, Introduction
Cardiovascular events are the leading cause of death in patients with chronic kidney disease (CKD) [1,2]. A variety of cardiovascular events, including coronary artery diseases and heart failure (HF), develop and deteriorate in the context of CKD progression [1]. Accordingly, several biomarkers have been proposed to predict the risk of major adverse cardiovascular events (MACEs) [3–5], though the practical value to date is limited in most cases among patients with CKD [5,6]. Rather, previous studies demonstrated the contribution of the nontraditional as well as the traditional cardiovascular risk factors related to the uremic condition to the pathogenesis of cardiovascular diseases in patients with CKD [7–9].
Hypocalcemia has been associated with mortality in various conditions. It is best illustrated in the cases of severe acute pancreatitis, where the decrease in serum calcium levels indicates the autodigestion of intra-abdominal adipose tissue, resulting in higher mortality [10–12]. The relation between hypocalcemia and the risk of death has also been suggested in other conditions, such as trauma [13,14], pneumonia [15,16], and pulmonary thromboembolism [17], though the mechanisms may differ from or be less clear than those in severe acute pancreatitis.
In this regard, low serum calcium levels have been associated with cardiovascular mortality. A previous study reported that hypocalcemia is associated with long-term mortality after acute myocardial infarction (MI) [18]. Another study demonstrated hypocalcemia as a risk factor for in-hospital mortality in patients with HF and CKD [19]. The impact of hypocalcemia on all-cause mortality has also been demonstrated in patients with end-stage kidney disease (ESKD) [20,21].
Interestingly, hypocalcemia develops frequently during the course of CKD progression, even before the initiation of kidney replacement therapy (KRT) [22]. The association of low serum calcium levels with the risk of MACE and cardiovascular mortality, nevertheless, has not been established in this population yet, especially among those at non-dialysis stages. The present study, therefore, aimed to investigate whether low serum calcium levels increase the risk of MACE and cardiovascular mortality in patients with non-dialysis CKD.
Methods
Study design
The present study analyzed the data from the KNOW-CKD (KoreaN Cohort Study for Outcome in Patients With Chronic Kidney Disease) [23], a prospective cohort study of patients with CKD at stages 1 to pre-dialysis 5. The study was designed and conducted in agreement with the Declaration of Helsinki. Nine tertiary hospitals in South Korea recruited the participants who voluntarily provided the informed consent from between 2011 and 2016 (NCT01630486 at https://www.clinicaltrials.gov), and acquired the approval for the study protocol from the Institutional Review Board at each participating center (Seoul National University Hospital, 1104-089-359; Seoul National University Bundang Hospital, B-1106/129-008; Severance Hospital, 4-2011-0163; Kangbuk Samsung Medical Center, 2011-01-076; The Catholic University of Korea, Seoul St. Mary’s Hospital, KC11OIMI0441; Gachon University Gil Hospital, GIRBA2553; Eulji General Hospital, 201105-01; Chonnam National University Hospital, CNUH-2011-092; Inje University Busan Paik Hospital, 11-091), as previously described [23,24]. A total of 2,238 participants were initially recruited and were closely monitored during the follow-up period to report major events, which were thoroughly cross-checked by the collaborating investigators. After excluding those lacking the data on follow-up duration (n = 31) and those lacking the baseline measurement of serum calcium or albumin levels (n = 19), a total of 2,188 patients were finally included for the analyses (Fig. 1). The study observation period ended on January 31, 2024, with a median duration of 9.2 years.
Data collection
Sociodemographic data, anthropometric measurements, comorbidities, and other medical information of each participant at the baseline were extracted from the electronic data management system of the KNOW-CKD. Blood and urine samples were collected following an overnight fast. Blood samples were centrifuged within 1 hour for serum separation. Blood and urine samples were delivered to the central laboratory of the KNOW-CKD (Lab Genomics) for further analyses. Serum creatinine was measured using an isotope dilution mass spectrometry traceable method [23] to calculate estimated glomerular filtration rate (eGFR) using the Chronic Kidney Disease Epidemiology Collaboration equation [25].
Exposure and study outcomes
Corrected calcium levels were calculated as follows: if serum albumin was ≥4.0 mg/dL, then corrected calcium is equal to serum calcium; if serum albumin was <4.0 mg/dL, then albumin-corrected calcium = (0.8 × [4.0 – measured serum albumin]) + measured serum calcium [21]. The participants were categorized into the three groups by the operationally defined corrected calcium levels: low (<8.5 mg/dL), normal, and high (≥9.5 mg/dL). The primary outcome of the current study was MACE, defined as the composite of cardiovascular death, nonfatal MI, and nonfatal stroke [3]. The secondary outcomes included the individual events of the primary outcome and all-cause death.
Statistical analysis
Continuous and categorical variables of the baseline characteristics by corrected calcium levels were compared using one-way analysis of variance and the chi-square test, respectively. The cumulative incidence of the study outcomes by corrected calcium levels was visualized by Kaplan-Meier survival curves and was subsequently compared for statistical significance by the log-rank test. The last visiting date was used as the censoring date for those with follow-up loss. To determine the independent association between corrected calcium levels and the risk of cardiovascular events, Cox regression analyses were conducted. The models in the Cox analyses were adjusted for the potential confounders as the following: Model 1 was adjusted for age, sex, eGFR and spot urine albumin-to-creatinine ratio (ACR); Model 2 was additionally adjusted for medical history, such as Charlson comorbidity index, the primary cause of CKD, smoking status, and medications (e.g., angiotensin converting enzyme inhibitors and angiotensin receptor blockers, diuretics, lipid lowering agents, and antiplatelet/anticoagulant agents), and anthropometric data, including body mass index (BMI) and systolic blood pressure (SBP); Model 3 was finally further adjusted for the baseline laboratory data, including hemoglobin, albumin, low-density lipoprotein cholesterol (LDL-C), fasting glucose, 25-hydroxyvitamin D (25(OH)D), and high-sensitivity C-reactive protein. Participants with any missing data were excluded from primary analyses. The results of Cox regression analyses were reported with hazard ratios (HRs) and 95% confidence intervals (CIs). The linear correlation between corrected calcium levels (as a continuous variable) and the risk of cardiovascular events was illustrated by the penalized spline curve. To validate the robustness of the primary analyses, we planned a series of sensitivity analyses. First, we used the cause-specific hazard regression model, where the incident ESKD before the occurrence of the study outcomes was considered as a competing risk, because the reporting for the occurrence of cardiovascular events and all-cause death was stopped after the initiation of KRT according to the KNOW-CKD protocol [23]. Second, we repeated the Cox regression analysis after excluding the participants with serum albumin <3.5 g/dL, to minimize the impact of concomitant hypoalbuminemia on the overall study outcomes. Third, any missing data from 427 participants who were excluded from the primary analyses were replaced by multiple imputations using a random sampling method to impute 10 independent copies of the data to repeat the Cox regression analyses. Fourth, phosphorus and parathyroid hormone (PTH) levels were added to the model for further adjustment, which are also important components of CKD–mineral and bone disorder, but were not suitable for multiple imputations due to a high missing rate (i.e., 39.0% of the cases were missing for PTH levels). Fifth, we recategorized the participants into the five subgroups, including a group with corrected calcium ≥10.2 mg/dL, and repeated the Cox regression analysis to also evaluate the impact of hypercalcemia on the risk of MACE. Lastly, a 1:2 propensity score matching analysis between participants with age- and sex-adjusted or CKD stage-adjusted calcium levels <8.5 mg/dL (n = 108) and those with levels ≥8.5 mg/dL (n = 216) was performed to minimize the bias due to the huge gap in sample size between the groups. Prespecified subgroup analyses were conducted to examine whether the association between corrected calcium levels and the risk of cardiovascular events is modified by specific clinical contexts. These subgroups were defined by age (<60 or ≥60 years), sex (male or female), BMI (<25 or ≥25 kg/m2), eGFR (<45 or ≥45 mL/min/1.73 m2), and spot urine ACR (<300 or ≥300 mg/g). Two-tailed p-values below 0.05 were considered statistically significant. All statistical analyses were executed using IBM SPSS version 22.0 (IBM Corp.) and R version 4.1.1 (R Foundation for Statistical Computing).
Results
Baseline characteristics
The baseline characteristics of the participants significantly differed by corrected calcium levels (Table 1). The follow-up duration was significantly shortened in patients with low corrected calcium levels. Higher burden of comorbid conditions, higher prevalence of diabetes mellitus as the primary cause of CKD, more frequent use of diuretic medication, and higher SBP were inversely correlated with corrected calcium levels. Low levels of hemoglobin, albumin, total cholesterol, high-density lipoprotein cholesterol, LDL-C, triglycerides, and 25(OH)D were significantly related to low corrected calcium levels. Most importantly, the participants with low corrected calcium levels had significantly lower eGFR and heavier proteinuria. Overall, the analysis of the baseline characteristics revealed the correlation between low corrected calcium levels and clinically unfavorable features.
Association between low corrected calcium levels and adverse cardiovascular outcomes
The cumulative incidence of the study outcomes was visualized by a Kaplan-Meier survival curve with a log-rank test. The cumulative incidence of MACE (p = 0.004) (Fig. 2) and cardiovascular death (p = 0.002) (Supplementary Fig. 1, available online), but not nonfatal MI (Supplementary Fig. 2, available online) and nonfatal stroke (Supplementary Fig. 3, available online), significantly differed by corrected calcium levels. The cumulative incidence of all-cause death also significantly differed by corrected calcium levels (p < 0.001) (Supplementary Fig. 4, available online). Cox regression analyses demonstrated that low corrected calcium levels were independently and significantly associated with the risk of MACE (adjusted HR, 2.854; 95% CI, 1.439–5.659), cardiovascular death (adjusted HR, 5.256; 95% CI, 1.993–13.861), and all-cause death (adjusted HR, 1.902; 95% CI, 1.185–3.054), but not with the risk of nonfatal MI or nonfatal stroke (Table 2). Penalized spline curve displayed a linear, negative correlation between serum corrected calcium levels and the risk of MACE (Fig. 3).
Sensitivity and subgroup analyses
Cause-specific HRs for MACE (adjusted HR, 2.630; 95% CI, 1.377–5.022), cardiovascular death (adjusted HR, 4.808; 95% CI, 1.687–13.704) and all-cause death (adjusted HR, 1.696; 95% CI, 1.167–2.466) associated with low corrected calcium levels were all statistically significant in the competing risks analyses (Table 3). The associations of low corrected calcium levels with MACE (adjusted HR, 2.365; 95% CI, 1.151–4.862), cardiovascular death (adjusted HR, 4.900; 95% CI, 1.733–13.854) and all-cause death (adjusted HR, 1.734; 95% CI, 1.050–2.864) remained significant, even after excluding the participants with clinically significant hypoalbuminemia (Supplementary Table 1, available online). Cox regression analyses after multiple imputations also revealed significant associations of low corrected calcium levels with MACE (adjusted HR, 2.053; 95% CI, 1.059–3.982), cardiovascular death (adjusted HR, 3.862; 95% CI, 1.533–9.734) and all-cause death (adjusted HR, 1.637; 95% CI, 1.061–2.527) (Supplementary Table 2, available online). Low corrected calcium levels were significantly associated with the risk of MACE (adjusted HR, 4.582; 95% CI, 1.904–11.026) even after additionally adjusting for phosphorus and PTH levels (Supplementary Table 3, available online). Low corrected calcium levels were still significantly associated with the risk of MACE (adjusted HR, 2.349; 95% CI, 1.146–4.816) compared to the reference group (i.e., corrected calcium 9.0 to <9.5 mg/dL) after categorizing the participant into the five subgroups, while high corrected calcium levels (i.e., corrected calcium ≥10.2 mg/dL) was not significantly associated with the risk of MACE (Supplementary Table 4, available online). Low serum calcium levels were still significantly associated with the risk of MACE even in the propensity score matching analyses by age and sex (adjusted HR, 3.619; 95% CI, 1.120–11.690) (Supplementary Table 5, available online) or by CKD stage (adjusted HR, 3.442; 95% CI, 1.299–9.121) (Supplementary Table 6, available online). Only the age of the participants (p for interaction = 0.046), but not sex, BMI, eGFR, or spot urine ACR, significantly modified the association between low corrected calcium levels and the risk of MACE (Table 4).
Discussion
The present study discovered that hypocalcemia is significantly associated with the risk of adverse cardiovascular outcomes in patients with non-dialysis CKD. Especially, hypocalcemia increased the risk of both cardiovascular and all-cause death, and this association remained robust regardless of the absence of hypoalbuminemia. Hypercalcemia (i.e., corrected calcium ≥10.2 mg/dL) was not significantly associated with the risk of MACE in the present study.
Hypocalcemia is one of the classical features among patients with CKD. A previous study reported the prevalence of hypocalcemia (defined as corrected calcium <8.4 mg/dL) among the patients at stage 3 CKD as approximately 10%, which rises up to 30% in patients with stage 5 non-dialysis CKD [26]. With the addition of calcimimetic treatment to manage secondary hyperparathyroidism, the prevalence of hypocalcemia in patients with ESKD further increases [27]. Despite its high prevalence among patients with CKD and ESKD, the clinical implications of hypocalcemia in relation to cardiovascular outcomes have been forgotten so far.
The association of hypocalcemia with the outcomes in patients with ESKD has been previously reported [20,21,28], all of which primarily focused on all-cause mortality. Yamaguchi et al. [20] reported that low ionized calcium levels, despite normal or high calcium levels in patients undergoing incident hemodialysis, are associated with increased risk of all-cause mortality. The Dialysis Outcomes and Practice Patterns Study (DOPPS) also reported the association between hypocalcemia (i.e., uncorrected or corrected calcium levels ≤7.5 mg/dL) and a greater mortality among the subpopulation without evidence of hypoalbuminemia (i.e., albumin levels >3.8 g/dL) [28]. Similarly, both low (<8.5 mg/dL) and high (≥10.2 mg/dL) corrected calcium levels were associated with excess mortality in patients on maintenance dialysis [21]. In contrast, the association between hypocalcemia and the outcomes in patients with non-dialysis CKD has been still elusive. As the current study addresses the association of hypocalcemia with all-cause mortality as well as MACE including cardiovascular mortality, to the best of our knowledge, this study is the first to demonstrate the impact of hypocalcemia on the cardiovascular outcomes in patients with CKD at non-dialysis stages.
Our finding is of particular importance in that targeting lower calcium levels to avoid hypercalcemia among patients with CKD has long been advocated based on the epidemiological studies [28–30]. Those observational studies consistently reported the association of serum calcium levels in the higher range of normal with the greater risk of vascular calcification or deaths from all-cause and cardiovascular events. The KDIGO (Kidney Disease: Improving Global Outcomes) clinical practice guideline on CKD-MBD updated in 2017, thus discourages maintaining serum calcium in the normal range by the correction of hypocalcemia unless it is symptomatic or severe [29]. This may, on the other hand, lead to the misconception that hypocalcemia in uremic patients is somewhat natural and negligible. In this regard, the current study highlights that hypocalcemia is a nontraditional risk factor for cardiovascular events and all-cause death in patients with CKD, and that its clinical implications should be equally emphasized.
It should also be noted that, on the other hand, the association of hypocalcemia with adverse cardiovascular outcomes does not necessarily favor calcium replacement in patients with CKD. It has been reported that oral calcium replacement does not directly lead to an increase in serum calcium levels [31] and that the use of calcium-based phosphate binders accelerates vascular calcification in this population [32]. We, therefore, suggest that CKD patients with hypocalcemia should be under more intense monitoring for the development of cardiovascular events, instead of correcting calcium levels to the normal range, which is not mutually exclusive to the recommendation from the current practice guideline.
The precise mechanism to explain the association between hypocalcemia and the risk of cardiovascular events remains unclear. One of the possible explanations may include the reduced vitamin D activity leading to the development of hypocalcemia in patients with CKD [33]. As the role of active vitamin D is critical for the homeostatic maintenance of serum calcium levels, hypocalcemia may represent the severity of the underlying vitamin D deficiency. Importantly, the association between vitamin D deficiency and adverse cardiovascular outcomes has been well-documented in previous observational studies [34,35]. The current study, however, included serum 25(OH)D levels as a co-variable in the regression models. Accordingly, it seems reasonable that the results shown here are independent of serum 25(OH)D levels. Furthermore, the effect of vitamin D replacement on the prevention of cardiovascular events remains to be proven [36]. Another theory may involve the interaction between serum albumin levels and total calcium concentration in plasma. Regardless of ionized calcium levels, serum total calcium levels are affected by serum albumin levels, as a significant portion of total calcium is bound to albumin [37]. As the most important surrogate of malnutrition, the association of hypoalbuminemia with all-cause as well as cardiovascular mortality has been validated [38,39]. One may argue that, therefore, the findings presented in the current study are a reflection of the concomitant malnutrition in patients with adverse cardiovascular outcomes. The current study, however, adopted albumin-corrected calcium levels as the main exposure to minimize the effect of serum albumin levels both on total calcium levels and the study outcomes. Even in the sensitivity analysis that excluded the participants with clinically meaningful hypoalbuminemia (i.e., <3.5 g/dL), the association between hypocalcemia and adverse cardiovascular outcomes remained significant. Rather, the effect of hypocalcemia on the cardiovascular outcomes seems more direct and independent of the other confounding factors. The role of hypocalcemia in QT prolongation is well-established [40], which in turn contributes to the development of life-threatening arrhythmia, such as Torsade de Pointes [41]. Regardless of the primary cardiovascular events, the risk of ventricular arrhythmia increases during the acute phase of MI and/or HF [42,43]. The presence of hypocalcemia as a comorbid condition, thus, may deprive patients with CKD of the chance of recovery from the primary cardiovascular events by promoting the development of life-threatening arrhythmia. In accordance with this theory, we found that hypocalcemia is significantly associated with cardiovascular and all-cause mortality, but not with nonfatal MI or nonfatal stroke, from the analyses of the secondary outcomes, implying the pivotal role of serum calcium levels during the critical care of CKD patients.
Another potential mechanism linking hypocalcemia to adverse cardiovascular outcomes involves the impairment of cardiac contractility. Calcium ion plays a central role in cardiac excitation-contraction coupling, where extracellular calcium influx through L-type calcium channels triggers calcium-induced calcium release from the sarcoplasmic reticulum, ultimately enabling myocardial contraction [44,45]. Hypocalcemia has been demonstrated to reduce cardiac contractility, as evidenced by decreased left ventricular stroke work index, ejection fraction, and cardiac index [46]. Case reports have documented reversible dilated cardiomyopathy caused by severe hypocalcemia, with dramatic recovery of cardiac function following calcium correction [46,47]. In the context of CKD, where the cardiovascular system is already compromised by multiple traditional and nontraditional risk factors, the superimposed negative inotropic effect of hypocalcemia may further reduce cardiac reserve and contribute to the development of HF or fatal outcomes during acute cardiovascular events. This mechanistic link provides additional biological plausibility to explain why hypocalcemia is significantly associated with cardiovascular and all-cause mortality in patients with CKD.
Some limitations are to be acknowledged. First, the current study is based on a prospective cohort, thereby the causality between hypocalcemia and adverse cardiovascular outcomes cannot be confirmed. Lower eGFR or the presence of proteinuria/albuminuria is a well-known nontraditional risk factor for cardiovascular events. Thus, it is also possible that the hypocalcemia may simply be a prognostic marker of more severe underlying disease status, rather than an independent risk factor, because the statistical adjustment may not be perfect to overcome the impact of the significantly different levels of eGFR or proteinuria/albuminuria at the baseline. The results presented in the current study, however, are enough to define hypocalcemia as a nontraditional risk factor for cardiovascular events and all-cause death in patients with CKD, as its robustness is supported by a series of sensitivity analyses including competing risks analyses and multiple imputation methods. Second, due to the lack of data on serum ionized calcium levels from the KNOW-CKD, the actual impact of hypocalcemia on the cardiovascular outcomes may be over- or under-estimated. Hypocalcemia determined by serum total calcium levels does not accurately represent the impact of serum calcium levels, as its function is mainly mediated by albumin-unbound calcium ions. We, thus, adopted albumin-corrected calcium levels as the main exposure, and additionally conducted a sensitivity analysis that excluded the participants with clinically significant hypoalbuminemia, to minimize the effect of serum albumin levels both on total calcium levels and the study outcomes. Third, the study participants are relatively homogeneous, who are ethnically Korean residing in South Korea. The extension of the current findings to the other population, therefore, needs caution, though the tentative mechanism to link hypoalbuminemia and adverse cardiovascular outcomes, such as QT prolongation and the subsequent risk of fatal arrhythmia, is universally accepted. Fourth, the primary result of the current study excluded PTH levels as a potential confounder, because of its high missing rate, though a sensitivity analysis demonstrated that the addition of PTH, as well as phosphorus levels, did not alter the major finding of the study. Fifth, some medications affecting calcium levels, such as cinacalcet, were not included for the adjustment. Yet, cinacalcet is exclusively prescribed for patients with ESKD in South Korea, and is not allowed for the participants of the current study. Lastly, the power of statistical analysis might be limited by a huge gap in the sample size between the groups, though a propensity score matching analysis was additionally performed to minimize the bias.
In conclusion, we report that hypocalcemia is significantly associated with the risk of adverse cardiovascular outcomes in patients with non-dialysis CKD. As a nontraditional risk factor for cardiovascular events and all-cause death in this population, the presence of hypocalcemia should urge more intense monitoring for the development of cardiovascular events.
PDF Links
PubReader
ePub Link
XML Download
Full text via DOI
Download Citation
Supplement table 1
Print
