Article Text

Download PDFPDF

Original research article
Sex differences in left ventricular remodelling, myocardial fibrosis and mortality after aortic valve replacement
  1. Anvesha Singh1,
  2. Tarique Al Musa2,
  3. Thomas A Treibel3,
  4. Vassiliou S Vassiliou4,5,
  5. Gabriella Captur3,
  6. Calvin Chin6,
  7. Laura E Dobson2,
  8. Silvia Pica3,
  9. Margaret Loudon7,
  10. Tamir Malley4,
  11. Marzia Rigolli7,
  12. James Robert John Foley2,
  13. Petra Bijsterveld2,
  14. Graham R Law8,
  15. Marc Richard Dweck9,
  16. Saul G Myerson7,
  17. Sanjay K Prasad10,
  18. James C Moon3,
  19. John P Greenwood2,
  20. Gerry P McCann1
  1. 1 Cardiovascular Sciences, University of Leicester and the NIHR Leicester Biomedical Research Centre, Glenfield Hospital, Leicester, UK
  2. 2 Cardiovascular Sciences, Multidisciplinary Cardiovascular Research Centre and The Division of Biomedical Imaging, Leeds Institute for Cardiovascular and Metabolic Medicine, University of Leeds, Leeds, UK
  3. 3 Cardiovascular Sciences, Barts Health NHS Trust and University College London, London, UK
  4. 4 Cardiovascular Sciences, Imperial College London, Royal Brompton Hospital, London, UK
  5. 5 University of East Anglia and Norfolk and Norwich University Hospitals, Norwich, Norfolk, United Kingdom
  6. 6 Cardiovascular Medicine, National Heart Center Singapore, Singapore, Singapore
  7. 7 Cardiovascular Sciences, University of Oxford Centre for Clinical Magnetic Resonance Research, Oxford, UK
  8. 8 Medical Statistics, School of Health and Social Care, University of Lincoln and Multidisciplinary Cardiovascular Research Centre and The Division of Biomedical Imaging, Leeds Institute for Cardiovascular and Metabolic Medicine, University of Leeds, Lincoln and Leeds, UK
  9. 9 Cardiovascular Sciences, Centre for Cardiovascular Science, University of Edinburgh, Edinburgh, UK
  10. 10 Cardiology, Royal Brompton Hospital, London, UK
  1. Correspondence to Dr Marc Richard Dweck, Cardiovascular Sciences, Centre for Cardiovascular Science, University of Edinburgh, Edinburgh EH8 9YL, UK; marcdweck{at}


Objectives To investigate sex differences in left ventricular remodelling and outcome in patients undergoing surgical or transcatheter aortic valve replacement (SAVR/TAVR).

Methods In this multicentre, observational, outcome study with imaging core-lab analysis, patients with severe aortic stenosis (AS) listed for intervention at one of six UK centres were prospectively recruited and underwent cardiovascular magnetic resonance imaging. The primary endpoint was all-cause mortality and secondary endpoint was cardiovascular mortality.

Results 674 patients (425 men, 249 women, age 75±14 years) were included: 399 SAVR, 275 TAVR. Women were older, had higher surgical risk scores and underwent TAVR more frequently (53% vs 33.6%, p<0.001). More men had bicuspid aortic valves (BAVs) (26.7% vs 14.9%, p<0.001) and demonstrated more advanced remodelling than women. During a median follow-up of 3.6 years, 145 (21.5%) patients died, with no significant sex difference in all-cause mortality (23.3% vs 20.5%, p=0.114), but higher cardiovascular mortality in women (13.7% vs 8.5%, p=0.012). There were no significant sex-related differences in outcome in the SAVR or TAVR subgroups, or after excluding those with BAV. Factors independently associated with all-cause mortality were age, left ventricular ejection fraction (LVEF), BAV (better) and myocardial fibrosis detected with late gadolinium enhancement (LGE) in men, and age, LVEF and LGE in women. Age and LGE were independently associated with cardiovascular mortality in both sexes.

Conclusions Men demonstrate more advanced remodelling in response to a similar severity of AS. The higher cardiovascular mortality observed in women following AVR is accounted for by women having less BAV and higher risk scores resulting in more TAVR. LGE is associated with a worse prognosis in both sexes.

  • aortic stenosis
  • AVR
  • CMR

This is an open access article distributed in accordance with the Creative Commons Attribution 4.0 Unported (CC BY 4.0) license, which permits others to copy, redistribute, remix, transform and build upon this work for any purpose, provided the original work is properly cited, a link to the licence is given, and indication of whether changes were made. See:

View Full Text

Statistics from


Male and female patients remodel differently in response to pressure overload/ischaemia induced by aortic stenosis (AS). Remodelling is defined as a change in shape, structure or function of the heart. While echocardiographic studies suggest more concentric remodelling in women,1 2 recent cardiovascular magnetic resonance (CMR) imaging studies have confirmed higher left ventricular (LV) volumes, mass index, mass/volume, lower EF and more late gadolinium enhancement (LGE), a marker of focal fibrosis, in men.3–5 LGE is associated with adverse prognosis following aortic valve replacement (AVR).6–8

Despite more advanced remodelling in men, women appear to have worse outcomes following surgical AVR (SAVR)9–11 in some studies but not others.12 13 Recent studies using transcatheter AVR (TAVR) have demonstrated a survival benefit in women.14 15 Female sex has been found to be an independent predictor of symptom onset in AS16 and a predictor of cardiovascular mortality in our recently published multicentre study.17 The aim of this study was to investigate the apparent discrepancy between remodelling and outcome between sexes in patients undergoing SAVR/TAVR.


Study design

In this multicentre, longitudinal, observational outcome study conducted in the UK,17 patients with severe AS listed for intervention at one of six cardiothoracic surgical units were prospectively recruited. Patients>18 years of age with severe AS (one of the following: aortic valve area (AVA)<1 cm2, peak pressure gradient >64 mmHg, mean pressure gradient (MPG) >40 mmHg) who had undergone CMR for research purposes were included. The primary endpoint was all-cause mortality. The secondary endpoint was cardiovascular disease-related mortality, as defined by diagnosis on the UK death certificate. Data were collected on baseline characteristics (demographics, medical and drug history), surgical risk scores (STS V.2.73 and EuroSCORE II) and aortic valve gradients and area from transthoracic echocardiography, at the time of CMR.

CMR acquisition

CMR was performed on 1.5/3.0 T scanners using standardised protocols including cine imaging for ventricular volumes and function, phase-contrast velocity-encoded imaging for valve haemodynamics and LGE imaging for myocardial scar assessment.7 18 All participating centres have previously published single-centre studies in AS, with image quality and CMR pulse sequence parameters.16 18–21

Image analysis and data management

The details of data management and image analysis have been published.17 Anonymised data were collected and managed using Research Electronic Data Capture (REDCap) software.22 All deaths were identified through the UK National Health Service National Spine Database. Cardiovascular mortality was established in all deceased from the death certificates and adjudicated by two blinded readers (PB and JPG). All CMR scans were re-analysed in core-lab fashion,17 with each centre reporting a single component for all patients, after training and reproducibility assessment, and using standardised operating procedures, on CVI42 software (Circle Calgary, Canada). The full-width-half-maximum technique was used to quantify LGE.

Statistical analysis

Normality was assessed using the Shapiro-Wilk test, histograms and Q-Q plots using SPSS V.24.0 software (Statistical Package for the Social Sciences, Chicago, Illinois, USA). For continuous data, mean±SD for normally distributed data and median(IQR) for non-parametric data are presented. Categorical variables are expressed as counts and percentage. Data between the sexes were compared using independent t-test or Mann-Whitney U test. The Χ2 test was used for categorical variables. P values <0.05 were considered statistically significant. Univariate associates of outcomes were determined using Cox proportional hazards models, with the inclusion of sex-interaction variable into the model. Variables for the multivariable models were selected based on statistical significance (p<0.10) and clinical relevance, while avoiding co-linear variables (LV volumes were not included in addition to left ventricular ejection fraction (LVEF), as they are used in its calculation). As Society of Thoracic Surgeons (STS) score incorporates most clinical and demographic variables, this was not included in the initial model, but the added effect of LGE was tested in separate models. Survival was evaluated using the Kaplan-Meier method and compared between sexes using the log-rank test. The index date was the date of CMR. HRs were expressed as mean±95% CI.


Baseline characteristic

From 703 patients who underwent CMR, 29 were managed medically, and were excluded from further analysis (15 men and 14 women). This left 674 patients, 425 men and 249 women: mean age 75±14 years, AVA index 0.38±0.14 cm2/m2, MPG of 46±18 mmHg. Female patients were older, with higher STS and EuroSCORES (table 1). The prevalence of coronary artery disease and bicuspid aortic valve (BAV) was higher in men. AS severity was similar (no statistically significant difference in trans-thoracic echocardiogram (TTE)-measured pressure gradients and corrected AVA). Men demonstrated higher body surface area-corrected ventricular volumes, LV mass and mass/volume; marginally lower EF and a greater prevalence and amount of LGE (figure 1). A greater proportion of women underwent TAVR (53% vs 33.6%, p<0.001). There was no statistically significant difference in age or AS severity of the men and women in either SAVR-only or TAVR-only subgroup.

Table 1

Baseline characteristics

Figure 1

An example of a male (top row) and female (bottom row) patient with similar degree of aortic stenosis, showing cine stills (a-c, e, f), LGE imaging (d, h) and Continuous Wave Doppler through the aortic valve on echocardiography (g). AVAI, aortic valve area index (to BSA); LGE, late-gadolinium enhancement; LVEDVI, left ventricle end-diastolic volume index; LVMI, left ventricle mass index; MPG, mean pressure gradient; PPG, peak pressure gradient.


During a median follow-up of 3.6 years (IQR 2.6–5.9 years, range 9.9 years), 145 (21.5%) patients died: 87 (20.5%) men and 58 (23.3%) women (table 2). There was no significant difference in all-cause mortality, but a higher incidence of cardiovascular mortality in women (13.7% vs 8.5%, p=0.012 on log-rank test) (figure 2). There were no significant sex differences in all-cause or cardiovascular mortality in the TAVR-only or SAVR-only subgroup (figure 3, table 2).

Table 2

Outcomes for all patients, SAVR group and TAVR group

Figure 3

Kaplan-Meier survival curves for SAVR subgroup (top panel), TAVR subgroup (middle panel) and trileaflet aortic valve patients only (bottom panel), showing all-cause mortality (left panel) and cardiovascular mortality (right panel). All p>0.05 on Log-rank test. SAVR, surgical aortic valve replacement; TAVR, transcatheter aortic valve replacement.

Figure 2

Kaplan-Meier survival curves for male and female patients: (a) all-cause mortality and (b) cardiovascular mortality.

Factors associated with outcome

As we have previously shown sex to be independently associated with cardiovascular mortality in the overall cohort,17 univariate associations were assessed with the inclusion of the variable’s interaction with sex in the regression model. Common associates of all-cause mortality (table 3) included age, atrial fibrillation, coronary disease, surgical risk scores, left atrial volume, LV end-systolic and stroke volumes, left and right ventricular EF (L/RVEF) and the presence/amount of LGE. In addition, BAV morphology and non-infarct pattern LGE were significant in men, while previous coronary intervention and infarct-pattern LGE were significant in women. AS severity was not associated in either sex.

Table 3

Univariate associations with all-cause mortality

The following variables were entered into a multivariate regression model in a single step: age, AF, BAV, CAD, LVEF, RVEF, LGE presence, LAVI and LVMI, as well as the sex-interaction variable with BAV and LVMI. The independently associated variables were age, LVEF and LGE in both sexes, with the addition of BAV in men (table 4). The same variables remained independent on entering all variables of interest and their sex-interaction variables into a backward selection model.

Table 4

Univariate associations with cardiovascular mortality

Univariate associations with cardiovascular mortality (table 5)were similar, with the addition of diabetes for both sexes, lack of association of coronary disease and non-infarct pattern LGE in men, and fewer remodelling parameters in women. Independent associations with cardiovascular mortality included age and LGE in both sexes, when all variables shown in table 4 were entered into the model in a single step, with the addition of AF in men, and AF and diabetes in women on using a backward selection model.

Table 5

Table 5Multivariable associations with all-cause mortality and cardiovascular mortality

We also performed multivariable analysis with stepwise selection of LGE in addition to STS score, and both remained independently associated with all-cause and cardiovascular mortality in both sexes, with LGE providing incremental prognostic information.

Exclusion of BAV patients

We excluded BAV to remove the bias of the younger BAV subgroup in men (mean age 63.3±11.4 vs 75.6±8.8 years, p<0.01 and lower incidence of diabetes, AF and coronary disease). This left 313 male and 212 female patients, with similar sex differences in remodelling parameters (higher mass, volumes and mass/volume in men), but differences in LVEF, RVEF and LV stroke volume index no longer being significant (online supplementary table 1). There was no significant sex difference in all-cause or cardiovascular mortality (table 2, figure 2, online supplementary table 2). Univariate associations, corrected for sex-interactions, are presented in online supplementary tables 3,4. On multivariable analysis entering all variables in a single step, independent associations of all-cause mortality were age and LGE in both sexes, with the addition of AF in women when using a backward selection model. Independent associations of cardiovascular mortality were age, AF and LGE in men, with the addition of diabetes in women. The same variables remained significant on backward stepwise selection (online supplementary table 5).


This large multicentre CMR study confirms sex differences in LV remodelling. Although there was no significant difference in all-cause mortality, cardiovascular mortality was higher in women. However, these observed differences are accounted for by men being younger with more BAV, while women having more TAVR likely due to higher risk profile, possibly reflecting differences in the referral practices for male and female patients. LGE was independently associated with all-cause and cardiovascular mortality in both sexes.

Sex differences in LV remodelling

We have confirmed findings from previous single-centre studies utilising CMR3–5 that for a similar severity of AS, men demonstrate more advanced LV remodelling, with larger indexed volumes, mass, mass/volume and lower EF, in addition to more focal fibrosis (LGE). While women were thought to demonstrate more concentric remodelling based on older TTE measurements of higher relative wall thickness,23 24 this was traditionally based on a single basal slice, often using M-mode, which has many assumptions about the shape and symmetry of the LV. CMR overcomes many of these limitations and is now regarded as the gold standard for quantitative LV assessment, and the finding of greater concentric remodelling (ie, higher mass/volume) in men has now been confirmed in other CMR studies.4 Despite this seemingly more maladaptive response in men, female sex has been associated with both earlier symptom onset16 and worse mortality in AS.9–11 Putative mechanisms for these differences include higher wall stress in women due to less adaptive concentric remodelling for a similar degree of pressure overload, which may contribute to earlier symptoms. In fact, in this study, a greater proportion of women had New York Heart Association (NYHA) III–IV symptoms than men.

Referral patterns for intervention

There were more men in the SAVR group (71%), a difference that persisted even after removing those with BAVs (69%), with the proportions being almost equal for TAVR. This discrepancy in referral for surgical intervention has been noted in both historical single-centre studies (68% in a retrospective analysis of consecutive procedures10) and a multicentre analysis of an American national database (63%).25 This is also confirmed by the UK national cardiac surgery database reporting 60% of patients undergoing AVR being men, rising to around 70% for AVR and coronary artery bypass graft.26 On the contrary, the proportions are almost equal for TAVR in both single-centre studies14 and larger registries.27 Both a disparity in referral for testing and referral for surgery have been suggested,25 28 possibly due to greater incidence of comorbidities or risk scores at presentation in female participants.10 11 27 Other data suggest a greater benefit of TAVR in high-risk female patients,15 27 possibly leading to more women being referred for TAVR. There is also the possibility of perceived and/or real higher incidence of patient-prosthesis mismatch in women leading to a bias towards TAVR referral, and higher cardiovascular mortality. The female patients in our study were indeed older, more symptomatic (greater proportion of NYHA III–IV symptoms) and had higher surgical risk scores at the time of intervention.

Factors associated with outcome

Age and LGE were the common factors associated independently with all-cause and cardiovascular mortality in both sexes, with LVEF also significant for all-cause mortality. LGE also provided incremental prognostic information to STS score, which incorporates many clinical variables. The fact LVEF was not associated with cardiovascular outcomes may relate to the fact that few patients had impaired LV function. As expected, there was a greater proportion of BAVs in the male subgroup. BAV is thought to represent a different pathology to degenerative trileaflet AS, with better survival post AVR in a large age and sex matched cohort.29 Despite its exclusion, age and LGE remained significant for all-cause mortality, where as LGE was significant for cardiovascular mortality in only men, with the addition of diabetes in women and AF in both sexes, as the younger and fitter BAV patients were excluded.

Is earlier intervention warranted?

Given that women in our multicentre study were older, more symptomatic, with higher risk scores, and more frequently referred for TAVR, differences in care pathways for male and female patients may exist. It is possible that due to more advanced remodelling, men tend to get referred earlier for a similar degree of AS. In addition, subjective differences in interpretation and acknowledgement of symptoms may add to this potential referral bias. The association of LGE with poor outcome even after intervention in both sexes, supports the need for earlier intervention in AS, before fibrosis develops, and the need for trials to establish the best stratification tools. The EVOLVED (Early Valve Replacement guided by Biomarkers of Left Ventricular Decompensation in Asymptomatic Patients with Severe AS) trial is underway, which specifically addresses this question (NCT03094143).

There are limitations to our study.17 This was an observational study of patients at surgical centres with an interest in CMR, potentially introducing selection bias. Certain patient groups with a contraindication to CMR were excluded (advanced renal failure and non-compatible devices). LGE was not performed in a small minority of patients (n=61), and T1 mapping and biomarker analysis are also lacking. We did not objectively measure frailty or exercise capacity. The prevalence of occult amyloidosis, which has been shown to be present in 4%–8% of severe AS patients,30 is also unknown, as biopsy or radiolabelled scans were not performed as part of this study.


Men demonstrate more advanced remodelling in response to a similar degree of AS. The higher cardiovascular mortality observed in women following AVR is accounted for by women having less BAV and higher risk scores resulting in more TAVR. LGE is associated with a worse prognosis in both sexes.

Key questions

What is already known on this subject?

  • Sex differences in remodelling and post-aortic valve replacement (AVR) outcome have been reported in aortic stenosis (AS), with men demonstrating more advanced remodelling and worse outcomes in women in some studies.

What might this study add?

  • In this large multicentre, prospective longitudinal outcome study of patients undergoing cardiovascular magnetic resonance imaging before surgical/transcatheter AVR (SAVR/TAVR), during a median follow-up of 3.6 years (IQR 2.6–5.9 years), there was no significant difference in all-cause mortality (23.3% vs 20.5%, p=0.114), but higher cardiovascular mortality in women (13.7% vs 8.5%, p=0.012). However, this finding did not persist after accounting for type of intervention and prevalence of bicuspid aortic valve. A greater proportion of women underwent TAVR (53% vs 33.6%, p<0.001), with higher risk scores. Late gadolinium enhancement presence was associated with adverse prognosis in both sexes, even after AVR.

How might this impact on clinical practice?

  • Clinicians need to be aware of a possible bias in not referring female patients with severe AS for intervention and that when other risk factors are accounted for, cardiovascular and all-cause mortality are similar in men and women. Further studies are needed to define the best stratification tools in AS, with possible sex-specific cut-offs to define severity and timing of intervention.


View Abstract


  • Correction notice Since this article was first published online, the middle initial P has been added to the author John Greenwood.

  • Contributors We confirm that all authors have made significant contribution to this work and have approved the manuscript.

  • Funding This study was in part supported by the British Heart Foundation (University of Leeds, Greenwood JP [PG/11/126/29321]; University of Leicester, McCann GP [PG/07/068/2334]; University of Oxford, Myerson SG [FS/10/015/28104]; University of Edinburgh, Dweck MR [FS/10/026]; University College London, Moon JC [FS/08/028/24767]), the National Institute for Health Research (University of Leicester, McCann PDF 2011-04-51; University College London, Treibel TA [DRF-2013-06-102]), including via its Biomedical Research Centre and Clinical Research Facility programmes, as well as Rosetrees Trust. The views expressed are those of the author(s) and not necessarily those of the UK National Health Service, National Institute for Health Research or Department of Health.

  • Competing interests None declared.

  • Patient consent for publication All patients gave written informed consent.

  • Ethics approval The study was approved by the UK National Research Ethics Service (13/NW/0832), conformed to the principles of the Helsinki Declaration.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.