Kidney transplantation considerably improves the prognosis of patients with ESRD; yet in renal transplant recipients (RTR) the annual rate of fatal or nonfatal cardiovascular events remains much higher compared with the general population.1,2 Kidney transplantation restores renal function but incompletely mitigates cardiovascular risk. This appears to be mainly a result of the markedly accelerated vascular calcification in RTR. Vascular calcification progresses substantially even in stable RTR and is an established predictor of morbidity and mortality.3–5 The so-called traditional risk factors tobacco use, diabetes, obesity, hypertension, and dyslipidemia (i.e., the Framingham risk factors) have been identified as independent predictors of cardiovascular disease after kidney transplantation.6
Biomineralization is a tightly regulated and location-specific process, determined by both inhibitors and promoters of calcification. In serum, precipitation of supersaturated calcium and phosphate is prevented by the formation of primary calciprotein particles (CPPs).7,8 Primary CPPs spontaneously undergo topological rearrangement and convert to secondary CPPs.7 The formation of CPPs can be regarded as a defense mechanism against ectopic calcifications. Circulating CPPs can be detected in blood in pathologic situations like CKD, and their levels have been associated with aortic stiffness and vascular calcification.9,10
Recently, a novel in vitro blood test was developed that monitors the calcification propensity, i.e., the transformation time (T50) of primary to secondary CPPs in serum.11 A long delay of T50 indicates a high residual capacity of the patient’s serum to prevent the formation of secondary CPPs and is therefore indicative of an intact endogenous defense against calcification.
In this study, we aimed to quantify calcification propensity in stable RTR and to determine its association with all-cause mortality, cardiovascular mortality, and graft failure risk by using the CPP maturation time (T50) in serum. Furthermore, we compared the predictive value of serum T50 for the risk of mortality with the established “traditional” cardiovascular risk factors, i.e., the “Framingham risk factors” and the calcium-phosphate product. Finally, we performed mediation analyses to identify potential causal pathways shared among T50 and two established nontraditional cardiovascular risk factors in RTR: N-terminal–pro brain natriuretic peptide (NT-proBNP)12 and fibroblast growth factor 23 (FGF23).13
All chronically kidney transplanted patients from the University Medical Center Groningen, transplanted for more than 1 year, were recruited for this study (n=699, 57% male, age 53±13 years). The median time after kidney transplantation upon inclusion in the cohort was 5.4 (1.9–12.1) years. Serum T50 was normally distributed and mean serum T50 was 286±62 minutes (Supplemental Figure 1). Baseline patient characteristics according to tertiles of T50 are presented in Table 1. Calcification propensity was significantly increased (i.e., lower serum T50) in RTR who had received a kidney from a deceased donor compared with RTR with a living donor (278±62 versus 299±58minutes, respectively; t test P<0.001). Furthermore, there was a borderline significant difference between preemptive and nonpreemptive transplant recipients (296±57 versus 284±62 minutes, respectively; t test P=0.07). Multivariable linear regression analyses revealed that serum magnesium, albumin, venous bicarbonate, and parathyroid hormone (PTH) were positively associated, and phosphate, hemoglobin, and the use of calcineurin inhibitors (CNI) or vitamin K antagonists were inversely associated with serum T50 (Supplemental Table 1; model R2=0.41). Of note, in multivariate analyses renal function and dialysis vintage were not independently associated with serum T50.
Serum T50 and Mortality
During a median follow-up of 3.1 (interquartile range [IQR] 2.7–3.9) years, 81 out of 699 patients died (12%). Kaplan–Meier analysis revealed an increased all-cause mortality with tertiles of decreasing serum T50 (Figure 1A; log-rank P<0.001); first tertile n=43 (19%), second tertile n=24 (10%), and third tertile n=14 (6%); chi-squared test P<0.001.
Also, in univariable Cox regression analysis of serum T50 as a continuous variable, serum T50 was significantly associated with all-cause mortality (hazard ratio [HR], 1.69; 95% confidence interval [95% CI], 1.37 to 2.09; P<0.001 per SD decrease, Table 2). Hazard ratios were consistent in subgroup analyses (Figure 2). This association remained significant after adjustment for potential confounders, including recipient age and gender, eGFR (Chronic Kidney Disease Epidemiology Collaboration [CKD-EPI]), albuminuria, the traditional risk factors (current smoking, systolic BP, body mass index, current diabetes and LDL cholesterol), high-sensitivity C-reactive protein, CNI use, dialysis vintage and type of kidney transplantation (living versus deceased donor) (final model HR, 1.43; 95% CI, 1.11 to 1.85; P=0.006 per SD decrease) (Table 2). When analyzed according to tertiles of T50, patients in the lowest T50 tertile (i.e., higher calcification propensity) were at a considerably higher mortality risk (final model HR, 2.86; 95% CI, 1.41 to 5.80) than patients in the middle (final model HR, 1.58; 95% CI, 0.76 to 3.32) or higher T50 tertile (reference, P[trend]=0.002).
During follow-up, 38 (47%) of the 81 patients who died, died due to cardiovascular causes. Other causes of death were infection (24%), malignancy (16%), and miscellaneous and other causes (14%). Kaplan–Meier analysis revealed an increased cardiovascular mortality with tertiles of decreasing serum T50 (Figure 1B; log-rank P<0.001). In Cox regression analysis, serum T50 was significantly associated with cardiovascular mortality independent of potential confounders (final model HR, 1.55; 95% CI, 1.04 to 2.29; P=0.03 per SD decrease) (Table 2).
To compare the performance of serum T50 with other biochemical measures of calcification as individual predictors of all-cause mortality, separate Cox regression analyses were performed for each variable. As indicated in Figure 3, serum T50 displayed the strongest association with all-cause mortality of all tested parameters, both in unadjusted and adjusted models. To identify possible pathophysiological pathways of T50, the associations between T50 and FGF23, a phosphaturic hormone or NT-proBNP, a marker of cardiac dysfunction, and all-cause mortality, was tested using mediation analyses (Supplemental Figure 2). The indirect effect of T50 on mortality via FGF23 was not significant. In contrast, the indirect effect of T50 on mortality via NT-proBNP was significant (P<0.05) with a magnitude of 10.7%.
Predictive Value of Serum T50 for All-Cause Mortality
The discrimination performance of different models was tested with Harrell’s concordance statistic (c-statistic). Serum T50 alone had a c-statistic (95% CI) of 0.67 (0.61 to 0.73). When added to a basic model of recipient age, gender, and eGFR (c-statistic, 0.72; 95% CI, 0.66 to 0.78), the serum T50 yielded higher Harrell’s c values than Framingham risk factors or calcium-phosphate product: c-statistics 0.75 (95% CI, 0.70 to 0.80); 0.73 (95% CI, 0.67 to 0.79); 0.72 (95% CI, 0.66 to 0.78), respectively; see Supplemental Table 2. The model containing serum T50, but not the model containing Framingham risk factors, improved significantly according to the integrated discrimination improvement index (IDI, 1.6%; P=0.03 versus 0.8%, P=0.3; see Supplemental Table 2). Net reclassification improvement (NRI) analysis showed that addition of serum T50 to a basic model of recipient age, gender, and eGFR improved classification in both higher and lower risk categories (NRI 14%, P=0.002; see Table 3), whereas addition of traditional risk factors or calcium-phosphate product did not improve classification (data not shown, NRI –2%, P=0.70; NRI 0.7%, P=0.80, respectively).
Serum T50 and Graft Failure
After a median follow-up of 3.1 (IQR, 2.6–3.8) years, 45 of 699 patients (6%) developed graft failure. The main reason for the allograft loss was chronic transplant dysfunction (71%). Other causes for graft failure were relapse of the original kidney disease (9%), infection (9%), vascular disease (4%), acute rejection (4%), and unknown cause (2%). The patients who developed graft failure had lower serum T50 compared with the patients who did not develop graft failure (235±68 versus 289±60, respectively, P<0.001). Competing risks analyses using a subdistribution proportional hazards model indicated that all-cause mortality was not a competing risk for graft failure. Kaplan–Meier analysis revealed a graded increase in risk of death-censored graft failure according to descending tertiles of serum T50 (Figure 1C; log-rank P<0.001). In univariable analyses of serum T50 as a continuous variable, serum T50 was significantly associated with death-censored graft failure (HR, 2.61; 95% CI, 1.94 to 3.50; P<0.001 per SD decrease). This association remained significant after adjustment for potential confounders, including recipient age and gender, eGFR (CKD-EPI), and albuminuria or CNI use, dialysis vintage, and the type of donor (Supplemental Table 3). Furthermore, patients in the lowest T50 tertile were at a higher graft failure risk compared with patients in the higher T50 tertile (HR, 6.38; 95% CI, 2.63 to 15.48; P<0.001) (Supplemental Table 3). Serum T50 displayed significant associations with graft failure in adjusted models in separate Cox regression analyses (Supplemental Figure 3).
Independent Replication Cohort
Renal transplant recipients (n=198, 68% male, age 52.9±13.0 years) were included at 5.2 (IQR, 1.8–10.2) years after kidney transplantation. Diabetes mellitus was present in 24%, mean body mass index was 26.1±4.4 kg/m2, mean systolic BP was 137±19 mmHg and mean diastolic BP was 85±10 mmHg, 9% of the patients were transplanted preemptively, median dialysis vintage was 17 (7–36) months, mean eGFR was 49.2±18.9 ml/min per 1.73 m2, mean phosphate was 1.05±0.26 mmol/L, mean (corrected) calcium was 2.45±0.14 mmol/L and median PTH was 6.3 (4.3–9.7) pmol/L. At study inclusion, immunosuppressive treatment was mainly based on a calcineurin-inhibitor (77%) combined with prednisone (93%). All but eight patients received at least one antihypertensive drug (96%).
During 5.1 (IQR, 4.9–5.3) years of follow-up, 28 (14%) patients died and 14 (7%) patients developed graft failure. In univariable Cox regression analyses, serum T50 was significantly associated with all-cause mortality (HR, 1.68; 95% CI, 1.03 to 2.74; P=0.04 per SD decrease) and death-censored graft failure (HR, 3.80; 95% CI, 1.53 to 9.45; P=0.004 per SD decrease).
The measurement of the CPP maturation time (T50), an integrated functional measure of calcification propensity, is a novel approach in medicine. T50 quantifies the effects of calcification-promoting and -inhibiting forces in serum by monitoring the spontaneous transformation of spherical primary CPPs, which contain amorphous calcium phosphate, to spindle-shaped secondary CPPs, which contain crystalline calcium phosphate.11 The major findings of this study are that serum T50 is associated with all-cause and cardiovascular mortality and graft failure in stable RTR. In both unadjusted and adjusted models, these associations were already observed after 3 years of follow-up and were independent of established risk factors. The associations of T50 with all-cause mortality and graft failure were confirmed in an independent cohort. Corroborating and extending these findings, calcification propensity had additive value for patient survival prediction in addition to recipient age, gender, and eGFR.
The in vitroT50 measurement is very stable intraindividually.14 Within-subject variance was 19% and between-subject variance was 81%, i.e., compared with serum phosphate, serum T50 measurement yielded a higher intraclass correlation coefficients (0.63; 95% CI, 0.34 to 0.71; and 0.81, 95% CI, 0.68 to 0.91, respectively), indicating a higher stability of T50 than serum phosphate in individuals over time.14
Serum phosphate, magnesium, venous bicarbonate and albumin were the strongest determinants of T50 in our multivariable linear regression model. This model explained about 40% of the variation of T50. Amorphous calcium phosphate is formed as the first mineral phase and consists of so-called Posner clusters (Ca9[PO4]6),15 when calcium and phosphate concentrations are raised in aqueous solutions. This amorphous mineral phase is then spontaneously transformed via octacalcium phosphate (Ca8H2[PO4]6) to hydroxyapatite (Ca10[PO4]6[OH]2).16In vitro data demonstrate that the molecules found in our multivariate analysis have direct accelerating (phosphate) or delaying (magnesium, bicarbonate, and albumin) effects on this transformation process.11,17–19 Although PTH was also an independent determinant of serum T50, PTH did not differ among the tertiles of serum T50. The explanation for this seeming discrepancy is probably that continuous analysis has more statistical power to detect associations than categorical analysis. Interestingly, total as well as albumin-corrected calcium did not have a strong impact on T50, a result also in line with previous in vitro and clinical data.11,14 After multivariable modeling, serum T50 lost its significant association with renal function in stable RTR, corroborating data previously obtained in CKD patients.14 Of note, the multivariable model did not include dialysis vintage, indicating that T50 is independent of time of previous uremic exposure in ESRD before transplantation.
Our results extend recently published findings in predialysis CKD patients, where T50 was associated with all-cause mortality.14 In the renal transplant population, several parameters of calcium-phosphate metabolism have been associated with adverse outcome.20 In addition, traditional cardiovascular risk factors in the general population (i.e., the Framingham risk factors) are predictive of cardiovascular events and therefore should also be considered in RTR according to current guidelines.6 Interestingly, these risk factors had no predictive value in our study. This indicates that calcification propensity T50 as an integrated value representing non-traditional risk factors may be of clinical relevance as a prognostic novel biomarker in RTR.
Pathophysiologically, the close association of T50 with all-cause and cardiovascular mortality is potentially related to its association with naturally occurring CPPs in renal patients. CPPs have been found in different stages of preclinical and clinical CKD and it is prudent to assume that these particles can also be encountered in RTR.9,21,22 Calcification propensity measured as T50, had a higher predictive value for future events and survival than the direct determination of circulating CPPs.14
T50 might share common pathophysiological pathways with established markers of mineral metabolism or impairment of cardiac function. We, therefore, tested whether the association between T50 and mortality was modulated by FGF23, an important regulator of mineral metabolism, or NT-proBNP, a strong predictor of mortality in RTR.12 Although no effect of FGF23 was found, NT-proBNP had a small but significant modulating effect on the association between T50 and mortality. This indicates that the mechanism by which T50 may exert its effects includes a pathophysiologic pathway and mechanism partially shared with NT-proBNP. We propose that a higher calcification propensity, by promoting vascular calcification over time, may predispose to an adverse hemodynamic state (volume overload, cardiac dysfunction), as reflected by higher NT-proBNP levels.23,24 Our data suggest that this pathway underlies at least in part the association between a lower T50 and a higher mortality risk.
Interestingly, T50 was also associated with graft failure in two independent cohorts, indicating that a tight control of calcification propensity may also be of high relevance for single organs or organ systems. A hypothetical causal relation may exist through the stiffness of the renal vasculature and intrarenal resistance, a marker currently used in renal ultrasound imaging for the prediction of renal aging and rejection.25,26 Alternatively, but not mutually exclusively, the mineralization propensity within the tubular system might be related to T50 and to the progression of renal failure. This hypothesis was initially proposed as the precipitation-calcification hypothesis, and is currently being reconsidered in the field of renal failure progression and premature aging.27–29
T50 might be of clinical use for the early detection of patients at high mortality risk. Furthermore, serum T50 may be of use as a guide for the treatment of renal patients. Therapeutic strategies to improve serum calcification propensity should be identified in future clinical trials. Based on our analysis of T50 determinants and previous in vitro data, interventions lowering serum phosphate (i.e., dietary restriction and/or phosphate binders) or supplementation of magnesium or bicarbonate may be suitable to improve serum T50 as an intermediate end point.
Our study has several limitations and strengths. Although prospective, it is of observational nature and therefore a causality of the associations found, although likely, cannot be proven. The prespecified all-cause mortality risk categories (<5%/ 5%–10%/ >10%) used in the NRI analysis have not been validated. The absolute meaning of these risk categories should be interpreted with caution. Furthermore, no clinical data on vascular calcification as an intermediate end point were available. On the other hand, strengths include the complete follow-up and uniform single-center handling of samples along with the hard and clinically relevant end points (all-cause and cardiovascular mortality, and graft failure) and the confirmation of the major findings in a replication cohort.
In conclusion, increased serum calcification propensity (i.e., lower serum T50) is a novel in vitro diagnostic tool and potent predictor and functional biomarker of all-cause and cardiovascular mortality and of graft failure in long-term RTR, which substantially improves mortality prognostication. Intervention studies based on T50 measurements are needed to clarify whether therapeutic targeting of serum T50 improves patient and graft survival after kidney transplantation.
Research Design and Subjects
For this longitudinal cohort study, RTR with a functioning graft for over 1 year were recruited. A total of 707 out of 817 eligible RTR (87%) who visited the outpatient clinic of the University Medical Center Groningen, The Netherlands between November 2008 and June 2011 were included.30,31 This cohort design yielded a cohort representative of our outpatient clinic of prevalent RTR with a median time after transplantation of 5.4 (IQR, 1.9–12.1) years. T50 was finally measured in 699 RTR with available serum samples (99%). The Institutional Review Board approved the study protocol (METc 2008/186), which adhered to the Declaration of Helsinki.
Clinical End Points
The primary end point of this study was all-cause mortality and the secondary end points were cardiovascular mortality (defined as death due to cerebrovascular disease, ischemic heart disease, heart failure, or sudden cardiac death) and death-censored graft failure (defined as restart of dialysis or retransplantation). End points were recorded until the end of May 2013; median follow-up was 3.1 (IQR, 2.7–3.9) years. There was no loss to follow-up.
Clinical and Laboratory Measurements
Relevant transplant characteristics such as date of transplantation, donor characteristics, HLA mismatches, dialysis vintage and acute rejection were extracted from the local University Medical Center Groningen renal transplantation database. Cause of death or graft failure was obtained from patient records. Further details of the cohort have been published previously.30,31 In short, upon entry into the cohort, blood was drawn in the morning after completion of a 24-hour urine collection. Routine plasma and urine analyses, including electrolytes, creatinine, and albumin, were performed using standard laboratory procedures. Serum cholesterol, triglycerides, alkaline phosphatase, hemoglobin A1c, high-sensitivity C-reactive protein, and NT-proBNP were also directly analyzed using standard laboratory procedures. Serum samples were stored at –80°C. The eGFR was calculated using the CKD-EPI equation.32 Serum calcium was adjusted for hypoalbuminemia (<40 g/L); corrected calcium=serum calcium (mmol/L)+0.02×(40–serum albumin [g/L]). Serum intact FGF23 levels were determined using a commercially available ELISA kit (Kainos Laboratories, Inc., Tokyo, Japan). Intra-assay and interassay coefficients of variation are <10% and <14%, respectively.33 Serum magnesium was measured using the xylidyl blue method.34
Serum T50 Measurement
The T50 calcification propensity test has been described previously.11 For the present study, the test was performed with minor changes that did not materially change the test when compared with the original description. Here, serum samples were measured in triplicate in 384-well plates at 37°C over 600 minutes in a Nephelostar nephelometer (BMG Labtech, Ortenberg, Germany). All serum samples were measured in a blinded manner. Stock solutions were a calcium and a phosphate solution. The pH was adjusted to 7.40 at 37°C in both solutions. For measurement, 35 μl of calcium solution was mixed with 40 μl serum, and then 25 μl of the phosphate solution was added. Data analyses of nonlinear regression curves were performed using Microsoft Excel software to determine the half-maximal precipitation time (T50). The analytical coefficients of variation of standards precipitating at 120, 260, and 390 minutes were 7.8%, 5.1%, and 5.9%, respectively.
Independent Replication Cohort
Between August 2006 and February 2007, all RTR from the outpatient clinic of the Department of Nephrology at the University Hospital Bern were screened for inclusion in this cohort study. Patients were included if they had been transplanted at least 6 months before inclusion and revealed a stable renal graft function (serum creatinine stable within a range of ±20% in the last 3 months) with no signs of acute rejection. Patients with cardiac arrhythmia, hemodynamic significant transplant renal artery stenosis, hydronephrosis > grade 2 or compression of the transplant by adjacent masses, were excluded from the study. A total of 222 RTR were screened and met the general inclusion criteria. Of these, 15 refused consent and seven were excluded because of screening failure. The follow-up was 52 months. Two patients were lost during follow-up. The study was approved by the local ethics committee and registered with the Cochrane renal group database (http://www.cochrane-renal.org, CRG110600098).
Routine laboratory values were obtained using standard techniques. The GFR was estimated according to the CKD-EPI formula.32 Serum calcification propensity, T50, was measured as described above.
The data are presented as mean±SD, median (IQR), and number (percentage) as indicated. A P value <0.05 (two-tailed) was considered statistically significant. Statistical analyses were performed using SPSS 20.0 for Windows (IBM SPSS, Chicago, IL), STATA Statistical Software: Release 11 (StataCorp., College Station, TX) and GraphPad Prism version 5.00 for Windows (GraphPad Software, San Diego, CA).
Variable distribution was tested with histograms and probability plots. For illustrative purposes, the study population was subdivided into tertiles of T50 to visualize associations with serum T50. P values for differences in T50 tertiles were assessed with ANOVA for normally distributed continuous data, the Kruskal–Wallis test for non-normally distributed data and the chi-squared test for nominal data. Univariable and subsequent multivariable linear regression analyses were used to identify independent determinants of T50. Non-normally distributed variables were transformed to the natural log to fulfill criteria for linear regression analyses. Multivariable linear regression models were constructed using backward selection (Pout>0.05) including variables that were significantly associated with T50 in univariable analysis.
Tertiles of serum T50 were tested for associations with all-cause mortality and death-censored graft failure by Kaplan–Meier analysis with log-rank testing. Associations of T50 with all-cause mortality, cardiovascular mortality or graft failure were further tested by Cox proportional hazards regression analysis with stepwise adjustments for relevant covariates. Non-normally distributed variables were transformed to the natural log before entering the Cox proportional hazards regression analysis models. The full model for all-cause or cardiovascular mortality included adjustment for age, gender, renal function, albuminuria, the Framingham risk factors, high-sensitivity C-reactive protein, CNI use, dialysis vintage, and type of kidney transplantation (living or deceased). The models for graft failure included adjustment for (recipient) age and gender (model 2) plus renal function, albuminuria (model 3), or CNI use, dialysis vintage, and the type of kidney transplantation (living or deceased) (model 4). Cox regression models were built stepwise to keep the number of covariates accurate in relation to the number of events and to avoid over fitting.35
In additional sensitivity analyses, associations of T50 with all-cause mortality were tested by Cox proportional hazards regression analysis in subgroups. For continuous variables the subgroups were based on below or above mean or median. To compare the performance of serum T50 with serum corrected calcium, serum phosphate, serum magnesium, serum PTH and calcium-phosphate product as individual risk factors for all-cause mortality or graft failure, separate Cox regression analyses were performed for each variable and adjusted for known risk factors of mortality or graft failure, respectively. For each exposure, the first or last tertile served as the reference group (hazard ratio set at 1) depending on which tertile had the lowest risk of an event.
We assessed model discrimination using Harrell’s concordance statistic (c-statistic), the NRI and the integrated discrimination improvement index.36,37
Harrell’s c-statistic corresponds to the area under the receiver-operating curve for proportional hazards models. Harrell’s c-coefficient is the proportion of all usable subject pairs in which the predictions and outcomes are concordant. The value “1” implies a perfect discrimination, whereas the value “0.5” implies a performance comparable to chance. The NRI provides reclassification tables constructed separately for patients with and without events and quantifies the correct movement between categories: upwards for events and downward for non-events. For the NRI we used the following classifications: low (<5%), intermediate (5%–10%), or high (>10%) all-cause mortality risk. Addition of serum T50, traditional risk factors or calcium-phosphate product to the basic model including recipient age and gender and eGFR (CKD-EPI) was tested. Improved classification was defined as upward movement in the risk categories for participants who have died and downward movement for participants who survived. The integrated discrimination improvement index represents a continuous measure without a priori–defined risk categories. Competing risks analyses were performed using a subdistribution proportional hazards model.38
Possible mediation by FGF23 or NT-proBNP on the association between T50 and all-cause mortality was examined (Supplemental Figure 2). The Preacher and Hayes method was used to test magnitude and significance of mediation.39,40 First, the total effect of T50 on mortality was estimated using logistic regression analysis. Second, the indirect effect of T50 on mortality via FGF23 or NT-proBNP was calculated. Third, the significance of the indirect effect was assessed with bias-corrected bootstrap confidence intervals with 2000 repetitions. Finally, the magnitude of mediation was calculated by dividing the coefficient of the indirect effect by the total effect. Significant mediation (P<0.05) was proven if zero was not between the lower and upper boundary of the 95% confidence interval of the indirect effect.
A.P. and W.J.-D. are co-founders of Calcisco AG, Bern, Switzerland.
The generation of the cohort was made possible by a grant from the Dutch Top Institute Food and Nutrition. M.H.B. is supported by personal grants from the Dutch Kidney Foundation (grant no. KJPB.08.07) and The Netherlands Organization for Scientific Research (Veni grant). M.H.B. and C.A.K. are supported by a consortium grant from the Dutch Kidney Foundation (NIGRAM consortium, grant no. CP10.11). This project was supported by a grant from the National Center for Competence in Research, NCCR kidney.ch to A.P. We thank Beatrix Blanchard for her valuable technical help.
Parts of this study were presented during the Late breaking clinical trials 2 session at the 51st ERA-EDTA Congress, Amsterdam, The Netherlands, May 31 to June 3, 2014 and during the poster session at the sixth congress of the German Society of Nephrology (DGfN), Berlin, Germany September 6–9, 2014 (independent replication cohort).
Published online ahead of print. Publication date available at www.jasn.org.
This article contains supplemental material online at http://jasn.asnjournals.org/lookup/suppl/doi:10.1681/ASN.2014070670/-/DCSupplemental.
1. Jardine AG, Gaston RS, Fellstrom BC, Holdaas H: Prevention of cardiovascular disease in adult recipients of kidney transplants. Lancet 378: 1419–1427, 2011
2. Ojo AO: Cardiovascular complications after renal transplantation
and their prevention. Transplantation 82: 603–611, 2006
3. Maréchal C, Coche E, Goffin E, Dragean A, Schlieper G, Nguyen P, Floege J, Kanaan N, Devuyst O, Jadoul M: Progression of coronary artery calcification and thoracic aorta calcification in kidney transplant recipients. Am J Kidney Dis 59: 258–269, 2012
4. Hernández D, Rufino M, Bartolomei S, González-Rinne A, Lorenzo V, Cobo M, Torres A: Clinical impact of preexisting vascular calcifications on mortality after renal transplantation
. Kidney Int 67: 2015–2020, 2005
5. Nguyen PT, Henrard S, Coche E, Goffin E, Devuyst O, Jadoul M: Coronary artery calcification: a strong predictor of cardiovascular events in renal transplant recipients. Nephrol Dial Transplant 25: 3773–3778, 2010
6. Kidney Disease: Improving Global Outcomes (KDIGO) Transplant Work Group: KDIGO clinical practice guideline for the care of kidney transplant recipients. Am J Transplant 9[Suppl 3]: S1–S155, 2009
7. Heiss A, Jahnen-Dechent W, Endo H, Schwahn D: Structural dynamics of a colloidal protein–mineral complex bestowing on calcium phosphate a high solubility in biological fluids. Biointerphases 2: 16–20, 2007
8. Heiss A, DuChesne A, Denecke B, Grötzinger J, Yamamoto K, Renné T, Jahnen-Dechent W: Structural basis of calcification inhibition by alpha 2-HS glycoprotein/fetuin-A. Formation of colloidal calciprotein particles
. J Biol Chem 278: 13333–13341, 2003
9. Hamano T, Matsui I, Mikami S, Tomida K, Fujii N, Imai E, Rakugi H, Isaka Y: Fetuin–mineral complex reflects extraosseous calcification stress in CKD. J Am Soc Nephrol 21: 1998–2007, 2010
10. Smith ER, Ford ML, Tomlinson LA, Rajkumar C, McMahon LP, Holt SG: Phosphorylated fetuin-A-containing calciprotein particles
are associated with aortic stiffness and a procalcific milieu in patients with pre-dialysis CKD. Nephrol Dial Transplant 27: 1957–1966, 2012
11. Pasch A, Farese S, Gräber S, Wald J, Richtering W, Floege J, Jahnen-Dechent W: Nanoparticle-based test measures overall propensity for calcification in serum. J Am Soc Nephrol 23: 1744–1752, 2012
12. Oterdoom LH, de Vries AP, van Ree RM, Gansevoort RT, van Son WJ, van der Heide JJ, Navis G, de Jong PE, Gans RO, Bakker SJ: N-terminal pro-B-type natriuretic peptide and mortality in renal transplant recipients versus the general population. Transplantation 87: 1562–1570, 2009
13. Baia LC, Humalda JK, Vervloet MG, Navis G, Bakker SJ, de Borst MHNIGRAM Consortium: Fibroblast growth factor 23 and cardiovascular mortality after kidney transplantation. Clin J Am Soc Nephrol 8: 1968–1978, 2013
14. Smith ER, Ford ML, Tomlinson LA, Bodenham E, McMahon LP, Farese S, Rajkumar C, Holt SG, Pasch A: Serum calcification propensity
predicts all-cause mortality in predialysis CKD. J Am Soc Nephrol 25: 339–348, 2014
15. Posner AS, Betts F: Synthetic amorphous calcium phosphate and its relation to bone mineral structure. Acc Chem Res 8: 273–281, 1975
16. Habraken WJ, Tao J, Brylka LJ, Friedrich H, Bertinetti L, Schenk AS, Verch A, Dmitrovic V, Bomans PH, Frederik PM, Laven J, van der Schoot P, Aichmayer B, de With G, DeYoreo JJ, Sommerdijk NA: Ion-association complexes unite classical and non-classical theories for the biomimetic nucleation of calcium phosphate. Nat Commun 4: 1507doi:10.1038/ncomms2490, 2013
17. Termine JD, Posner AS: Calcium phosphate formation in vitro
. I. Factors affecting initial phase separation. Arch Biochem Biophys 140: 307–317, 1970
18. Termine JD, Peckauskas RA, Posner AS: Calcium phosphate formation in vitro
. II. Effects of environment on amorphous-crystalline transformation. Arch Biochem Biophys 140: 318–325, 1970
19. Heiss A, Eckert T, Aretz A, Richtering W, van Dorp W, Schäfer C, Jahnen-Dechent W: Hierarchical role of fetuin-A and acidic serum proteins in the formation and stabilization of calcium phosphate particles. J Biol Chem 283: 14815–14825, 2008
20. Schaeffner ES, Födinger M, Kramar R, Sunder-Plassmann G, Winkelmayer WC: Prognostic associations of serum calcium, phosphate and calcium phosphate concentration product with outcomes in kidney transplant recipients. Transpl Int 20: 247–255, 2007
21. Matsui I, Hamano T, Mikami S, Fujii N, Takabatake Y, Nagasawa Y, Kawada N, Ito T, Rakugi H, Imai E, Isaka Y: Fully phosphorylated fetuin-A forms a mineral complex in the serum of rats with adenine-induced renal failure. Kidney Int 75: 915–928, 2009
22. Smith ER, Cai MM, McMahon LP, Pedagogos E, Toussaint ND, Brumby C, Holt SG: Serum fetuin-A concentration and fetuin-A-containing calciprotein particles
in patients with chronic inflammatory disease and renal failure. Nephrology (Carlton) 18: 215–221, 2013
23. Jouni H, Rodeheffer RJ, Kullo IJ: Increased serum N-terminal pro-B-type natriuretic peptide levels in patients with medial arterial calcification and poorly compressible leg arteries. Arterioscler Thromb Vasc Biol 31: 197–202, 2011
24. Coutinho T, Turner ST, Mosley TH, Kullo IJ: Biomarkers associated with pulse pressure in African-Americans and non-Hispanic whites. Am J Hypertens 25: 145–151, 2012
25. Schwenger V, Keller T, Hofmann N, Hoffmann O, Sommerer C, Nahm AM, Morath C, Zeier M, Krumme B: Color Doppler indices of renal allografts depend on vascular stiffness of the transplant recipients. Am J Transplant 6: 2721–2724, 2006
26. Radermacher J, Mengel M, Ellis S, Stuht S, Hiss M, Schwarz A, Eisenberger U, Burg M, Luft FC, Gwinner W, Haller H: The renal arterial resistance index and renal allograft survival. N Engl J Med 349: 115–124, 2003
27. Lau K: Phosphate excess and progressive renal failure: the precipitation-calcification hypothesis. Kidney Int 36: 918–937, 1989
28. Kuro-o M: Klotho, phosphate and FGF-23 in ageing and disturbed mineral metabolism. Nat Rev Nephrol 9: 650–660, 2013
29. Stenvinkel P, Larsson TE: Chronic kidney disease: a clinical model of premature aging. Am J Kidney Dis 62: 339–351, 2013
30. van den Berg E, Engberink MF, Brink EJ, van Baak MA, Joosten MM, Gans RO, Navis G, Bakker SJ: Dietary acid load and metabolic acidosis in renal transplant recipients. Clin J Am Soc Nephrol 7: 1811–1818, 2012
31. van den Berg E, Pasch A, Westendorp WH, Navis G, Brink EJ, Gans RO, van Goor H, Bakker SJ: Urinary sulfur metabolites associate with a favorable cardiovascular risk profile and survival benefit in renal transplant recipients. J Am Soc Nephrol 25: 1303–1312, 2014
32. Levey AS, Stevens LA, Schmid CH, Zhang YL, Castro AF 3rd, Feldman HI, Kusek JW, Eggers P, Van Lente F, Greene T, Coresh JCKD-EPI (Chronic Kidney Disease Epidemiology Collaboration): A new equation to estimate glomerular filtration rate. Ann Intern Med 150: 604–612, 2009
33. Heijboer AC, Levitus M, Vervloet MG, Lips P, ter Wee PM, Dijstelbloem HM, Blankenstein MA: Determination of fibroblast growth factor 23. Ann Clin Biochem 46: 338–340, 2009
34. Bohuon C: Mircodetermination of magnesium in various biological media. Clin Chim Acta 7: 811–817, 1962
35. Harrell FE Jr, Lee KL, Mark DB: Multivariable prognostic models: issues in developing models, evaluating assumptions and adequacy, and measuring and reducing errors. Stat Med 15: 361–387, 1996
36. Pencina MJ, D’Agostino RBS Sr, D’Agostino RB Jr, Vasan RS: Evaluating the added predictive ability of a new marker: from area under the ROC curve to reclassification and beyond. Stat Med 27: 157–172, discussion 207–212, 2008
37. Steyerberg EW, Vickers AJ, Cook NR, Gerds T, Gonen M, Obuchowski N, Pencina MJ, Kattan MW: Assessing the performance of prediction models: a framework for traditional and novel measures. Epidemiology 21: 128–138, 2010
38. Noordzij M, Leffondré K, van Stralen KJ, Zoccali C, Dekker FW, Jager KJ: When do we need competing risks methods for survival analysis in nephrology? Nephrol Dial Transplant 28: 2670–2677, 2013
39. Preacher KJ, Hayes AF: SPSS and SAS procedures for estimating indirect effects in simple mediation models. Behav Res Methods Instrum Comput 36: 717–731, 2004
40. Hayes AF: Beyond Baron and Kenny: Statistical mediation analysis in the new millennium. Commun Monogr 76: 408–420, 2009