Articles Angiotensin receptor blockers and β blockers in Marfan syndrome: an individual patient data meta-analysis of randomised trials Alex Pitcher, Enti Spata, Jonathan Emberson, Kelly Davies, Heather Halls, Lisa Holland, Kate Wilson, Christina Reith, Anne H Child, Tim Clayton, Matthew Dodd, Marcus Flather, Xu Yu Jin, George Sandor, Maarten Groenink, Barbara Mulder, Julie De Backer, Arturo Evangelista, Alberto Forteza, Gisela Teixido-Turà, Catherine Boileau, Guillaume Jondeau, Olivier Milleron, Ronald V Lacro, Lynn A Sleeper, Hsin-Hui Chiu, Mei-Hwan Wu, Stefan Neubauer, Hugh Watkins, Hal Dietz, Colin Baigent, on behalf of The Marfan Treatment Trialists’ Collaboration Summary Lancet 2022; 400: 822–31 Background Angiotensin receptor blockers (ARBs) and β blockers are widely used in the treatment of Marfan syndrome Published Online to try to reduce the rate of progressive aortic root enlargement characteristic of this condition, but their separate and August 29, 2022 joint effects are uncertain. We aimed to determine these effects in a collaborative individual patient data meta-analysis https://doi.org/10.1016/ of randomised trials of these treatments. S0140-6736(22)01534-3 See Comment page 790 Methods In this meta-analysis, we identified relevant trials of patients with Marfan syndrome by systematically The Heart Centre, John Radcliffe Hospital, Oxford University searching MEDLINE, Embase, and CENTRAL from database inception to Nov 2, 2021. Trials were eligible if they Hospitals NHS Foundation involved a randomised comparison of an ARB versus control or an ARB versus β blocker. We used individual patient Trust, Oxford, UK data from patients with no prior aortic surgery to estimate the effects of: ARB versus control (placebo or open control); (A Pitcher DPhil, X Y Jin FRCP); ARB versus β blocker; and indirectly, β blocker versus control. The primary endpoint was the annual rate of change MRC Population Health Research Unit, Nuffield of body surface area-adjusted aortic root dimension Z score, measured at the sinuses of Valsalva. Department of Population Health, University of Oxford, Findings We identified ten potentially eligible trials including 1836 patients from our search, from which seven trials Oxford, UK (E Spata MSc, and 1442 patients were eligible for inclusion in our main analyses. Four trials involving 676 eligible participants Prof J Emberson PhD, K Davies BSc, H Halls BSc, compared ARB with control. During a median follow-up of 3 years, allocation to ARB approximately halved the L Holland PhD, annual rate of change in the aortic root Z score (mean annual increase 0·07 [SE 0·02] ARB vs 0·13 [SE 0·02] control; Prof C Baigent FMedSci); absolute difference –0·07 [95% CI –0·12 to –0·01]; p=0·012). Prespecified secondary subgroup analyses showed that Clinical Trial Service Unit and the effects of ARB were particularly large in those with pathogenic variants in fibrillin-1, compared with those without Epidemiological Studies Unit, Nuffield Department of such variants (heterogeneity p=0·0050), and there was no evidence to suggest that the effect of ARB varied with Population Health, University β-blocker use (heterogeneity p=0·54). Three trials involving 766 eligible participants compared ARBs with β blockers. of Oxford, Oxford, UK (E Spata, During a median follow-up of 3 years, the annual change in the aortic root Z score was similar in the two groups Prof J Emberson, K Davies, (annual increase –0·08 [SE 0·03] in ARB groups vs –0·11 [SE 0·02] in β-blocker groups; absolute difference 0·03 H Halls, L Holland, K Wilson DPhil, C Reith PhD, Prof C Baigent); [95% CI –0·05 to 0·10]; p=0·48). Thus, indirectly, the difference in the annual change in the aortic root Z score Royal Brompton and Harefield between β blockers and control was –0·09 (95% CI –0·18 to 0·00; p=0·042). Hospitals Unit, Guy’s and St Thomas’ NHS Trust and Interpretation In people with Marfan syndrome and no previous aortic surgery, ARBs reduced the rate of increase of Department of Surgery and Oncology, Imperial College the aortic root Z score by about one half, including among those taking a β blocker. The effects of β blockers were London, London, UK similar to those of ARBs. Assuming additivity, combination therapy with both ARBs and β blockers from the time of (A H Child FRCP); Clinical Trials diagnosis would provide even greater reductions in the rate of aortic enlargement than either treatment alone, which, Unit, Department of Medical if maintained over a number of years, would be expected to lead to a delay in the need for aortic surgery. Statistics, London School of Hygiene & Tropical Medicine, London, UK (Prof T Clayton MSc, Funding Marfan Foundation, the Oxford British Heart Foundation Centre for Research Excellence, and the UK M Dodd MSc); Faculty of Medical Research Council. Medicine and Health Sciences, University of East Anglia, Norwich, UK Copyright © 2022 The Author(s). Published by Elsevier Ltd. This is an Open Access article under the CC BY 4.0 (Prof M Flather MBBS); license. Children’s Heart Centre, British Columbia’s Children’s Hospital, Vancouver, BC, Canada Introduction purpose, but their use is based mainly on the results of (Prof G Sandor MBChB); Marfan syndrome is a genetic disorder, usually caused by observational studies8,9 and one small randomised trial.10 Department of Paediatrics, pathogenic variants in the fibrillin-1 (FBN1) gene that The discovery that transforming growth factor β (TGFβ) Faculty of Medicine, University causes progressive enlargement of the aortic root. If dysregulation is implicated in the pathogenesis of some of British Columbia, Vancouver, BC, Canada unchecked, aortic enlargement in Marfan syndrome can aortic aneurysms led to the hypothesis that angiotensin (Prof G Sandor); lead to life-threatening aortic dissection, sometimes in receptor blockade (which attenuates TGFβ activity) might Academic Medical Center, early adulthood.1–7 Initial treatment is aimed at slowing slow aortic root growth in Marfan syndrome.11,12 Favourable University of Amsterdam, aortic root growth, and β blockers are widely used for this results in animal models11 and in small observational 822 www.thelancet.com Vol 400 September 10, 2022 Articles Amsterdam, Netherlands Research in context (M Groenink PhD, Prof B Mulder PhD); Center for Evidence before this study Added value of this study Medical Genetics and We searched Embase, MEDLINE, and CENTRAL from inception This meta-analysis of individual patient data from randomised Department of Cardiology, until Nov 2, 2021, for randomised trials that assessed the trials, which followed a protocol that was agreed and Ghent University Hospital, Ghent, Belgium effects of angiotensin receptor blocker (ARB) versus control published before any analyses, showed that ARBs reduced the (Prof J De Backer PhD); Servei de or ARB versus β blockers in patients with Marfan syndrome. rate of aortic root enlargement by about one half, and that Cardiologia, Hospital Cochrane search filters were used in Embase and Medline to this effect was generalisable to different types of patients. In Universitari Vall d´Hebron, identify randomised trials, and the terms “Marfan” or particular, ARBs were effective even among those already Barcelona, Spain (Prof A Evangelista PhD); “Marfan syndrome” were used in all three databases taking a β blocker. The estimated effect of ARBs was Hospital Puerta de Hierro, (appendix p 2). The searches were not restricted to English significantly greater among those with a pathogenic variant in Majadahonda, Spain language publications. We identified a number of fibrillin-1, than those without such a fibrillin-1 variant, (A Forteza PhD); Department of Cardiology, Hospital randomised trials of ARB versus placebo (or open control) providing biological support for the effect. β blockers were Universitari Vall d’Hebron, in which the aim was to estimate the effects of treatment estimated to have a similar beneficial effect as ARBs. CIBER-CV, Vall d’Hebron institut with ARBs on aortic root size in patients with Marfan de Recerca, Universitat Implications of all the available evidence syndrome. Several meta-analyses of published data from Autònoma de Barcelona, If tolerated, the combination of a β blocker and an ARB could Barcelona, Spain these trials were also identified, but none were able to define reduce the rate of enlargement of the aortic root by at least (G Teixido-Turà PhD); Université the effects of ARBs in different circumstances, including Paris Cité and Université one half, and potentially by much more than this. Although this according to age, sex, or blood pressure, but also according to Sorbonne Paris Nord, Inserm meta-analysis of trials did not have sufficient power to assess whether or not a β blocker was part of existing treatment, U1148, LVTS, effects on the need for surgery (and it is unlikely that any F-75018 Paris, France and whether or not a diagnosis of Marfan syndrome had been randomised study will ever be done to study this question (Prof C Boileau PhD, confirmed by genotyping. There were no large, randomised Prof G Jondeau PhD, directly), as elective surgery is almost always driven by aortic root trials assessing the efficacy of a β blocker for O Milleron MD); Service de size and rate of expansion, our results suggest that long-term Marfan syndrome, despite the fact that such treatment is Cardiologie (Prof C Boileau, combination therapy could eventually reduce such outcomes. Prof G Jondeau, O Milleron); used widely for this condition. CRMR Syndrome de Marfan et apparentés (Prof C Boileau, Prof G Jondeau, O Milleron), studies in humans12–14 led to several randomised trials in an assessment of the effects of β-blocker therapy given AP-HP Hôpital Bichat-Claude patients with Marfan syndrome, the first of which was alone and the effects of combined ARB and β-blocker Bernard, F-75018, Paris, France; Department of Cardiology, published in 2013.15–24 Combining these trials could therapy. Boston Children’s Hospital, increase the precision in estimates of treatment effect and Boston, MA, USA (R V Lacro MD, increase the power of subgroup analyses. In 2012, we Methods L A Sleeper ScD); Department of established a collaborative group (the Marfan Treatment Study design and outcomes Pediatrics, Harvard Medical School, Boston, MA, USA Trialists’ [MTT] Collaboration) to do a meta-analysis of In this meta-analysis, we identified relevant trials by (R V Lacro, L A Sleeper); individual patient data from all relevant Marfan syndrome systematically searching MEDLINE, Embase, and Department of Pediatrics, trials. A protocol was subsequently agreed for the CENTRAL from database inception to Nov 2, 2021. Trials Taipei Tzu Chi Hospital, rationale, design, and conduct of the meta-analysis.25 were eligible if they involved a randomised comparison Buddhist Tzu Chi Medical Foundation, Since 2012, most of the participating trial groups in the of an ARB versus control or an ARB versus β blocker in New Taipei City, Taiwan MTT Collaboration have reported their results,15–18,20–24 and patients with Marfan syndrome and if patient-level data (H-H Chiu PhD); Department of several meta-analyses of published data have been were available. A description of the search terms is in the Pediatrics and Adult Congenital reported.26–28 Two of these meta-analyses26,28 did not appendix (p 2). The review of search results was done Heart Center, National Taiwan University Hospital, include the recent UK-based AIMS trial,23 but one that independently by two authors (including HH, KD, KW, Taipei, Taiwan did include this trial27 concluded that angiotensin and LH) and adjudicated by either AP or CBa. Relevant (Prof M-H Wu PhD); Oxford receptor blockers (ARBs) are effective when used alone trials were also sought through enquiry with authors of Centre for Clinical Magnetic or when added to a β blocker. Since this meta-analysis collaborating trials.25 Each trial was reviewed for eligibility Resonance Research, Division of Cardiovascular Medicine included only published data, it was subject to biases that by reviewing its protocol and methods, and clarification (Prof S Neubauer MD), Radcliffe might arise from selective reporting of findings in was sought as necessary by discussion with the authors Department of Medicine, publications, and it was not able to harmonise definitions of the relevant trial. Bias was accounted for by including University of Oxford, of aortic root size for all trials or explore treatment effects only properly randomised, unconfounded trials in the Oxford, UK (A Pitcher, Prof H Watkins DPhil); Howard in detail (eg, among particular prognostic subgroups main analyses. Sources of variability within and between Hughes Medical Institute and such as those with a confirmed FBN1 pathogenic variant). studies was controlled by a harmonised definition of Department of Genetic The present report describes a meta-analysis in which patients to be included in the analyses, and explored by Medicine, Johns Hopkins the availability of individual patient data removes these prespecified subgroup analyses. The protocol for our University School of Medicine, Baltimore, MD, USA limitations, allowing a more complete assessment of study has been published.25 The primary aims of this (Prof H Dietz MD); European ARBs in Marfan syndrome. By prespecifying the use of meta-analysis were to estimate the effects of ARB and Reference Network for Rare indirect comparisons of trials of an ARB versus control β blockers on the change in aortic root size in patients Multisystemic Vascular Disease and of an ARB versus β blocker, our report also provides with Marfan syndrome and no previous aortic root (VASCERN), HTAD Rare Disease www.thelancet.com Vol 400 September 10, 2022 823 Articles Working Group, Amsterdam, surgery. Thus, patients with previous aortic root surgery An indirect assessment of the effect of β blockers The Netherlands (M Groenink, in the identified trials were excluded from analyses. The compared with controls was calculated using indirect Prof B Mulder); European Reference Network for Rare primary comparisons involved only the unconfounded comparisons of trial results31 as follows: if d1 (with Multisystemic Vascular Disease trials (ie, trials that had no protocol-mandated differences variance v1) is the difference in mean annual rate of (VASCERN), HTAD Rare Disease between randomised groups other than those created by change in aortic root Z scores estimated from the three Working Group, Ghent, the randomised allocations), but a sensitivity analysis trials that compared ARB with β blockers and d2 (with Belgium (Prof J De Backer); European Reference Network included one trial22 in which there were different dosing variance v2) is the difference in mean annual rate of for Rare Multisystemic Vascular strategies for β blockers in the ARB and control groups change in aortic root Z scores estimated from the Disease (VASCERN), HTAD Rare of the study. The prespecified primary outcome was the four trials that compared ARB with controls, then an Disease Working Group, annual rate of change of body surface area (BSA)-adjusted indirect estimate of the effect of β blockers is provided by Barcelona, Spain (Prof A Evangelista, aortic root dimension Z score, measured at the sinuses d2 – d1 (which has variance equal to v1 + v2). This indirect G Teixido-Turà); European of Valsalva. The secondary outcome was the annual rate analysis assumes that the effects of ARBs and β blockers Reference Network for Rare of change of the absolute aortic root dimension are additive (that is, the effect of an ARB is the same if a Multisystemic Vascular Disease measured at the sinuses of Valsalva. Other secondary β blocker is given or not and the effect of a β blocker is (VASCERN), HTAD Rare Disease Working Group, Paris, France aims were to assess those effects across different the same irrespective of whether an ARB is given or not). (Prof C Boileau, Prof G Jondeau, subgroups defined on the basis of patients’ characteristics For the primary comparisons, a two-sided p value less O Milleron) at baseline; to assess effects on cardiovascular outcomes, than 0·05 was considered significant. Analyses were Correspondence to: including aortic dissection, aortic root surgery and done using SAS version 9.4 and R version 3.5.0. The Prof Colin Baigent, MRC death, as well as the composite of these three outcomes; MTT database is a research database, and did not involve Population Health Research Unit, Clinical Trial Service Unit and and to assess effects on a range of other outcomes and accessing or otherwise processing patient-identifiable Epidemiological Studies Unit, measures that were sufficiently complete to permit information, and hence did not require ethical approval. Nuffield Department of meaningful analyses (appendix p 4). Population Health, University of Role of the funding source Oxford, Oxford OX3 7LF, UK colin.baigent@ndph.ox.ac.uk Data analysis The funders of the study had no role in study design, See Online for appendix We requested individual participant data on relevant data collection, data analysis, data interpretation, writing characteristics including age, sex, previous aortic of the report, or the decision to submit. surgery, family history of Marfan syndrome, pathogenic variants in relevant genes, ectopia lentis, β-blocker Results usage, physical measurements, and blood pressure. We We identified ten trials with a total of 1836 patients as also collected data on aortic measurements, surgery, being potentially eligible for the study (table 1; appendix dissection, and death during the trial. The data analysis p 7).15–24 Three trials, with a total of 324 patients, were not methods used in our study have been published included in the primary analyses: one trial (262 patients) elsewhere.25 Z scores were calculated using the method was published as an abstract only and was unable to used by each trial (as reported in their main analysis) contribute data;32 one trial (34 patients) was published except in cases in which values were provided directly but unable to contribute data;16 and one trial (28 patients) by the trialist. Secondary analyses using the methods of contributed data but was found to be confounded Campens and colleagues29 and Pettersen and colleagues30 owing to protocol-mandated adjustment in β-blocker were done to explore the effect of using an alternative doses in the control group22 (this trial contributed only method to estimate aortic root Z score. A two-stage to a sensitivity analysis). Of the seven remaining meta-analysis approach was used. For each patient, a trials, 70 (4·6%) of the 1512 randomised patients were linear slope of the annual rate of change (from baseline) excluded because they had had previous aortic root of the BSA-adjusted Z score was calculated. The surgery. The main analyses therefore include individual difference in mean slopes between treatment groups data from 1442 participants in seven trials.15,17,18,20–24 (and its SE) was then calculated for each trial and Data for the primary analysis were available from four standard inverse-variance-weighted methods were used trials of ARB versus control, including 676 patients to estimate the overall inverse-variance-weighted (excluding 70 patients with previous aortic root surgery at average difference in slopes across all trials. A random enrolment; 353 assigned to ARB and 323 to control).15,18,20,23 effects meta-analysis, which assumes that the The mean age of participants in these trials was 29 years underlying set of trials are representative of an (SD 14), 367 (54%) were female and 507 (75%) were underlying population of possible trial designs, was also receiving β blockers at baseline (all trials allowed patients done as a sensitivity analysis. Patients with missing data to remain on their β blockers). Overall, 526 (83%) of on rate of change in aortic root Z score were excluded. 630 genotyped individuals had an FBN1 pathogenic variant To allow for multiple subdivisions of the data, only (table 2). The mean Z score at the sinuses of Valsalva at summary effect estimates are presented with 95% CIs; baseline was 3·76 (SD 2·14) in patients allocated to ARB all other effect estimates (such as results from individual and 3·64 (SD 1·94) in patients allocated to control. The trials or in particular patient subgroups) are presented mean annual change of the Z score during follow-up was with 99% CIs. 0·07 (SE 0·02) in the ARB group and 0·13 (SE 0·02) 824 www.thelancet.com Vol 400 September 10, 2022 Treatment comparison (daily Number of Median Median Measures Data provided Main outcome measures doses) randomly age, years follow- collected, to MTT assigned up, months secretariat patients months Body surface area Aortic dimension at the sinuses of Z score at the sinuses of Valsalva (mm) Valsalva Method Baseline, mean Methods Baseline, Method Baseline, (SD) mean (SD) mean (SD) ARB vs control (placebo or open control) Marfan Sartan18 Losartan (50–100 mg) vs 299 26 42 0, 6, 12, 18, Yes * 1·83 (0·27) Echo, end diastole, and 38·8 (5·8) Roman and 3·60 placebo 24, 30, 36, 42, leading edge to leading colleagues33 (2·15) 48, 54, 60 edge www.thelancet.com Vol 400 September 10, 2022 COMPARE20 Losartan (50–100 mg) vs 233 35 36 0, 36 Yes Haycock and 2·02 (0·24) MRI, end diastole, and 44·3 (5·2) † 4·42 open control colleagues34 inner edge to outer edge (1·75) AIMS23 Irbesartan (150–300 mg) vs 192 18 46 0, 12, 24, 36, Yes DuBois and 1·71 (0·40) Echo, peak systole, and 34·4 (5·6) Devereux 3·24 placebo 48, 60 Du Bois35 inner edge to inner edge and (2·04) colleagues36 Ghent Marfan15 Losartan (25–100 mg) vs 22 36 36 0, 6, 12, 24, 36 Yes DuBois and 1·98 (0·24) Echo and leading edge 41·2 (3·5) † 3·55 placebo Du Bois35 to leading edge (1·04) Taiwanese trial22 Losartan plus β blocker vs 28 13 35 0, 6, 12, 18, Yes (but only Unknown 1·39 (0·38) Echo and inner edge to 32·9 (6·1) † 2·07 β blocker‡ 24, 30, 35 included in inner edge (1·89) sensitivity analyses) Italian trial19 Losartan (100 mg adults or 160§ Unknown 48 0, 12, 24, 36, No (trial yet to Unknown ·· ·· ·· ·· ·· ≤1·4 mg/kg children) plus 48 publish its full nebivolol (10 mg adults or results) ≤0·16 mg/kg children) vs nebivolol (10 mg adults or ≤0·16 mg/kg children) ARB vs β blocker PHN24 Losartan (0·4–1·4 mg/kg) vs 608 11 36 0, 6, 12, 24, 36 Yes Haycock and 1·28 (0·48) Echo, systole, and inner 33·6 (7·1) Sluysmans 4·32 atenolol (0·5–4·0 mg/kg) colleagues34 edge to inner edge and Colan37 (1·35) LOAT21 Losartan (12·5–100 mg) vs 140 26 36 0, 36 Yes DuBois and 1·75 (0·36) MRI and end diastolic 36·1 (6·2) Devereux 3·17 atenolol (12·5–100 mg) Du Bois35 frame and (2·21) colleagues36 Canadian trial17 Losartan (25 mg) vs atenolol 18 17 12 0, 3, 6, 9, 12 Yes DuBois and 1·77 (0·22) Echo and trailing edge 39·0 (5·8) † 3·55 (25–50 mg) Du Bois35 to leading edge (1·93) Italian trial19 Losartan (100 mg adults or 155§ Unknown 48 0, 12, 24, 36, No (trial yet to Unknown ·· ·· ·· ·· ·· ≤1·4 mg/kg children) vs 48 publish its full nebivolol (10 mg adults or results) ≤0·16 mg/kg children) Boston, USA Losartan (100 mg) vs atenolol 34 35 6 0, 6 No (trialist Unknown ·· ·· ·· ·· ·· trial16 (50 mg) unable to provide data) For each trial, the imaging method used to estimate aortic root dimension was the method used in that trial’s primary analysis. ARB=angiotensin receptor blocker. MTT=Marfan Treatment Trialists. *Data not provided by the trialist so estimated using the DuBois and Du Bois34 method. †Data not provided by the trialist so estimated using the Devereux and colleagues35 method). ‡Adults were randomly assigned to losartan (25–100 mg per day) plus β blocker (at a low maintenance dose [50 mg atenolol once daily or 20 mg propranolol twice daily]) vs β blocker (atenolol or propranolol, maximum 150 mg per day), and children were randomly assigned to losartan (0·7 mg/kg per day, to a maximum of 50 mg per day) plus β blocker (low maintenance dose, 1 mg/kg per day atenolol or propranolol) vs β blocker (atenolol or propranolol, maximum 2 mg/kg per day). §Number reported in trialists’ abstract was number analysed (the total number of patients randomly assigned across all three groups of the Italian trial was 262). Table 1: Characteristics of ten trials of angiotensin receptor blockers in patients with Marfan syndrome Articles 825 Articles trials (heterogeneity p=0·11; figure 1), within these ARB vs control ARB vs β blocker trials, there was significant heterogeneity of treatment ARB (n=353) Control (n=323) ARB (n=384) β blocker (n=382) effects with an ARB between the 490 participants Median follow-up, years 3·0 (2·9–4·0) 3·0 (3·0–4·0) 3·0 (3·0–3·0) 3·0 (3·0–3·0) with a documented pathogenic variant in FBN1, Mean age, years 28·8 (14·7) 28·3 (13·8) 13·9 (9·9) 13·9 (9·7) compared with the 95 participants who did not have a Age, years pathogenic variant in FBN1 (figure 2; appendix p 8; <16 75 (21%) 67 (21%) 258 (67%) 254 (66%) heterogeneity p=0·0050). There was no significant ≥16 to <25 80 (23%) 78 (24%) 82 (21%) 88 (23%) heterogeneity of the effects of an ARB in any of the ≥25 to <40 114 (32%) 119 (37%) 37 (10%) 31 (8%) other prespecified subgroups. In particular, the mean ≥40 84 (24%) 59 (18%) 7 (2%) 9 (2%) annual change in aortic root Z score was similar Gender irrespective of whether or not patients were taking a Male 164 (46%) 145 (45%) 218 (57%) 212 (55%) β blocker at baseline (heterogeneity p=0·54). Female 189 (54%) 178 (55%) 166 (43%) 170 (45%) Secondary analyses were done to assess the sensitivity of Family history of Marfan syndrome the findings to different methods of aortic root size Yes 100 (28%) 82 (25%) 187 (49%) 188 (49%) measurement. For the secondary outcome of absolute No 164 (46%) 158 (49%) 111 (29%) 116 (30%) aortic dimension, the mean annual change was 0·38 mm Unknown 89 (25%) 83 (26%) 86 (22%) 78 (20%) (SE 0·04) in patients allocated ARB and 0·52 mm (SE 0·04) Family history of aortic dissection in patients allocated control, resulting in a mean difference Yes 6 (2%) 1 (<1%) 55 (14%) 58 (15%) of –0·14 mm (95% CI –0·26 to –0·02; p=0·025; appendix No 4 (1%) 4 (1%) 258 (67%) 254 (66%) p 9). Findings were similar when the analysis included one confounded trial22 in which there were different dosing Unknown 343 (97%) 318 (98%) 71 (18%) 70 (18%) strategies for β blockers in the ARB and control groups Presence of FBN1 during the study (mean difference –0·08 [95% CI Yes 270 (76%) 256 (79%) 135 (35%) 145 (38%) –0·13 to –0·03]; p=0·0009; appendix p 10), and were also No 59 (17%) 45 (14%) 27 (7%) 20 (5%) similar when the Z scores were calculated using the Unknown 24 (7%) 22 (7%) 222 (58%) 217 (57%) method described by Campens and colleagues29 (mean Ectopia lentis difference –0·04 [95% CI –0·07 to –0·01]; p=0·0071; Yes 84 (24%) 67 (21%) 129 (34%) 133 (35%) appendix p 11) or by Pettersen and colleagues30 (mean No 116 (33%) 108 (33%) 150 (39%) 134 (35%) difference –0·04 [95% CI –0·07 to –0·01]; p=0·017; Unknown 153 (43%) 148 (46%) 105 (27%) 115 (30%) appendix p 12). Current β-blocker use 265 (75%) 242 (75%) 0 0 Individual participant data were available from Aorta at the sinuses of Valsalva three trials of ARB versus β blockers, including 766 patients Mean dimension, mm 39·0 (6·8) 38·9 (6·5) 34·2 (7·0) 34·2 (7·1) (384 assigned to ARB and 382 to β blocker).17,21,24 The mean Mean Z score 3·76 (2·14) 3·64 (1·94) 4·18 (1·71) 4·03 (1·50) age of participants was 14 years (SD 10); 336 (44%) of 766 were Other baseline measures female, and 280 (86%) of 327 genotyped individuals had a Mean weight, kg 67·6 (19·7) 69·6 (21·3) 45·0 (23·7) 46·5 (23·7) pathogenic variant in FBN1 (table 2). The baseline mean Mean height, cm 178 (15) 179 (15) 155 (31) 156 (32) Z score was 4·18 (SD 1·71) in patients allocated to ARB and Mean systolic blood pressure, 117 (16) 117 (15) 102 (16) 102 (16) 4·03 (SD 1·50) in patients allocated to β blocker. The mean mm Hg annual change of the Z score during follow-up was –0·08 Mean diastolic blood 70 (11) 70 (10) 62 (11) 62 (11) pressure, mm Hg (SE 0·03) in the ARB group and –0·11 (SE 0·02) in the β-blocker group, and the mean difference in the change in Mean heart rate, beats per 64 (14) 65 (14) 78 (18) 77 (17) min Z scores between ARB groups and β-blocker groups was Mean body surface area, m² 1·83 (0·32) 1·86 (0·33) 1·36 (0·49) 1·40 (0·50) not significant (0·03 [95% CI –0·05 to 0·10]; p=0·48]; Mean BMI, kg/m² 20·9 (4·6) 21·3 (5·3) 17·3 (4·0) 17·5 (4·0) figure 1). There were no significant differences in aortic Z score when using other methods29,30 (appendix pp 11–12). Data are n (%), median (IQR), or mean (SD). 70 patients with previous aortic root surgery at enrolment were excluded: two in the ARB group and five in the placebo group from Ghent Marfan and 27 in the ARB group and 36 in the control Similarly, there were no significant differences between group from COMPARE. ARB=angiotensin receptor blocker. FBN1=fibrillin-1. the two groups in other measures of change in aortic dimensions, including absolute aortic dimension Table 2: Baseline characteristics by randomised allocation (appendix p 9). There was some evidence of heterogeneity in the Z score difference between ARB and β blockers in the control group, corresponding to a mean depending on family history of aortic dissection (favouring difference of –0·07 (95% CI –0·12 to –0·01; p=0·012; ARB in the 110 patients with such a family history, figure 1), which represents an approximate halving in the heterogeneity p=0·010), but otherwise no evidence of annual rate of change in the aortic root Z score in the ARB heterogeneity in any of the other prespecified subgroups group compared with the control group. (figure 3). Although there was no evidence of heterogeneity In the trials of ARB versus control, there was no between the overall results of the four contributing significant difference in the proportion of patients who 826 www.thelancet.com Vol 400 September 10, 2022 Articles ARB Comparator Difference, mean (95% CI or 99% CI) n Mean (SE) n Mean (SE) ARB vs control (χ23=6·1; p=0·11) Marfan–Sartan 145 −0·03 (0·04) 143 −0·03 (0·04) 0·00 (−0·14 to 0·15) COMPARE 77 0·11 (0·02) 67 0·17 (0·03) −0·07 (−0·15 to 0·02) AIMS 94 0·03 (0·05) 85 0·19 (0·04) −0·16 (−0·32 to 0·00) Ghent Marfan 10 0·32 (0·24) 5 0·02 (0·08) 0·30 (−0·35 to 0·94) Overall 326 0·07 (0·02) 300 0·13 (0·02) −0·07 (−0·12 to −0·01) p=0·012 ARB vs β blocker (χ22=5·5; p=0·064) PHN 299 −0·06 (0·04) 294 −0·12 (0·02) 0·06 (−0·05 to 0·17) LOAT 57 −0·14 (0·05) 55 −0·05 (0·05) −0·10 (−0·28 to 0·09) Canada 8 0·52 (0·24) 9 0·03 (0·21) 0·49 (−0·33 to 1·30) Overall 364 −0·08 (0·03) 358 −0·11 (0·02) 0·03 (−0·05 to 0·10) p=0·48 99% CI or 95% CI −0·5 −0·25 0 0·25 0·5 0·75 Favours ARB Favours comparator Figure 1: Annual rate of change of body surface area-adjusted aortic root dimension Z score at the sinuses of Valsalva Indirect effect of β blocker vs control is –0·09 (95% CI –0·18 to 0·00); p value=0·042 (β blocker minus control). ARB=angiotensin receptor blocker. had the composite outcome of aortic dissection, aortic a random effects model: the absolute differences for each root surgery, or death during study follow-up of these were similar to those derived from our (30 [8%] of 353 ARB vs 27 [8%] of 323 control, p=0·86). prespecified method of analysis (a so called fixed effect Nor was there any evidence of difference in this analysis), but with wider CIs (as would be expected; composite outcome in the trials of ARB versus β blockers appendix p 6). Finally, we prespecified that for baseline (21 [5%] of 384 ARB vs 14 [4%] of 382 β blockers, p=0·23; groups defined by age, systolic blood pressure, diastolic appendix p 3). blood pressure, and BSA-adjusted Z score we would do Combining the results of the four trials of ARB versus additional interaction tests in which these factors were control with the three trials of ARB versus β blockers considered as continuous rather than categorical allowed us to indirectly assess β blockers versus control. variables. For all eight interaction tests (four in each of With such an analysis, the difference in mean absolute the two groups of trials) there was no good evidence that annual change in the aortic root Z score between the effect on aortic root dimension Z score varied β blocker and control was –0·09 (95% CI –0·18 to 0·00; significantly depending on the baseline characteristic p=0·042; ie, similar to the direct estimate when (the smallest of these interaction p values was 0·05, comparing ARB with control). Since the trials of ARB which is not significant given the multiple tests that were versus control and ARB versus β blockers included done). patients within different age ranges, in a post-hoc exploratory analysis we examined the indirect comparison Discussion of β blockers versus control separately in patients Marfan syndrome affects about one in 5000 people, younger than 16 years and patients aged 16 years or older. which corresponds to a global total of 1·6 million people.38 We found no evidence that the effect of a β blocker Marfan syndrome causes a dramatically increased risk of differed in those younger than 16 years compared with aortic complications, mainly aortic dissection, commonly those aged 16 years and older (heterogeneity p=0·09). resulting in premature death or disability: a 2018 study We also prespecified a range of other secondary reported 11 fatal aortic complications among 412 patients analyses in our published protocol.25 These included with Marfan syndrome, compared to only six such assessments of aortic dimension at locations other than complications among 41 500 age-matched and sex- the sinuses of Valsalva, assessments using different matched controls over a comparable period (hazard ratio imaging methods, and analyses of haemodynamic 194·6 [95% CI 67·4–561·7]).7 Prophylactic surgery to variables (eg, blood pressure) and other physical replace the aortic root is recommended in cases in which measurements. The numbers of patients available for a large or rapidly expanding aneurysm presents an such analyses varied, depending on what was included in imminent risk of aortic dissection, but such surgery is each trial’s case report forms, and the results are itself associated with morbidity, occasional mortality, and summarised in the appendix (p 4). None of these results is not available in all health-care systems. Effective are qualitatively inconsistent with the main findings. medical therapy that is well-tolerated by both children Additionally, we prespecified exploratory analyses using and adults could delay or prevent the need for surgery. www.thelancet.com Vol 400 September 10, 2022 827 Articles ARB Control Difference, mean (95% CI or 99% CI) n Mean (SE) n Mean (SE) Age, years (χ21=0·6; p=0·43) <16 72 −0·02 (0·06) 65 0·13 (0·06) −0·13 (−0·34 to 0·08) ≥16 254 0·08 (0·02) 235 0·13 (0·02) −0·06 (−0·13 to 0·01) Gender (χ21=0·5; p=0·47) Female 174 0·08 (0·02) 162 0·14 (0·02) −0·05 (−0·13 to 0·04) Male 152 0·04 (0·02) 138 0·12 (0·02) −0·08 (−0·19 to 0·02) Family history of Marfan syndrome (χ21=1·2; p=0·28) Yes 92 0·07 (0·02) 80 0·08 (0·04) −0·10 (−0·24 to 0·04) No 156 −0·02 (0·03) 152 0·08 (0·03) −0·01 (−0·15 to 0·12) Unknown 78 0·11 (0·02) 68 0·17 (0·03) −0·07 (−0·15 to 0·02) Presence of FBN1 (χ21=7·9; p=0·0050) Yes 252 0·06 (0·02) 238 0·14 (0·02) −0·08 (−0·15 to −0·01) No 52 0·14 (0·04) 43 0·05 (0·03) 0·09 (−0·05 to 0·23) Unknown 22 0·04 (0·10) 19 0·29 (0·13) −0·25 (−0·67 to 0·16) Ectopia lentis (χ21=0·1; p=0·71) Yes 76 0·10 (0·03) 60 0·12 (0·02) −0·07 (−0·18 to 0·04) No 104 0·09 (0·02) 97 0·19 (0·03) −0·09 (−0·19 to 0·01) Unknown 146 −0·03 (0·04) 143 −0·03 (0·04) 0·00 (−0·14 to 0·15) β blocker (χ21=0·4; p=0·54) Yes 247 0·06 (0·02) 225 0·13 (0·02) −0·07 (−0·15 to 0·01) No 79 0·10 (0·03) 75 0·13 (0·03) −0·04 (−0·15 to 0·07) BSA-adjusted Z score (χ21=0·2; p=0·63) <4·5 220 0·07 (0·02) 207 0·09 (0·02) −0·03 (−0·11 to 0·04) ≥4·5 106 0·07 (0·03) 93 0·14 (0·04) −0·06 (−0·18 to 0·06) Systolic blood pressure, mm Hg (χ21=0·1; p=0·80) <130 235 0·04 (0·02) 224 0·09 (0·02) −0·05 (−0·12 to 0·03) ≥130 63 0·12 (0·02) 52 0·15 (0·03) −0·06 (−0·20 to 0·08) Diastolic blood pressure, mm Hg (χ21=0·3; p=0·57) <80 240 0·05 (0·02) 228 0·12 (0·02) −0·07 (−0·15 to 0·00) ≥80 58 0·10 (0·03) 47 0·13 (0·04) −0·03 (−0·18 to 0·12) Overall 326 0·07 (0·02) 300 0·13 (0·02) −0·07 (−0·12 to −0·01) p=0·01 99% CI or 95% CI −0·5 −0·25 0 0·25 0·5 Favours ARB Favours control Figure 2: Annual rate of change of BSA-adjusted aortic root dimension Z score at the sinuses of Valsalva, by subgroups (ARB vs control) ARB=angiotensin receptor blocker. BSA=body surface area. This meta-analysis shows that among patients with gene. Given that FBN1 variant status is a marker of the Marfan syndrome who had had no previous aortic certainty of a diagnosis of Marfan syndrome, this finding surgery, an ARB reduces the rate of increase of the aortic is what might be expected if an ARB is effective at root Z score by around one half and that this effect seems slowing root expansion in this condition. The second to be in addition to any effects of β blockers (which are observation was that all of the prespecified methods for discussed later). The robustness of our findings on the estimating change in root size yielded significant results, effects of an ARB is reinforced by several observations so that our findings were not dependent on the that, since they depended on the availability of individual performance of a particular method; this, again, is as participant data, went beyond the results of previous might be expected if an ARB is effective. meta-analyses of tabular data.26–28 The first of these Our analyses are most informative about the effects observations is that, although in general there was little of an ARB, because these analyses involved meta- evidence of heterogeneity of the effect of ARBs among analyses of trials making direct comparisons of an ARB the prespecified subgroups, ARBs had a significantly versus control. Our results are less definitive for greater effect on aortic root dimension Z score among β blockers than ARBs, because the β-blocker analyses patients with a known pathogenic variant in the FBN1 depended on indirect comparisons of two groups of 828 www.thelancet.com Vol 400 September 10, 2022 Articles ARB β blocker Difference, mean (95% CI or 99% CI) n Mean (SE) n Mean (SE) Age, years (χ21=3·2; p=0·075) < 16 250 −0·02 (0·04) 244 −0·14 (0·02) 0·06 (−0·06 to 0·18) ≥16 114 −0·09 (0·03) 114 −0·04 (0·03) −0·06 (−0·18 to 0·06) Gender (χ21=1·5; p=0·22) Female 155 −0·14 (0·05) 159 −0·12 (0·03) −0·04 (−0·21 to 0·13) Male 209 −0·05 (0·02) 199 −0·11 (0·02) 0·05 (−0·04 to 0·13) Family history of Marfan syndrome (χ21=0·5; p=0·48) Yes 182 −0·03 (0·06) 183 −0·14 (0·02) 0·09 (−0·07 to 0·26) No 109 −0·06 (0·04) 112 −0·10 (0·03) 0·04 (−0·09 to 0·17) Unknown 73 −0·15 (0·04) 63 −0·06 (0·04) −0·09 (−0·24 to 0·05) Family history of aortic dissection (χ21=6·6; p=0·010) Yes 54 −0·22 (0·05) 56 −0·09 (0·04) −0·11 (−0·28 to 0·06) No 252 0·00 (0·04) 247 −0·12 (0·02) 0·10 (−0·02 to 0·23) Unknown 58 −0·14 (0·05) 55 −0·05 (0·05) −0·10 (−0·28 to 0·09) Presence of FBN1 (χ21=1·0; p=0·31) Yes 128 −0·16 (0·05) 136 −0·16 (0·03) 0·00 (−0·18 to 0·17) No 20 −0·10 (0·07) 17 0·15 (0·09) −0·16 (−0·52 to 0·20) Unknown 216 −0·09 (0·02) 205 −0·10 (0·02) 0·01 (−0·07 to 0·09) Ectopia lentis (χ21=1·1; p=0·30) Yes 125 −0·12 (0·03) 131 −0·13 (0·03) 0·02 (−0·09 to 0·12) No 148 0·03 (0·07) 127 −0·09 (0·03) 0·11 (−0·09 to 0·31) Unknown 91 −0·14 (0·03) 100 −0·10 (0·04) −0·03 (−0·16 to 0·11) BSA-adjusted Z score (χ21=2·8; p=0·10) <4·5 237 −0·10 (0·02) 242 −0·10 (0·02) 0·01 (−0·06 to 0·07) ≥4·5 127 0·28 (0·06) 116 −0·14 (0·04) 0·16 (−0·07 to 0·39) Systolic blood pressure, mm Hg (χ21=0·0; p=0·83) <130 341 −0·07 (0·03) 337 −0·10 (0·02) 0·02 (−0·08 to 0·12) ≥130 19 −0·07 (0·05) 19 −0·17 (0·09) 0·00 (−0·27 to 0·27) Diastolic blood pressure, mm Hg (χ21=1·5; p=0·22) <80 333 −0·07 (0·03) 331 −0·12 (0·02) 0·04 (−0·06 to 0·13) ≥80 27 0·58 (0·02) 25 −0·03 (0·08) −0·10 (−0·37 to 0·17) Overall 364 −0·08 (0·03) 358 −0·11 (0·02) 0·03 (−0·05 to 0·10) p=0·48 99% CI or 95% CI −0·5 −0·25 0 0·25 0·5 Favours ARB Favours β blocker Figure 3: Annual rate of change of BSA-adjusted aortic root dimension Z score at the sinuses of Valsalva, by subgroups (ARB vs β blocker) ARB=angiotensin receptor blocker. BSA=body surface area. trials (four comparing an ARB vs control and and 28·5 years for ARB vs control), we assessed the three comparing an ARB vs a β blocker). Our indirect separate effects of a β blocker in those younger than estimate of the effect of β blockers is consistent with 16 years and those aged 16 years and older in stratified the results of one small trial that directly compared a analyses. Although we found no evidence that effects of β blocker with no treatment.10 However, our estimates β blockers varied significantly depending on age, there rely on the assumption that the effects of ARBs and was little power to detect any true heterogenity that β blockers are independent of each other (ie, that the might have been present. effects of each drug on change in aortic Z score are The clinical significance of our results is informed by additive). Our observation of no significant hetero­ sample size calculations done by the Pediatric Heart geneity of the effect of ARBs depending on concomitant Network investigators24 who assumed that, in an adult use of β blockers suggests this assumption is a Marfan syndrome population with a mean age of reasonable one, but does not prove the assumption to 20 years and Z score of 4·3, the threshold for aortic be correct. Since the trials of an ARB versus control and surgery would be reached in about 15 years (when the of an ARB versus a β blocker differed substantially in Z score is 7·3 and the aortic root diameter is 5·04 cm). average age at entry (13·9 years for ARB vs β blockers Reducing this annual rate of change from 0·20 to 0·12 www.thelancet.com Vol 400 September 10, 2022 829 Articles (ie, annual reduction of 0·08) would therefore increase Ghent Marfan trial. AE, AF, and GT-T were investigators on the the expected time to surgery by about 10 years (since, at LOAT trial. CBo, GJ, and OM were investigators on the Marfan Sartan trial. HD, RVL, and LAS were investigators on the US National Institutes an annual increase in Z score of 0·12 rather than 0·20, it of Health (NIH) National Heart, Lung, and Blood Institute (NHLBI) would take 25 rather than 15 years for the Z score to Pediatric Heart Network (PHN) trial. H-HC and M-HW were increase from 4·3 to 7·3). Such an annual reduction is investigators on the Taiwanese trial. AP, JE, CBa, SN, and HW conceived consistent with the absolute changes in Z score in our and designed the experiment. HH, KD, KW, LH, and ES did the systematic searches, data collection, and processing. KD, JE, and ES did analyses (–0·07 for an ARB and –0·09 for a β blocker). the statistical analyses. CR provided clinical input and contributed to A limitation of our analysis is that, despite making every project oversight. AP and CBa drafted the first version of the effort to obtain all available trial datasets, not all were manuscript. All authors critically revised the manuscript for important available for individual data analysis: one published trial16 intellectual content. All authors accept full responsibility for the content of the paper and have seen and approved the final version. did not contribute data but was very small (34 participants followed up for only 6 months) and its conclusions were Declaration of interests AP has received grants paid to his institutions from the Marfan consistent with the results of the meta-analysis, so it Foundation, the British Heart Foundation, the UK National Institute for would not have influenced our conclusions. Another trial, Health and Care Research Biomedical Research Centre, the Gibson of moderate size (n=262, follow-up 48 months), was not fund, Oxford University Hospitals Charitable Fund, and the Academy of available from the investigators.32 The main findings of Medical Sciences Clinical Lecturer Starter Grant scheme (which is administered by the Academy on behalf of the Academy, the Wellcome this second trial from which data were unavailable have, Trust, the UK Medical Research Council [MRC], the British Heart however, been reported in abstract form and showed that Foundation, Arthritis Research UK, Prostate Cancer UK, and the Royal the combination of an ARB (losartan) and a β blocker College of Physicians). CR has received institutional grants from the British Heart Foundation and UK MRC. CBa has received institutional (nebivolol) reduced progression as compared with either grants from the British Heart Foundation, UK MRC, NIHR, and treatment alone (p=0·009),32 which is consistent with the Boehringer Ingelheim (EMPA-KIDNEY trial). main findings of the meta-analysis. Of the trials included Data sharing in our meta-analysis, one used irbesartan23 and losartan The NIH NHLBI PHN Marfan Study dataset was used in preparation of was used in the others, hence the amount of data available this work. Data were downloaded from https://www. for irbesartan was scarce compared with that for losartan. pediatricheartnetwork.org/login/ on Oct 15, 2019. Other individual participant datasets were provided to the Marfan Treatment Trialists’ No trials randomly allocated participants to prespecified (MTT) secretariat in Oxford, UK, by the trialists. Data sharing requests for ARB dosing strategies or to different agents. Consequen­t­ these trials should be sent to the principal investigators of the trials, who ly, it was not possible to explore whether any particular are: MF (m.flather@uea.ac.uk) for the AIMS trial; GS (gsandor@telus. ARB selection or dosing strategy was superior to any other net) for the Canadian Trial; MG (m.groenink@amc.uva.nl) for the COMPARE trial; JDB (Julie.DeBacker@UGent.be) for the Ghent Marfan (and similar limitations apply to β blockers). The trial; GT-T (gteixido@vhebron.net) for the LOAT trial; GJ (guillaume. generalisability of our findings to older adults with Marfan jondeau@aphp.fr) for the Marfan-Sartan trial; RL (ron.lacro@cardio. syndrome is somewhat uncertain, as only 11% of the chboston.org) for the PHN trial; and M-HW (wumh@ntu.edu.tw) for the randomly assigned patients were aged 40 years or older Taiwanese trial. and only 6% were aged 50 years or older. Finally, even in Acknowledgments this meta-analysis of all eligible and available trials, the The work of the MTT is supported by the Marfan Foundation, the Oxford British Heart Foundation Centre for Research Excellence, number of patients who had major clinical outcomes was and the UK Medical Research Council (MC_UU_00017/4). too small to provide sufficient statistical power to detect AP acknowledges funding support from the Marfan Foundation, the benefit on such outcomes over the relatively short duration British Heart Foundation, including from the Oxford British Heart of the trials. Foundation Centre for Research Excellence (RE/13/1/30181), the UK National Institute for Health and Care Research Biomedical Research In summary, in these trials of patients with Centre, Oxford, the Gibson fund, Oxford University Hospitals Charitable Marfan syndrome, ARBs reduced the rate of enlarge­ Fund, and from an Academy of Medical Sciences Clinical Lecturer ment of the aortic root by about one half, including Starter Grant scheme, which is administered by the Academy on behalf among those already taking a β blocker. The effect was of the Academy, the Wellcome Trust, the Medical Research Council, the British Heart Foundation, Arthritis Research UK, Prostate Cancer UK, particularly large among patients with a pathogenic and the Royal College of Physicians. We would like to thank all the FBN1 variant, strengthening the main finding. The participants who originally took part in the included trials, without effects of β blockers were similar in magnitude to those whom this research would not have been possible. We would like to of ARBs. Moreover, for ARB versus control, there was thank the clinicians, scientists, and all those who worked on the original trials. JDB would like to thank Laura Muiño Mosquera for her no evidence that the effect size depended on use of contributions to the Ghent Marfan trial. We would also like to thank β blockers. Our findings therefore suggest that, if Samantha Briggs and Clare Mathews for their valuable administrative tolerated, the combination of a β blocker and ARB assistance in preparing this report. would reduce the rate of enlargement of the aortic root References by at least one half, and potentially by much more than 1 McKusick VA. The cardiovascular aspects of Marfan’s syndrome: a heritable disorder of connective tissue. Circulation 1955; 11: 321–42. this which, if maintained over a sustained period, would 2 Keane MG, Pyeritz RE. Medical management of Marfan syndrome. be expected to delay the need for surgery substantially. Circulation 2008; 117: 2802–13. Contributors 3 Dietz HC, Cutting GR, Pyeritz RE, et al. Marfan syndrome caused AHC, TC, MD, MF, and XYJ were investigators on the AIMS trial. by a recurrent de novo missense mutation in the fibrillin gene. Nature 1991; 352: 337–39. GS was an investigator on the Canadian trial. MG and BM were investigators on the COMPARE trial. JDB was an investigator on the 830 www.thelancet.com Vol 400 September 10, 2022 Articles 4 Baetens M, Van Laer L, De Leeneer K, et al. Applying massive 22 Chiu HH, Wu MH, Wang JK, et al. Losartan added to β-blockade parallel sequencing to molecular diagnosis of Marfan and therapy for aortic root dilation in Marfan syndrome: a randomized, Loeys-Dietz syndromes. Hum Mutat 2011; 32: 1053–62. open-label pilot study. Mayo Clin Proc 2013; 88: 271–76. 5 De Paepe A, Devereux RB, Dietz HC, Hennekam RC, Pyeritz RE. 23 Mullen M, Jin XY, Child A, et al. Irbesartan in Marfan syndrome Revised diagnostic criteria for the Marfan syndrome. (AIMS): a double-blind, placebo-controlled randomised trial. Lancet Am J Med Genet 1996; 62: 417–26. 2019; 394: 2263–70. 6 Loeys BL, Dietz HC, Braverman AC, et al. The revised Ghent 24 Lacro RV, Dietz HC, Sleeper LA, et al. Atenolol versus losartan in nosology for the Marfan syndrome. J Med Genet 2010; 47: 476–85. children and young adults with Marfan’s syndrome. N Engl J Med 7 Groth KA, Stochholm K, Hove H, Andersen NH, Gravholt CH. 2014; 371: 2061–71. Causes of mortality in the Marfan syndrome (from a Nationwide 25 Pitcher A, Emberson J, Lacro RV, et al. Design and rationale Register Study). Am J Cardiol 2018; 122: 1231–35. of a prospective, collaborative meta-analysis of all randomized 8 Halpern BL, Char F, Murdoch JL, Horton WB, McKusick VA. controlled trials of angiotensin receptor antagonists in A prospectus on the prevention of aortic rupture in the Marfan syndrome, based on individual patient data: Marfan syndrome with data on survivorship without treatment. a report from the Marfan Treatment Trialists’ Collaboration. Johns Hopkins Med J 1971; 129: 123–29. Am Heart J 2015; 169: 605–12. 9 Gersony DR, McClaughlin MA, Jin Z, Gersony WM. The effect of 26 Elbadawi A, Omer MA, Elgendy IY, et al. Losartan for preventing beta-blocker therapy on clinical outcome in patients with Marfan’s aortic root dilatation in patients with Marfan syndrome: syndrome: a meta-analysis. Int J Cardiol 2007; 114: 303–08. a meta-analysis of randomized trials. Cardiol Ther 2019; 8: 365–72. 10 Shores J, Berger KR, Murphy EA, Pyeritz RE. Progression of aortic 27 Al-Abcha A, Saleh Y, Mujer M, et al. Meta-analysis examining the dilatation and the benefit of long-term beta-adrenergic blockade usefulness of angiotensin receptor blockers for the prevention of in Marfan’s syndrome. N Engl J Med 1994; 330: 1335–41. aortic root dilation in patients with the Marfan syndrome. 11 Habashi JP, Judge DP, Holm TM, et al. Losartan, an AT1 Am J Cardiol 2020; 128: 101–06. antagonist, prevents aortic aneurysm in a mouse model of 28 Kang YN, Chi SC, Wu MH, Chiu HH. The effects of losartan versus Marfan syndrome. Science 2006; 312: 117–21. beta-blockers on cardiovascular protection in Marfan syndrome: 12 Brooke BS, Habashi JP, Judge DP, Patel N, Loeys B, Dietz HC 3rd. a systematic review and meta-analysis. J Formos Med Assoc 2020; Angiotensin II blockade and aortic-root dilation in Marfan’s 119: 182–90. syndrome. N Engl J Med 2008; 358: 2787–95. 29 Campens L, Demulier L, De Groote K, et al. Reference values for 13 Pees C, Laccone F, Hagl M, Debrauwer V, Moser E, echocardiographic assessment of the diameter of the aortic root and Michel-Behnke I. Usefulness of losartan on the size of the ascending aorta spanning all age categories. Am J Cardiol 2014; ascending aorta in an unselected cohort of children, adolescents, 114: 914–20. and young adults with Marfan syndrome. Am J Cardiol 2013; 30 Pettersen MD, Du W, Skeens ME, Humes RA. Regression 112: 1477–83. equations for calculation of Z scores of cardiac structures in a large 14 Mueller GC, Stierle L, Stark V, et al. Retrospective analysis of the cohort of healthy infants, children, and adolescents: an effect of angiotensin II receptor blocker versus β-blocker on aortic echocardiographic study. J Am Soc Echocardiogr 2008; 21: 922–34. root growth in paediatric patients with Marfan syndrome. Heart 31 Bhala N, Emberson J, Merhi A, et al. Vascular and upper 2014; 100: 214–18. gastrointestinal effects of non-steroidal anti-inflammatory drugs: 15 Muiño-Mosquera L, De Nobele S, Devos D, Campens L, meta-analyses of individual participant data from randomised trials. De Paepe A, De Backer J. Efficacy of losartan as add-on therapy to Lancet 2013; 382: 769–79. prevent aortic growth and ventricular dysfunction in patients with 32 Di Toro A, Klersy C, Giuliani L, et al. Losartan vs nebivolol vs the Marfan syndrome: a randomized, double-blind clinical trial. association of both on the progression of aortic root dilation in Acta Cardiol 2017; 72: 616–24. genotyped Marfan syndrome: 48 months open label randomized 16 Bhatt AB, Buck JS, Zuflacht JP, et al. Distinct effects of losartan controlled phase III trial. Eur Heart J 2019; 40 (suppl 1): 3742. and atenolol on vascular stiffness in Marfan syndrome. Vasc Med 33 Roman MJ, Devereux RB, Kramer-Fox R, O’Loughlin J. 2015; 20: 317–25. Two-dimensional echocardiographic aortic root dimensions in 17 Sandor GG, Alghamdi MH, Raffin LA, et al. A randomized, double normal children and adults. Am J Cardiol 1989; 64: 507–12. blind pilot study to assess the effects of losartan vs atenolol on the 34 Haycock GB, Schwarz GJ, Wisotsky DH. Geometric method for biophysical properties of the aorta in patients with Marfan and measuring body surface area: a height-weight formula validated in Loeys-Dietz syndromes. Int J Cardiol 2015; 179: 470–75. infants, children and adults. J Pediatr 1978; 93: 62–66. 18 Milleron O, Arnoult F, Ropers J, et al. Marfan Sartan: a randomized, 35 Du Bois D, Du Bois E F. A formula to estimate the approximate double-blind, placebo-controlled trial. Eur Heart J 2015; 36: 2160–66. surface area if height and weight be known. Arch Intern Med 1916; 19 Gambarin FI, Favalli V, Serio A, et al. Rationale and design of a trial 17: 863–71. evaluating the effects of losartan vs nebivolol vs the association of 36 Devereux RB, de Simone G, Arnett DK, et al. Normal limits in both on the progression of aortic root dilation in Marfan syndrome relation to age, body size and gender of two-dimensional with FBN1 gene mutations. J Cardiovasc Med (Hagerstown) 2009; echocardiographic aortic root dimensions in persons ≥15 years of 10: 354–62. age. Am J Cardiol 2012; 110: 1189–94. 20 Groenink M, den Hartog AW, Franken R, et al. Losartan reduces 37 Sluysmans T, Colan SD. Theoretical and empirical derivation of aortic dilatation rate in adults with Marfan syndrome: a randomized cardiovascular allometric relationships in children. J Appl Physiol controlled trial. Eur Heart J 2013; 34: 3491–500. 2005; 99: 445–57. 21 Forteza A, Evangelista A, Sánchez V, et al. Efficacy of losartan vs 38 The Marfan Foundation. Marfan syndrome. https://marfan.org/ atenolol for the prevention of aortic dilation in Marfan syndrome: conditions/marfan-syndrome (accessed May 11, 2022). a randomized clinical trial. Eur Heart J 2016; 37: 978–85. www.thelancet.com Vol 400 September 10, 2022 831