Influence of cusp morphology and sex on quantitative valve composition in severe aortic stenosis

Abstract

Aims

Aortic stenosis is characterized by fibrosis and calcification of the valve, with a higher proportion of fibrosis observed in women. Stenotic bicuspid aortic valves progress more rapidly than tricuspid valves, which may also influence the relative composition of the valve. We aimed to investigate the influence of cusp morphology on quantitative aortic valve composition quantified from contrast-enhanced computed tomography angiography in severe aortic stenosis.

Methods and results

Patients undergoing transcatheter aortic valve implantation with bicuspid and tricuspid valves were propensity matched 1:1 by age, sex, and comorbidities. Computed tomography angiograms were analysed using semi-automated software to quantify the fibrotic and calcific scores (volume/valve annular area) and the fibro-calcific ratio (fibrotic score/calcific score). The study population (n = 140) was elderly (76 ± 10 years, 62% male) and had a peak aortic jet velocity of 4.1 ± 0.7 m/s. Compared with those with tricuspid valves (n = 70), patients with bicuspid valves (n = 70) had higher fibrotic scores [204 (interquartile range 118–267) vs. 144 (99–208) mm³/cm², P = 0.006] with similar calcific scores (P = 0.614). Women had greater fibrotic scores than men in bicuspid [224 (181–307) vs. 169 (109–247) mm³/cm², P = 0.042] but not tricuspid valves (P = 0.232). Men had greater calcific scores than women in both bicuspid [203 (124–355) vs. 130 (70–182) mm³/cm², P = 0.008] and tricuspid [177 (136–249) vs. 100 (62–150) mm³/cm², P = 0.004] valves. Among both valve types, women had a greater fibro-calcific ratio compared with men [tricuspid 1.86 (0.94–2.56) vs. 0.86 (0.54–1.24), P = 0.001 and bicuspid 1.78 (1.21–2.90) vs. 0.74 (0.44–1.53), P = 0.001].

Conclusions

In severe aortic stenosis, bicuspid valves have proportionately more fibrosis than tricuspid valves, especially in women.

Graphical Abstract

Influence of sex and bicuspid valves on valve composition in patients with severe aortic stenosis.

Open in new tabDownload slide

Introduction

Aortic stenosis is caused by a complex cycle of pathological insults involving endothelial damage, lipid deposition, and inflammation, which eventually leads to fibrosis and calcification.¹ The latter is responsible for the characteristic features of leaflet thickening and restricted excursion. Among patients with tricuspid aortic valves, valve calcification has been extensively studied using computed tomography (CT) calcium scoring. It is an established prognostic marker in aortic stenosis^2–5 and provides valuable diagnostic utility in patients with discordant echocardiographic findings⁶ where sex-specific thresholds for severe aortic stenosis have been established.⁷ However, in certain patient populations, valve calcification can show discrepancies with aortic stenosis severity⁸ as it ignores the haemodynamic influence of fibrosis. Recent advances in CT angiography and image analysis software now allow the burden of both fibrosis and calcification to be simultaneously measured in the aortic valve. Interestingly, valve fibrosis appears to be of particular importance to the pathophysiology of aortic stenosis in female patients.^4,9–11

Bicuspid valves are common, affecting ∼1% of the population,^12,13 and patients are at increased risk of developing aortic stenosis.¹⁴ Whilst many patients are young, there is an increasing number of elderly patients with multiple comorbidities and bicuspid valves. These patients are often treated with transcatheter aortic valve implantation (TAVI). Whilst some studies have suggested that fibrosis may play a more prominent role in the pathogenesis of aortic stenosis in subjects with bicuspid valves, this has not been systematically investigated, especially in elderly patients. A better understanding of valve composition may improve diagnostic thresholds of grading aortic stenosis severity in bicuspid valves, optimize the timing of valve intervention, and enable the development and targeted deployment of medical therapies. In the present study, our aim was to compare valve composition quantified by CT angiography, in particular the contribution of fibrosis and calcification, in patients with stenotic bicuspid and tricuspid valves being assessed for TAVI.

Methods

Study population

Patients for this study were selected from a single-centre registry of consecutive patients who underwent TAVI between April 2015 and October 2020 at St Bartholomew’s Hospital, London, UK. All patients were reviewed by a multidisciplinary heart team and determined to be at high or prohibitive surgical risk; these patients were then assessed for TAVI feasibility by the same team. Patients with bicuspid valves were identified and propensity matched 1:1 to patients with tricuspid valves. Bicuspid valve morphology was defined by the Sievers classification using CT angiography.¹⁵ The bicuspid subtype was assessed independently by two readers (K.P. and G.E.). Covariates used for matching included age, sex, previous myocardial infarction or stroke, diabetes mellitus, chronic kidney disease, pulmonary disease, frailty score, and urgency of TAVI (based on whether the patient was admitted for acute decompensated aortic stenosis and had an urgent or elective TAVI). Chronic kidney disease was defined as an estimated glomerular filtration rate of <60 mL/min/1.73 m². Pulmonary disease was defined as any chronic lung disease that results in abnormal lung function tests. Multivessel coronary artery disease was defined as two or more epicardial coronary stenoses >70% stenosis or left main stem stenosis >50%. Frailty was defined as a Rockwood clinical frailty score of >5.¹⁶

The study was approved by a UK research ethics committee (North West: Greater Manchester South Research Ethics Committee, UK; reference number: 21/NW/0182) that waived the need for informed consent given its retrospective observational nature and secondary use of routinely collected data.

CT angiography protocol

Pre-interventional planning of CT angiography was performed on a Somatom FORCE scanner (Siemens Healthineers, Erlangen, Germany) using a peak tube voltage of 120 kV and collimation of 128 × 0.625 mm. A 100 mL bolus injection of iohexol (Omnipaque 300, GE Healthcare, Chicago, IL, USA) contrast was used with bolus triggering in the ascending aorta. Images were acquired craniocaudally, using a FLASH whole-body acquisition (lung apices down to the lesser trochanters). Patients did not receive rate-limiting medications for the purpose of their scans.

CT angiography image analysis

Semi-automated software (Autoplaque version 2.5, Cedars-Sinai Medical Center, Los Angeles, CA, USA) was used to quantify the tissue composition of the aortic valve using a mediastinal window from pre-TAVI CT angiograms¹¹ (Figure 1). The best diastolic phase was selected at 70% of the R-R interval. Multiplanar reconstructions were reoriented to the aortic valve plane and the annulus defined (see Supplementary data online, Figure S1). A valve complex region of interest was defined between the lower coronary ostium and the virtual basal ring formed by the hinge points of each aortic valve cusp. Then, using serial cross-sectional images orthogonal to the longitudinal axis of the ascending aorta, the valve was contoured around its perimeter using 15–20 adjustable points, excluding the aortic wall. Adaptive scan-specific Hounsfield unit (HU) thresholds for fibrotic and calcific tissue components were automatically identified for each patient using Gaussian mixture modelling by automatically fitting Gaussian functions to the HU histograms for blood pool and non-calcific and calcific aortic valve tissue. Thresholds for fibrotic (lower 99.7th percentile of blood pool HU) and calcific tissue (upper 99.7th percentile blood pool HU) were automatically generated (see Supplementary data online, Figure S2). The analysis was performed by two experienced observers (K.P.P. and A.L.) and overseen by two experienced mentors (K.G. and D.D.).

To adjust for differences in patient size and annulus dimensions, the tissue volume was indexed to the aortic valve annular area to provide the fibrotic and calcific scores (see Supplementary data online, Figure S1). This indexing step allows for an objective comparison between bicuspid and tricuspid valves, which differ in size and have been described before.^17,18 In a secondary analysis, we employed a second alternative method to correct for valve size instead of indexing the fibrotic and calcific valve volumes to the average sinus of Valsalva diameter. The average sinus of Valsalva diameter was obtained using three cusp-to-commissure diameters for all valves except Type 0 bicuspid valves, which had two measurements: cusp-to-cusp and commissure-to-commissure diameters (see Supplementary data online, Figure S2).

The sum of the fibrotic and calcific scores provided the fibro-calcific score. The fibrotic score divided by the calcific score provided the fibro-calcific ratio. Our previous study has demonstrated excellent reproducibility for valve composition assessment.¹¹

Echocardiography

Transthoracic echocardiography was performed by the British Society of Echocardiography–accredited physiologists prior to TAVI and within 3 months of CT angiography. As per the British Society of Echocardiography guidelines,¹⁹ measurements of the peak aortic jet velocity, mean gradient, and aortic valve area were performed for each patient. Where possible, left ventricular ejection fraction was calculated using the Simpson biplane method. All patients went through a multidisciplinary team meeting to confirm the severity of aortic stenosis and decide on the optimal management strategy.

Statistical analysis

All patients with bicuspid valves were propensity score matched 1:1 to patients with tricuspid valves using matching factors determined a priori (age, sex, and cardiovascular risk factors) and a greedy nearest neighbour algorithm without replacement, which were determined and entered into a logistic regression. Normality of continuous variables was evaluated using the Shapiro–Wilk test and presented using the mean ± standard deviation for normally distributed variables and median (interquartile range) for non-normally distributed variables. Frequencies are presented as number (percentage). Baseline characteristics, including demographics, comorbidities, CT, and echocardiography variables, were compared between bicuspid and tricuspid patients. Aortic valve scores were compared between all bicuspid and tricuspid patients, with further subanalyses performed according to sex. Interobserver variability of valve tissue composition measurements was independently assessed in a random sample of 10 patients by two observers (K.P. and A.L.) (see Supplementary Results). A two-sided P-value of <0.05 was deemed statistically significant. All analyses were performed using SPSS version 28.0 (SPSS, Chicago, IL, USA).

Results

Overall, 1874 patients were registered in the study database. Seventy-four patients with bicuspid aortic valves were identified, of whom four were excluded due to insufficient image quality. Twenty-three patients had a Type 0 valve (no raphe) and 41 patients a Type 1 valve (single raphe). Only one patient had a Type 2 valve (two raphes). Further bicuspid classifications are provided in Supplementary Results. These 70 patients with bicuspid valves were propensity matched to 70 patients with tricuspid valves. The study population was predominantly elderly men with severe aortic stenosis. As expected, bicuspid valves had larger aortic root dimensions compared with tricuspid valves. Baseline patient characteristics are presented in Table 1 and Supplementary data online, Table S1.

The analysis time per scan ranged from 3 to 6 min. Measurements of tissue volumes showed excellent interobserver repeatability with interclass correlation coefficients of 0.928 (0.718–0.982) for fibrotic tissue volume, 0.999 (0.997–1.000) for calcified tissue volume, and 0.985 (0.939–0.996) for fibro-calcific tissue volume, with no fixed or proportional biases and very good limits of agreements (see Supplementary data online, Figure S3).

Tissue composition by valve subtype

Compared with those with tricuspid valves, patients with bicuspid valves (n = 70) had higher fibrotic scores [144 (99–208) vs. 204 (118–267) mm³/cm², P = 0.006] and higher fibro-calcific scores [326 (249–416) vs. 389 (273–516) mm³/cm², P = 0.015] but similar calcific scores [152 (100–230) vs. 172 (91–267) mm³/cm², P = 0.614] (Figure 2). The fibro-calcific ratio was similar between tricuspid and bicuspid valves: 1.03 (0.56–1.70) vs. 1.32 (0.56–2.23), P = 0.191. Consistent findings were observed when the alternative method for indexing was used and when assessing non-indexed fibrotic, calcific, and fibro-calcific volumes (Figure 2).

There were no demonstrable differences in calcific, fibrotic, or fibro-calcific scores between patients with Type 0 and Type 1 valves, although there was an apparent trend for higher calcific scores in patients with a Type 0 valve [216 (171–284) vs. 139 (63–210) mm³/cm², P = 0.051; Table 2].

Valve tissue composition by sex

Women had greater fibrotic scores than men in bicuspid [224 (181–307) vs. 169 (109–247) mm³/cm², P = 0.042] but not tricuspid valves [184 (94–253) vs. 133 (99–187) mm³/cm², P = 0.232]. Men had greater calcific scores than women in both bicuspid [203 (124–355) vs. 130 (70–182) mm³/cm², P = 0.008] and tricuspid valves [177 (136–249) vs. 100 (62–150) mm³/cm², P = 0.004]. Among both valve types, women had a greater fibro-calcific ratio compared with men [tricuspid 1.86 (0.94–2.56) vs. 0.86 (0.54–1.24), P = 0.001 and bicuspid 1.78 (1.21–2.90) vs. 0.74 (0.44–1.53), P = 0.001] (Table 3). Similar results were obtained when fibrotic volumes were indexed to the sinus of Valsalva diameters (see Supplementary data online, Table S2).

Discussion

To our knowledge, this is the first contrast-enhanced CT angiography study to evaluate quantitative fibro-calcific valve composition in patients with severe aortic stenosis according to valve morphology. Our primary findings are that bicuspid valves have more fibrotic and total tissue but a similar extent of calcification compared with tricuspid valves. We also extend prior findings of sex-specific differences in valve tissue composition by demonstrating a greater fibro-calcific ratio in women compared with men, regardless of valve morphology. This has implications for the evaluation of aortic valve disease severity.

The greater degree of valvular fibrosis we observed on CT in bicuspid vs. tricuspid valves is consistent with a prior histological study.²⁰ There are several possible explanations for these findings. First, bicuspid valves are associated with increased valvular inflammation and neovascularization compared with tricuspid valves.^21,22 The fibrotic response to inflammation could accelerate the development of aortic stenosis. Second, bicuspid valves induce turbulent blood flow and higher tissue stress, concentrated in the abnormally large cusps and at the raphe. The aortic valve tissue may respond by increasing in thickness to buffer or to compensate for mechanical stress.²³ Once stenosis is present, the clinical course for bicuspid valves appears to be similar to that for calcific tricuspid valves, in which calcific deposits predominate at the cusp bases.²⁴

The influence of sex on aortic valve calcification is well established in tricuspid valves.^4,9 Here, we extend these findings to bicuspid valves, demonstrating that women have more fibrosis than men, whilst men have more calcification than women. The sex differences in the fibro-calcific ratio were similar regardless of valve morphology. Sex-related differences in the pathophysiology of aortic stenosis therefore appear to apply to bicuspid valves just as they do to tricuspid valves and suggest that different pharmacological treatment strategies may be required in men and women.

Valve fibrosis has previously been measured using histology of explanted bicuspid valves rather than in vivo imaging. These pathological studies have reported discrepant findings, with one demonstrating no differences¹⁰ and another showing more fibrosis among bicuspid compared with tricuspid valves.²⁰ Differences in the prevalence of comorbidities (such as hypertension and coronary artery disease), the severity of aortic stenosis, and age between the two studies are likely to have accounted for these. Our study has systematically accounted for these confounders.

Two previous studies have compared calcification between bicuspid and tricuspid valves with discrepant results.^8,10 One study showed higher calcification with bicuspid compared with tricuspid valves, although aortic stenosis severity was also higher in the bicuspid cohort.¹⁰ Another study showed more calcification in tricuspid compared with bicuspid valves. However, patients with bicuspid valves were over 20 years younger and had fewer comorbidities including diabetes mellitus, hypertension, coronary artery disease, and dyslipidaemia.⁸ We have accounted for the confounding imposed by aortic stenosis severity, patient demographics, and the presence of comorbidities and confirmed that fibrosis is indeed greater in bicuspid than tricuspid valves in both men and women.

There are several clinical implications of our findings. First, for the detection of earlier disease, valvular fibrosis often precedes calcification in the natural history of aortic stenosis.¹ The detection and quantification of fibrosis using CT may allow for the identification of earlier disease, which may influence the frequency of patient follow-up. Second, the grading of aortic stenosis severity is currently based on haemodynamic data from echocardiography and the Agatston score. Our study suggests that bicuspid valves have more fibrosis compared with tricuspid valves, especially among women. Previously, we have demonstrated improved diagnostic accuracy by incorporating fibrotic quantification to calcification among a population with predominantly tricuspid aortic stenosis.^11,17 Future studies need to ascertain whether the same is true for bicuspid valves. Third, although there is no established medical therapy for the treatment of aortic stenosis, trials are underway with several drugs targeting calcium metabolism: EAVaLL (NCT02109614) and CALCIFIA (NCT01000233). Whilst these have the potential to slow the progression of aortic stenosis, certain cohorts such as women and patients with bicuspid valves who have more fibrotic tissue may benefit less than men and tricuspid valves, respectively. Targeting mechanisms earlier in the natural history of aortic stenosis, such as inflammation or fibrosis, may lead to a greater benefit in such patients.

Bicuspid valves have larger aortic roots and annuli compared with tricuspid valves.²⁵ Similarly, men have large valves than women. For both these reasons, it is important to index fibrotic and calcific volumes for the valve annular area—a consistent measurement that is now made to guide valve sizing in all patients undergoing TAVI. The resulting fibrotic and calcific scores subsequently allow comparisons across different patient populations, although many of the observations we made in this study also held true for the unadjusted fibrotic and calcific volumes and when indexing was performed based on the sinus of Valsalva diameters.

Limitations

We acknowledge the limitations in our single-centre study. We could only include elderly patients undergoing TAVI who were at intermediate-to-high surgical risk. Patients undergoing surgical aortic valve replacement were not included as they do not undergo routine planning of CT angiography. Thus, our findings may not be generalizable to younger and lower-surgical-risk populations with aortic stenosis. Finally, we do not have histological validation for our CT angiography–derived valve tissue volumes, although this has been successfully demonstrated in previous studies investigating the fibro-calcific score.¹⁷

Conclusions

In patients with severe aortic stenosis, bicuspid valves had higher quantitative measures of fibrosis with similar measures for calcification when compared with tricuspid valves. There are important sex-specific differences in valve composition, with men having more calcific tissue and women having more fibrotic tissue, regardless of valve morphology. Differences therefore exist in the pathophysiology of bicuspid valve disease that may have important implications for detecting earlier disease, grading the severity of aortic stenosis, and developing novel pharmacological interventions.

Supplementary data

Supplementary data are available at European Heart Journal - Cardiovascular Imaging online.

Funding

None declared.

Data availability

Data for this study are not available for external review due to ethical restrictions.

References

Author notes