Severe Coronary Artery Calcifications in Chronic Kidney Disease Patients, Coupled with Inflammation and Bone Mineral Disease Derangement, Promote Major Adverse Cardiovascular Events through Vascular Remodeling

Cardiovascular (CV) diseases remain the leading cause of morbidity/mortality in patients with chronic kidney disease (CKD) [1]. Exposure to uremia-specific factors including inflammation, oxidative stress, endothelial dysfunction, disturbances in mineral metabolism, and calcification of arteries and heart valves may all trigger the onset and development of atherosclerosis and associated CV diseases, which may partly explain the increased CV mortality observed in these patients [1]. Therefore, identifying new pathways as well as timing of occurrence appear are crucial steps toward the development of potentially more effective therapies.

Vascular calcification (VC) is an active process akin to bone formation and results not only from an elevation in calcium/phosphorus serum levels but also from the dysregulated induction of osteogenesis. The differentiation of vascular smooth muscle cells into osteoblast-like cells [2] results in a bone matrix synthesis. Then, the loss of mineralization inhibitors, which could be estimated using the calcification propensity score, leads to VCs. Active inducers of VC in CKD mainly include accumulation of uremic toxins, mineral disorders (such as hypercalcemia, high inorganic phosphate and intact parathyroid hormone [iPTH] levels, low vitamin D₃, fibroblast growth factor-23 [FGF-23]) [3], and inflammation [4, 5]. Among mineralization inhibitors, pyrophosphates, α-Klotho, and active matrix gla protein (MGP) are the most studied molecules [6‒8]. Of note, systems controlling bone metabolism including Wnt signaling pathway inhibitors such as sclerostin (SOST) and the osteoprotegerin (OPG)/RANK/RANK ligand complex also constitute regulators of central significance in VC [9, 10].

Interestingly, growth differentiation factor-15 (GDF-15), a member of the transforming growth factor-β (TGF-β) superfamily, constitutes another molecule of interest. Indeed, correlations of GDF-15 with CV and non-CV mortality have been documented, and a primary role in multiple CV diseases, including heart failure, cardiac hypertrophy, and coronary heart disease, has been ascribed to it [11‒13]. Regarding its implication in VC, only indirect evidence is currently available with experimental studies reporting promotion of osteoclast and inhibition of osteoblast differentiation [14, 15]. Therefore, the present study aimed to explore the potential additive values of a detrimental environment (combining a micro-inflammatory state, disturbances in mineral metabolism, and arterial calcifications) on major adverse CV events (MACE)/CV mortality in a prospective cohort of non-dialysis CKD (ND-CKD) patients in order to develop targeted therapies.

The study was conducted according to the principles of the Declaration of Helsinki and in compliance with International Conference on Harmonization/Good Clinical Practice regulations. According to the French Law, this study (ClinicalTrials.gov Identifier: NCT00608998 and NCT02808572) has been registered at “Ministère de la Santé et des Solidarités” after approval by the Nîmes University Hospital’s Ethics Committee. All patients gave their written informed consent.

ND-CKD patients at various stages of kidney disease, issued from the outpatient general nephrology consultation of the Montpellier Lapeyronie university hospital, were enrolled in this prospective observational study. Inclusion criteria were age ≥18 years and the presence of CKD defined, in agreement with the National Kidney Foundation [16], as either kidney damage or glomerular filtration rate (GFR) <60 mL/min/1.73 m² for ≥3 months. Kidney damage was defined as pathologic abnormalities or markers of damage, including abnormalities in blood or urine tests or imaging studies. Detailed medical history including age, gender, weight, height, cause of CKD, diabetes mellitus, hypertension, past or current smoking, the presence of CV disease was recorded (see the online suppl. Materials; for all online suppl. material, see https://doi.org/10.1159/000542418).

At inclusion, blood samples were collected as part of our regular CKD patient management plan, centrifuged, and supernatant was stored at −80°C until measurement of the following parameters: creatinine, calcium, phosphate, 25(OH) vitamin D₃, iPTH, albumin, high-sensitive C reactive protein (hsCRP), bone ALP, dephosphorylated uncarboxylated (dp-uc) MGP, total uncarboxylated (t-uc) MGP, tartrate-resistant acid phosphatase 5b (TRAP5b), OPG, SOST, FGF-23, α-Klotho, and GDF-15. Assays were sequentially done with less than 1-year intervals after freezing.

Evaluation of routine biological parameters including creatinine, calcium, phosphate, 25(OH) vitamin D₃, iPTH, albumin, and hsCRP was performed on a cobas 8000 analyzer (Roche, Meylan, France). The GFR was estimated using the CKD-EPI equation. Bone ALP, dp-ucMGP, t-ucMGP, and TRAP5b were assessed using automated IDS-iSYS methods (Immunodiagnostic Systems, Boldon, UK). OPG, SOST, FGF-23 (both intact and C-terminal fragments), and α-Klotho were determined by enzyme-linked immunosorbent assay (OPG: Biovendor Laboratory Medicine, Brno, Czech Republic; SOST: TECOmedical Quidel Corporation, San Diego, CA, USA; FGF-23: Immutopics International Inc, San Clemente, CA, USA; soluble α-Klotho: IBL, Minneapolis, MN, USA). GDF-15 was measured by electrochemiluminescence immunoassay (Elecsys GDF-15, Roche Diagnostics, Mannheim) on a cobas analyzer.

At inclusion, patients underwent multidetector computed tomography. Regarding data acquisition, all multidetector computed tomography scans derived from a multidetector-row spiral CT (Lightspeed VCT, General Electric Medical System, Milwaukee, WI, USA). Prospective ECG-triggered step-scan was performed using with 2.5-mm collimation width ×64 detectors so that the center of the temporal window corresponded to 70% of the R-R interval. The scanning parameters were a gantry rotation speed of 0.35 s/rotation, 120 kV, and 300 mA 8X2.5 collimation and 20 mm table feed per rotation. The matrix size was 512 × 512 pixels, and the display field of view was 25 cm. The reconstruction kernel was standard. The temporal resolution was 250 ms.

Regarding image evaluation, the calcium score was calculated using a semiautomatic software (Smartscore version 3.5, Advantage Window 4.4 workstation, General Electric, Milwaukee, WI, USA). Coronary calcification was defined as a plaque of ≥4 pixels (area = 1.37 mm²) with a density of ≥130 Hounsfield units. Quantitative calcium scores were calculated according to the method described by Agatston et al. [17]. Coronary calcium scoring was performed by either a physician or computed tomography technician with specific training for the methodology described above. The presence of severe coronary artery calcification (CAC) was defined by a CAC ≥300 and an insignificant or moderate CAC by a CAC <300 [18].

Patients were included between 2004 and 2018 and were followed up to the first MACE, CV death, transplantation, dialysis initiation or were censored at 26th April 2018, whichever came first. Deaths were recorded using medical charts. Components of MACE included acute coronary syndrome or ischemic heart disease; coronary revascularization procedures (angioplasty or bypass grafting); valvular cardiopathy if decompensation; heart failure; acute pulmonary edema; stroke (either ischemic or hemorrhagic); transient ischemic attack; arteriopathy of the lower limb (above stage II); lower limb, aorta, or renal artery revascularization procedures.

Characteristics of the population were described by using proportions for categorical variables and means and standard deviation for quantitative variables. Biological parameters were divided into tertiles.

Proportional hazard Cox models were performed to estimate hazard ratios (HRs) and their confidence intervals (CIs) for studying the associations between baseline biological parameters (considered as both categorical and continuous variables) and the risk of MACE/CV mortality. In the case of multiple events during the follow-up period, only the first event was considered in the survival analysis. Proportional hazards assumption was tested using the Schoenfeld residuals. Three models were successively performed: without adjustment (model 0); adjustment for age and gender (model 1); adjustment for age, gender, diabetes, and history of CV events (model 2).

To determine whether CAC and CV risk biomarkers were independently associated with MACE/CV mortality, all variables were simultaneously entered into a stepwise proportional hazard Cox model with potential confounders: gender, age, CV history, diabetes, smoker status, body mass index associated with MACE/CV mortality at p < 0.10. The best fitting model was selected using the Akaike Information Criterion (AIC).

To investigate potential additive effects of biological variables and CAC on MACE/CV mortality occurrence, composite indexes were created, categorizing patients based on high levels of CAC and either low or high levels of iPTH, hsCRP, 25(OH) vitamin D₃, FGF-23, α-Klotho, OPG, SOST, dp-ucMGP, and GDF-15. The cutoffs of variables were chosen as tertiles values strongly associated with MACE/CV mortality. The associations between these composite indexes and the risk of MACE/CV mortality were described using the Kaplan-Meier method and tested for statistical significance using the log-rank test.

Logistic regression models were performed to estimate odds ratio and their 95% CI for studying the associations between CAC extent and vascular risk biomarkers. The significance level was set a p < 0.05. All analyses were carried out with SAS software, version 9.4 (SAS Institute, Cary, NC, USA).

Four hundred and twenty-five patients were included in the study. Table 1 depicts their clinical and biological characteristics at inclusion. Among the 425 patients, 152 (35.8%) patients had severe CAC (score ≥ 300).

Causes of CKD were glomerulonephritis (n = 65), autosomal polycystic kidney disease (n = 38), diabetic kidney disease (n = 40), diabetic and hypertensive nephropathy (n = 40), angiosclerosis and hypertensive nephropathy (n = 131), infectious/obstructive interstitial nephropathy (n = 16), nephrectomy following renal neoplasia (n = 4), genetic/congenital cause (n = 4), necrotizing angiitis (n = 7), unknown cause (n = 27), other cause (n = 53).

During the follow-up (median time of 3.61 years [25th–75th percentiles = 1.92–6.70]), MACE/CV mortality occurred in 69 patients (55 [79.7%] were men): acute coronary syndrome or ischemic heart disease (myocardial infarction, unstable angina, stress testing or dobutamine positive test) (n = 14), coronary revascularization procedures (angioplasty or bypass grafting) (n = 6), valvular cardiopathy if decompensation (n = 3), heart failure and acute pulmonary edema (n = 8), stroke (either ischemic or hemorrhagic) or transient ischemic attack (n = 10), arteriopathy of the lower limb (above stage II) (n = 9), lower limb revascularization procedures (n = 10), aorta or renal artery revascularization procedures (n = 6), CV death (n = 3). The risk of MACE events or CV death increased in older patients (HR = 1.04, 95% CI = [1.01–1.06], p = 0.0003), male (HR = 2.75, 95% CI = [1.53–4.95], p = 0.0007), with diabetes mellitus (HR = 3.34, 95% CI = [2.07–5.38], p < 0.0001), history of CV events (HR = 3.34, 95% CI = [1.93–5.79], p < 0.0001), and a lower estimated GFR (HR = 1.02, 95% CI = [1.00–1.03], p = 0.01).

The relationships between the presence of CAC, biological parameters, and MACE/CV death occurrence are presented in Table 2 (parameters being considered as tertiles). Only high levels of CAC, hsCRP, and iPTH were associated with a higher MACE/CV mortality occurrence (in both unadjusted analysis [model 0] and after adjustments for age and gender [model 1]). However, when diabetes and history of CV events confounders were entered in the model (model 2), only iPTH remained associated with MACE/CV mortality. Significant associations were observed between high levels of SOST or GDF-15 and high occurrence of MACE/CV mortality in the unadjusted analysis, but they did not reach significance after adjustments. Using continuous variables, CAC, hsCRP, and iPTH were significantly associated with MACE/CV mortality whatever the adjustment was (see online suppl. Table 1).

In order to identify whether variables significantly associated with MACE occurrence (including CAC, hsCRP, and iPTH) were independent from each other, these parameters were introduced with classical risk factors such as gender, diabetes, history of CV events, smoking status, and body mass index into a multivariable model using a stepwise procedure. As reported in Table 3, CAC, hsCRP, and iPTH remained statistically significant.

Finally, Figure 1 illustrates Kaplan-Meier survival analyses comparing composite indexes for MACE/CV mortality, demonstrating an additive effect of CAC and bone remodeling biomarkers. Interestingly, this result is confirmed by the increased risk of MACE/CV mortality observed with the combination of high levels of CAC (≥300) and high levels of hsCRP (>1.30 mg/L; HR = 2.01, 95% CI = [1.13–3.57]), iPTH (>65 pg/mL; HR = 2.95, 95% CI = [1.61–5.40]), FGF-23 (>166.84 RU/mL; HR = 2.26, 95% CI = [1.18–4.30]), dp-ucMGP (>838.5 pmol/L; HR = 2.81, 95% CI = [1.45–5.45]), OPG (>5.68 pmol/L; HR = 1.86, 95% CI = [1.05–3.30]), SOST (>0.76 ng/mL; HR = 2.01, 95% CI = [1.12–3.62]), or GDF-15 (>1,657 pg/mL; HR = 1.97, 95% CI = [1.14–3.41]) after adjustment for all confounders (model 2). Similarly, an increased risk of MACE/CV mortality was observed with high levels of CAC (≥300) and low levels of 25(OH) vitamin D₃ (≤22.0 ng/mL; HR = 2.63, 95% CI = [1.27–5.42]) or α-Klotho (≤575.49 pg/mL; HR = 2.31, 95% CI = [1.31–4.07]) after adjustment for all confounders (model 2) (see online suppl. Table 2 and Fig. 1).

This study aimed at analyzing the predictive values of CAC and biomarkers of CV risk on MACE/CV mortality in a sample of 425 ND-CKD patients. Our findings indicate that CAC stands out as a significant independent factor for MACE/CV mortality, exhibiting a HR comparable to that of diabetes, history of CV events, iPTH, and hsCRP. Furthermore, the predictive efficacy of CAC was notably heightened in conjunction with elevated levels of hsCRP, iPTH, FGF-23, dp-ucMGP, OPG, SOST, GDF-15 and diminished levels of 25(OH) vitamin D₃ and α-Klotho.

When analyzing the predictive value of all biomarkers on MACE/mortality occurrence, only hsCRP and iPTH were strongly independently associated with MACE/CV mortality after adjustment for age and gender; all other biomarkers of CV risk not being predictive per se of adverse events. Interestingly, the presence of an inflammatory environment, hyperparathyroidism, and bone remodeling molecules involved in CV risk (including high FGF-23, OPG, SOST, dp-ucMGP, and GDF-15 along with low 25(OH) vitamin D₃ and α-Klotho) dramatically worsened the predictive value of CAC after full adjustment (model 2). Thus, one can hypothesize that VC conspire with inflammation, hyperparathyroidism and factors of bone/vascular calcification remodeling to promote MACE/CV mortality.

The role of VC in atherosclerosis and subsequent poor outcome remains a matter of ongoing controversy, being recognized as either protective or deleterious. Some studies have reported that calcified plaques are more stable than non-calcified plaques [19]. Conversely, others could evidence that calcified plaques independently predict combined vascular events [20]. Several factors may explain such contradictory findings, including the phenotype of the calcification and the surrounding environment which may exert differential roles in plaque homeostasis [21]. Microcalcifications and spotty calcifications may represent an active stage of VC correlated with inflammation and may increase the risk of plaque rupture in the fibrous plaque [21]. By contrast, macrocalcifications would contain rather reparative macrophages including osteoclast-like cells and fewer inflammatory cells [21] and would be a typical feature of more stable plaques and asymptomatic disease. Of note, both types of calcifications can coexist during plaque progression. In CKD, two main uremia-related pathways may simultaneously contribute to the risk of MACE: ongoing VC remodeling leading to spotty calcifications and inflammatory process within established plaques. Our results align with this hypothesis. First, the increased risk of MACE associated with the combination of VC turnover biomarkers and CAC highlights the detrimental implication of coronary artery remodeling. Numerous studies have demonstrated the involvement of procalcifying agents such as FGF-23 [22‒24], OPG [9, 10, 25], SOST [10, 25], and dp-ucMGP, the inactive form of MGP [26] and their potential synergistic effects on VC [10, 25]. Our results (see online suppl. Table 3) corroborate the association of CAC with vascular bone remodeling factors and suggest that elevated levels of FGF-23, SOST, OPG, and dp-ucMGP worsen CAC prognosis (Fig. 1 and online suppl. Table 2). Interestingly, unfunctional MGP has been shown to facilitate initial calcium crystal deposition and early nidus formation [27]. In addition, GDF-15 emerged as another biomarker strongly associated with CAC and MACE. GDF-15, a stress-inducible cytokine upregulated by several inflammatory or stress-related proteins [28], exhibits a paradoxical and controversial role in CV diseases. Indeed, studies in animal models suggest that GDF-15 may prevent or reduce plaque formation [29]. Conversely, clinical studies indicate that high GDF-15 levels are associated with heart failure [12] and with the severity of CAC in both the general population and male patients with end stage kidney disease [30]. Notably, Mayer et al. [13] recently demonstrated that stable coronary heart disease patients with high levels of both GDF-15 and dp-ucMGP faced an increased risk of MACE. In the same manner, the inverse relationship between vitamin D [31, 32] or α-Klotho [33] and CAC has been previously documented [34]. Our findings confirm these data and clearly demonstrate that high levels of 25(OH) vitamin D₃ or α-Klotho protect against MACE (Fig. 1 and online suppl. Table 2). The protective effect of circulating α-Klotho aligns with its beneficial role in VC and heart disease observed in animal models using recombinant Klotho [35, 36].

The surrounding inflammatory environment may also substantially contribute to the risk of MACE associated with severe CAC. A highly pro-inflammatory milieu can promote the recruitment and activation of neutrophils, leading to SMC death via neutrophil extracellular traps [37] and subsequent necrotic core enlargement [38]. The enrichment of inflammatory cells and the expansion of the necrotic core can increase plaque instability and ultimately lead to plaque rupture [39].

Interestingly, as recently well described [40‒42], macrophages, the main inflammatory cells involved in atherosclerotic plaques, can exert a dual role in the disease progression, due to their remarkable plasticity and functional heterogeneity [42]. Originally, two subpopulations have been defined: classically activated (M1-like phenotype) and alternatively activated (M2-like phenotype) macrophages with respective pro- and anti-inflammatory properties [43]. M1 macrophages produce pro-inflammatory cytokines (such as TNF-α, IL-1β, IL-6, etc.) that promote the differentiation of vascular smooth muscle cells into osteoblasts and stimulate the production of reactive oxygen species and bone morphogenetic proteins [41]. M2 macrophages, in turn, present different properties, sometimes opposite, to M1 macrophages. They secrete anti-inflammatory factors (including IL-10) that help reduce the inflammation by inhibiting the expression of pro-inflammatory cytokines and by reducing the production of reactive oxygen species [41]. M1 macrophages are associated with unstable plaques, while M2 macrophages are more common in asymptomatic lesions and in stable zones of plaques [43]. M1 polarization is dependent on inflammatory conditions [41]. The preferential M1- or M2-like phenotyping shift may also be dependent on the calcified milieu as calcium phosphate supplementation can promote M1 polarization [42]. Results of our study clearly demonstrated that inflammation and hyperparathyroidism could drive a shift toward a pro-inflammatory M1-like phenotype, worsening the prognostic value of CAC for CV events (Fig. 2).

Of note, the M1-M2 polarization and vascular-bone remodeling pathways are closely interconnected. In osteosarcoma biopsies, a positive correlation between M1-polarized macrophages and OPG expression has been observed [44]. This relationship between OPG, as a surrogate of OPG/RANK/RANKL pathway dysregulation, and the M1 phenotype aligns with our previous findings showing that OPG levels predict poor outcomes in hemodialysis patients, but only in an inflammatory context [45]. Similarly, the interplay between bone remodeling factors (SOST and RANKL) and macrophage polarization has been demonstrated during alveolar bone repair [46]. On the other hand, previous studies have evidenced that active vitamin D can downregulate macrophage transition to the M1 phenotype [47, 48]. Similarly, α-Klotho has been shown to promote microglial polarization toward the M2 phenotype, which may exhibit anti-inflammatory properties [49]. It could be assumed that the effect of α-Klotho on M2 polarization in microglia could also apply to macrophages. All these findings are consistent with our results reporting a poor outcome in patients with deficient levels of 25(OH) vitamin D₃ and low levels of circulating α-Klotho. Additional anti-inflammatory properties of vitamin D and α-Klotho [50] may also be highlighted in this population [51‒53]. In summary, the original findings of this study aimed to introduce the concept of “multiple hits” contributing to the occurrence of MACE in patients with CKD.

Our study acknowledges certain limitations. First, being a single-center study, there may be limitations in generalizability. Additionally, an extended follow-up duration could have strengthened the observed relationships between biomarkers and MACE/CV mortality. The inclusion of a broader array of biomarkers, such as pro/anti-inflammatory cytokines, could have further validated the involvement of inflammatory processes in the occurrence of MACE/CV mortality. In addition, some parameters including 25(OH) vitamin D₃ and dc-ucMGP could not be measured in the entire patient population. Lastly, a qualitative exploration of plaques would have yielded supplementary information, enhancing the understanding of the assumptions made regarding the nature of calcifications.

In conclusion, our findings emphasize that the adverse prognosis associated with CAC in CKD is exacerbated by a detrimental environment, encompassing inflammation or surrogate biomarkers indicative of active bone matrix synthesis or mineralization, such as iPTH, FGF-23, OPG, and SOST. Conversely, vitamin D and α-Klotho may offer protective effects on CAC-related poor outcomes, potentially through mechanisms involving M2 macrophage polarization or anti-inflammatory pathways. This study introduces the original concept of “multiple hits” contributing to the occurrence of MACE. Consequently, effective therapeutic strategies aimed at improving CV outcomes in CKD patients should focus on mitigating the harmful environment that fosters plaque vulnerability, curtailing chronic microinflammation, and normalizing mineral metabolism and vascular bone remodeling.

The authors would like to acknowledge the nurses from Policlinique at Lapeyronie Hospital, especially Anne Bebengut, Hélène Lebihan, Violaine Barral, and Lauranne Talbot for patient appointment scheduling and blood withdrawals.

This study protocol was reviewed and approved by the Nîmes University Hospital’s Ethics Committee CPP SUD MEDITERRANEE III, Approval No. 2013.02.03. All patients gave their written informed consent.

The results presented in this paper have not been published previously in whole or part, except in abstract format.

This work was partly supported by a grant from Ministère de la Santé (PHRC-UF 7853).

Conceptualization: Marion Morena, Leila Chenine, Anne-Marie Dupuy, Hélène Leray-Moragues, Kada Klouche, Hélène Vernhet, Bernard Canaud, and Jean-Paul Cristol. Data curation: Marion Morena, Anne-Marie Dupuy, Anne-Sophie Bargnoux, and Kada Klouche. Formal analysis: Isabelle Jaussent. Funding acquisition, project administration, and supervision: Bernard Canaud and Jean-Paul Cristol. Investigation: Marion Morena, Anne-Marie Dupuy, and Anne-Sophie Bargnoux. Methodology: Isabelle Jaussent, Bernard Canaud, and Jean-Paul Cristol. Resources: Marion Morena, Isabelle Jaussent, Leila Chenine, Anne-Marie Dupuy, Anne-Sophie Bargnoux, Hélène Leray-Moragues, Kada Klouche, Hélène Vernhet, and Bernard Canaud. Validation: Marion Morena, Isabelle Jaussent, Anne-Marie Dupuy, and Anne-Sophie Bargnoux. Visualization: Marion Morena. Writing – original draft: Marion Morena and Jean-Paul Cristol. Writing – review and editing: Marion Morena, Isabelle Jaussent, Leila Chenine, Anne-Marie Dupuy, Anne-Sophie Bargnoux, Hélène Leray-Moragues, Kada Klouche, Hélène Vernhet, Bernard Canaud, and Jean-Paul Cristol.

The data that support the findings of this study are not publicly available because they contain information that could compromise the privacy of research participants but are available from the corresponding author (J.P.C.) upon reasonable request.