Gla-Rich Protein, Magnesium and Phosphate Associate with Mitral and Aortic Valves Calcification in Diabetic Patients with Moderate CKD

Accelerated and premature cardiovascular calcification is a hallmark of chronic kidney disease (CKD) patients. Valvular calcification (VC) is a critical indicator of cardiovascular disease and all-cause mortality in this population, lacking validated biomarkers for early diagnosis. Gla-rich protein (GRP) is a cardiovascular calcification inhibitor recently associated with vascular calcification, pulse pressure, mineral metabolism markers and kidney function. Here, we examined the association between GRP serum levels and mitral and aortic valves calcification in a cohort of 80 diabetic patients with CKD stages 2–4. Mitral and aortic valves calcification were detected in 36.2% and 34.4% of the patients and associated with lower GRP levels, even after adjustments for age and gender. In this pilot study, univariate, multivariate and Poisson regression analysis, show that low levels of GRP and magnesium (Mg), and high levels of phosphate (P) are associated with mitral and aortic valves calcification. Receiver operating characteristic (ROC) curves showed that the area under the curve (AUC) values of GRP for mitral (0.762) and aortic (0.802) valves calcification were higher than those of Mg and P. These results suggest that low levels of GRP and Mg, and high levels of P, are independent and cumulative risk factors for VC in this population; the GRP diagnostic value might be potentially useful in cardiovascular risk assessment.


Introduction
Chronic kidney disease (CKD), diabetes mellitus, and atherosclerosis are the clinical conditions that most contribute towards the development of cardiovascular calcification [1,2], which is a strong predictor of cardiovascular risk, while cardiovascular disease (CVD) is the most common cause of death in CKD patients [3]. Cardiovascular calcification can occur at different sites within the vascular tree, including the vessel wall at the media or intimal layers, and heart valves, with different impacts on cardiovascular outcomes. Intimal calcification reflects atherosclerotic plaque burden, may influence plaque rupture,

Evaluation of Mitral and Aortic Valve Calcification
The assessment of valve calcification was performed using echocardiography, following the KDIGO guidelines. Accordingly, the Work Group suggests that in patients with CKD (G3a-G5D), an echocardiogram can be used to detect the presence or absence of VC as a reasonable alternative to CT-based imaging [30]. A good correlation between echocardiographic measurements of valve area and the valvular calcium Agatston score was previously demonstrated [31]. Echocardiographic evaluation was obtained using standard M mode and two-dimensional images (Vivid 7 Dimension-GE Healthcare Ultrasound; GE Healthcare, Waukesha, WI, USA). Offline analysis was obtained using the workstation Echopac PC'08 version 7.0.0 GE Vingmed Ultrasound (GE Healthcare), and the measurements were consistently obtained by the same physician. Images were digitally stored and analyzed by two independent experienced cardiologists. Quality control procedures included blind rereading and patient reexamination to allow assessment of intra-reader variability, inter-reader variability, and intra-patient variability. An average of three measurements was used for each variable. The subjects were categorized according to the presence of valvular calcification. The evaluation of the aortic valve calcification was performed using a semi-quantitative assessment as proposed by Lullo et al. [32]: non-calcified = score 1 (partial calcification on single cusp) and score 2 (partial calcification on two cusps); calcified = score 3 (extended calcification on two cusps) and score 4 (extended calcification on all three cusps). The extent of mitral valve calcification was determined according to the Wilkins calcification scores as: non-calcified = grade 1 and 2; calcified = grade 3 and 4 [33].

Statistical Analysis
Descriptive results were presented using mean and standard deviation (±SD) for continuous variables with normal distribution, using the Kolmogorov-Smirnov test. Twosample t-tests were used to assess differences in subgroups defined by mitral valve calcified/non calcified and aortic valve calcified/non calcified for continuous measures. Subjects were split into two groups according to the median value of serum GRP: GRP ≥ 0.9 ng/mL and GRP < 0.9 ng/mL, to determine the percentage of calcified/non calcified mitral valve and calcified/non calcified aortic valve. Chi-square tests were used to test for association between mitral valve calcified/non calcified and aortic valve calcified/non calcified. Partial correlations were used to analyse relationships between aortic and mitral valve calcification with GRP and renal function (eGRF), adjusted by sex and age groups. Univariate logistic regression analysis was used to identify independent factors associated with aortic and mitral valve calcification. Statistically significant variables were analysed in multivariate logistic regression models to assess the main predictive risk factors for aortic and mitral valve calcification. Potential confounding factors offered to the logistic regression models included age, gender, eGFR, FGF-23, TNFα, P, calcium x phosphate (CaxP), α-Klotho and GRP. The exponentials of the model parameters were the adjusted odds ratio (ORa) to other variables of the model, with 95% confidence interval. A modified Poisson regression with robust error variance estimation to calculate adjusted prevalence ratios (aPR) was used to estimate the cumulative relative risk of aortic and mitral valve calcification. Variables included in these multivariate analysis were age, eGRF, TNFα, calcium, phosphorus, CaxP, PTH, GRP, 1.25(OH)2D3 vitamin D, FGF-23 and α-Klotho. The receiver operating characteristic curves (ROC) were constructed to assess the sensitivity and specificity of risk factors for valvular calcifications. The median of risk factors was used to determine the best cut-off for the ROC curve. The null hypothesis was rejected below the level of 5%. Statistical analysis was performed with SPSS (version 17.0).

Results
This study enrolled 80 consenting diabetic patients meeting the inclusion criteria with stage 2-4 CKD (stage 2, n = 23; stage 3, n = 39; stage 4, n = 18), 28.7% females, mean age of 56 ± 8.1 years (range: 41-65). All variables had a normal distribution. The mean GRP levels was 0.9 ± 0.56 ng/mL (range, 0.19-2.6 ng/mL). The subjects were classified into groups according to the non-calcified or calcified valves (mitral valve calcified/non calcified and aortic valve calcified/non calcified). Table 1 describes the patients' main clinical and biochemical characteristics, as a function of mitral and aortic valves calcification, including osteo-mineral markers and known risk factors for CVD. In total, 36.2% of patients presented mitral valves calcification (65.5% (n = 19) males and 34.4% (n = 10) females), and 36.2% had aortic valves calcification (72.4% (n = 21) males and 27.6% (n = 8) females). The groups with calcified mitral and aortic valves displayed significantly lower levels of eGFR (p < 0.0001 for both groups), GRP (p < 0.0001 for both groups), Mg (p = 0.029 and p = 0.001, respectively) and α-Klotho (p = 0.002 for both groups), while higher levels of P (p = 0.001 for both groups), PTH (p = 0.025 and p = 0.030, respectively), FGF-23 (p < 0.0001 for both groups) and TNFα (p = 0.037 for both groups), as compared to the respective noncalcified groups. No differences were found between groups regarding age, gender (f-m), hemoglobin, albumin, ACR, duration of disease, HgA1c, Ca, CaxP, 1.25(OH)2 Vitamin D and BP (Table 1). The association between levels of GRP and aortic and mitral valves calcification using the chi-square test showed that higher levels of GRP were associated with a significant higher percentage of patients without aortic (96.7%) and mitral (87%) valves calcification, and to a lower percentage of patients with aortic (3.3%) and mitral (13%) valves calcification (Figure 1a,b, respectively). The association between levels of GRP and aortic and mitral valves calcification using the chi-square test showed that higher levels of GRP were associated with a significant higher percentage of patients without aortic (96.7%) and mitral (87%) valves calcification, and to a lower percentage of patients with aortic (3.3%) and mitral (13%) valves calcification ( Figure 1a,b, respectively). Percentage of patients with calcified and non-calcified aortic (a) and mitral (b) valves, across the median GRP serum levels (<0.9; ≥0.9 ng/mL); the chi-square test was used to evaluate this association (** p < 0.0001). (a,b) Percentage of patients with calcified and non-calcified aortic (a) and mitral (b) valves, across the median GRP serum levels (<0.9; ≥0.9 ng/mL); the chi-square test was used to evaluate this association (** p < 0.0001).
The relationships between GRP serum levels and Mg and TNFα, which were not previously reported for this cohort, were assessed using the simple linear regression. GRP serum levels were positively associated with levels of Mg (β = −0.466; p < 0.0001) and inversely associated with TNFα (β = −0.243; p = 0.03) (Figure 2a,b, respectively).

Discussion
In this work we show that decreased levels of serum GRP are strongly associated with increased risk of aortic and mitral valve calcification in a population of adult diabetic patients with mild to moderate CKD. In addition, decreased levels of Mg and increased levels of P were also found independently associated with increased risk of both aortic and mitral valve calcification. Moreover, the simultaneous occurrence of decreased GRP and Mg and increased P levels are cumulative risk factors for valvular calcification (VC) in this population. Between these three factors, GRP showed the highest diagnostic value for aortic and mitral valves calcification. To our knowledge, this is the first clinical study showing an association between circulating GRP levels and valvular calcification. We recently reported that low levels of GRP were strongly associated with increased vascular calcification, pulse pressure and increased levels of the calcification promotors P, FGF-23 and CaxP, in this same patient cohort [24]. Here we show that GRP serum levels are positively associated with Mg, a known vascular calcification inhibitor [34,35]. The importance of P and Mg in vascular calcification and cardiovascular risk have been extensively demonstrated at both epidemiological and mechanistic levels, both in the general population and CKD, with hyperphosphatemia and hypomagnesia associated with increased vascular calcification, cardiovascular events and mortality [34][35][36][37][38][39]. In relation to VC, hyperphosphataemia has been associated with aortic and mitral valves calcification, including in moderate CKD patients [12,[40][41][42][43]. Low serum Mg was found associated with the prevalence and incidence of aortic valve calcification in an adult population without known CVD and CKD [44]. In moderate CKD, low serum Mg was associated with abdominal aortic calcification, increased pulse pressure, mitral valve calcification and increased intima media thickness [45][46][47]. The findings that decreased levels of GRP and Mg and increased levels of P are independent and cumulative risk factors of VC, are in line with the current knowledge of their involvement on the pathophysiological mechanisms of cardiovascular calcification, and reinforce the crucial role of bone mineral metabolism in cardiovascular risk. Importantly, both GRP and Mg have been shown to inhibit high P-induced calcification of vascular smooth muscle cells (VSMCs), suggesting a tight and important relationship between these three factors. Mg status was shown to modify the risk of P-induced progression to end-stage kidney disease [48], also suggesting decreased levels of Mg and increased levels of P as cumulative risk factors in CKD progression.
Mechanistically, P is known to function as a primary stimulus for the osteochondrogenic differentiation of vascular and valvular cells [38,49]. Conversely, both GRP and Mg function as calcification inhibitors through several, and interestingly, coincident molecular processes, such as the inhibition of VSMCs osteochondrogenic differentiation through down-regulation of osteogenic and up-regulation of contractile markers, the inhibition of calciprotein particles (CPP) maturation, and capacity to act as anti-inflammatory agents [21][22][23]34,35,50,51].
Overall, these data clearly indicate GRP and Mg as essential protective factors in the calcification milieu, providing the biological rational for decreased levels of GRP and Mg as independent risk factors of VC. Additionally, although a relation between GRP and Mg at the cell level and possible interactions and/or intersections in their mechanisms of action are currently unknown, our findings that low levels of GRP and Mg and high levels of P are cumulative risk factors for VC, together with their simultaneous involvement in common cardiovascular calcification molecular processes, suggest a synergistic effect affecting cardiovascular calcification outcomes.
Our results showing a strong association between GRP and eGFR, as previously demonstrated [24], and with mitral and aortic VC, but a weaker association between eGFR and VC, suggest that GRP is involved in the pathophysiology of VC regardless of the stage of renal disease. This is in line with recent results showing an association between GRP and coronary artery calcification (CAC) in non-CKD patients with atrial fibrillation and heart failure [52].
Of note, therapies aiming to counteract hyperphosphatemia using phosphate binders have shown limited results in the suppression of CAC [53], while more recent clinical trials demonstrated that magnesium oxide is able to slow progression of CAC but not of thoracic aorta calcification [54]. This reinforces the notion that despite general assumption that mechanisms of calcification are similar within the vascular tree, specific factors can differ-ently drive and affect calcification depending on the location, type of cells and environment, and that additional efforts are required to discover novel biomarkers and interventional agents to fight cardiovascular calcification. In calcified aortic valve disease, GRP was found highly accumulated at sites of mineral deposition and foam cells, and suggested to be associated with osteoblast-like VICs after myofibroblast-VIC differentiation [21], indicating a specific GRP action in VC.
It should be noted that in our study, although intact FGF23 was found as an independent risk factor of both aortic and mitral valves calcification, it was not identified as a cumulative risk factor for valves calcification. Many epidemiological studies have demonstrated that elevated levels of FGF23 are associated with renal function declining, increased cardiovascular morbidity and mortality, and higher aortic and coronary calcification scores [55][56][57][58]. However, a recent meta-analysis study suggests that there is no causal relation between FGF23 levels and cardiovascular risk, although this is still under investigation [59]. An important issue concerning the determination of FGF23 levels relates with the lack of consensus on a gold-standard method for dosing this hormone. FGF23 circulates as a full-length or intact protein (iFGF23), which constitutes the biologically active molecule, and as carboxi-terminal fragments (cFGF-23), whose activity remains controversial. ELISAs are available for both iFGF23 and cFGF23, but most of the published studies have only measured one of these forms. In some studies comparing iFGF23 and cFGF23, cFGF23 has been suggested as more sensitive in detecting eGFR decline in a non-CKD population [60], and increased risk of overall graft loss in kidney transplant recipients [61]. In addition, cFGF23 was suggested to mediate the association between iron deficiency and mortality in renal transplant recipients [62], and associated with red cell distribution width in CKD patients with heart failure [63]. In patients with atherosclerotic cardiovascular disease, levels of both iFGF23 and cFGF23 were associated with vascular calcification [64]. Clearly, additional comparative studies are required to further elucidate the different performances of iFGF23 and cFGF23 within specific populations and specific clinical features.
Major limitations of our study include the small sample size, requiring additional and larger studies to strengthen the evidence on the clinical relevance of GRP in VC, measurements of serum GRP at a single point, and the absence of reference intervals for GRP levels in a healthy population. Additionally, this study included subjects with mild to moderate CKD from a single centre, and may not be representative of kidney disease of other etiologies.

Conclusions
This study shows that decreased circulating levels of GRP and Mg and increased levels of P are independent and cumulative risk factors for mitral and aortic valves calcification in diabetic patients with moderate CKD. The high diagnostic value found for GRP in VC, together with previous knowledge that levels of GRP are also associated with vascular calcification and bone mineral metabolism [25], suggest GRP as a novel marker for cardiovascular calcification of potential clinical utility for cardiovascular risk assessment, calling for additional molecular and clinical research.

Patents
The tools and methods described in this manuscript are included in a Patent Cooperation Treaty (PCT) patent application PCT/PT2009000046.  Informed Consent Statement: Informed consent was obtained from all subjects involved in the study.
Data Availability Statement: Data available on request due to privacy restrictions. The data presented in this study are not available, because they are in the process of analysis for results publication.