Contrast-enhanced computed tomography for quantifying mitral annular calcification in mitral valve surgery and association with outcomes

Abstract

Aims

Evaluation of mitral annular calcification (MAC) on non-contrast computed tomography (CT) has been associated with clinical outcomes. There is no accepted methodology for evaluating MAC using contrast CT. This study aimed to validate MAC quantification using contrast CT and evaluate its clinical impact in patients undergoing mitral valve (MV) surgery.

Methods and results

Between 2011 and 2020, 303 patients with native MV dysfunction who underwent contrast CT prior to surgery were included. In a validation cohort of 50 patients undergoing both contrast and non-contrast CT, Agatston scores were correlated to derive contrast-specific thresholds. Baseline clinical parameters and a multi-parametric MAC score (range: 2–12) were recorded for the entire cohort. The contrast-enhanced CT MAC score consisted of Agatston score (1 point: <650 AU; 2 points: 650–1700AU; 3 points: 1700–2800 AU; 4 points: >2800 AU), MAC circumferential angle (1 point: <90°; 2 points: 90–<180°; 3 points: 180–<270°; 4 points: 270–360°), trigone involvement (1 point: single trigone; 2 points: both trigones), myocardial infiltration (1 point), and left ventricular outflow tract involvement (1 point). Associations between MAC score and clinical outcomes were evaluated. Higher MAC scores predicted increased total operating time (P < 0.001), cross-clamp time (P < 0.001), intensive care unit stay (P < 0.001), open surgical approach (P < 0.001), in-hospital complications (P < 0.001), and pacemaker implantation (P = 0.055). Contrast-enhanced CT MAC score independently predicted all-cause mortality (P < 0.001).

Conclusion

A multiparametric MAC score on contrast CT enables assessment of MAC severity, and is associated with operative approach and postoperative outcomes in patients with MV dysfunction undergoing surgery.

Graphical Abstract

Graphical Abstract.

Prognostic value of contrast-enhanced CT–derived mitral annular calcification score in patients undergoing mitral valve surgery. CT, computed tomography; MAC, mitral annular calcification; IQR, interquartile range; h, hours; min, minutes; ICU, intensive care unit.

Introduction

Mitral annular calcification (MAC) is a degenerative process characterized by progressive calcium deposition within the mitral annulus.¹ Although it predominantly affects the posterior annulus, involvement of the anterior annulus may also occur, and in advanced cases, the calcification can extend into the adjacent myocardium and left ventricular outflow tract (LVOT).² MAC alters the structural integrity and dynamic motion of the mitral leaflets and is frequently associated with mitral valve (MV) dysfunction, including stenosis and regurgitation.³ The prevalence of MAC is incompletely defined but is estimated to range from 7% to 23%, increasing markedly with age and cardiometabolic risk factors such as atherosclerosis and chronic kidney disease.⁴ Accurate assessment of MAC severity is clinically important, as MAC has been independently associated with mortality and adverse events including stroke, arrhythmias, and myocardial infarction.⁵ Although MV intervention improves outcomes in patients with MAC-related valve dysfunction, these procedures are technically challenging, and optimal timing remains uncertain.⁶

A major barrier to improving outcomes is the absence of consensus regarding the definition and grading of MAC severity. While echocardiography has traditionally been used for detection and characterization of MAC, it may be limited by acoustic shadowing and variability in qualitative definitions.^7,8 More recently, non-contrast cardiac CT has emerged as a complementary imaging modality capable of providing quantitative and reproducible assessment of MAC burden. Xu et al. proposed a multiparametric non-contrast CT–based MAC score that demonstrated associations with both surgical outcomes and long-term prognosis.^9,10 In clinical practice, contrast-enhanced, electrocardiogram-gated CT is frequently performed in patients with MV dysfunction as part of pre-procedural evaluation for MV surgery, enabling assessment for candidacy for robotic assisted vs. open repair approach.¹¹ This modality provides an additional opportunity for quantitative MAC scoring^12,13; however, quantification of MAC using the Agatston method on contrast-enhanced CT has not been validated in MAC. In this study, we evaluate calcium scores derived from contrast-enhanced CT in patients with MAC undergoing MV surgery, validate a multiparametric MAC scoring system, and assess its association with clinical outcomes.

Methods

We conducted a retrospective cohort study by reviewing the patients who underwent MV surgery at our centre. Patients with MAC identified on both pre-operative transthoracic echocardiography and contrast-enhanced cardiac CT were identified and included in the final study cohort (see Supplementary data online, Figure S1). In addition, a validation cohort was identified consisting of patients with MAC who underwent both preoperative contrast-enhanced and non-contrast cardiac CT within a 1-year interval.

Baseline clinical and echocardiographic characteristics were extracted from electronic medical records. All patients in the study cohort were managed by experienced cardiothoracic surgeons with substantial expertise in the surgical management of complex MAC and MV disease. Details regarding the type of MV intervention (open vs. robotic) and any concomitant procedures, including coronary artery bypass grafting (CABG), concomitant surgical atrial fibrillation ablation using the MAZE procedure, or left atrial appendage (LAA) ligation, were recorded. Surgery decisions were made by consensus between experienced imaging cardiologists and cardiothoracic surgeons. The present study was conducted in accordance with the Declaration of Helsinki Ethical Principles and approved by the institutional review board (19-1055).

Follow-up period and clinical outcome

Six clinical outcomes were evaluated in the cohort, including all-cause mortality, major adverse cardiovascular events (MACE), defined as myocardial infarction, coronary revascularization, stroke, or hospitalization for heart failure, along with the need for permanent pacemaker implantation after surgery, total operative time, aortic cross-clamp time, length of stay in the intensive care unit (ICU), and a composite endpoint of in-hospital complications (30-day mortality, stroke, acute kidney injury, major bleeding, prolonged ventilation, reoperation, or sepsis). Mortality and cause-of-death information, when available, were obtained from electronic medical records and cross-referenced with online obituary records when available. Follow-up duration was measured from the date of MV surgery to the last known clinical encounter or death.

Echocardiographic assessment

Comprehensive transthoracic echocardiographic examinations were performed by experienced sonographers using Vivid E9 (GE Healthcare, Chicago, IL), EPIQ 7C, or EPIQ CVx (Philips Medical Systems, Boston, MA). Valvular dysfunction was assessed in accordance with contemporary European Society of Cardiology (ESC) and European Association of Cardiovascular Imaging (EACVI) guidelines.^14,15 Peak and mean transmitral gradients were obtained using continuous-wave Doppler. Mitral regurgitation (MR) volume was quantified using the continuity equation. MAC severity was assessed from the parasternal short-axis view at the level of the mitral annulus and qualitatively categorized based on circumferential involvement: <90° (grade 1), 90° to <180° (grade 2), 180° to <270° (grade 3), and 270° to circumferential involvement (grade 4). Grades 3 and 4 indicate severe MAC.²

Contrast CT-based MAC score

CT scans were performed utilizing dedicated multi-detector cardiac computed tomography scanners (Siemens Somatom Force Scanner, Siemens Definition AS Scanner, Siemens Definition Flash Scanner, Siemens Healthineers, Erlangen, Germany; Philips iCT 256 Scanner, Koninklijke Philips N.V., Amsterdam). All patients underwent contrast-enhanced, ECG-gated cardiac CT acquisition following administration of iodinated contrast (Omnipaque 300 or 350). Image reconstruction and analysis were performed using dedicated post-processing software (Aquarius iNtuition, version 4.6.1; TeraRecon Inc., Durham, NC). MAC scoring was performed using the previously described multiparametric MAC score, which ranges from 2 to 12 points and incorporates five components: two quantitative parameters—(i) Agatston calcium score and (ii) MAC circumferential angle—and three qualitative parameters—(iii) LVOT extension, (iv) myocardial infiltration, and (v) involvement of the fibrous trigones (Figure 1).¹⁰ Myocardial infiltration was defined as an extension of calcification into the adjacent myocardial tissue. Trigone involvement was defined as the presence of calcification involving the right and/or left fibrous trigone, assessed on short-axis CT images. LVOT extension was defined as extension of mitral annular calcification into the left ventricular outflow tract, assessed using both short- and long-axis CT views.² The total MAC score was calculated as the sum of these components. MAC score was further categorized into severe MAC (scores 9–12) and non-severe MAC (scores 2–8). Calcium scoring was quantified using a modified Agatston method based on the 4 SD thresholding technique.¹⁶ A circular region of interest was placed in the ascending aorta blood pool at the level of the pulmonary artery bifurcation to determine the mean and standard deviation (SD) of the blood pool attenuation in Hounsfield units (HU). Using the Agatston methodology with a threshold defined as ≥4 SD above the mean blood pool attenuation, Agatston score was calculated as the cumulative calcium burden along the mitral annulus using multi-planar reconstructions on cardiac CT. The software subsequently computed the MAC calcium score, as well as associated calcium volume and mass.

Figure 1.

Multiparametric MAC score based on contrast-enhanced CT imaging.

Statistical analysis

Continuous variables were expressed as mean ± SD for normally distributed data and as median with interquartile range (IQR) for skewed distributions. Normality was assessed using the Kolmogorov–Smirnov test, with P < 0.05 indicating non-normal distribution. Categorical variables were summarized as counts and percentages.

To validate Agatston score cut-offs for use in contrast-enhanced CT, a validation cohort of 50 patients with MAC who underwent both non-contrast, ECG-gated CT and contrast-enhanced, ECG-gated CT within a 1-year interval were analysed. Agatston scores were quantified on both scans. Pearson correlation was used to assess the linear association between scores obtained from contrast-enhanced vs. non-contrast CT. Linear regression analysis was subsequently performed to evaluate the relationship between the two measurements. Based on these results, new contrast-enhanced CT Agatston cut-offs were derived for incorporation into the MAC scoring system.

After construction of the novel MAC score, multivariable regression models—logistic or linear, depending on whether the outcome was binary or continuous—were used to evaluate its association with clinical outcomes. For each outcome, a baseline model including age, sex, hypertension, diabetes, and prior cardiac surgery was built. The MAC score was then added to the baseline model to determine whether it provided incremental prognostic value. Regression coefficients (B), standard errors (SE), and P values were reported. To further assess prognostic performance, a Cox regression analysis was performed for time to all-cause mortality, adjusted for age and sex due to the limited number of events. Kaplan–Meier curves were used to estimate survival, and MAC score groups were compared by the log-rank test. Differences in MAC score according to type of MV intervention (open vs. robotic) and the occurrence of postoperative complications were evaluated using boxplots, with comparisons made using the Mann–Whitney U test. A two-sided P < 0.05 was considered statistically significant.

For reproducibility assessment, two independent investigators, blinded to clinical data, evaluated the MAC score in 20 randomly selected patients on two separate occasions. Inter- and intra-observer agreement were evaluated using the standard error of the mean (SEM), denoted as SEM_inter and SEM_intra, respectively. All statistical analyses were performed using SPSS version 23.0 (IBM Corp., Armonk, NY).

Results

Clinical characteristics

Between January 2011 and September 2019, a total of 11 296 patients underwent MV surgery at our institution. Of these, 303 patients had MAC identified on both pre-operative echocardiography and contrast-enhanced CT and were included in the final cohort. Patients without MAC on echocardiography, those lacking pre-operative contrast-enhanced CT, those with only non–contrast-enhanced CT, inadequate imaging quality, or prior MV intervention were excluded (see Supplementary data online, Figure S1). Baseline clinical, echocardiographic, and computed tomographic characteristics are summarized in Tables 1 and 2.

Table 1.

Baseline clinical characteristics of the study cohort

Characteristics	Overall	Mild MAC	Moderate MAC	Severe MAC	P value
n	303	192	69	42
Age	68.0 (61.0–75.0)	66.0 (57.0–73.0)	72.0 (64.0–78.0)	73.0 (69.0–78.8)	<0.001
Male	156 (51.5%)	103 (53.6%)	34 (49.3%)	19 (45.2%)	0.563
Height, cm	170.2 (162.6–177.9)	171.2 (162.6–180.3)	170.2 (162.6–178.0)	165.1 (157.5–177.8)	0.067
Weight, kg	78.5 (65.8–90.9)	77.6 (65.6–88.6)	80.7 (66.7–98.4)	80.0 (63.3–90.4)	0.656
Body mass index, kg/m2	26.6 (23.6–30.3)	26.2 (23.3–29.1)	27.9 (24.0–30.8)	28.8 (23.2–35.2)	0.035
Comorbidities
Heart failure	88 (29.0%)	48 (25.0%)	21 (30.4%)	19 (45.2%)	0.031
Hypertension	185 (61.1%)	106 (55.2%)	46 (66.7%)	33 (78.6%)	0.011
Hyperlipidaemia	163 (53.8%)	98 (51.0%)	37 (53.6%)	28 (66.7%)	0.184
Diabetes mellitus	52 (17.2%)	27 (14.1%)	14 (20.3%)	11 (26.2%)	0.124
Atrial fibrillation	98 (32.3%)	58 (30.2%)	25 (36.2%)	15 (35.7%)	0.578
Coronary heart disease	87 (28.7%)	49 (25.5%)	18 (26.1%)	20 (47.6%)	0.014
Mitral valve prolapse	136 (44.9%)	104 (54.2%)	29 (42.0%)	3 (7.1%)	<0.001
Peripheral artery disease	15 (5.0%)	8 (4.2%)	1 (1.4%)	6 (14.3%)	0.007
Smoker	140 (46.2%)	90 (46.9%)	27 (39.1%)	23 (54.8%)	0.264
Stroke/TIA	33 (10.9%)	16 (8.3%)	8 (11.6%)	9 (21.4%)	0.047
Chronic lung disease	65 (21.5%)	34 (17.7%)	10 (14.5%)	21 (50.0%)	<0.001
Chronic kidney disease	56 (18.5%)	27 (14.1%)	12 (17.4%)	17 (40.5%)	<0.001
Prior cardiac surgery	70 (23.1%)	37 (19.3%)	17 (24.6%)	16 (38.1%)	0.030
Valve surgery	23 (7.6%)	10 (5.2%)	6 (8.7%)	7 (16.7%)	0.037
Coronary artery bypass grafting	15 (5.0%)	7 (3.6%)	4 (5.8%)	4 (9.5%)	0.264
Medication
Aspirin	137 (45.2%)	83 (43.2%)	33 (47.8%)	21 (50.0%)	0.643
ACE-I/ARBs	119 (39.3%)	78 (40.6%)	29 (42.0%)	12 (28.6%)	0.304
Beta-blockers	137 (45.2%)	83 (43.2%)	31 (44.9%)	23 (54.8%)	0.396
Diuretics	114 (37.6%)	66 (34.4%)	23 (33.3%)	25 (59.5%)	0.007
Statin	147 (48.5%)	81 (42.2%)	34 (49.3%)	32 (76.2%)	<0.001
Oral anticoagulants	74 (24.4%)	41 (21.4%)	15 (21.7%)	18 (42.9%)	0.011
Labs
Albumin	4.2 (3.9–4.5)	4.3 (4.0–4.5)	4.2 (3.9–4.4)	4.0 (3.4–4.3)	0.001
Haemoglobin	13.4 (12.0–14.6)	13.6 (12.5–14.8)	13.0 (12.0–14.1)	11.7 (10.3–13.6)	<0.001
NT-proBNP, pg/mL	404.5 (157.8–1091.0)	306.5 (125.5–978.5)	781.0 (288.5–1932.5)	414.0 (307.0–629.0)	0.070
Serum creatinine	0.9 (0.8–1.1)	0.9 (0.8–1.1)	0.9 (0.8–1.1)	1.1 (0.8–1.2)	0.024
Mitral valve surgery
Repair	211 (69.6%)	162 (84.4%)	41 (59.4%)	8 (19.0%)	<0.001
Replacement	92 (30.4%)	37 (19.3%)	30 (43.5%)	25 (59.5%)	<0.001
Robotic	46 (15.2%)	43 (22.4%)	2 (2.9%)	1 (2.4%)	<0.001
Open	257 (84.8%)	149 (77.6%)	67 (97.1%)	41 (97.6%)	<0.001
Total operation time, min	274.2 (216.0–360.0)	252.0 (204.0–330.0)	266.4 (228.0–354.0)	366.0 (308.4–466.5)	<0.001
Cross-clamp time, min	81.0 (62.0–115.8)	76.0 (59.0–104.0)	93.0 (69.0–143.0)	111.5 (79.0–136.2)	<0.001
ICU stay, h	46.5 (24.5–95.1)	35.6 (24.1–73.0)	51.0 (24.7–94.8)	97.8 (64.2–143.2)	<0.001

P values <0.05 have been bolded.

ACE-I/ARB, angiotensin-converting enzyme inhibitor or angiotensin receptor blocker; CABG, coronary artery bypass grafting; ICU, intensive care unit; NT-proBNP, N-terminal pro–B-type natriuretic peptide; TIA, transient ischaemic attack.

Table 2.

Echocardiographic and computed tomographic parameters of the study cohort

Imaging characteristics	Overall	Mild MAC	Moderate MAC	Severe MAC	P value
n	303	192	69	42
Echocardiographic characteristics
LV ejection fraction, %	60.0 (56.0–66.0)	60.0 (56.0–66.0)	63.0 (55.0–67.0)	60.0 (54.5–66.0)	0.564
LVED volume index, mL/m2	61.1 (48.0–75.7)	60.7 (47.9–75.8)	64.2 (45.4–76.6)	59.9 (49.8–71.3)	0.995
LVES volume index, mL/m2	22.8 (18.2–31.3)	23.1 (18.7–31.8)	22.3 (15.5–30.5)	22.1 (18.3–31.0)	0.445
LVED diameter, cm	4.9 ± 0.9	4.9 ± 0.9	4.9 ± 0.8	5.0 ± 1.0	0.918
LVES diameter, cm	3.2 (2.7–3.8)	3.3 (2.7–3.9)	3.4 (2.8–3.8)	2.9 (2.7–3.3)	0.091
Septal wall diameter, cm	1.2 (1.0–1.4)	1.2 (1.0–1.4)	1.3 (1.1–1.5)	1.2 (0.9–1.4)	0.209
Posterior wall diameter, cm	1.1 (1.0–1.3)	1.0 (0.9–1.2)	1.2 (1.0–1.3)	1.0 (0.9–1.2)	0.003
LVMI (g/m2)
LA volume index, mL/m2	50.0 (39.0–66.0)	50.0 (39.0–67.0)	50.5 (39.0–64.0)	49.0 (43.2–59.8)	0.980
RVSP, mm Hg	39.0 (30.0–49.0)	39.0 (28.0–48.0)	41.5 (34.5–49.0)	36.5 (27.0–51.0)	0.114
TAPSE	2.6 (2.0–12.0)	2.9 (2.2–12.4)	2.2 (1.9–4.5)	2.2 (1.8–3.0)	0.007
S′	10.6 (3.0–13.3)	10.0 (3.0–13.0)	12.3 (8.0–15.0)	10.0 (2.5–12.0)	0.066
MV peak gradient	9.1 (6.0–14.0)	9.2 (6.1–13.7)	8.3 (5.4–12.6)	9.2 (6.0–19.8)	0.330
MV mean gradient	3.4 (2.3–5.4)	3.3 (2.3–5.0)	3.9 (2.8–6.0)	4.0 (2.3–7.0)	0.380
MV-VTI	30.0 (23.5–39.5)	29.5 (23.0–38.2)	30.0 (25.5–39.8)	31.0 (24.1–44.3)	0.312
MV diameter	2.6 ± 0.6	2.6 ± 0.7	2.5 ± 0.6	2.3 ± 0.6	0.015
MR vol (mL)	52.2 (34.6–65.8)	54.0 (38.5–66.9)	44.2 (25.7–63.8)	48.7 (28.2–58.5)	0.362
Moderate or severe mitral stenosis	61 (20.1%)	23 (12.0%)	14 (20.3%)	24 (57.1%)	<0.001
Significant mitral regurgitation (≥3+)	204 (67.3%)	144 (75.0%)	43 (62.3%)	17 (40.5%)	<0.001
MAC grade
Mild	123 (43.9%)	107 (62.2%)	15 (22.4%)	1 (2.4%)	<0.001
Moderate	92 (32.9%)	57 (33.1%)	30 (44.8%)	5 (12.2%)
Severe	65 (23.2%)	8 (4.7%)	22 (32.8%)	35 (85.4%)
RV dysfunction
None	248 (81.8%)	153 (79.7%)	57 (82.6%)	38 (90.5%)	0.147
Mild	26 (8.6%)	21 (10.9%)	3 (4.3%)	2 (4.8%)
Moderate to severe	12 (4.0%)	6 (3.1%)	4 (5.8%)	2 (4.8%)
Contrast CT parameters
LVOT extension/involvement of aorto-mitral curtain	61 (20.1%)	10 (5.2%)	18 (26.1%)	33 (78.6%)	<0.001
Involvement of one trigone	33 (10.9%)	2 (1.0%)	15 (21.7%)	16 (38.1%)	<0.001
Involvement of additional trigone	26 (8.6%)	0 (0.0%)	4 (5.8%)	22 (52.4%)	<0.001
Infiltration of myocardium	30 (9.9%)	2 (1.0%)	10 (14.5%)	18 (42.9%)	<0.001
Agatston score, AU	956.0 (295.8–2950.0)	947.0 (278.0–2554.5)	858.5 (307.0–2260.8)	2167.0 (458.0–5433.0)	0.155
Mitral annular calcification volume, mm3	706.0 (190.5–2237.0)	707.0 (190.0–2149.0)	585.0 (175.0–1597.5)	1638.0 (286.0–4503.0)	0.120
Mitral annular calcification mass, mg	484.0 (119.5–1305.0)	484.0 (125.5–1122.5)	586.0 (127.0–1417.5)	478.0 (109.0–1988.0)	0.900
Circumferential angle, degrees	90.2 (39.6–197.8)	82.0 (41.4–171.6)	90.7 (32.6–222.5)	114.0 (57.0–228.0)	0.563

P values <0.05 have been bolded.

AU, Agatston units; LVED, left ventricular end-diastolic; LVES, left ventricular end-systolic; LVMI, left ventricular mass index; LA, left atrial; RVSP, right ventricular systolic pressure; S′, systolic tricuspid annular velocity; VTI, velocity–time integral.

The median age was 68 years (IQR 61–75), and 49.5% were female. Hypertension (61.1%) and hyperlipidaemia (53.8%) were common, and 46.2% were current or former smokers. Prior cardiac surgery was present in 23.1%, including 23 patients with previous valve surgery and 15 patients with coronary artery bypass grafting. Common medications included aspirin (45.2%), beta-blockers (45.2%), and statins (48.5%). Median laboratory values at the time of surgical evaluation included haemoglobin 13.4 g/dL (IQR 12.0–14.6), albumin 4.2 g/dL (3.9–4.5), and creatinine 0.9 mg/dL (0.8–1.1).

On echocardiography, the median left ventricular ejection fraction (LVEF) was 60% (IQR 56–66). Peak and mean transmitral gradients were 9.1 mmHg (IQR 6.0–14.0) and 3.4 mmHg (IQR 2.3–5.4), respectively. Median regurgitant volume was 52.2 mL (IQR 34.5–65.9), 20.1% had moderate or severe mitral stenosis and 4 67.3% had significant (≥3+) mitral regurgitation. MAC severity was classified as mild in 40.6%, moderate in 30.4%, and severe in 21.5% of cases. Median right ventricular systolic pressure was 39 mmHg (IQR 30–49), tricuspid annular plane systolic excursion (TAPSE) was 2.6 cm (IQR 1.9–3.0), and 12 patients had moderate or severe right ventricle (RV) dysfunction.

Validation of Agatston score in contrast-enhanced CT

To validate Agatston score cut-offs on contrast-enhanced CT, we identified a subset of 50 patients from the final cohort of 303 patients who underwent both contrast and non-contrast cardiac CT within a 1-year interval prior to surgery. The median time between scans was 77 days (IQR 11–152). MAC burden was quantified on both studies. Echocardiographic and computed tomographic characteristics of the validation cohort is demonstrated in Supplementary data online, Table S1. There was a strong correlation for Agatston scores obtained from contrast and non-contrast CT on the Pearson analysis (P < 0.001). Simple linear regression showed a significant linear relationship, with a slope (B) of 0.539 and an intercept of 95.603 (r = 0.932, P < 0.001). Scatter plots with fitted regression line are demonstrated in Figure 2. Using the regression-derived calibration, the established non-contrast thresholds (<1000, 1000–<3000, 3000–5000, and >5000 AU) corresponded to lower values on contrast-enhanced CT. These translate to <650, 650–<1700, 1700–<2800, and >2800 AU, respectively, allowing MAC severity to be graded consistently across both scan types.¹⁰  Figure 3 is a representative example showing correlation of MAC quantification using non-contrast CT and contrast-enhanced CT.

Figure 2.

Correlation between contrast-enhanced and non-contrast CT-derived Agatston scores of mitral annular calcification.

Figure 3.

MAC quantification using cardiac CT, showing correlation between non-contrast and contrast-enhanced CT imaging.

MAC score on contrast-enhanced CT

On contrast-enhanced CT, the median Agatston score was 956 Agatston units (AU) (IQR 293–2950), with corresponding MAC volume of 706 mm³ (189–2287) and MAC mass of 484 mg (115–1312). The median circumferential MAC angle was 90.2° (IQR 39.0–198.5). Extension of MAC into the LVOT was observed in 61 patients (20.1%), involvement of one trigone in 33 patients (10.9%), both trigones in 26 patients (8.6%), and myocardial infiltration in 30 patients (9.9%). The median overall MAC score was 3 (IQR 2–6).

Interobserver and intraobserver variability

MAC scoring reproducibility was assessed in 20 randomly selected cases evaluated by two independent observers, each blinded to their own prior measurements and to those of the other observer. Interobserver agreement was excellent, with an intraclass correlation coefficient (ICC) of 0.983 for the continuous MAC score (95% CI: 0.956–0.994) and a corresponding percentage SEM of 6.0%. When MAC severity was categorized into mild, moderate, and severe classes, interobserver agreement remained high, with an ICC of 0.905 (95% CI: 0.763–0.962) and an SEM of 11.9%.

Intraobserver reproducibility was similarly excellent for the continuous MAC score, with an ICC of 0.992 (95% CI: 0.980–0.997) and an SEM of 4.1%. Agreement for categorical MAC severity remained high, with an ICC of 0.905 (95% CI: 0.763–0.962) and an SEM of 10.8%.

MV intervention

All patients underwent MV intervention a median of 7 days after CT imaging (IQR 3–42). MV repair was performed in 211 patients (69.6%), and MV replacement in 92 patients (30.4%). An open surgical approach was used in 257 patients (84.8%), whereas 46 patients (15.2%) underwent robotic surgery. All patients who had robotic surgery underwent MV repair. During the surgery, 73 patients had MAC debridement, of which 13 patients had extensive MAC debridement, and 8 patients (2.6%) had annular reconstruction with an autologous or bovine pericardial patch. Patients undergoing robotic surgery were more often male (65.2% vs. 34.8%, P = 0.043). Concomitant surgical procedures were performed in 204 patients (67.3%), and it was more common in patients undergoing open surgery compared to robotic surgery (90.2% vs. 58.7%, P < 0.001) (see Supplementary data online, Table S2). Of these, 86 patients underwent MAZE, pulmonary vein isolation (PVI) or LAA closure. Among those undergoing MV replacement, no postoperative paravalvular leaks were identified. Among patients with MV repair, 3 patients had moderate residual MR after repair. In the entire cohort, median total operation time was 275 min (IQR 216–360), and cross-clamp time was 81 min (IQR 62–116).

MAC score differed significantly according to surgical approach and surgical technique (Figure 4A). Patients undergoing robotic MV repair had the lowest MAC scores, whereas those undergoing open MV replacement exhibited the highest MAC burden. Compared with robotic repair, MAC scores were significantly higher in patients undergoing open repair [median, 3 (IQR, 2–5) vs. 2 (IQR, 2–2) P = 0.001] and open replacement [median, 7 (IQR, 3–10) vs. 2 (IQR, 2–2); P < 0.001]. In addition, MAC scores were significantly higher in patients undergoing open replacement compared with open repair [median, 7 (IQR, 3–10) vs. 3 (IQR, 2–5); P < 0.001].

Figure 4.

MAC score distribution by (A) type of MV intervention and (B) occurrence of post-operative complications.

Patient outcomes

A summary of the study outcomes is presented in Supplementary data online, Table S3. The median ICU length of stay was 46.5 h (IQR 24.5–95.4). At least one post-operative complication occurred in 73 patients (24.1%), including 30-day mortality in three patients, stroke in 10, acute kidney injury in 33, major bleeding in 10, prolonged mechanical ventilation in 23, sepsis in one, and reoperation in 11 patients. Among the 11 reoperations, the most common indication was mediastinal bleeding requiring chest exploration and/or washout (n = 7). Other indications included postoperative cardiac herniation (n = 1), haemopericardium (n = 1), haemothorax (n = 1), and atrioventricular groove disruption resulting in haemorrhagic shock (n = 1). Aetiologies of 30-day mortality included cardiac arrest due to multi-organ system failure in two patients and haemorrhagic shock in one patient. Patients with post-operative complications had significantly higher MAC scores compared to those with no complications (P < 0.001, Figure 4B). During follow-up, 32 deaths were confirmed, with follow-up duration ranging from the day of surgery to 9.6 years (median 1.4 years). The median duration to death was 336 days (IQR 70–1829 days). Major adverse cardiovascular events occurred in 40 patients, and 29 patients required permanent pacemaker implantation during the follow-up period with a median MAC score of 5 (IQR 2–9).

Clinical utility of MAC score

Regression analyses evaluating the association between MAC score and clinical outcomes are summarized in Table 3. For each endpoint, the MAC score was incorporated into a baseline model of relevant clinical variables. In linear regression analysis, the MAC score was an independent predictor of total operating time (P < 0.001), aortic cross-clamp time (P < 0.001), and length of ICU stay (P < 0.001) after adjustment for age, sex, hypertension, diabetes, and prior cardiac surgery. Quantitatively, each one-point increase in MAC score was associated with an additional 0.12 h of total operating time, 4.9 min of cross-clamp time, and 0.6 days in ICU (see Supplementary data online, Figure S2). In binary logistic regression, higher MAC scores were associated with a greater likelihood of undergoing an open surgical approach [odds ratio (OR), 1.733; 95% CI, 1.292–2.326; P < 0.001] and occurrence of post-operative complications (OR, 1.193; 95% CI, 1.088–1.308; P < 0.001) after adjustment for the same baseline covariates. The MAC score also showed a trend towards independently predicting the need for postoperative pacemaker implantation when adjusted for age and sex, although it did not reach statistical significance (OR, 1.127; 95% CI, 0.997–1.273; P = 0.055). Cox regression demonstrated that higher MAC severity was an independent predictor of all-cause mortality after adjusting for age and sex (hazard ratio, 1.214; 95% CI, 1.076–1.371; P < 0.001). Kaplan–Meier analysis demonstrated significantly higher mortality in patients with severe MAC compared to those with mild or moderate MAC (Log-rank P < 0.001, Figure 5).

Table 3.

Regression coefficients for association between MAC score and clinical outcomes

Outcome	Baseline model	MAC	P value
Total operating time, ha	3.309 (0.635)	0.123 (0.037)	<0.001
Aortic cross-clamp time, mina	62.293 (16.113)	4.948 (0.902)	<0.001
Days in ICUa	2.295 (2.417)	0.563 (0.138)	<0.001
Open surgery (vs. robotic)a	−1.568 (0.894)	0.550 (0.150)	<0.001
In-hospital complicationsa	−5.595 (1.114)	0.176 (0.047)	<0.001
Pacemaker requirementb	−4.030 (1.319)	0.120 (0.062)	0.055
MACEb	−5.529 (1.303)	0.033 (0.056)	0.555
All-cause mortalityb	–	0.210 (0.056)	<0.001

P values <0.05 have been bolded.

^aAdjustment for age, sex, hypertension, diabetes mellitus, Prior cardiovascular surgery.

^bAdjustment for age.

Figure 5.

Kaplan–Meier analysis for all-cause mortality for the entire cohort stratified by MAC severity.

Discussion

This cohort study of 303 patients with MAC who underwent both echocardiography and contrast-enhanced CT prior to MV surgery yielded several important findings. First, we demonstrate that MAC quantification using contrast-enhanced CT is both feasible and valid, and that the novel contrast-enhanced CT-based MAC score is highly reproducible and can be reliably applied in preoperative imaging. Second, patients undergoing open MV surgery exhibited significantly higher MAC scores, compared to those undergoing robotically assisted surgery. Third, the MAC score was an independent predictor of total operative time, aortic cross-clamp time, ICU length of stay, postoperative complications, and showed a trend towards predicting the need for permanent pacemaker implantation. Finally, greater MAC severity was associated with an increased risk of all-cause mortality. By extending a previously validated MAC scoring system to contrast-enhanced CT,¹⁰ our findings broaden its applicability to a larger population of patients who routinely undergo contrast CT as part of preoperative planning. In addition, contrast enhancement provides complementary anatomic information not available on non-contrast CT, including the spatial relationship of calcification to adjacent structures such as the left circumflex coronary artery, which is highly relevant for operative risk assessment. Collectively, these findings suggest that our contrast-enhanced CT–based MAC score may serve as a broadly applicable tool for preoperative risk stratification in patients with MAC.

MAC is increasingly encountered in patients with MV dysfunction, particularly in the context of ageing populations. Echocardiography remains the first-line imaging modality for detecting MAC and assessing mitral MV.² On transthoracic echocardiography (TTE), MAC appears as an echodense band involving the mitral annulus, often accompanied by acoustic shadowing.¹⁷ Several echocardiographic definitions have been proposed to characterize MAC severity, including qualitative grading based on the circumferential extent of calcification and alternative methods that classify severity according to maximal calcification thickness measured from the anterior to the posterior edge of MAC.^7,8 However, these approaches are limited by the inability of echocardiography to distinguish fibrosis from calcification. Furthermore, such grading systems do not adequately capture the full morphologic spectrum and extent of MAC. The mitral annulus is a complex three-dimensional structure, and MAC may involve only a portion of the annulus or extend into adjacent structures such as the LVOT, myocardium, or fibrous trigones.²

Given these limitations, CT has emerged as an effective, non-invasive imaging modality for the evaluation of MAC. With superior spatial resolution and three-dimensional reconstruction capabilities, CT allows detailed assessment of the distribution, depth, and extent of calcification, as well as involvement of adjacent structures. Prior studies have demonstrated the role of contrast-enhanced CT in MAC characterization and grading, using various semi-quantitative definitions.^13,18–20 However, similar to echocardiographic classifications, many of these methodologies rely on subjective assessments and lack reproducibility.²¹ More recently, Guerrero et al. proposed a scoring system incorporating average calcification thickness, circumferential extent, involvement of one or both trigones, and leaflet involvement, which predicted valve embolization and migration risk during transcatheter MV replacement.¹² Despite its utility, this score was limited to a highly selected transcatheter MV replacement population. Furthermore, it did not account for critical MAC-related features such as LVOT extension or myocardial infiltration, findings that significantly influence surgical planning and outcomes. Additionally, there was no reproducible quantitative assessment of MAC using the Agatston method.

Quantitative calcium assessment using the Agatston method offers several advantages over qualitative and thickness-based approaches. The Agatston score reflects both calcium area and density, correlating strongly with the risk of haemodynamic obstruction and valvular stiffening. For the aortic valve and coronary arteries, Agatston scoring has consistently outperformed qualitative evaluation in predicting haemodynamic consequences.^22,23 Recently, quantitative assessment of MAC using the Agatston method has demonstrated excellent reproducibility and clinical utility based on non-contrast CT.^21,24,25 Building on this principle, we proposed a multiparametric CT-based MAC score incorporating Agatston calcium burden, MAC angle, and morphologic features, demonstrating strong associations with surgical outcomes, and increased long-term mortality in patients undergoing MV surgery.^9,10 However, the Agatston method is originally based on non-contrast enhanced CT,²⁶ and application of Agatston-based MAC quantification utilizing contrast-enhanced CT has not been previously investigated. To our knowledge, this study is the first to quantify MAC calcium burden using contrast-enhanced CT. We demonstrate a high level of agreement between Agatston scores obtained from non-contrast and contrast-enhanced CT and provide a regression-derived formula to convert between the two measurements. Furthermore, we applied the previously validated multiparametric MAC score to a large surgical cohort and confirmed its predictive value for both perioperative outcomes and long-term mortality.

Surgical MV intervention in patients with MAC is complex and presents unique technical challenges that require advanced surgical techniques. Recent American Association for Thoracic Surgery expert consensus guidelines recommend routine preoperative CT imaging in all patients with MAC to adequately assess the extent and distribution of calcification.²⁷ Our study demonstrates the clinical utility of contrast-enhanced CT–based MAC score in procedural planning and risk stratification. Higher MAC scores were strongly associated with longer cross-clamp and total operative times, reflecting the increased technical complexity of calcium debridement and prosthesis implantation in patients with more extensive calcification. Adequate debridement is essential to ensure adequate suture placement and optimal prosthesis seating, and our findings show that the novel MAC score can help predict procedural complexity. Furthermore, patients with higher MAC scores were more likely to undergo open surgical replacement, whereas lower MAC scores were associated with robotic approaches and valve repair. This aligns with current recommendations favouring repair in patients with limited calcification and replacement in those with extensive MAC.²⁷ Higher MAC scores also predicted a greater likelihood of postoperative complications. Although conventional surgical risk scores estimate morbidity and mortality for patients undergoing cardiac surgery, they do not incorporate MAC as an independent risk factor and therefore may underestimate perioperative risk in this high-risk population. Incorporating MAC severity into preoperative assessment may enhance procedural planning, improve patient selection, and ensure that expected outcomes align with operative strategy.

Our study has several limitations. First, it is a single-centre retrospective cohort, which introduces inherent selection bias. Second, our institution is a quaternary referral centre with highly experienced cardiothoracic surgeons and excellent surgical outcomes, which may limit the generalizability of these findings to centres with different case volumes or expertise. In addition, a Cavitron Ultrasonic Surgical Aspirator was not routinely used for surgical management of mitral annular decalcification in our cohort. Furthermore, detailed intraoperative anaesthesia and myocardial protection variables, including the cardioplegia protocol and delivery route, were not collected as part of the original study protocol and therefore could not be evaluated. Consequently, the potential influence of these factors on postoperative outcomes could not be assessed. Third, the validation cohort consisted of a relatively small number of patients (n = 50), which may limit the robustness of the derived thresholds for contrast-enhanced CT–based calcium scoring. Additionally, the validation cohort was part of the overall analysis cohort, and the calcium score cut-offs were not externally validated in an independent population. Therefore, the generalizability of these cut-offs may be limited, and external validation in independent cohorts is warranted. Furthermore, the interval between non-contrast and contrast-enhanced CT scans may have introduced variability, as MAC is a progressive degenerative process and calcification burden could change over time. However, in approximately 75% of patients, both scans were performed within a 5-month interval, mitigating this concern. Additionally, the timing of MACE was not collected as part of the original study protocol, precluding time-to-event analyses. Finally, the modified Agatston method used to quantify MAC in our study may be influenced by technical factors, including contrast attenuation, bolus timing, iodine concentration, aortic region-of-interest placement, and scanner characteristics. Consequently, the derived calcium score values and cut-offs may be protocol-specific, which may limit their generalizability to institutions with different imaging protocols. Additionally, blooming artefacts on contrast-enhanced CT may overestimate the extent of mitral annular calcification, potentially affecting the accuracy of calcium quantification. Nevertheless, the strong agreement observed between contrast-enhanced and non-contrast Agatston scores provides reassurance regarding the reliability of this approach.

In conclusion, MAC quantification using contrast-enhanced CT is feasible and demonstrates strong correlation with measurements obtained from non-contrast CT. The proposed MAC score enables a systematic and reproducible evaluation of MAC severity and is associated with operative complexity, postoperative complications, and long-term outcomes in patients undergoing MV surgery.

Supplementary data

Supplementary data are available at European Heart Journal - Imaging Methods and Practice online.

Author contributions

Mohammad Al Zein (Data curation, Formal analysis, Investigation, Visualization, Writing—original draft [lead]), Yuichiro Okushi (Data curation, Formal analysis, Investigation, Methodology [equal], Writing—original draft, Writing—review & editing [supporting]), Shinya Unai (Investigation, Methodology, Validation [equal], Writing—review & editing [supporting]), Gösta B. Pettersson (Investigation, Methodology, Validation [equal], Project administration, Writing—review & editing [supporting]), A. Marc Gillinov (Investigation, Methodology, Project administration, Validation, Writing—review & editing [supporting]), Per Wierup (Investigation, Methodology, Project administration, Writing—review & editing [supporting]), Richard A. Grimm (Investigation, Methodology [equal], Resources, Writing—review & editing [supporting]), Brian P. Griffin (Investigation, Methodology [supporting], Project administration, Resources, Writing—review & editing [equal]), and Bo Xu (Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Project administration, Resources, Supervision, Validation, Visualization, Writing—original draft, Writing—review & editing [lead])

Funding

We would like to acknowledge the generous support of Mr. David Wright and Ms. Melanie Alder for this important research work.

Data availability

The data reported in this study cannot be shared with external institutions, because the data belong to the Cleveland Clinic.

Lead author biography

Professor Bo Xu is a multimodality imaging cardiologist in the Section of Cardiovascular Imaging, Robert and Suzanne Tomsich Department of Cardiovascular Medicine, Sydell and Arnold Miller Family Heart, Vascular and Thoracic Institute. His research interests include the multimodality imaging assessment of mitral annular calcification and related valvular dysfunction, the role of cardiac imaging in infective endocarditis, and the role of artificial intelligence in cardiovascular imaging.