Abstract
Coronary artery calcifications (CACs) are observed in most patients with CKD on dialysis (CKD-5D). CACs frequently progress and are associated with increased risk for cardiovascular events, the major cause of death in these patients. A link between bone and vascular calcification has been shown. This prospective study was designed to identify noninvasive tests for predicting CAC progression, including measurements of bone mineral density (BMD) and novel bone markers in adult patients with CKD-5D. At baseline and after 1 year, patients underwent routine blood tests and measurement of CAC, BMD, and novel serum bone markers. A total of 213 patients received baseline measurements, of whom about 80% had measurable CAC and almost 50% had CAC Agatston scores>400, conferring high risk for cardiovascular events. Independent positive predictors of baseline CAC included coronary artery disease, diabetes, dialysis vintage, fibroblast growth factor-23 concentration, and age, whereas BMD of the spine measured by quantitative computed tomography was an inverse predictor. Hypertension, HDL level, and smoking were not baseline predictors in these patients. Three quarters of 122 patients completing the study had CAC increases at 1 year. Independent risk factors for CAC progression were age, baseline total or whole parathyroid hormone level greater than nine times the normal value, and osteoporosis by t scores. Our results confirm a role for bone in CKD–associated CAC prevalence and progression.
Coronary artery calcifications (CACs) coexisting with low bone mass or volume are frequently seen in patients with CKD on dialysis (CKD-5D).1–5 Patients with CKD stages 4 and 5 have an annual mortality of 22.8%,6 and cardiovascular disease is the cause of death in 60%–70% of patients with CKD-5D.7 CACs have been found to be a strong predictor of cardiovascular disease,8 are observed frequently, and progress rapidly in patients with CKD-5D.8,9 It is also known that age and intake of calcium–containing phosphate binders are associated with CAC in patients with CKD-5D.9–12 In addition, low bone volume and bone turnover abnormalities assessed by bone histology predict CAC in patients with CKD-5D.1–3 However, for CAC progression, no information is available on the use of noninvasive bone mass assessment using dual energy x-ray absorptiometry (DXA) or quantitative computed tomography (QCT) or the use of traditional and novel serum biochemical parameters. This study was designed to fill these gaps in identifying patients at risk for CAC progression with its inherent effect on cardiovascular events and mortality.
Results
Baseline Clinical, Biochemical, Bone Mineral Density, and CAC Measures
In total, 213 patients were enrolled in the study and had baseline measurements between August of 2009 and April of 2013. Baseline patient clinical characteristics are shown in Table 1. Prevalence of diabetes, coronary artery disease (CAD), and use of medications were similar to the US Renal Data System published rates.6
Baseline clinical characteristics of patients with CKD-5D who completed the study and those who did not
Biochemical parameters are shown in Table 2. They included the traditional parameters parathyroid hormone (PTH) and bone–specific alkaline phosphatase (BSAP), the novel marker fibroblast growth factor-23 (FGF23), the bone resorption parameter tartrate–resistant acid phosphatase-5b (TRAP-5b), and the bone formation parameters sclerostin, Dickkopf-1 (DKK1), and procollagen type 1 N-terminal propeptide (P1NP). Median blood concentrations of PTH were elevated, which was expected for patients on chronic dialysis. BSAP concentrations were in the normal range, TRAP-5b and sclerostin were elevated, DKK1 was low, and P1NP was mostly normal, whereas FGF23 was extremely elevated in almost all patients.
Serum baseline biochemical characteristics of patients with CKD-5D who completed the study and those who did not
Prevalence of osteoporosis by any site and method is shown in Table 1. One third (32.8%) of patients who completed the study were osteoporotic by at least one site or method. CAC Agatston scores and square root of coronary artery calcification volume (CAC SQRV) scores are also shown in Table 1. Close to one half of the patients had baseline CAC Agatston scores>400, placing them at high risk for cardiovascular events.
Univariate Relationships and Multivariable Predictors of Baseline CAC
Baseline CAC SQRV correlated with age (Table 3) but did not vary significantly with sex, race, smoking history, or body mass index. CAC SQRV was higher in patients with a history of CAD (35.5 versus 15.9; P<0.001), patients with diabetes (26.9 versus 15.9; P<0.001), and patients reporting no exercise (none: 25.2 versus two or more times per week: 16.9; P=0.002). There were no differences in baseline CAC SQRV in patients treated versus not treated with active vitamin D, calcium–containing phosphate binders, statins, or BP medication. Patients treated with cinacalcet had higher CAC SQRV than those not treated (25.3 versus 19.3; P=0.05).
Spearman’s Rho correlation between variables and baseline CAC SQRV along with standardized regression coefficients (n=213)
Baseline CAC SQRV correlated with FGF23 (Rho=0.185; P=0.04) and was elevated in patients with high PTH (whole PTH>450 pg/ml: 30.4 versus 19.7; P=0.02; total PTH>540 pg/ml: 23.9 versus 19.2; P=0.06) but not associated with any other serum biochemical measures, including HDL, LDL, serum albumin, calcium, phosphorus, and calcium×phosphorus. Baseline CAC SQRV was higher in patients who were osteoporotic than in patients who were nonosteoporotic diagnosed by QCT or DXA at any measured site, except for DXA at the spine. The results at the individual sites in patients who were osteoporotic by QCT at the hip showed CAC SQRV of 33.8 versus 19.0 in patients who were nonosteoporotic (P<0.001); QCT spine was 30.1 versus 18.6 (P<0.001), QCT femoral neck was 32.4 versus 19.6 (P=0.004), DXA hip was 29.7 versus 19.4 (P<0.01), and DXA femoral neck was 26.4 versus 19.5 (P=0.04). CAC SQRV correlated with absolute values of bone mineral density (BMD) by QCT of the total hip, femoral neck, and spine and DXA of the hip and femoral neck but not the spine (Table 3).
Multivariable predictors of baseline CAC SQRV by descending order of standardized coefficient were CAD (0.34; P<0.001), diabetes (0.25; P<0.001), dialysis vintage (0.24; P<0.001), natural log FGF23 (0.23; P<0.001), age (0.22; P<0.001), and spinal BMD by QCT (−0.13; P<0.05). The multivariable linear regression was highly significant in predicting CAC SQRV with an adjusted R2 of 0.39.
Changes in CAC, BMD, and Biochemical Parameters at 1 Year
One-year follow-up visits and measurements occurred through May of 2014. Of 213 patients receiving baseline scans, 26 patients died, 10 patients were transplanted, 4 patients moved out of the area, 8 patients were not able to participate because of changes in health status, 42 patients withdrew from the study or could not participate because of lack of transportation, and 1 patient did not tolerate positioning in the computed tomography (CT) scanner, leaving 122 patients with baseline and 1-year measurements for analysis. Sex, race, age, medications, dialysis vintage, all serum biochemical values, high-risk CAC status (Agatston score>400), and presence of osteoporosis were compared between the patients who completed the study and those who did not. There were no significant differences found (Tables 1 and 2).
Over 1 year of study, medications did not change significantly (Table 4). The serum biochemical parameters FGF23, sclerostin, and DKK1 increased significantly over the year, whereas P1NP decreased significantly. No other significant changes in serum biochemical parameters were observed on average (Table 5).
Medications of patients with CKD-5D at baseline and 1 year
Serum biochemical characteristics of patients with CKD-5D at baseline and 1 year
The percentage of patients losing >2% of BMD after 1 year and differences in CAC are shown in Tables 6 and 7. Over one third of patients experienced >2% bone loss. The 2% threshold was chosen, because it has been shown to be associated with a significant increase in fracture risk.13–15 It is outside the coefficient of variation of both the DXA and QCT methods for BMD determination.
BMD changes after 1 year
Changes in CAC at 1 year
Median CAC Agatston score progression was 87, and mean change in CAC SQRV was 3.9 (both P<0.001). This 25% progression is well above the 15% that has been shown to be associated with higher mortality risk8 and also well above the 3% interscan variability and 1% intraobserver variability for the 64-slice multislice computed tomography (MSCT) technology used in this study. Three quarters of patients experienced CAC progression. One-year CAC SQRV progression correlated with baseline CAC SQRV (Rho=0.185; P=0.04) and age (Rho=0.263, P=0.004) but did not vary with sex, race, history of diabetes, history of CAD, or body mass index.
CAC SQRV progression correlated with baseline TRAP-5b (Rho=0.203; P=0.04) and whole and total PTH (Rho=0.235; P=0.02 and Rho=0.231; P=0.02, respectively) but not any of the other measured biochemical parameters. Progression of CAC SQRV was higher in patients who were osteoporotic by any site or method (7.5 versus 2.1; P=0.001).
With adjustment for age, predictors for CAC progression were osteoporosis (β=4.6; 95% confidence interval [95% CI], 1.8 to 7.5; P=0.002) (Figure 1), high total PTH (>540 pg/ml; β=7.1; 95% CI, 2.8 to 11.3; P=0.001), and elevated whole PTH (>450 pg/ml; β=6.9; 95% CI, 2.4 to 11.4; P=0.003) (Figure 2). TRAP-5b was not an independent predictor after adjustment for age.
Higher CAC progression in osteoporotic patients. Mean CAC SQRV progression (±SEM) at 1 year in patients with CKD-5D who were osteoporotic and nonosteoporotic stratified by age groups. After adjustment for age, osteoporosis was a significant predictor of CAC progression (β=4.6; 95% CI, 1.8 to 7.5; P=0.002). Osteoporosis was determined by DXA or QCT t scores of the total hip, spine, or femoral neck.
Higher CAC progression in patients with parathyroid hormone (PTH) levels greater than 450 pg/mL. Mean CAC SQRV progression (±SEM) at 1 year in patients with CKD-5D stratified by baseline whole PTH levels and age. After adjustment for age, whole PTH levels that were nine times higher than normal predicted CAC progression (β=6.9; 95% CI, 2.4 to 11.4; P=0.003). Patients over 65 years of age are not shown because of a low number of patients with high PTH in this age group.
Differences with Respect to Treatment Modalities
CAC progression did not vary with treatment using antihypertensives or statins. Also, there were no significant differences between baseline or changes in total PTH, CAC, or FGF23 in patients treated with vitamin D analogs compared with those not treated (Table 8). Similarly, patients treated with calcium–containing phosphate binders did not significantly differ from patients treated with noncalcium–containing phosphate binders with respect to baseline or changes after 1 year in these parameters. Patients treated with cinacalcet had higher baseline PTH, higher baseline FGF23, and more CAC progression.
Baseline and 1-year changes in total PTH, CAC, and FGF23 in patients with CKD-5D stratified by medications taken
Changes in CAC and FGF23 in patients who had total PTH within the recommended range at baseline and 1 year had a median CAC progression of 20 and a median increase in FGF23 of 839 relative units/ml. Patients who did not stay within the recommended range at 1 year had worse CAC progression and higher increase in FGF23, although these differences were not significant (Table 9).
Changes in CAC and FGF23 in patients whose total PTH was within the recommended range at baseline stratified by their total PTH at 1 year
However, the design of the study was not aimed at evaluating the effect of different treatment modalities or PTH control on CAC progression. Rather, the aims of the study were to prospectively follow changes in CACs in patients managed according to published guidelines and find predictors for CAC progression. Elevated PTH and osteoporosis were identified as independent predictors.
Discussion
The results of this study show, in agreement with prior studies, that close to one half of patients with CKD-5D had baseline CACs conferring high risk for cardiovascular events.9,16,17 Furthermore, in keeping with prior reports, baseline CAC correlated with age and dialysis vintage2,18–24 and was higher in patients with diabetes25–28 and patients with a history of CAD.29,30 Contrary to other studies,2,20,25 we did not find a sex difference in baseline CAC. Baseline CAC was lower in patients who reported exercising compared with those who did not. This is a novel finding in patients with CKD-5D but in agreement with results in subjects without CKD.31
Use of DXA for measurement of BMD has been considered of limited usefulness for patients with CKD.32,33 This was mostly related to the occurrence of extraosseous calcifications that might be included in the measurement results. However, DXA measurements have recently been shown to be useful for assessment of osteoporosis in patients with CKD,1,34 and other investigators have found BMD measurements by DXA or QCT to be predictive of fractures in patients with CKD.35,36 We used QCT, which excludes extraosseous calcifications, in addition to DXA for the measurement of BMD in this study.
In this study, BMD of the total hip, femoral neck, and spine measured by QCT or DXA (except for DXA spine) is shown for the first time in patients with CKD-5D to correlate with baseline CAC as well as diagnosis of osteoporosis by t scores. This adds to cross-sectional studies on the relationship between CAC and bone volume assessed invasively by bone biopsy,1,2,5 one cross-sectional report using DXA of the spine in patients with CKD not on dialysis,4 and a study showing a relationship between vascular calcifications and fractures.37 An inverse relationship between CAC prevalence and BMD by DXA has also been shown in patients who are osteoporotic with normal kidney function and patients with HIV or metabolic syndrome.38–41
In this study, a positive correlation was found between CAC SQRV and FGF23, which is in agreement with several published reports on vascular calcifications42–46 but not in keeping with another report on patients with CKD not on dialysis.29 Baseline FGF23 remained an independent predictor for baseline CAC after multivariable adjustment for age, CAD, diabetes, dialysis vintage, and BMD.
The presented results in patients with CKD-5D do not confirm prior reports of relationships between baseline CAC and serum concentrations of phosphorus and calcium in patients with CKD before requiring dialysis and pediatric patients.29,47–49
These presented baseline findings are important for our understanding of the various factors contributing to the prevalence of CAC in patients with CKD-5D, but there is a great need for the clinician to learn about factors that help predict CAC progression in these patients. This prospective study addresses this need.
In agreement with prior studies, three quarters of patients experienced CAC progression over the 1-year course of this study.26,30,50–53 This progression was many fold higher than the intraobserver and interscan variabilities of 1% and 3%, respectively, for the current 64-slice MSCT technology used in this study. We confirm the risk factors for CAC SQRV progression: age and baseline CAC. In addition, this study provides the novel information that, after adjustment for age, both whole and total PTH greater than nine times the normal value (that is, outside the range recommended by Kidney Disease Improving Global Outcomes [KDIGO]) predict CAC progression in patients on dialysis. These results extend data showing that serum PTH levels are positively associated with CAC at baseline in patients on dialysis.20,54 PTH concentrations, although measured regularly, are usually not used to predict CAC progression. Although this study was not designed to test the influence of PTH control on CAC progression, we observed that patients who started and stayed within recommended PTH ranges over the course of the study had the least CAC progression. On the basis of the presented findings, we recommend that clinicians measure PTH to assess not only bone turnover but also risk of CAC progression in patients with CKD-5D. This recommendation is easy to implement given PTH measurements are already included in the dialysis bundle payments.
Another independent predictor of CAC progression shown by our study is osteoporosis by t scores. Osteoporosis determination by BMD measured with DXA or QCT technology has been shown to be useful in predicting fractures in patients with CKD.35,36 However, osteoporosis is rarely addressed in the routine management of patients with CKD-5D. If addressed, it is considered a risk factor only for fractures. Notably, current guidelines are not supportive of preventative surveillance for osteoporosis using easily available DXA measurements. Our results show that osteoporosis diagnosed by DXA is a useful noninvasive tool for predicting CAC progression with its known associated increased risk for cardiovascular events.55 This adds a reason for measurement of bone mass in CKD, which has been called for by others.56–61
Our results do not confirm increased CAC progression in patients who were treated with calcium–containing phosphate binders. However, this study was designed to identify predictors of CAC progression in a broadly representative clinical population and thus, does not allow us to draw conclusions on the effects of treatment modalities on CAC progression. We observed greater CAC progression in patients receiving cinacalcet, contrary to other reports, but our patients on cinacalcet had higher PTH levels to begin with; also, we found PTH to be an independent predictor of CAC progression. As a next step, additional controlled prospective studies are needed to evaluate the effects of individual PTH-influencing therapies on CAC progression.
Other limitations to this study are those inherent to any prospective clinical study (i.e., a certain percentage of patients is excluded because of their inability to participate). We excluded approximately 20% of the screened patients because of severe comorbidity or impaired mental status. This is acceptable and not unexpected given the high patient morbidity and mortality. The listed exclusion criteria regarding bone-modifying medications or illicit drug use did not result in any patient exclusions. Also inherent to a prospective clinical trial is the limited ability to address disease mechanisms. There may be parallel pathogenetic abnormalities affecting both bone and coronary vessels, and potential interactions cannot be addressed. Our results, however, ascribe an often underappreciated role to bone in CKD–associated CAC progression. Measurements of bone mass and factors regulating bone loss should be integrated into research in patients with CKD-mineral and bone disorder. This complex study was limited to 1 year, and longer-term relationships need to be studied. Also, additional prospective studies are needed to evaluate potential benefits of antiosteoporosis therapies on preventing, retarding, or reversing CAC progression in patients with CKD-5D.
Concise Methods
Participants
Patients with CKD-5D from 38 participating dialysis centers across Kentucky were screened, consented, and enrolled into this prospective institutional review board-approved study. The investigators adhered to the Declaration of Helsinki in the conduct of the study and registered the study with ClinicalTrials.gov (NCT00859672). Inclusion criteria were age 18 years or older, chronic maintenance dialysis of at least 3 months in duration, mental competence, willingness to participate in the study, and calcidiol levels within the normal range. Exclusion criteria included pregnancy, systemic illnesses or organ diseases that may affect bone (except diabetes mellitus), clinical conditions that may limit study participation (e.g., respiratory distress and infections), chronic alcoholism and/or drug addiction, participation in a study of an investigational drug during the past 90 days, planning to move out of the area within 1 year, on the active transplant list, and treatment within last 6 months with drugs that may affect bone metabolism (except for treatment with calcitriol, vitamin D analogs, and/or calcimimetics). At baseline and after 1 year, patients underwent measurement of CAC, BMD, and routine and novel serum biochemical parameters. Demographic and clinical parameters were also recorded at the same times. The treating nephrologist determined all treatments on the basis of standardized practice following Kidney Disease Outcomes Quality Initiative and KDIGO recommendations. There were no interventions by the investigators.
Determinations of Blood Parameters
Serum calcium, phosphorus, hemoglobin, albumin, bicarbonate, HDL, and LDL were measured using routine laboratory techniques. Calcidiol was determined by liquid chromatography-tandem mass spectrometry: the intra- and interassay coefficients of variation were <12.9% and <14.0%, respectively.
The following biochemical parameters were measured because of their novelty or their correlation with bone parameters in cross-sectional prior studies: serum whole and total PTH (commonly used to assess bone turnover abnormalities, which have been shown to be associated with bone volume62); BSAP and P1NP, markers of osteoblastic activity63,64; TRAP5-b, a marker of osteoclast activity65; sclerostin, a protein produced by osteocytes66,67 and expressed at bone formation sites68,69; DKK-1 (found in bone and other tissues,70 and like sclerostin, it leads to increased bone formation and bone volume when knocked out71,72); and FGF23, which is involved in mineral metabolism with a role in bone mineralization/remodeling.73–75
Plasma total and whole PTH levels were measured by radioimmunometric assays (Scantibodies Laboratory Inc., Santee, CA). The intra- and interassay coefficients of variation for total PTH were <5% and <7%, respectively, and the intra- and interassay coefficients of variation for whole PTH were <6% and <8%, respectively. BSAP levels were measured using an enzyme-linked immunosorbent assay (EIA) (Metra BAP EIA; Quidel, San Diego, CA). The intra- and interassay coefficients of variation were <6% and <8%, respectively. P1NP levels were measured using an ELISA (USCNK, Wuhan, China); the intra- and interassay coefficients of variation were <9% and <10%, respectively. TRAP-5b levels were determined using an EIA (MicroVue; Quidel, Santa Clara, CA).The intra- and interassay coefficients of variation were <2.2% and <3%, respectively. Serum sclerostin levels were measured using an EIA (Tecomedical Group, Sissach, Switzerland). The intra- and interassay coefficients of variation were <3.1% and <3.5%, respectively. DKK1 was determined by an ELISA (Biomedica, Vienna, Austria). The intra- and interassay coefficients of variation were <8% and <12%, respectively. Serum FGF23 levels were measured by an ELISA (EMD Millipore, Billerica, MA). The intra- and interassay coefficients of variation were <8% and <12%, respectively.
Measurements of BMD
For assessment of BMD, we used DXA, because it is the most widely used tool for assessment of bone mass and fracture risk in the general population.76 We also used QCT because of its stronger correlation with histologically determined bone volume.77 BMD of the spine, femoral neck, and total hip were assessed using both methods by the same International Society of Clinical Densitometry-certified operator using the same machines for the duration of the study. iDXA (GE Medical Systems Lunar, Madison, WI) was used for DXA; the coefficients of variation for DXA measurements were 1.35% for spine and 0.52% for hip. SOMATOM Sensation 16 was used for QCT using the QCT PRO software (Mindways Software Inc., Austin, TX). The coefficients of variation for QCT measurements were 0.80% for spine and 0.82% for hip. Changes in BMD were stratified into those losing >2% versus nonlosers.13,14 This 2% value is outside our error of DXA and QCT, and it is the threshold defined by the International Society for Clinical Densitometry.15
Measurement of CAC
CAC was assessed using MSCT of the heart. Noncontrast CT was performed on a Dual Source 64-Slice CT Scanner (Siemens AG, Erlangen, Germany). Images were acquired from the carina to the left ventricular apex. Scan parameters were electrocardiogram gating, 64×0.6 collimation, 120 kVp, 80 mAs/rotation, 0.33-seconds gantry rotation time, and 3-mm slice thickness. Images were analyzed on a three-dimensional workstation using calcium scoring software (HeartView CT; Siemens AG). Calcifications were identified as a plaque of ≥1 mm2 with a density of ≥130 Hounsfield units and quantified using the previously described Agatston scoring method.78 Intraobserver error for interpretation of images was established to be <1%. This was determined through repeated interpretation 2–4 weeks apart. In addition, interscan variability was determined (through repeat measurements 30 minutes apart) to be <3%. CAC was also determined at baseline and 1 year using CAC SQRV, an analytic method that accounts for interscan variability.79
Statistical Analyses
Results are given as means±SDs or medians (interquartile ranges). Univariate analyses of continuous variables were performed using Mann–Whitney U tests, t tests, or ANOVA as appropriate; for categorical variables, Fisher’s exact or chi-squared tests were used. Bivariate correlations were calculated using Spearman’s Rho test. Multivariable linear regression was used to assess predictive relationships with CAC SQRV at baseline and changes in CAC SQRV at 1 year. Significance for analyses was set at P<0.05. SPSS statistical software (version 22; IBM Corporation, Armonk, NY) was used for all calculations.
Disclosures
None.
Acknowledgments
The authors thank Kimberly McLaughlin and Nedda Hughes for their diligent and knowledgeable work in patient enrollment and data collection and Teresa Sexton for her work quantifying quantitative computed tomography and dual energy x-ray absorptiometry measurements.
Research reported in this publication was supported by National Institute of Arthritis and Musculoskeletal and Skin Diseases of the National Institutes of Health Award R01-080770, the Kentucky Nephrology Research Trust, and National Center for Advancing Translational Sciences, National Institutes of Health Grant UL1-TR000117.
The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health or the Kentucky Nephrology Research Trust.
Footnotes
Published online ahead of print. Publication date available at www.jasn.org.
- Copyright © 2015 by the American Society of Nephrology