Disease Activity in Mitral Annular Calcification: A Multimodality Study

Background: Mitral annular calcification (MAC) is associated with cardiovascular events and mitral valve dysfunction. However, the underlying pathophysiology remains incompletely understood. In this prospective longitudinal study, we used a multimodality approach including positron emission tomography, computed tomography, and echocardiography to investigate the pathophysiology of MAC and assess factors associated with disease activity and progression. Methods: A total of 104 patients (age 72±8 years, 30% women) with calcific aortic valve disease, therefore predisposed to MAC, underwent 18F-sodium fluoride (calcification activity) and 18F-Fluorodeoxyglucose (inflammation activity) positron emission tomography, computed tomography calcium scoring, and echocardiography. Sixty patients underwent repeat computed tomography and echocardiography after 2 years. Results: MAC (mitral annular calcium score >0) was present in 35 (33.7%) patients who had increased 18F-fluoride (tissue-to-background ratio, 2.32 [95% CI, 1.81–3.27] versus 1.30 [1.22–1.49]; P<0.001) and 18F-Fluorodeoxyglucose activity (tissue-to-background ratio, 1.44 [1.37–1.58] versus 1.17 [1.12–1.24]; P<0.001) compared with patients without MAC. MAC activity (18F-fluoride uptake) was closely associated with the local calcium score and 18F-Fluorodeoxyglucose uptake, as well as female sex and renal function. Similarly, MAC progression was closely associated with local factors, in particular, baseline MAC. Traditional cardiovascular risk factors and calcification activity in bone or remote atherosclerotic areas were not associated with disease activity nor progression. Conclusions: MAC is characterized by increased local calcification activity and inflammation. Baseline MAC burden was associated with disease activity and the rate of subsequent progression. This suggests a self-perpetuating cycle of calcification and inflammation that may be the target of future therapeutic interventions.

M itral annular calcification (MAC) is a common finding on cardiovascular imaging studies with an estimated prevalence ranging from 8% to 42% depending on age of the population studied and analysis method. 1,2 Often associated with aortic, coronary artery, and aortic valve calcification (AVC), 2 MAC has been linked to increased atherosclerotic burden, 2 incident stroke, 3 and cardiovascular mortality. 4 Although MAC is associated with endothelial damage, lipid infiltration, and progressive valve calcification, [5][6][7] the pathophysiology of MAC remains incompletely understood and medical therapies to halt its progression are lacking. MAC also has functional consequences, helping to drive progressive mitral stenosis and mitral regurgitation, the severe stages of which can only be remedied through surgical or, potentially, percutaneous intervention. 8,9 Several epidemiological studies have investigated risk factors for MAC, finding similar determinants as for calcific aortic valve disease, including age, obesity, smoking, and serum phosphate. 10,11 Important differences have also been observed, with MAC showing female predominance 7 and a stronger association with chronic kidney disease and dysregulated mineral metabolism. 12,13 An association with low bone mineral density (BMD) has been suggested but remains unproven. 14 Despite the various investigations into risk factors for MAC incidence and prevalence, no studies to date have evaluated risk factors governing disease activity.
Hybrid positron emission and computed tomography (PET-CT) allows for the simultaneous noninvasive evaluation of disease activity and heart valve anatomy. 15,16 CT provides a detailed assessment of calcium burden, and PET can measure the activity of specific disease processes dependent on the availability of suitable tracers. 18 F-Sodium fluoride ( 18 F-fluoride) is a marker of vascular and valvular calcification activity 17 that has been used to investigate vascular atherosclerosis and aortic stenosis. 15,18 18 F-Fluorodeoxyglucose ( 18 F-FDG) has been used to measure vascular inflammation because of its accumulation within tissue macrophages. 19 In this clinical imaging study, our aim was to use a state-of-the-art multimodality imaging approach to investigate activity and inflammation to elucidate factors associated with disease prevalence, activity, and progression.

METHODS
Patients aged >50 years with calcific aortic valve disease were recruited as previously described and formed the study cohort. 15 All had AVC on CT and were, therefore, deemed prone to developing calcific valve disease. Exclusion criteria included insulin-dependent diabetes mellitus, blood glucose >200 mg/dL, end-stage renal disease, metastatic malignancy, and life expectancy <2 years. The study cohort was divided into patients who did (score>0) or did not (score=0) have MAC on CT to assess factors associated with MAC prevalence. Clinical data were ascertained based on detailed history and clinical examination. Fasting serum biomarkers were analyzed as previously described. Lipoprotein(a) was measured using chemiluminescent immunoassays as previously described. 20 A detailed echocardiographic exam was performed under standardized conditions according to a formal protocol as previously described. 15 The Institutional Review Board of the University of Edinburgh approved the protocol, and participants provided written informed consent. Study data can be made available to other researchers on request to the corresponding author.
A control cohort without evidence of heart valve calcification (CT calcium score of 0 in mitral annulus and aortic valve) was also included to determine the normal range of 18 F-fluoride PET uptake in the mitral annulus, with the highest 18 F-fluoride tissue-to-background ratio (TBR max ) values defining the upper limit of normal and differentiating between study cohort patients who did (PET+) and did not have (PET−) increased PET activity.

CLINICAL PERSPECTIVE
Mitral annular calcification (MAC) is associated with cardiovascular events and mitral valve dysfunction, but its pathophysiology is incompletely understood. We employed a multimodality imaging approach including 18 F-fluorodeoxyglucose and 18 F-sodium fluoride positron emission tomography, computed tomography calcium scoring, and echocardiography to investigate the pathology of MAC and elucidate the factors associated with its prevalence, disease activity, and disease progression. Patients who had MAC (34% of patients) had increased inflammatory and calcification activity by positron emission tomography imaging in the mitral annulus. Furthermore, calcification activity was most closely associated with computed tomography-MAC calcium score, inflammation, female sex, and renal dysfunction. Similarly, MAC progression on repeat computed tomography scans after 2 years was closely associated with baseline MAC, with the fastest rate of progression found in those with high baseline computed tomography-MAC scores and the highest calcification activity. By contrast, traditional cardiovascular risk factors and calcification activity in bone or remote atherosclerotic areas were not associated with disease activity nor progression. This suggests that MAC activity and progression are characterized by a vicious cycle of established calcium, injury and inflammation within the valve that prompts further calcification activity. These findings support the concept that therapeutic strategies targeting MAC will need to focus on breaking this vicious calcification cycle.

PET-CT Imaging
PET-CT scans of the heart and aorta were performed with a hybrid scanner (Biograph mCT, Siemens Medical Systems, Erlangen, Germany). Two scans were performed at least 24 hours apart, 60 minutes after administration of 18 F-fluoride 125 MBq and 90 minutes after 18 F-FDG 200 MBq. ECGgating was not used, and all counts were used for analysis. All patients were asked to adhere to a carbohydrate-free diet for 24 hours preceding their 18 F-FDG scan to suppress myocardial uptake, as previously described. 15 Patients were given a list of foods (high in fat and low in carbohydrate) to eat and also those to avoid. An ECG-gated breath-hold CT scan (noncontrast-enhanced, 40 mA/rot [CareDose], 100 kV) of the heart was performed for calcium scoring.

Image Analysis: CT
Mitral annulus, aortic valve, coronary artery, and aortic CT calcium scores were determined using dedicated analysis software (VScore, Vital Images, Minnetonka, and OsiriX Lite version 8.5.1, OsiriX Imaging Software, Geneva, Switzerland). Agatston scores were calculated using a threshold of 130 Hounsfield units. 21 MAC on (CT-MAC) was defined as calcium score >0 Agatston units (AU) in the mitral annulus.

Image Analysis: PET
Mitral annular 18 F-fluoride and 18 F-FDG PET activity were quantified according to a standardized protocol using OsiriX. Regions of interest were drawn around maximal areas of 18 F-fluoride and 18 F-FDG activity to obtain the maximum standardized uptake values (SUV max ), which were divided by blood pool uptake values in the right atrium (2 cm 2 area) to obtain TBR max values. Given the difficulty in determining the exact borders of the mitral annulus, SUV mean , and TBR mean values were not quantified.
Uptake of 18 F-fluoride and 18 F-FDG in the aortic valve, aorta, and coronary arteries was measured as previously reported (Data Supplement). 15 BMD and 18 F-fluoride bone uptake were measured in 4 thoracic vertebrae as detailed previously. 18 Briefly, 0.5 cm 2 regions of interest were drawn within the cancellous bone. The average Hounsfield unit density within those regions was used as a relative measure of BMD. 22 Maximum 18 F-fluoride SUV values were quantified in the same regions of interest. Myocardial 18 F-FDG uptake was assessed by recording the maximum SUV in the left ventricular septum. A diffuse pattern of myocardial 18 F-FDG uptake accompanied by SUV ≥5.0 indicated failed myocardial suppression. 15 Patients with failed suppression were excluded from the analysis of FDG data, but not from analysis of 18 F-fluoride data.

Repeatability Studies
All CT and PET quantifications were independently performed in a blinded fashion by 2 trained observers (M.G. Trivieri and D. Massera). Disagreements were resolved by consensus with involvement of a third observer (R. Abgral).

Image Analysis: Echocardiography
Examination of the mitral valve apparatus was performed in a blinded fashion by one cardiologist (J. Andrews). At least 3 diastolic transmitral continuous-wave Doppler envelopes were traced to obtain an average diastolic transmitral gradient. Mitral regurgitation severity was assessed according to the American Society of Echocardiography guidelines. 23 No adjustment for heart rate was performed because 87% of patients had a heart rate of <80 bpm.

Disease Progression Studies
A subset of study participants underwent repeat CT and echocardiography using the same protocol and equipment 2 years after initial imaging. Mitral annular disease progression was assessed using the annualized change in CT calcium score and transmitral pressure gradient.

Statistical Analysis
Continuous variables are reported as mean±SD or median (interquartile range [IQR]) and were compared with the unpaired Student t test, Wilcoxon rank-sum or Kruskal-Wallis tests, as appropriate. Categorical variables are reported as proportions and analyzed with the χ 2 or Fishers exact test. Correlations were calculated using Spearman correlation coefficients. Data are presented by presence or absence of CT-MAC or mitral annular 18 F-fluoride activity or were dichotomized at the median CT-MAC calcium score. Bland-Altman mean differences and limits of agreement were obtained. Intraclass correlation coefficients were calculated with 2-way mixed-effects models. Multivariable linear and logistic regression models were used to identify predictors of MAC prevalence and 18 F-fluoride activity. Logarithmic transformation of 18 F-fluoride uptake was performed to achieve a normal distribution. Initially, all variables with P<0.2 in bivariate comparisons were included in the model, as well as important cardiovascular risk factors (age, sex, hypertension, diabetes mellitus, smoking, low-density lipoprotein cholesterol, and prior cardiovascular disease). Subsequently, a backward stepwise selection process was used with age and sex forced into the model. Separately, 18 F-FDG TBR max was added to the model to identify FDG as a predictor of 18 F-fluoride uptake. Multiple linear and multinomial logistic regression models were used to identify predictors of MAC progression. All analyses were performed with STATA 14.2 (StataCorp LP, College Station, TX). A 2-tailed P<0.05 was used to define statistical significance.

Patient Population
The study cohort comprised 104 patients (mean age 72±8 years, 30% women; baseline characteristics are presented in Tables 1 and 2). The median transmitral mean diastolic pressure gradient was 1.4 (IQR, 1.0-2.1) mm Hg (Data Supplement). In addition, a control cohort of 17 subjects without heart valve calcification was included (68±8 years; Data Supplement). The effective radiation dose per patient was 9.7±1.2 mSv (CT conversion factor 0.014 mSv/mGy/cm

Factors Associated with MAC Prevalence
The median baseline MAC-CT calcium score was 0 (IQR, 0-316) AU and was higher in women (283 [0-1082] AU) compared with men (0 [0-0] AU; P=0.001). Overall, 35 (33.7%) patients had MAC on CT (CT+; 837 [300-2129] AU), who were older, twice as likely to be female, had more AVC, lower BMD, and reduced estimated glomerular filtration rate (eGFR) compared with patients without MAC (CT−). Both groups had extensive cardiovascular disease risk factor burden ( Table 1). In a multiple logistic regression model, female sex and AVC calcium score were statistically significantly associated with MAC prevalence (Table 3).

Mitral Annular Inflammatory Activity ( 18 F-FDG PET)
Thirty-three patients (32%) met criteria for failed myocardial suppression of physiological 18 F-FDG uptake and were excluded from further analysis of FDG data only. In the remaining patients, median mitral annular 18

MAC Activity ( 18 F-Fluoride PET)
Median mitral annular 18  between mitral annular 18 F-fluoride activity and baseline CT-MAC score (r=0.79, P<0.001; Figure 1A) while a moderate correlation was observed with 18 F-FDG uptake (r=0.32, P=0.001; Figure 1B). By comparison, modest or no correlations were observed between mitral annular 18 F-fluoride uptake and uptake in other areas (aorta, r=0. 23 (Figure 1C). Compared with PET− patients, PET+ patients were older, more likely to be female, had more AVC, lower BMD, and eGFR (Table 1). In a multiple linear regression model, CT-MAC and AVC calcium scores, female sex, and eGFR demonstrated a statistically significant association with MAC disease activity. When 18 F-FDG TBR max was added to the model, significant predictors of MAC 18 F-fluoride activity were baseline CT-MAC and 18 F-FDG TBR max in the subset of patients with successful myocardial suppression (Table 5).

Disease Progression in Mitral Annular Calcification
Sixty patients in the study cohort underwent repeat echocardiography and CT after a median of 741 (IQR, 726-751) days (Figure 2 includes examples of 3 patients). The annual progression rate of CT-MAC calcium score was 2 (0-166) AU per year. The strongest associations of MAC progression were observed with baseline CT-MAC (r=0.82, P<0.001; Figure 3A), 18 F-fluoride (r=0.75, P<0.001; Figure 3B) and 18 F-FDG activity (r=0.48; P<0.002). Women tended to have a higher rate of MAC progression (34 [0-409] AU/y) than men (0 [0-68] AU/y; P=0.083). There was no association between baseline eGFR and MAC progression (r=−0.13; P=0.308) nor differences in the rate of MAC progression between those with and without advanced   Figure 3C).

DISCUSSION
We used state-of-the-art multimodality imaging to investigate MAC, providing novel insights into the pathophysiology of this common condition and factors associated with its prevalence, disease activity, and progression. We confirmed that MAC is characterized by both calcification and inflammatory activity that increases proportionally to the baseline MAC burden. Importantly, while female sex, renal dysfunction, and local inflammatory activity were associated with MAC disease activity, the strongest correlate was the local burden of calcium already present within the valve annulus. Similar observations were made with respect to progression, with the fastest progression observed in patients with the largest baseline burden of MAC. We, therefore, suggest that once established, MAC activity and progression are characterized by a vicious cycle of established calcium, injury, and inflammation within the valve that prompts further calcification activity. These findings support the concept that therapeutic strategies targeting MAC will need focus on breaking this vicious calcification cycle.
Despite its high prevalence, contribution to mitral valve dysfunction and adverse prognosis, 4 the pathobiology of MAC remains incompletely understood. Moreover, therapeutic options are limited since effective medical therapy is lacking and surgical intervention is made complicated by its presence. 24 There is, there- fore, an urgent need to illuminate the pathophysiology underlying MAC and to identify novel therapeutic strategies to prevent its clinical sequelae. 9 We describe a new multimodality imaging approach to help address this need. First, we have applied CT calcium scoring to define the presence of MAC and to quantify disease prevalence, burden, and progression. Second, we used 18 F-FDG to measure inflammatory activity. Although 18 F-FDG was only interpretable in two-thirds of patients, our data clearly demonstrate that MAC is an inflammatory condition with the 18 F-FDG PET signal increasing in proportion to baseline disease severity. Finally, we used 18 F-fluoride PET as marker of calcification activity demonstrating a close association with subsequent progression and building upon a growing body of literature using 18 F-fluoride to image developing cardiovascular microcalcification. The use of a cohort of patients with calcific aortic valve disease provided a patient population at high risk of developing MAC, as evidenced by the particularly high prevalence. This gave us the opportunity to assess disease activity and progression in patients with established MAC, but also in patients who subsequently developed MAC during follow-up. It also provided insights into why certain patients with aortic stenosis develop MAC, while others do not, with female sex, renal impairment, and advanced AVC appearing to be of particular importance in this population.

Factors Associated With Disease Activity in MAC
Using 18 F-fluoride PET, we demonstrated that calcification activity in the mitral annulus is closely related to the local inflammatory signal provided by 18 F-FDG imaging. This is consistent with histological studies of excised mitral valves demonstrating increased expression of pro-calcific cells and mediators adjacent to T-lymphocytic infiltrates and suggests that calcium deposition is closely related to inflammatory activity. 5,6 However, MAC activity was, in fact, most closely associated with the baseline CT-MAC calcium score. Similar results were observed for progression: patients with rapid disease progression and highest disease activity were those with the highest baseline CT calcium scores. Indeed, baseline MAC was the strongest predictor of MAC progression, here replicating the findings from the Multiethnic Study of Atherosclerosis. 7 We believe our concordant data on MAC disease activity and progression have important therapeutic implications. The findings are remarkably similar to observations made in aortic stenosis, where it has been suggested that calcium within the valve increases mechanical stress and injury leading to inflammation and increased calcification activity. 25 A similar self-perpetuating cycle of calcification inducing further calcification  might also underlie MAC. The development of effective medical therapy in both conditions is, therefore, likely to require strategies that interrupt this cycle without impacting bone health. Studies are currently underway testing such therapies in patients with aortic stenosis (SALTIRE2, NCT02132026) providing an opportunity to investigate their impact on bystander MAC.

Study Limitations
Our study cohort comprised participants with calcific aortic valve disease. Although this ensured high proportions of prevalent and incident MAC, our results may not directly apply to patients with isolated mitral valve disease or other conditions known to be associated with MAC. Moreover, our sample size was modest, precluding more detailed examination of determinants and consequences of microcalcification and inflammation. In addition, one-third of patients met criteria for failed myocardial FDG suppression and were excluded from the analysis of FDG data. Further studies exploring the role of PET-CT in larger samples and different patient populations are warranted. Such studies may benefit from the use of contrast CT to better investigate the spatial distribution of PET uptake within the mitral annulus and to improve interobserver reproducibility. In addition, advanced imaging processing technologies such as adaptive thresholding may improve uptake delineation, and ECG-gating of the PET acquisition may reduce image blurring because of cardiac motion.