomised, retrospective studies by Ladouceur 2007 and Salim 1994 analysed the effect of beta‐blockade on aortic dilatation in participants with Marfan’s syndrome. Although Ladoceur and colleagues focused on a paediatric sub‐population (mean age at diagnosis: 6.7 years) and Salim and colleagues included an older, adolescent sub‐population less than 21 years of age, both studies reported that beta‐blocker therapy reduced the rate of aortic dilatation. We note that Salim 1994 had a small number of participants in the control group (N = 13), composed of patients who could not or would not take beta‐blockade therapy, compared to the beta‐blocker treatment group (N = 100). Beta‐blockers used in both studies were atenolol, nadolol and propranolol. In contrast, another retrospective study, Tierney 2007, demonstrated no significant difference between beta‐blocker therapy and no treatment over a six‐ to seven‐year follow‐up in a population less than 18 years old. Additionally there was no significant difference in clinical endpoints reached or adverse symptoms reported. Although specific exercise testing was not routinely used to assess optimal beta‐blockade dosing, the study authors report that the heart rate and heart rate z‐scores were significantly lower in the treatment group at the time of last follow‐up.
We identified two prospective studies assessing beta‐blocker mono‐therapy compared to no treatment (Rossi‐Foulkes 1999; Tahernia 1993). Tahernia 1993 was a small study of six participants that reported that people with Marfan’s syndrome (N = 3) taking beta‐blockade therapy had no progression of aortic root dilatation. Rossi‐Foulkes 1999, a non‐randomised, prospective study compared treatment with beta‐blocker or calcium channel blocker with no treatment in children with Marfan syndrome and reported a beneficial impact of drug therapy on the absolute and relative rates of aortic root growth.
Legget 1996 and Silverman 1995 were two retrospective studies. Although their primary endpoints did not include measuring the effects of beta‐blocker therapy in Marfan’s syndrome, they did include a subanalysis of the efficacy of beta‐blocker treatment in their published reports. Silverman 1995, a large retrospective, multi‐centre study of 417 participants with Marfan’s syndrome evaluating life‐expectancy in this population suggested that the 119 participants on a wide variety of types and dosages of beta‐blocker therapy conferred benefit for survival. The study did not report on aortic root changes. Legget 1996 studied various clinical and echocardiographic predictors of outcome in Marfan’s syndrome and reported no change in aortic ratio between beta‐blocker therapy (N = 30) and no therapy (defined as never received or received for less than one year, N = 53), over a mean follow‐up of four years.
Though the above studies suggest agreement with the included Shores 1994 RCT in this review that beta‐blockers may decrease the rate of aortic dilatation, these studies must be interpreted with caution given their limitations, differences in study design and high risk of bias.
Beta‐blocker compared to other anti‐hypertensives
The data comparing beta‐blocker therapy to other anti‐hypertensives are more robust. Although our review focus is beta‐blocker monotherapy, for the reasons outlined in the introduction, we provide a brief summary and outline of the literature below and in Table 5. We excluded the RCTs listed in this table from our review as they did not include a no‐treatment control group.
| Study | Study design | Patient population | Results | Conclusions |
|
|
|
|
| |
|
|
|
|
| |
|
|
|
|
| |
|
|
|
|
| |
|
|
|
|
| |
|
|
|
|
|
ACEI: ACEI: angiotensin‐converting enzyme inhibitor; ARB: angiotensin receptor blockers; BP: blood pressure
Phomakay 2014 was a retrospective study of 67 people, comparing the effects of beta‐blockers versus ACE inhibitors versus no treatment on aortic root growth rate in Marfan syndrome for an average of 7.6 years. A normal control group was also included. Beta‐blockers used were atenolol, metoprolol and propranolol, while the ACE inhibitors were lisinopril, enalapril and captopril. They reported that beta‐blocker therapy resulted in near normalisation of aortic growth velocity (AGV), while ACE inhibitors did not significantly attenuate AGV. Two participants experienced aortic dissections, one from the beta‐blocker group and the other from the ACE‐inhibitor group. In contrast, Yetman 2005, a prospective, non‐randomised trial assessing the effects of enalapril compared to beta‐blocker therapy in 57 participants over three years demonstrated that enalapril was superior to beta‐blocker in improving aortic distensibility and stiffness, with associated slower rate of aortic growth. Additionally, the study authors reported an increased number of adverse effects and participants undergoing aortic root replacement from the beta‐blocker group. One death occurred in the beta‐blocker group with documented ventricular tachyarrhythmia.
Three small and relatively short RCTs (Bhatt 2015; Sandor 2015; Williams 2012) evaluated the haemodynamic and biophysical aortic effects of pharmacological therapy in Marfan’s syndrome. Williams 2012 assessed the efficacy of atenolol, perindopril and verapamil in 18 participants with a mean age of 30.4 years and reported that all drug groups reduced peripheral and central pressure. Atenolol further delayed aortic wave travel. No significant change in aortic diameter was observed over 18 weeks of follow‐up. Bhatt 2015 and Sandor 2015 both compared the effects of losartan and atenolol and although they reported differing effects of atenolol on pulse wave velocity, concluded that atenolol and losartan had different mechanisms of action on aortic function, indicating a role for both in the treatment for Marfan’s syndrome. Neither study reported a significant change in aortic root dilatation for either treatment group over a study duration of 6 to 12 months.
Two RCTs (Forteza 2016; Lacro 2014) have recently been published, comparing the effects of losartan and atenolol over a three‐year period. Neither had a no‐treatment control group. Lacro 2014 recruited 604 participants distributed between 21 clinical centres and found no significant difference in the rate of aortic root dilatation between losartan and beta‐blocker treatment in children and young adults (age range: 6 to 25 years). They reported a possible higher rate of adverse events in participants treated with beta‐blockade, but there was no difference in adverse clinical outcomes between the two groups. Similarly, Forteza 2016, a randomised, parallel, double‐blind study reported no significant difference in the progression of aortic root and ascending aortic diameters between losartan and atenolol. Further, they demonstrated that aortic root diameter increased significantly in both groups. No serious adverse effects were observed in either treatment group.
Comparison with other reviews
Two meta‐analyses have been conducted on the efficacy of beta‐blocker therapy versus no treatment in people with Marfan syndrome (Gao 2011; Gersony 2007).
-
Gao 2011 focused on a subpopulation of children and adolescents with Marfan syndrome and included five non‐randomised studies (Ladouceur 2007; Rossi‐Foulkes 1999; Salim 1994; Tahernia 1993; Tierney 2007) to assess the effectiveness of beta‐blocker therapy on aortic dilatation and clinical outcomes (death, cardiovascular surgery, aortic dissection or rupture). Out of a study population of 392 children and adolescents less than 18 years old, their meta‐analysis reported a significantly decreased rate of aortic dilatation with treatment (SMD ‐1.30, 95% CI ‐2.11 to ‐0.49, P = 0.002). There was no significant difference in clinically important endpoints (death, cardiovascular surgery, or aortic dissection or rupture). Beta‐blockers used were mainly atenolol or propranolol. The authors acknowledge an important limitation of their analysis in that the measured aortic change was not normalised to body size as these data were not reported in their selected studies. However, they recognise that adiposity is often reduced in their young study population and state that reporting on absolute diameter is appropriate.
-
In contrast, Gersony 2007 report from their meta‐analysis that there is no evidence of clinical benefit from long‐term beta‐blockade in people with Marfan syndrome. Their analysis imposed no restriction on age and included six studies, one of which was the RCT included in this review (Shores 1994), two non‐randomised, prospective studies, and three studies that were not designed to observe the effect of beta‐blocker therapy. Clinical endpoints were defined as aortic dissection or rupture, cardiovascular surgery or death. Aortic root dimensions were not included in the analysis. Beta‐blockers included in the studies were mainly atenolol and propranolol, but one study (Silverman 1995) additionally included nadolol and metoprolol. Out of 433 participants in the treatment group and 369 participants in the untreated group, participants on beta‐blocker therapy were more likely to reach a clinical endpoint using a fixed‐effects model (odds ratio (OR) 1.50, 95% CI 1.05 to 2.16). However, when a random‐effects model was applied, no statistical significance was reached for treatment effect (OR 1.54, 95% CI 0.99 to 2.40). The authors acknowledge that the length of time on beta‐blocker treatment was not included in three selected studies and the presence or non‐presence of advanced aortic disease was not included in the analysis.
These two meta‐analyses report contradicting conclusions on the efficacy of beta‐blocker therapy but they are both limited by the heterogeneous design of the included studies and the combination of randomised and non‐randomised data. Gao 2011's conclusion that beta‐blocker therapy can significantly slow the progression of aortic dilatation is in keeping with our single included RCT.
A recent review by Singh 2016 provides a comprehensive summary on recent clinical drug trials and clinical implications in Marfan syndrome. This publication presents data comparing the various groups of anti‐hypertensive medications that have been examined for their prophylactic effectiveness on aortic dilatation including beta‐blockers, ARBs with and without baseline beta‐blocker therapy, and ACE inhibitors. The report concludes that medical therapy in Marfan syndrome should be individualised according to patient tolerance and various risk factors including age and family history of aortic dissection. The authors of Singh 2016 recommend that those with aortic root dilatation should receive therapy with adequate doses of either a beta‐blocker or ARB, and if severe, a combination of these therapies should be considered. They state that the evidence for prophylactic medication in people without aortic dilatation is less clear.
In summary, the above comparisons with other identified studies and reviews are suggestive that beta‐blocker therapy may decrease the rate of aortic root dilatation in Marfan syndrome. However, our review specifically sought to examine the effect of beta‐blockers in the prevention of aortic dissection, the significant contributor to morbidity and mortality in this disease, and this outcome along with additional clinical adverse events remain inadequately addressed.
Study flow diagram
Risk of bias summary: review authors' judgements about each risk of bias item for each included study
Comparison 1 Beta‐blocker versus no treatment, Outcome 1 All‐cause mortality.
Comparison 1 Beta‐blocker versus no treatment, Outcome 2 Clinically important non‐fatal endpoints as defined by Shores 1994.
Comparison 1 Beta‐blocker versus no treatment, Outcome 3 Acute aortic dissection.
Comparison 1 Beta‐blocker versus no treatment, Outcome 4 Aortic regurgitation.
Comparison 1 Beta‐blocker versus no treatment, Outcome 5 Congestive heart failure.
Comparison 1 Beta‐blocker versus no treatment, Outcome 6 Cardiovascular surgery.
Comparison 1 Beta‐blocker versus no treatment, Outcome 7 Aortic root diameter > 6 cm.
| Beta‐blockers compared with placebo, no treatment or surveillance only for preventing aortic dissection in Marfan syndrome | ||||||
| Patient or population: people with confirmed diagnosis of Marfan syndrome Settings: outpatient Intervention: beta‐blockers Comparison: placebo or no treatment or surveillance only | ||||||
| Outcomes Follow‐up: mean 9.3 years in control group and 10.7 years in treatment group | Anticipated absolute effects* (95% CI) | Relative effect | No of participants | Quality of the evidence | Comments | |
| Risk with placebo, no treatment or surveillance only | Risk with beta blockers | |||||
| All‐cause mortality | 53 per 1000 | 13 per 1000 (1 to 250) | RR 0.24 (0.01 to 4.75) | 70 | ⊕⊕⊝⊝ | |
| Clinically important non‐fatal end point as defined byShores 1994(aortic dissection, aortic regurgitation, congestive heart failure, or cardiovascular surgery) | 316 per 1000 | 249 per 1000 (117 to 534) | RR 0.79 (0.37 to 1.69) | 70 | ⊕⊕⊝⊝ | |
| Acute aortic dissection | 105 per 1000 | 62 per 1000 (13 to 319) | RR 0.59 (0.12 to 3.03) | 70 | ⊕⊕⊝⊝ | |
| Aortic regurgitation | 53 per 1000 | 63 per 1000 (9 to 419) | RR 1.19 (0.18 to 7.96) | 70 | ⊕⊕⊝⊝ | |
| Congestive heart failure | 53 per 1000 | 63 per 1000 (9 to 419) | RR 1.19 (0.18 to 7.96) | 70 | ⊕⊕⊝⊝ | |
| Cardiovascular surgery | 105 per 1000 | 62 per 1000 (13 to 319) | RR 0.59 (0.12 to 3.03) | 70 | ⊕⊕⊝⊝ | |
| Aortic root diameter > 6 cm | 26 per 1000 | 31 per 1000 (2 to 480) | RR 1.19 (0.08 to 18.24) | 70 | ⊕⊕⊝⊝ | |
| *The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). | ||||||
| GRADE Working Group grades of evidence High quality: we are very confident that the true effect lies close to that of the estimate of the effect. | ||||||
| 1Downgraded one step for risk of bias (high risk of selection, performance, detection, attrition and selective outcome reporting bias). | ||||||
| Initial aortic diameter | Control male (N = 19) | Control female (N = 19) | Treatment male (N = 20) | Treatment female (N = 12) |
| Measured (mm) ± SD | 31.1 ± 6.9 | 29.4 ± 6.8 | 36.7 ± 9.3 | 31.2 ± 5.3 |
| Expected (mm) ± SD | 24.6 ± 3.9 | 23.3 ± 3.2 | 25.5 ± 4.1 | 23.3 ± 4.2 |
| Ratio | 1.27 ± 0.19 | 1.27 ± 0.26 | 1.43 ±0.26 | 1.37 ± 0.2 |
| SD: standard deviation | ||||
| BP measured as mmHg ±SD | Control male (N = 19) | Control female (N = 19) | Treatment male (N = 20) | Treatment female (N = 12) |
| SBP at baseline | 118 ± 14 | 110 ± 13 | 115 ± 13 | 115 ± 14 |
| DBP at baseline | 72 ± 11 | 70 ± 10 | 73 ± 10 | 69 ± 13 |
| SBP during optimal dose | Not reported | Not reported | 108 ± 15 | 108 ± 8 |
| DBP during optimal dose | Not reported | Not reported | 66 ± 11 | 63 ± 7 |
| End of treatment | Not reported | Not reported | Not reported | Not reported |
| BP: blood pressure; DBP: diastolic blood pressure; SBP: systolic blood pressure | ||||
| Adverse effect | Beta‐blocker group (N = 30) | Control |
| Third‐degree heart block | 1 | Not reported in Shores 1994, however, the study author reported that no adverse events occurred in the control group (personal communication) |
| First‐degree heart block | 3 | |
| Lethargy | 8 | |
| Depression | 1 | |
| Insomina | 4 | |
| Dream disturbance | 3 | |
| Mild bronchospasm | 1 | |
| Accentuated effects of alcohol | 1 | |
| Total with one or more effects | 10 |
| Study | Study design | Study population | Results | Conclusions |
|
|
|
|
| |
|
|
|
|
| |
|
|
|
|
| |
|
|
|
|
| |
|
|
|
|
| |
|
|
|
|
| |
|
|
|
|
| |
|
|
|
|
| |
| ACEI: angiotensin‐converting enzyme inhibitor; AGV: aortic growth velocity; BP: blood pressure; CT: computerised tomography; MRI: magnetic resonance imaging | ||||
| Study | Study design | Patient population | Results | Conclusions |
|
|
|
|
| |
|
|
|
|
| |
|
|
|
|
| |
|
|
|
|
| |
|
|
|
|
| |
|
|
|
|
| |
| ACEI: ACEI: angiotensin‐converting enzyme inhibitor; ARB: angiotensin receptor blockers; BP: blood pressure | ||||
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 1 All‐cause mortality Show forest plot | 1 | 70 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.24 [0.01, 4.75] |
| 2 Clinically important non‐fatal endpoints as defined by Shores 1994 Show forest plot | 1 | 70 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.79 [0.37, 1.69] |
| 3 Acute aortic dissection Show forest plot | 1 | 70 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.59 [0.12, 3.03] |
| 4 Aortic regurgitation Show forest plot | 1 | 70 | Risk Ratio (M‐H, Fixed, 95% CI) | 1.19 [0.18, 7.96] |
| 5 Congestive heart failure Show forest plot | 1 | 70 | Risk Ratio (M‐H, Fixed, 95% CI) | 1.19 [0.18, 7.96] |
| 6 Cardiovascular surgery Show forest plot | 1 | 70 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.59 [0.12, 3.03] |
| 7 Aortic root diameter > 6 cm Show forest plot | 1 | 70 | Risk Ratio (M‐H, Fixed, 95% CI) | 1.19 [0.08, 18.24] |
