Distribution of diffuse myocardial fibrosis is uneven and associated with left ventricular function in severe aortic stenosis

Abstract

Background

Aortic stenosis (AS) leads to left ventricular (LV) remodelling and fibrosis. Myocardial fibrosis can be focal and irreversible, associated with poor prognosis, or diffuse and potentially reversible after surgery. Cardiac magnetic resonance imaging (CMR) shows promise in quantifying diffuse myocardial fibrosis (DMF), but methods vary in precision. The aim of this study was to investigate the presence and distribution of DMF from myocardial biopsies and CMR tissue characteristics in severe AS. Secondarily, explore the association between DMF and LV function.

Methods

Forty-three patients with severe AS underwent CMR and transthoracic echocardiography within 1 week before surgical aortic valve replacement. CMR included balanced steady-state free-precession cine images, T1 relaxometry with Modified Look-Locker Inversion recovery and extracellular volume calculations. Endomyocardial biopsies were sampled.

Results

Histological analysis of 192 biopsies showed a median LV collagen volume fraction (CVF) of 19%, peaking in segment 1, according to the segment model recommended by the American Heart Association (AHA). There were significant moderate correlations between CVF and extracellular volume in AHA segments 1 (r² = 0.54, p<0.01) and 16 (r² = 0.44, p=0.03), where the DMF had the highest prevalence. Functional assessments demonstrated correlations between CVF and global circumferential and radial strain (r² 0.31, p=0.04 respectively r²−0.33, p=0.03), as well as LV ejection fraction (r²−0.34, p=0.03). T1 relaxation time correlated with GLS (r² 0.42, p<0.01), mitral annular plane systolic excursion (r²−0.39, p=0.01) and mean S’ LV (r²−0.45, p<0.01) from transthoracic echocardiography.

Conclusion

CVF varied across LV segments and was significantly highest at the base of the heart, suggesting the start location for fibrosis. Systolic functional data correlated with CVF and T1 relaxation time.

WHAT IS ALREADY KNOWN ON THIS TOPIC

• Diffuse myocardial fibrosis (DMF) is common in severe aortic stenosis and may be partially reversible; cardiac magnetic resonance imaging (CMR) offers non-invasive assessment but its accuracy compared with histology remains uncertain.

WHAT THIS STUDY ADDS

• Multisite biopsy sampling revealed highest histological fibrosis in the basal segment, moderately correlating with segmental CMR. DMF was associated with subtle systolic dysfunction.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

• Supports further research into regional CMR analysis as a tool to better reflect DMF and enhance understanding of myocardial remodelling in aortic stenosis.

Background

Aortic stenosis is the most common valve disease in the Western world. Its prevalence increases with age, reaching 12% in individuals aged 75 and older. Among these, 3%–4% have severe symptomatic stenosis and require aortic valve replacement (AVR).¹,³

The narrowing of the aortic valve results in pressure overload of the left ventricle (LV), which compensates by hypertrophy and changes in the connective tissue with increased collagen deposits resulting in diffuse fibrosis. Diffuse myocardial fibrosis (DMF) has been shown to be reversible, at least to some extent, after surgery.^{4 5} Focal fibrosis, on the contrary, is irreversible and increases postoperative long-term mortality and morbidity.⁶

Endomyocardial biopsy is considered the standard for quantifying DMF. However, some recently published studies using cardiac magnetic resonance imaging (CMR) have found relaxometry and extracellular volume fraction (ECV) to correlate with histological findings of DMF.⁷,⁹ Nonetheless, the question which method provides the most accurate tissue characterisation on a 3T MRI scan is not yet settled.⁷,¹¹

Evaluation of what parameters are important prognostic factors in asymptomatic AS patients are shown to be difficult.¹² Increasing attention is being paid to evaluating DMF as an objective early marker for LV remodelling, especially in patients who do not exhibit any symptoms.¹³

DMF has been shown to display the highest collagen content in the endocardium, gradually decreasing towards the mid-myocardium in severe AS.¹⁴ As most previous studies rely on a single biopsy, typically obtained from the septal region, this approach may introduce sampling bias and limits the ability to capture a potentially heterogeneous distribution of fibrosis within the LV.⁷¹⁵,¹⁷

Our aim was to investigate the presence and distribution of DMF in myocardial biopsies of the left ventricle from patients with severe AS undergoing surgical valve replacement. Further, to explore the findings of DMF in relation to tissue characteristics from CMR and functional data from echocardiography and CMR, we hypothesised that DMF was unevenly distributed throughout the LV and related to impaired LV function and structural changes.

Patients and methods

Patients older than 18 years with severe aortic stenosis scheduled for surgical AVR were recruited to this exploratory cohort between 2014 and 2020. Patients were accepted for surgical intervention according to current guidelines.^{18 19}

Exclusion criteria were concomitant heart valve disease other than AS, ischaemic heart disease, atrial fibrillation or other supraventricular tachycardias, congenital heart disease (except bicuspid aortic valve), haemodynamic instability, previous cardiac surgery, symptomatic lung disease and contraindications for CMR.

Patients who met the inclusion criteria and none of the exclusion criteria were included only after giving written informed consent to participate, following recommendations in the Declaration of Helsinki and its later amendments. The study was registered at ClinicalTrials.gov (ID: NCT02790008).

The study protocol included transthoracic echocardiography (TTE) and gadolinium-enhanced CMR within 1 week preoperatively, followed by myocardial biopsies at the time of surgery. The protocol was ethically approved by the Regional Ethical Review Board, Linköping (No. 2011/105-31).

Endomyocardial biopsies

Left ventricular endomyocardial biopsies 1.2 mm in diameter were taken from five predefined locations in each patient (figure 1). The biopsy sites were selected to represent the base, mid and apical regions of the LV as well as the septal and lateral segments. One surgeon conducted the sampling using a bioptome under direct visual control, avoiding sampling sites that posed an increased risk.

Figure 1. Left ventricular endomyocardial biopsies were taken at five predefined locations marked in yellow. Segmental numbers according to AHA recommendations.

Samples were dyed with Ladewig's Trichrome. Analysis was performed using an image quantification programme, QuPath,²⁰ for verification of the collagen distribution. The collagen volume fraction (CVF) was automatically quantified from the total biopsy area. All samples were analysed blinded to the CMR findings. Samples were excluded if they were found to contain superficial (mainly endocardial) tissue or tissue fragments only.

CMR protocol

CMR imaging, conducted on a Philips Ingenia 3T scanner, included a multislice short-axis cine acquisition using balanced steady-state free-precession, as well as native and post gadolinium contrast 2D acquisitions using Modified Look-Locker Inversion recovery (MOLLI). Acquisitions were obtained at the apical, midventricular and basal slice positions.²¹,²³ Multiple late gadolinium enhanced (LGE) images using a phase-sensitive inversion recovery sequence were collected across the entire short-axis stack.²⁴

The MOLLI acquisitions provided one 2D slice per breath-hold (17 heartbeats). For native T1 measurements, a (3)3 (3)5 acquisition was used and for post-contrast images a 4 (1)3 (1)2) acquisition. A flip angle of 35 degrees was used together with a repetition time of 2.4 msec and an echo time of 1.13 msec.

Image analysis

Quantitative images (T1 maps) from MOLLI were generated immediately on the scanner console. Manual contouring of the endo- and epicardial borders on the T1 maps was performed using the freely available software Segment v.3.0 and v.4.0.²⁵ Based on these manual contours, 16/17-segment bull’s-eye plots were automatically created by the analysis software. The 17 segments corresponded with the myocardial segment model recommended by the American Heart Association.²⁶ We excluded the apex (segment 17) from the T1, ECV and from feature tracking (FT) analysis for global circumferential strain (GCS). For global longitudinal strain (GLS), the apical segment 17 was included. CMR image interpretation and post-processing were conducted according to SCMR recommendations.^{21 27} All manual segmentations were performed by a single physician and validated by an expert research CMR physician (SCMR level 3 certified). Interobserver variability for chamber volume was 8.48 (4.02) mL and 5.46 (7.61) mL for end diastolic volume and end systolic volume, respectively, with SD within parenthesis, all within 95% CI.

For each patient, an average value for T1 relaxation time pre and post contrast was given for the area of each segment. The ECV map was derived from T1 native and post-contrast T1 maps at 20’ and the patient’s haematocrit level.^{27 28} ECV was automatically calculated in the Segment software. A visual assessment was performed to determine the presence or absence of LGE. All short-axis cine-scans were manually segmented for right ventricle and LV volumes as well as for ejection fraction (EF) and LV mass, indexed to body surface area (Mosteller formula). Strains were derived from the FT module in Segment, which estimates GLS from three long-axis cine images, while GCS and global radial strain (GRS) were derived from the short-axis cine.²⁹

Echocardiography

The severity of aortic stenosis (aortic valve area (AVA), mean pressure gradient and Doppler peak velocity), as well as biventricular systolic and diastolic function, was assessed by TTE according to current European and American recommendations.^{1 30 31} For evaluation of LV diastolic dysfunction in excess of grade II, a decision support form based on TTE was used.^{31 32}

Statistical analysis

Data analyses were performed using SPSS (IBM Corp. Released 2021. IBM SPSS Statistics for Windows V.28.0. Armonk, NY: IBM Corp)

Continuous variables were expressed as means (±SD) or medians (quartiles) when the distribution was skewed. Categorical variables were reported as counts and percentages. All tests were two-sided, and a p value <0.05 was considered statistically significant. For comparison of data for descriptive reasons, Student’s T-test and Mann Whitney U were used as appropriate. Analysis of variance and Wilcoxon signed ranks test were used as applicable. Pearson correlation coefficient or Spearman’s correlation test was used to examine the relationship between TTE, CMR and CVF as appropriate.

For correlation testing between CVF, T1 and ECV, T1/ECV was sampled by region of interest for each segment instead of coverage of the entire ventricular short-axis slice on the T1/ECV map, as described by Child et al.³³

Results

Between April 2014 and February 2020, a total of 233 patients were assessed for eligibility to participate in the study. Out of these, 49 patients fulfilled the inclusion criteria, did not have any exclusion criteria and gave their consent to participate. Biopsy sampling was avoided in six patients due to perceived risk and were therefore excluded from this report. Patient characteristics, histology, CMR and TTE data are shown in tables 1 and 2.

Table 1. Demographics.

n=43
Age (years)	68 (49–84)
Sex (M/F)	28/15 (65/35)
BPs (mmHg)	138 (±17)
BPd (mm Hg)	79 (±10)
BSA (m2)	2.0 (±0.2)
Diabetes	6 (14.0)
Smoker/previous smoker	10 (23.3)
CVI	1 (2.3)
NYHA class
I	3 (7.0)
II	39 (90.5)
III	1 (2.3)
Medication
Diuretics*	6 (14.0)
Beta-blocker	15 (34.9)
ACE-inhibitor/ARB	12 (27.9)
Calcium inhibitor	10 (23.3)

Continuous variables are presented as mean (+-SD) or medians (quartiles) if not normally distributed. Categorical variables are presented as number (%).

^*Any diuretic.

ARB, angiotensin receptor blocker; BPs/d, systolic/diastolic blood pressure; BSA, body surface area; CVI, cerebrovascular insult; NYHA, New York Heart Association Functional Classification.

Histology

A total of 192 biopsies were collected from 43 patients. 25 biopsies were excluded due to only containing tissue fragments (n=5) or primarily endocardial tissue (n=20). The 167 biopsies included in the analysis represented 78% of the total possible biopsies calculated (n=215). The median histological CVF in the LV was 19%. The CVF was highest in AHA segment 1, with a median of 55% (table 2). CVF in this area was significantly higher than CVF in segment 11 and 16 (table 3). There was a variation between patients in the degree of fibrosis observed in the biopsies (figure 2). Among patients with at least one biopsy from the base, mid and apical sections of the LV, two-thirds displayed the highest CVF in segment 1 compared with the other segments.

Table 2. Histology, TTE and CMR data.

	n=43
Histology
CVF (%)	19 (12–30)
CVF segment 1 (%)	55 (30-89)
CVF segment 10 (%)	12 (6–25)
CVF segment 11 (%)	10 (5–20)
CVF segment 14 (%)	13 (9–27)
CVF segment 16 (%)	20 (5–33)
TTE
AVAi (cm2/m2)	0.35 (±0.1)
AoV max (m/s)	4.6 (4.3–4.8)
Ao ΔP (mm Hg)	52 (48-57)
LVEF (%)	58 (56-62)
MAPSE (mm)	11 (±2)
GLS (%)	−17 (±3)
S’ LV (cm/s)	6.3 (±1.4)
E/A	0.88 (0.78–1.17)
E/E’	12.7 (± 3.5)
PVs/PVd	1.5 (±0.5)
Deceleration time (ms)	245 (201-301)
LAAi (cm2/m2)	11.2 (±1.8)
CMR
LVMi (g)	76 (±18)
LVEF (%)	66 (59-70)
CI (l/min)	3.5 (±0.7)
LVEDVi (mL/m2)	51 (±13)
LVEDDi (mm/m2)	25 (23–27)
T1 native (ms)	1197 (1155–1236)
ECV (%)	24.2 (22.7–28.1)
MAPSE (mm)	11 (±2)
GLS FT (%)	−13.0 (±2.9)
GCS FT (%)	−18.4 (±4.1)
Radial strain FT (%)	53 (±15)

Continuous variables are presented as mean (±SD) or medians (quartiles) if not normally distributed.

AoV max, transvalvular peak velocity; AVAi, aortic valve area index; CI, cardiac index; CMR, cardiac magnetic resonance imaging; CVF, collagen volume fraction; ECV, extracellular volume; FT, feature tracking; GCS, global circumferential strain; GLS, global longitudinal strain; LAAi, left atrial area index; LVEDDi, left ventricular end-diastolic diameter index; LVEDVi, left ventricular end-diastolic volume index; LVEF, left ventricular ejection fraction; LVMi, left ventricular mass index; MAPSE, mitral annular plane systolic excursion; PVs/PVd, pulmonary venous flow velocity systole/diastole; S’ LV, peak systolic annular velocity left ventricle; TTE, transthoracic echocardiography; ΔP, mean transvalvular pressure gradient.

Table 3. Significant differences in CVF between the LV segments (s).

Total N	15
Test statistic	27, 49
Degree of freedom	4
Asymptotic sig. (two-sided test)	<0.001

Sample 1-sample 2	Test statistic	Std. error	Std. test statistic	Adj. sig.*
s.1-s.10	−1.57	0.58	−0.71	0.07
s.1-s.11	−2.97	0.58	−5.14	<0.01
s.1-s.14	−1.3	0.58	−2.25	0.24
s.1-s.16	−1.83	0.58	−3.18	0.01
s.10-s.11	−1.4	0.58	−2.42	0.15
s.10-s.14	0.27	0.58	0.46	1
s.10-s.16	−0.27	0.58	−0.46	1
s.11-s.14	−1.67	0.58	−2.89	0.04
s.11-s.16	−1.13	0.58	−1.96	0.5
s.14-s.16	−0.53	0.58	−0.92	1

Related-samples Friedman’s two-way analysis of variance by ranks summary.

Wilcoxon signed-rank tests.

^*Significance values have been adjusted by the Bonferroni correction for multiple tests.

CVF, collagen volume fraction; LV, left ventricle.

Figure 2. a. Median CVF for each biopsied area. Interquartile range within paragraphs. b. Median T1 relaxation time per segment. Significant difference between T1 in lateral and septal segments, p=0.013. Significant differences between T1 in base and mid segments and base and apical segments, p=0.002 resp. p=0.05. c. Median ECV per segment. No significant differences in ECV were found between different areas of the left ventricle. Interquartile range within paragraphs calculated as Q3 minus Q1.

CMR compared with histology

The overall T1 relaxation time for the cohort was 1197 ms, and the ECV was 24.2%. Septal T1 relaxation time was significantly longer than that of the lateral T1 time (p<0.01). The basal segments exhibited longer T1 times compared with both the mid (p<0.01) and apical segments (p=0.05). ECV values remained consistent throughout the LV (figure 2). When investigating the relationship between histology and CMR, we found no correlation between median CVF and ECV (r² = 0.27, p=0.10) or T1 (r²= 0.02, p=0.88). AHA segmental analysis revealed a correlation between CVF and ECV in segment 1 (r² = 0.54, p<0.01) and segment 16 (r² = 0.44, p=0.03) (figure 3, online supplemental table 1). Additionally, we found that CVF in segments 1 and 16 significantly correlated with mean ECV (r² = 0.41, p=0.04 and r² = 0.44, p=0.02, respectively) (online supplemental table 1).

Figure 3. a. Correlations between CVF segment 1 and ECV segment 1. b. Correlations between CVF segment 16 and ECV segment 16.

Functional implications

To assess the functional implications of DMF, the study compared structural parameters and functional data where median CVF correlated to LVEF from CMR (r² −0.34, p=0.03) (figure 4).

Figure 4. Significant correlations were found between CVF and LVEF by CMR.

CVF also correlated with GCS (r² 0.31, p=0.04) and GRS (r² −0.33, p=0.03), and there was a tendency towards a correlation with GLS (r² 0.29, p=0.07), all derived from FT CMR. The impact on GCS was primarily seen in the basal segments (r² 0.31, p=0.04). Patients with CVF above median showed a significantly lower GLS from FT CMR analysis (p=0.04) (online supplemental table 2).

Patients having the highest level of CVF in the mid or apical parts of the myocardium had lower mitral annular plane systolic excursion (MAPSE) (p<0.01) and mean LV S’ (p=0.02) by TTE and smaller indexed AVA (AVAi) (p<0.01) compared with patients with the highest level of CVF in segment 1, despite no difference in median CVF (p=0.88) (online supplemental table 3). This was also seen when patients with one or more LGE areas were excluded from the analysis.

Only four patients were found to have diastolic dysfunction ≥grade 2 according to the decision support form used. Indicators for diastolic dysfunction were analysed separately, and we found that median CVF correlated to the left atrial area index (LAAi). Larger LAAi was related to higher ECV in segment 1 (onlinesupplemental tables 2 4).

We found that longer T1 relaxation time correlated to lower GLS (r² 0.42, p<0.01), MAPSE (r² −0.39, p=0.01) and mean S’ LV (r² −0.45, p<0.01) from TTE (online supplemental table 4).

Discussion

The objective of this exploratory study was to investigate the distribution of DMF in the LV and, secondarily, the association between DMF and LV function in patients with severe aortic stenosis. This was achieved by evaluating alterations in myocardial tissue characteristics and function using endomyocardial biopsies from multiple LV locations alongside echocardiographic and CMR imaging. The main findings were:

• CVF demonstrated significant regional variation within the LV with the highest levels observed in the basal segments.

• Systolic LV function correlated with both CVF and T1 relaxation time.

Regional distribution of DMF

This study represents one of the largest investigations to date using endomyocardial biopsy sampling from multiple LV locations in patients with severe AS. While prior studies have predominantly focused on septal biopsies,^{78 15},^{17 34} we performed systematic sampling from five distinct LV regions spanning from the basal to the apical segments and including both septal and lateral segments. This approach revealed pronounced regional variability in CVF, underscoring the limitation of using a single septal biopsy to represent global myocardial fibrosis.

We observed significantly higher CVF levels in AHA segment 1 compared with segments 11 and 16. A plausible explanation for this finding is the increased mechanical wall stress experienced by the basal segments attributable to the larger ventricular diameter in these regions, as described by the law of LaPlace. Elevated mechanical stress has previously been implicated as a trigger for the development of myocardial fibrosis and may thus account for the observed distribution of DMF.³⁵

Association between histological and imaging markers of fibrosis

The relationship between histologically quantified fibrosis (CVF) and non-invasive imaging parameters such as T1 relaxation time and ECV has been extensively investigated, with conflicting results. Several studies have reported correlations between CVF and T1 relaxation time or ECV,⁷,1113 17 36 37 whereas others, including larger biopsy cohorts, have failed to confirm these associations.^{14 38}

In our study, we observed a significant correlation between CVF and ECV in segments 1 and 16, consistent with previous findings.^{8 9 16 39 40} This suggests that regions with more pronounced fibrosis exhibit increased ECV reflecting changes in myocardial composition. However, given the known gradient of collagen from the endocardium to the mid-myocardium,¹⁴ it is likely that only areas with substantial fibrosis (high CVF) demonstrate a clear correlation with segmental ECV.

To minimise blood pool contamination, ROIs for T1 mapping and ECV quantification were carefully delineated within the endo- and epicardial borders.²⁷ Despite this, ECV likely underestimates subendocardial fibrosis, which may be more accurately captured by biopsy. This methodological consideration provides a plausible explanation for the segment-specific correlation between CVF and ECV.

Furthermore, we noted an increase in ECV throughout the LV when basal CVF was elevated, a finding not previously reported in the literature. The T1 relaxation time was longer in the basal segments compared with other locations following a similar pattern as CVF. However, no significant correlation was found between T1 relaxation time and CVF, likely due to technical limitations in T1 mapping, which can be influenced by multiple factors, as discussed below.

DMF and LV function

Current guidelines recommend AVR for patients with severe AS who have symptoms or evidence of LV impairment, typically indicated by an EF <50%.^{19 41} Recognising symptoms can be subjective and a decline in EF may be a late and irreversible manifestation.^{5 42} While previous studies have linked focal fibrosis to poor prognosis in AS, reversibility of DMF has been suggested.^{4 6 43} Therefore, it is of clinical importance to evaluate DMF in relation to LV function, early markers of decompensation and to understand its effect on the myocardium.

In this study, higher CVF levels were associated with impaired LV systolic function, as reflected by reduced GLS, GCS and radial strain. No direct association between ECV and systolic function was detected; however, increased ECV correlated with early signs of diastolic dysfunction, specifically an increase in LAAi. Although no additional echocardiographic markers of diastolic dysfunction were identified, the observed association between ECV and LAAi warrants further investigation.

Furthermore, a more extensive distribution of DMF involving mid-ventricular and near-apical regions was associated with impaired MAPSE and reduced AVAi. These findings suggest a progression of fibrosis from the basal segments towards the apex, accompanied by a gradual decline in LV function. Importantly, these associations remained consistent after exclusion of patients with LGE, indicating that the observed functional impairments are not solely attributable to replacement fibrosis.

Finally, prolonged T1 relaxation time was associated with impaired LV systolic function, as evidenced by lower GLS, MAPSE and mean peak systolic velocity (S’) on TTE. These findings are consistent with previous studies reporting associations between native T1 and myocardial dysfunction in AS.^{10 44}

Limitations and clinical implications

While this study provides novel insights into the regional distribution of DMF and its association with LV function in patients with severe AS, several limitations must be acknowledged when interpreting the findings.

A major methodological challenge was the alignment of biopsy sites with corresponding regions in T1 maps, which inherently complicates the spatial correlation between histological and CMR-derived data. The discrepancy in scale between the relatively small biopsy samples and the larger CMR regions of interest introduces an additional potential source of error. Furthermore, biopsies capture only a limited transmural extent of the myocardium and may not adequately reflect the full, possibly often heterogeneous, distribution of DMF throughout the LV.

The absence of biopsies from 23 segments, combined with the exclusion of an additional 25 samples due to insufficient quality (representing 22% of all possible biopsies), further limits the ability to robustly demonstrate associations between histological and imaging findings.

In addition, CMR-based techniques, particularly native T1 relaxation time and to a lesser extent ECV, are susceptible to technical variability. Factors such as the choice of T1 mapping sequence, magnetic field strength, scanner-specific characteristics and field homogeneity influence measurement accuracy and reproducibility. The considerable overlap in native T1 values between healthy and diseased myocardium further complicates clinical interpretation.^{14 21 45}

Nevertheless, despite these methodological constraints, ECV and T1 relaxation time remain promising non-invasive markers for the assessment of DMF in AS. Of particular relevance, ECV has demonstrated prognostic value and appears less susceptible to scanner-specific variations, supporting its potential clinical utility.⁹⁴⁵,⁴⁸

The limitations highlighted underscore the complexity of AS pathophysiology and its multifaceted impact on myocardial fibrosis and cardiac function. Although we established a significant correlation between ECV and histology, the correlation coefficient remained moderate.

Despite these limitations, this study uniquely includes a substantial number of biopsies from across the entire left ventricle. While the strict exclusion criteria limit the generalisability of the findings, they also help minimise confounding factors and ensure the homogeneity of the cohort. The association between DMF and both systolic and diastolic LV functional parameters provides a foundation for future investigations into the prognostic significance of regional DMF patterns in AS.

The present findings suggest that the regional distribution of DMF, rather than its overall burden, may be of importance for LV function. This adds nuance to current understanding and indicates potential limitations in how DMF is assessed for risk stratification. Longitudinal studies are therefore warranted to determine whether regional DMF distribution may predict myocardial recovery or inform the optimal timing of intervention.

Conclusion

In this study, DMF in patients with severe AS, quantified by CVF from endomyocardial biopsies, demonstrated an uneven distribution across the LV, with the highest levels observed in the basal segments. The correlation between segmental CVF and ECV, assessed by CMR, further supports this regional heterogeneity.

LV systolic function was inversely associated with CVF, as reflected by both CMR-derived and echocardiographic parameters. Furthermore, the observed associations between left atrial size, CVF and ECV suggest early diastolic impairment in the presence of advanced DMF.

Together, these findings highlight the clinical relevance of detecting DMF at an early stage in patients with aortic stenosis. They also provide a foundation for future studies aimed at exploring the mechanisms underlying the regional distribution of DMF and clarifying its prognostic and functional implications for both systolic and diastolic left ventricular performance.

Data availability statement

Data are available upon reasonable request.