Effect of the phosphate binder sucroferric oxyhydroxide in dialysis patients on endogenous calciprotein particles, inflammation, and vascular cells

ABSTRACT

Background

Calciprotein particles (CPPs), colloidal mineral-protein nanoparticles, have emerged as potential mediators of phosphate toxicity in dialysis patients, with putative links to vascular calcification, endothelial dysfunction and inflammation. We hypothesized that phosphate binder therapy with sucroferric oxyhydroxide (SO) would reduce endogenous CPP levels and attenuate pro-calcific and pro-inflammatory effects of patient serum towards human vascular cells in vitro.

Methods

This secondary analysis of a randomised controlled crossover study compared the effect of 2-week phosphate binder washout with high-dose (2000 mg/day) and low-dose (250 mg/day) SO therapy in 28 haemodialysis patients on serum CPP levels, inflammatory cytokine/chemokine arrays and human aortic smooth muscle cell (HASMC) and coronary artery endothelial cell (HCAEC) bioassays.

Results

In our cohort (75% male, 62 ± 12 years) high-dose SO reduced primary (amorphous) and secondary (crystalline) CPP levels {−62% [95% confidence interval (CI) −76 to −44], P < .0001 and −38% [−62 to −0.14], P < .001, respectively} compared with washout. Nine of 14 plasma cytokines/chemokines significantly decreased with high-dose SO, with consistent reductions in interleukin-6 (IL-6) and IL-8. Exposure of HASMC and HCAEC cultures to serum of SO-treated patients reduced calcification and markers of activation (IL-6, IL-8 and vascular cell adhesion protein 1) compared with washout. Serum-induced HASMC calcification and HCAEC activation was ameliorated by removal of the CPP-containing fraction from patient sera. Effects of CPP removal were confirmed in an independent cohort of chronic kidney disease patients.

Conclusions

High-dose SO reduced endogenous CPP formation in dialysis patients and yielded serum with attenuated pro-calcific and inflammatory effects in vitro.

Graphical Abstract

Graphical Abstract.

KEY LEARNING POINTS.

What is already known about this subject?

• Kidney failure carries an exceptionally high cardiovascular risk, much of which cannot be explained by traditional risk factors.

• Hyperphosphataemia is considered a major non-traditional risk factor, with accumulating evidence showing that circulating calciprotein particles, i.e. colloidal mineral-protein nanoparticles, might better explain the harmful effects of phosphate excess than direct effects of phosphate ions.

• The efficacy of the non-calcium-based phosphate binder sucroferric oxyhydroxide (SO) at lowering endogenous calciprotein particle levels and associated vascular sequelae is unknown.

What this study adds?

• In this analysis of a randomised controlled crossover study, SO therapy lowered serum calciprotein particle levels and attenuated systemic inflammation in prevalent dialysis patients.

• Lowering of calciprotein particle levels with SO diminished the ability of uraemic serum to induce vascular calcification and activate endothelial cells in vitro.

What impact this may have on practice or policy?

• Findings support a causal role for calciprotein particles in the development and progression of vascular disease.

• Calciprotein particles may be a better measure of phosphate toxicity than serum phosphate per se.

• This study provides mechanistic insight into how non-calcium-based phosphate binders may be beneficial via underrecognised anti-inflammatory effects.

INTRODUCTION

Kidney failure carries an exceptionally high cardiovascular risk that cannot be fully explained by traditional risk factors. Hyperphosphataemia is considered a major non-traditional risk factor [1, 2], with accumulating evidence suggesting that calciprotein particles (CPPs), circulating nano-sized aggregates of protein-bound calcium phosphate [3], mediate some of the toxic effects on the vasculature [4–9].

CPPs function as a buffer and cargo system for calcium and phosphate [10–12] in order to keep their concentrations in blood stable and prevent ectopic calcification[3, 13–15]. The nascent form of such particles are calciprotein monomers (CPMs), clusters of calcium and phosphate ions bound by fetuin-A, a liver-derived glycoprotein [10]. CPMs consolidate to form larger polymers with solid-phase minerals held as amorphous calcium phosphate (primary CPP; CPP-I) or crystalline hydroxyapatite (secondary CPP; CPP-II) [16]. Numerous factors influence this ripening process, including phosphate, which acts as the main promotor, and magnesium and fetuin-A as major inhibitors [11]. The balance of promoters and inhibitors present in a patient's serum determines its ability to resist conversion of CPP-I to CPP-II when supersaturating amounts of calcium and phosphate are added. This forms the basis of the in vitro T₅₀ test, which measures the time taken for 50% conversion. In conditions of disturbed mineral homeostasis and/or states of deficiency of calcification inhibitors, both common in chronic kidney disease (CKD) [17], CPM and CPP levels increase [3, 18, 19]. High CPP levels correlate with vascular stiffness [20] and calcification [18, 21], a higher risk for cardiovascular events [22] and death in an inflammation-dependent manner [23]. In contrast to CPMs, which appear mostly inert towards vascular cells in vitro [24], CPPs have been implicated as potential mediators of vascular injury inducing inflammation, endothelial dysfunction and vascular calcification [4–9].

We recently demonstrated that phosphate binder therapy with sucroferric oxyhydroxide (SO) improved T₅₀ times [25] in a randomised controlled crossover trial in haemodialysis (HD) patients [26]. Whether SO lowers endogenous CPP levels in HD patients has not yet been studied. Based on this work, we investigated the effects of SO on endogenous CPP levels using a novel fluorescent probe–based flow cytometric assay [3]. We hypothesised that better control of hyperphosphataemia with SO would lead to lower endogenous CPP levels and attenuate systemic inflammation, yielding serum with lower toxicity to vascular cells in vitro.

MATERIALS AND METHODS

Patients

All study participants gave written informed consent and clinical studies were conducted according with the Declaration of Helsinki.

SO interventional study

The design and results of the main trial were published previously [26]. Briefly, this single-centre (Elisabethinen Hospital, Linz, Austria), open-label, randomised controlled crossover study included 39 chronic HD patients who received either oral low-dose SO (250 mg/day) followed by high-dose SO (2000 mg/day) for 2 weeks each, or vice versa, with 2-week washout phases (no phosphate binder) before and after each treatment phase. Study visits were performed at the second and third dialysis session of the second week of each study phase. At each visit, blood samples were collected, centrifuged (1280 g for 15 minutes at room temperature) with aliquots stored at −80°C until batched analysis. Mean values for all biochemical parameters were calculated for each study phase.

The study was approved by the local ethics committee (A-VIII-16) and registered at the European Union clinical trials register (EudraCT 2016-004789-24) and ClinicalTrials.gov (NCT03010072).

Fetuin-A Levels in Systemic disease and Kidney Impairment (FLEKSI) observational cohort

Participants were enrolled (January 2014–September 2016) in the FLEKSI prospective observational study conducted at the Royal Melbourne Hospital [3]. For the present analysis, samples from three groups of participants were used: chronic HD patients, patients with CKD stages 3/4 and healthy volunteers. Details of the study design and inclusion and exclusion criteria have been described elsewhere [3]. The participants’ characteristics are given in Supplementary Table S1. The study was approved by the local ethics committee (2012.141).

Biochemical parameters

Samples were thawed (once) at 25°C, thoroughly vortexed and then centrifuged (1000 g for 1 minute at 4°C) prior to biochemical analysis. All measurements were performed in a blinded manner. Technical details of the assays used and their analytical imprecision are given in Supplementary Table S2.

In vitro studies

Source and cell culture conditions are summarised in Supplementary Table S3. Human sera were thawed at 37°C, cleared of cryoprecipitates by centrifugation (12 000 g for 20 minutes) and the supernatant was then filtered (0.22 µm; Millipore). Portions of the filtrate were used directly to supplement growth media (‘serum’) or were subjected to further centrifugation (30 000 g for 2 hours at 4°C) to remove the CPP-containing fraction (‘CPP-deficient serum’) prior to use [3]. While these centrifugation conditions readily sediment CPP-I and CPP-II under 30-nm diameter, they do not pellet matrix vesicles or calcifying exosomes [3]. Synthetic CPP-I and CPP-II were prepared from timed precipitation reactions of human serum enriched with supersaturating concentrations of calcium and phosphate as described in the Supplementary material.

Human aortic smooth muscle cell (HASMC) calcification

HASMC (passages 4–8) were seeded at 2 × 10⁴ cells/cm² in 24-well plates (Corning, Corning, NY, USA), grown to 80% confluence and serum-starved for 12 hours in M199 with 0.5% bovine serum albumin (BSA; Sigma-Aldrich, St. Louis, MO, USA) before switching to calcification medium containing M199/antibiotics supplemented with 2.5 mM sodium phosphate (pH 7.40) and 10% serum from controls or patients. Each patient serum was run in triplicate with all samples from the same patient run on a single plate. Cells were treated for 6 days with the media replaced after 3 days. Synthetic CPP-I and CPP-II were spiked into pooled extracellular vesicle-depleted human serum (1 × 10⁸/ml) and run on each plate as controls. Assessment of mineralisation was performed as previously described [27]. The calcium concentration was determined using a fluorometric probe (Ex/Em = 500/530 nm; #K409-100; BioVision, Milpitas, CA, USA). The total protein content was determined using the Pierce Micro BCA assay (Thermo Fisher Scientific, Waltham, MA, USA) and used for normalisation.

Human coronary artery endothelial cell (HCAEC) bioassay

The endothelial cell ‘biosensor’ assay was performed as previously described [28] with minor modifications. Briefly, HCAECs (passages 3–6) were seeded at 5 × 10³/cm² into 48-well plates (Corning), grown to confluence, serum-starved for 12 hours (M199, 0.5% BSA) and then switched to M199/antibiotics supplemented with 10% control or patient's serum to assess activation. Each serum sample was run in quadruplicate with all samples from the same patient run on a single plate. Lipopolysaccharide (25 ng/ml; #tlrl-eklps; InvivoGen, San Diego, CA, USA) and pooled extracellular vesicle-depleted human serum spiked with synthetic CPP-I and CPP-II (1 × 10⁸/ml) were run on each plate as positive controls. After 8 hours, cells were washed with saline, lysed and total RNA extracted (RNeasy Mini Prep kit; Qiagen, Hilden, Germany) with on-column DNase digestion (RNase-Free DNase Set; Qiagen). A total of 1 µg of RNA was used to synthesise complementary DNA (High-Capacity cDNA Reverse Transcription Kit; Applied Biosystems, Waltham, MA, USA). Quantitative real-time polymerase chain reaction was performed with triplicate technical replicates in a 384-well CFX cycler (Bio-Rad, Hercules, CA, USA) using the SsoAdvanced Universal SYBR Green Supermix (Bio-Rad) and qSTAR primer pairs (OriGene, Rockville, MD, USA; Table S4). Messenger RNA (mRNA) levels of TATA box binding protein were used for normalisation. Fold changes were calculated using the comparative Ct method (2^−ΔΔCt) relative to transcript levels in cells treated with pooled serum from controls.

Statistical analysis

As for the primary endpoint (T₅₀ time) in the main study, all secondary analyses were performed on the prespecified per-protocol population, which included patients adhering to ≥85% of the prescribed study medication (n = 28) [26]. No carry-over effect was detected for T₅₀ time in the main study [26], therefore data from both study sequences (high- to low-dose and low to high-dose SO) were merged.

Groupwise comparisons between SO treatment and washout phases were performed using the paired t-test or Wilcoxon matched pairs test, as appropriate. To compare more than two groups of the SO interventional study, repeated-measure analysis of variance (ANOVA) was performed with Bonferroni's correction for multiple testing. Multiple independent groups were compared by Brown–Forsythe and Welch ANOVA with Dunnett's T3 multiple comparisons test. Correlations were assessed using Spearman's or Pearson's correlation coefficient, as appropriate. Non-supervised hierarchal K-means clustering using Euclidian distance and average linkage (3 clusters, 5000 iterations) was used to assess clustering by patient (Morpheus; https://software.broadinstitute.org/morpheus). All analyses were performed using GraphPad Prism version 8 (GraphPad Software, San Diego, CA, USA).

RESULTS

Effect of SO on endogenous calciprotein monomer and calciprotein particle levels in dialysis patients

Characteristics of the per-protocol population of the SO interventional study [26] are shown in Table 1. The sequential formation of the three major fetuin-A-containing mineral colloid species is depicted in Fig. 1A. Two-week treatment with high-dose SO (2000 mg/day) significantly reduced endogenous levels of all three phases (CPM, CPP-I and CPP-II) compared with washout (Fig. 1B–D). A proportionately greater reduction was observed in CPP-I [median −62% (IQR −76 to −44), P < .0001] than in CPP-II [−38% (IQR −62 to −0.14), P < .001] and in CPM [−31% (−62–22), P < .01]. Low-dose SO (250 mg/day) therapy led to a modest but significant reduction in CPM [−21% (IQR −32–5), P < .05], but not in CPPs (Fig. 1E–G).

Table 1:

Baseline characteristics of the SO interventional study.

Characteristics	Per-protocol population (N = 28)
Age (years), mean ± SD	62 ± 12
Sex (male), n (%)	21 (75)
Dialysis vintage (months), mean ± SD	28 ± 14
Primary kidney disease, n (%)
Hypertensive/vascular	9 (32.1)
Diabetes mellitus	6 (21.4)
Glomerulonephritis	6 (21.4)
Autosomal dominant polycystic kidney disease	4 (14.3)
Other	3 (10.7)
Biochemical parameters, mean ± SD
Serum bicarbonate (mmol/l)	22.9 ± 2.3
Serum phosphate (mmol/l)	1.80 ± 0.34
Ionised serum calcium (mmol/l)	1.09 ± 0.06
Serum albumin (g/l)	39 ± 3
Parathyroid hormone (pg/ml)	315 ± 162
High-sensitivity C-reactive protein (mg/l), median (IQR)	4.2 (1.9–7.9)
Medication, n (%)
Sevelamer carbonate	23 (82.1)
Lanthanum carbonate	8 (28.6)
Calcium-containing phosphate binders	0
Vitamin K antagonists	5 (17.9)
Active vitamin D	20 (71.4)
Native vitamin D	1 (3.6)
Cinacalcet	10 (35.7)
Etelcalcetide	2 (7.1)
Erythropoiesis-stimulating agents	15 (53.6)

Figure 1:

Effect of SO on endogenous calciprotein monomer and calciprotein particle serum levels in dialysis patients. (A) Schematic showing the sequential formation and interrelationship of CPM, CPP-I and CPP-II from calcium phosphate ion clusters, fetuin-A and other proteins. (B–D) Effect of 2-week high-dose SO (2000 mg/day; n = 28) and (E–G) low-dose SO (250 mg/day; n = 27) therapy compared with the preceding phosphate binder washout phase on endogenous serum levels of (B, E) CPM, (C, F) CPP-I and (D, G) CPP-II at an individual patient level measured by fluorescent bisphosphonate probe-based assays. Medians are depicted as empty circles for the washout phases and filled circles for the SO treatment phases and IQRs are depicted as whiskers. The paired t-test was used for all analyses except for comparison of CPP-I between high-dose SO and washout, where Wilcoxon matched pairs test was used (ns, not significant; *P < .05, **P < .01, ***P < .001, ****P < .0001).

The size of CPP-II generated in serum after enrichment of serum with supersaturating amounts of calcium and phosphate has previously been associated with vascular calcification in patients with advanced CKD [29]. Since the mineral load associated with CPP-II should theoretically be a product of both particle number and size, we were interested to see the effect of SO therapy on both parameters. CPP-II size (hydrodynamic radius) was measured on samples following completion of the T₅₀ assay using three-dimensional cross-correlation dynamic light scattering (3D-DLS), and as a novel adjunct, we also estimated the size of native CPP-II detected by flow cytometry (FC). While estimates of the mean CPP-II size (FC) did not change between washout and high-dose SO therapy (146.4 ± 23.3 versus 140.2 ± 21.2 nm; P = .17; Supplementary Figure S1A), the CPP-II size (3D-DLS) decreased (231.7 ± 52.8 versus 214.5 ± 55.9 nm; P < .01; Supplementary Fig. S1B).

Association between changes in calciprotein monomer and calciprotein particle levels and mineral homeostasis in SO-treated dialysis patients

As previously reported [26], serum phosphate levels decreased and T₅₀ times increased significantly with high-dose SO therapy. In the current analysis, changes in serum phosphate in response to SO correlated strongly with changes in CPP-I {r = 0.64 [95% confidence interval (CI) 0.35–0.82], P < .001; Supplementary Fig. S2}, but not in levels of CPM, CPP-II or CPP-II size. Conversely, changes in T₅₀ times, parathyroid hormone, intact and C-terminal fibroblast growth factor 23 levels or their ratio, were not significantly correlated with changes in any of these parameters.

Effect of SO on patient serum-induced vascular smooth muscle cell (VSMC) calcification in vitro

Since CPP-II induces VSMC calcification in vitro, we hypothesised that lowering serum CPPs with high-dose SO should decrease the pro-calcific effects of uraemic serum towards VSMCs in culture [30]. Accordingly, we tested whether exposure of HASMCs to serum from SO-treated patients altered their propensity to calcify in a high-phosphate environment. Cell-associated calcium content was significantly lower in monolayers exposed to serum of SO-treated patients compared with washout serum (40.2 ± 15.4 versus 54.0 ± 18.6 µg/mg; P < .01; Fig. 2). Given that SO might modify multiple pro-calcific factors in serum in addition to CPPs, we repeated this experiment using serum that had first been depleted of endogenous CPP-I and CPP-II by high-speed centrifugation [3]. Compared with unfractionated ‘whole’ serum containing CPPs, treatment of HASMCs with CPP-deficient serum consistently induced less calcification (Fig. 2), corroborating the notion that CPPs might contribute to the pro-calcific effects of uraemic serum in vitro.

Figure 2:

Endogenous calciprotein particles mediate the effect of SO treatment on patient serum-induced VSMC calcification in vitro. Cell-associated calcium content of HASMCs was measured after 6 days of exposure to sera (10%) from chronic HD patients treated with high-dose SO (2000 mg/day) or no phosphate binder (washout) in the presence of elevated extracellular phosphate concentrations (2.5 mM). Serum samples were centrifuged prior to use (30 000 g for 2 hours at 4°C) to remove endogenous CPP-I and CPP-II (CPP-depleted serum) or used without additional centrifugation (whole serum). The box represents the median with IQR and the whiskers show minimum and maximum values. Repeated measures ANOVA (P < .0001) with consecutive groupwise comparison and Bonferroni's correction for multiple testing was applied (ns, not significant; **P < .01, ****P < .0001).

Suggestive of pro-calcific effects unique to CPP-II, only changes in endogenous CPP-II serum levels between high-dose SO and washout strongly correlated with the degree of calcification seen in vitro when cells were treated with the respective patient's serum [r = 0.81 (95% CI 0.63–0.91), P < .0001; Supplementary Fig. S3). Conversely, changes in CPP-II size (3D-DLS) were not associated with the extent of cell-associated calcium content [Spearman's r = 0.00 (95% CI −0.38–0.38), P > .99]. Treatment of HASMCs with synthetic CPP-I or CPP-II derived from pooled serum of dialysis patients or healthy volunteers from another cohort [16] supported this assumption, with consistently greater calcification observed with equivalent amounts of CPP-II compared with CPP-I (Fig. 3). Differences in CPP-II size, on the other hand, had a significant incremental effect (Supplementary Fig. S4). Intriguingly, however, these results also pointed to a greater calcifying potential of uraemic CPP-II compared with synthetic CPP-II of equivalent size and number created in a non-uraemic environment, implying additional conditioning of CPPs favouring HASMC calcification by exposure to the uraemic milieu.

Figure 3:

Synthetic calciprotein particles derived from pooled serum of healthy controls and dialysis patients induce calcification of VSMCs in vitro. Cell-associated calcium content of HASMCs was measured after 6 days of exposure to synthetic CPP-I and CPP-II (1 × 10⁸/ml) derived from human serum pools (n = 8) of healthy controls (empty circles) or dialysis patients (CKD5D) in the presence of elevated extracellular phosphate (P) concentrations (2.5 mM). Horizontal lines represent medians. P-values are from Brown–Forsythe and Welch ANOVA tests with Dunnett's T3 multiple comparisons test (**P < .01).

To explore these findings further, we tested the calcifying effect of serum obtained from a separate observational cohort (FLEKSI study) including healthy controls, patients with CKD 3/4 and dialysis patients [17]. Compared with controls, CKD patients had significantly higher endogenous CPP-I and CPP-II serum levels (Fig. 4A, B). The extent of calcification was greatest when HASMCs were exposed to sera of dialysis patients compared with that of non-dialysis-dependent CKD patients or controls (Fig. 4C). Similar to the results of our intervention cohort of SO-treated dialysis patients, removal of the CPP fraction by centrifugation markedly attenuated HASMC calcification, with attenuating effects proportionately greater in those on dialysis (Fig. 4D).

Figure 4:

Endogenous calciprotein particles of CKD patients mediate patient serum-induced VSMC calcification in vitro. (A) Endogenous serum levels of CPP-I and (B) CPP-II measured by a fluorescent probe-based flow cytometry assay in the FLEKSI cohort, including healthy adult controls (n = 25) and patients with CKD stage 3/4 (CKD; n = 21) or dialysis patients (CKD5D; n = 40). (C) Cell-associated calcification of HASMCs was measured after 6 days of exposure to serum (10%) from controls and patients with CKD stage 3/4 or dialysis patients (CKD5D) in the presence of elevated extracellular phosphate concentrations (2.5 mM). (D) Samples were centrifuged prior to use (30 000 g for 2 hours at 4°C) to remove endogenous CPP-I and CPP-II (CPP-depleted serum) or used without additional centrifugation (whole serum). The box represents the median with IQR and the whiskers represent minimum and maximum values. P-values in (A–C) are from Brown–Forsythe and Welch ANOVA tests with Dunnett's T3 multiple comparisons test. P-values in (D) are from paired t-tests (ns, not significant; *P < .05, **P < .01, ****P < .0001).

Effect of SO on plasma cytokine/chemokine levels in dialysis patients

SO treatment in 5/6 nephrectomised rats attenuated renal inflammation [31], but effects in humans and on systemic inflammatory markers are unknown. Using a combination of enzyme-linked immunosorbent assays and multiplex arrays, plasma concentrations of interleukin-1α (IL-1α), IL-1β, IL-18, IL-33, IL-6, IL-10, IL-17A, IL-23 and IL-8 were found to significantly decrease with high-dose SO, whereas IL-12 p70, monocyte chemoattractant protein-1 (MCP-1), tumour necrosis factor-α (TNF-α), interferon-α2 (IFN-α2) and IFN-γ did not change (Supplementary Table S5). The largest changes were seen for IL-1α [median −62% (95% CI −76 to −26), P < .0001], IL-8 [−46% (95% CI −73 to −17), P < .0001; Fig. 5A] and IL-6 [−31% (95% CI −51 to −1), P < .001; Fig. 5B]. This predominantly anti-inflammatory effect was reflected in significantly lower high-sensitivity C-reactive protein (hsCRP) levels with SO therapy [3.90 mg/l (95% CI 1.95–7.18) versus 2.45 (0.99–5.94); P < .05; Fig. 5C).

Figure 5:

Effect of SO treatment on markers of systemic inflammation in dialysis patients. Plasma levels of (A) IL-8, (B) IL-6 and (C) hsCRP in dialysis patients treated with high-dose SO (2000 mg/day) or no phosphate binder (washout). Medians are depicted as empty circles for the washout phases and filled circles for the SO treatment phases and IQRs are depicted as whiskers. The Wilcoxon matched pairs test was used for comparison between the two groups (*P < .05, ***P < .001, ****P < .0001). (D) Heatmap visualisation of relative changes (%) in plasma cytokine/chemokine levels in dialysis patients treated with high-dose SO compared with phosphate binder washout. Columns represent cytokines, rows represent patients. If serum cytokine levels were below the limit of detection at both time points, values were excluded (grey boxes). Clusters of patients (rows) were identified by non-supervised K-means clustering using Euclidian distance and average linkage (3 clusters, 5000 iterations) in Morpheus (https://software.broadinstitute.org/morpheus).

Heatmap visualisation of changes in plasma cytokine/chemokine levels revealed a markedly heterogeneous response to SO therapy, with a subgroup of patients responding with pronounced decreases in cytokine/chemokine levels (cluster 2, Fig. 5D), while the majority displayed only minor changes or a mixed response. Given the known pro-inflammatory effects of CPPs in vitro [32, 33] and in vivo [9, 34], we considered whether the response in cytokines/chemokines to SO therapy could be related to changes in CPP/CPM. However, exploratory subgroup analyses did not show any distinct pattern in changes of endogenous CPM/CPP load, CPP-II size, T₅₀ times or biochemical mineral markers (Supplementary Fig. S45).

Effect of SO on patient serum-induced endothelial cell activation in vitro

Given recent data strongly implicating CPPs as direct mediators of endothelial dysfunction [9, 35], we next investigated whether SO treatment altered the tendency of a patient's serum to activate endothelial cells (ECs) in vitro. We compared the gene expression of four established markers of EC activation (IL-6, IL-8, VCAM-1, ICAM-1) in HCAECs exposed to sera of SO-treated patients compared with washout. IL-6 and IL-8 mRNA levels decreased [median −11.6% (IQR −22.2–5.2), P < .05 and −14.7% (IQR −23.8 to −4.3), P < .01, respectively; Fig. 6A, B] upon exposure of HCAECs to sera of SO-treated patients. Still, mRNA levels remained higher than with control serum (Fig. 6), again emphasizing that binder therapy does not completely ameliorate the toxicity of uraemic serum. With respect to cell adhesion molecules, we observed lower mRNA levels of VCAM-1 [−9.2% (IQR −19.2–2.1), P < .01] but not ICAM-1 (Fig. 6C, D). Consistent with CPP-related effects, changes in serum CPP-I levels with SO therapy could be quantitatively correlated with the above reductions in IL-8 [Spearman's r = 0.38 (95% CI 0.00–0.67), P < .05] and VCAM-1 [Pearson's r = 0.43 (95% CI 0.06–0.69), P < .05] mRNA expression. Changes in the levels of CPM, CPP-II and CPP-II size, on the other hand, were not significantly associated with the same mRNA findings (all P > .05).

Figure 6:

Endogenous calciprotein particles mediate the effect of SO treatment on patient serum-induced endothelial cell activation in vitro. (A) mRNA levels of IL-6, (B)IL-8, (C)VCAM-1 and (D)ICAM-1 in HCAECs exposed to serum (10%) of dialysis patients treated with high-dose SO (2000 mg/day) or after phosphate binder washout (washout). Expression levels are set relative to transcript levels in cells treated with pooled serum from healthy adult controls depicted as triangles (mean = 1). Medians are depicted as empty circles for the washout phases and filled circles for the SO treatment phases and IQRs are depicted as whiskers. The paired t-test was used for comparison between the two groups (ns, not significant; *P < .05, **P < .01). (E) mRNA levels of IL-8 and (F)VCAM-1 in HCAECs after exposure towards sera (10%) from chronic HD patients treated with SO 2000 mg/day or no phosphate binder (washout). Serum samples were centrifuged prior to use (30 000 g for 2 hours at 4°C) to remove endogenous CPP-I and CPP-II (CPP-depleted serum) or were used without additional centrifugation (whole serum). The box represents the median with IQR and the whiskers represent minimum and maximum values. Repeated measure ANOVA (P < .0001) with consecutive groupwise comparison and Bonferroni's correction for multiple testing was applied (ns, not significant; *P < .05, **P < .01, ****P < .0001).

To better isolate effects due to CPPs, we again compared the expression of the same activation markers in response to treatment with whole unfractionated serum and the CPP-deficient fraction. IL-8 and VCAM-1 mRNA levels were lower when cells were exposed to CPP-deficient sera compared with unfractionated serum (Fig. 6E, F), suggesting that CPPs contribute to the induction of these markers in vitro.

To corroborate these findings, we also tested the effect of synthetic CPPs generated from pooled patient serum and samples from the FLEKSI cohort, as described above. Compared with CPPs derived from non-uraemic control serum, uraemic CPPs stimulated higher IL-6, IL-8, VCAM-1 and ICAM-1 expression in HCAECs (Fig. 7A–D) and greater cytokine secretion from five different human endothelial primary cultures and cell lines (Fig. 7E). When HCAECs were exposed to sera of patients from the FLEKSI cohort, mRNA levels of IL-8, VCAM-1, IL-6 and ICAM-1 were significantly higher in cells treated with serum from dialysis patients compared with non-dialysis-dependent CKD patients and controls (Fig. 8A, C, E, G). Once again, removal of the CPP fraction from patient serum significantly attenuated induction of all four mRNA targets compared with unfractionated samples (Fig. 8B, D, F, H) but still demonstrated higher expression levels than in HCAECs treated with control serum.

Figure 7:

Synthetic calciprotein particles derived from serum of healthy controls and dialysis patients induce endothelial cell activation and cytokine secretion in vitro. (A–D) Effect of synthetic CPP-I and CPP-II (1 × 10⁸/ml) derived from human serum pools (n = 8) of healthy controls or dialysis patients (CKD5D) on HCAEC mRNA levels of (A)IL-8, (B)VCAM-1, (C)IL-6 and (D)ICAM-1 relative to vehicle (saline)-treated cells (mean = 1). Horizontal lines represent medians. P-values in (A–D) are from Brown–Forsythe and Welch ANOVA tests with Dunnett's T3 multiple comparisons test (**P < .01, ***P < .001, ****P < .0001). (E) Heatmap showing the effect of synthetic CPP-I and CPP-II (1 × 10⁸/ml) derived from non-uraemic or uraemic human serum pools (n = 3) on cytokine secretion across five different human endothelial primary cultures/cell lines. Vehicle (saline) and lipopolysaccharide (LPS; 25 ng/ml)-treated cells served as comparators. Mean cytokine concentrations (pg/ml) were natural log-transformed and normalized to cell count (×10⁶). Experimental conditions are given in Supplementary Table S2. Cytokine concentrations below the respective limits of quantitation (LOQs) are indicated by grey boxes. EA.hy926, human umbilical vein cell line; HAEC, human aortic endothelial cell; HUVEC, human umbilical vein endothelial cell; MCP-1, monocyte chemoattractant protein-1; SK-HEP-1, human adenocarcinoma-derived endothelial cell line.

Figure 8:

Endogenous calciprotein particles of CKD patients lead to serum-induced endothelial cell activation in vitro. (A, C, E, G) Effect of serum (10%) from healthy adult controls (n = 25) and patients with CKD stages 3/4 (CKD 3/4; n = 20) or dialysis patients (CKD5D; n = 40) on HCAEC mRNA levels of (A)IL-8, (C)VCAM-1, (E)IL-6 and (G)ICAM-1 relative to controls (mean = 1). (B, D, F, H) Effects of CPP depletion on serum-induced HCAEC activation by patient group. Serum samples were centrifuged prior to use (30 000 g for 2 hours at 4°C) to remove endogenous CPP-I and CPP-II (CPP-depleted serum) or used without additional centrifugation (whole serum). The box represents the median with IQR and the whiskers represent minimum and maximum values. P-values in (A, C, E, G) are from Brown–Forsythe and Welch ANOVA tests with Dunnett's T3 multiple comparisons test. P-values in (B, D, F, H) are from paired t-tests (ns, not significant; *P < .05, **P < .01, ***P < .001, ****P < .0001).

DISCUSSION

In this secondary analysis of a randomized controlled crossover trial in HD patients [26] we demonstrated that high-dose SO therapy significantly decreased endogenous CPM, CPP-I and CPP-II serum levels and was associated with immunomodulatory effects. The SO-mediated effects on patient serum translated into lower VSMC calcification and EC activation in vitro, with CPPs revealed to be potential major determinants. These observations were confirmed in an independent cohort of CKD patients.

Whether phosphate binder therapy influences endogenous CPP levels has only been addressed by three studies [36–38], two of which included in dialysis patients. The Sevelamer Versus Calcium to Reduce Fetuin-A-Containing Calciprotein Particles in Dialysis randomized controlled trial (RCT) in 31 HD patients reported a 70% reduction in endogenous CPP-I levels with sevelamer compared with calcium carbonate, without significant changes in CPP-II levels [36]. The effect of SO on endogenous CPPs has only been studied in animals with CKD [31]. Consistent with previous findings, we observed the most pronounced changes in CPP-I following SO therapy with more modest changes in CPP-II [20, 23, 36]. Notably, both previous studies in dialysis patients used calcium carbonate as the comparator [36, 37], which because of exogenous calcium loading [39, 40] may have promoted CPP formation. In our cohort, we compared high- and low-dose SO with phosphate binder washout. The subtherapeutic SO dose (250 mg/day) was ineffective in lowering phosphate [26] and endogenous CPP levels. Nevertheless, it did reduce CPM levels, which may be a more sensitive measure of acute changes in mineral load given their intestinal origin [3, 13, 36, 37] and given the sigmoidal relationship between serum phosphate and CPM [19].

Beyond T₅₀ times, CPP-II size may provide additional information about mineral buffering capacity [25]. In a previous study of dialysis patients, Chen et al. [29] found that the size of CPP-II following enrichment with supersaturating amounts of calcium and phosphate (CPP-II size by 3D-DLS) was associated with the presence of vascular calcification in advanced CKD patients after adjusting for age, diabetes and serum calcium and phosphate levels and independent of T₅₀ times. In our study in SO-treated dialysis patients, estimates of endogenous CPP-II size did not change, while 3D-DLS revealed a smaller particle size formed ex vivo using the approach of Chen et al. [29]. In part, this apparent discrepancy may reflect the relative lack of precision in our estimates of endogenous CPP-II size or that such differences are only detectable under greater crystallisation pressure. Regardless, neither metric of CPP-II size could be quantitatively related to effects on VSMC calcification or endothelial activation in vitro. Indeed, we observed quite modest effects of increasing CPP-II size on VSMC calcification using synthetic particles. Hence the biological significance of this ex vivo phenomenon remains uncertain.

CPPs have emerged as potential mediators of vascular pathology. In vitro and animal studies consistently show that CPPs trigger inflammation [7, 32, 33], vascular calcification [7, 41] and endothelial dysfunction [8, 9, 33, 35]. In our cohort, SO therapy attenuated serum-induced VSMC calcification and EC activation in vitro. These effects were mitigated after removal of CPPs from the serum of patients of two independent cohorts, indicating that CPPs may be a major causative factor in vascular activation. The fact that only endogenous CPP-II levels in our cohort correlated with in vitro VSMC calcification is consistent with the finding that predominantly CPP-II induces vascular calcification [7]. On the other hand, synthetic CPP-I provoked stronger EC activation compared with CPP-II in vitro, consistent with previous studies [33]. In contrast to CPPs, CPM has been reported not to exert pro-inflammatory or cytotoxic effects in different cell lines in vitro [24]. Here the potential of serum derived from CKD patients to induce both VSMC calcification and EC activation was greater than the serum of healthy controls. A plausible explanation for uraemic serum being more toxic could be related to both the greater numbers of circulating CPPs in uraemia, but also compositional differences relating to mineral and protein content, lipid cargo and the presence of bioactive bacterial products (e.g. endotoxin) [16, 42]. Hyperphosphataemia may itself engender higher circulating levels of CPPs, accelerated transformation of CPP-I to CPP-II and larger CPP-II particles, all indicative of greater mineral stress. Although not captured in our bioassays, the uraemic milieu (e.g. inflammatory cytokines, uraemic toxins) may also modify the response of cells to CPPs, such that potentially injurious effects are amplified in those with CKD.

To our knowledge, this is the first study to document the effect of SO on systemic inflammation markers in dialysis patients and suggests that anti-inflammatory effects may be a more generic effect of non-calcium-containing phosphate binders than previously recognised. Pleiotropic effects on lipid metabolism or inflammation have been documented for sevelamer [43, 44] but have not yet been reported for patients treated with SO. Consistent with our observations in humans, two animal studies reported reduced renal and cardiac inflammation with SO [31, 45]. We observed significant reductions in 9 of 14 cytokine levels, especially in IL-1, IL-6 and IL-8, all of which may be involved in vascular pathology [46, 47]. The importance of targeting cytokines in this context may be highlighted by recent RCTs showing lower cardiovascular morbidity in patients with established cardiovascular disease and elevated hsCRP levels treated with anti-IL-1β or anti-IL-6 therapy [48, 49]. However, somewhat contrary to expectations, cytokine response could not be related to any CPP-related parameters, despite convincing effects of CPP-lowering on ameliorating endothelial cell inflammation. The multifactorial nature of inflammation in vivo, highly heterogeneous response to SO and small subgroups, superimposed on very high intraindividual variability even in health [50], may have all contributed to the ostensibly null finding here and in our view does not exclude the possibility of potentially beneficial effects of moderating CPP levels on the vasculature.

The prospective nature of the main trial with a crossover design permitted us to evaluate the impact of SO on endogenous CPP levels and systemic inflammation markers. Furthermore, our translational approach using HASMC and HCAEC bioassays allowed us to recapitulate the putative protective effects of phosphate lowering on the vasculature in vitro and provide evidence that CPPs may be mediating these effects. Importantly, we also confirmed these findings in an independent cohort of healthy controls and CKD patients. Nonetheless, we acknowledge that this secondary analysis has limitations. The main study was designed to investigate the effect of SO on T₅₀ and so all analyses related to serum CPP levels are exploratory. Given the trial design with 2-week treatment phases, longer-term changes in CPP load or inflammation markers could not be assessed, but this analysis provides important proof of concept. Although our findings are consistent with a causal role of CPPs in the development and progression of vascular disease in CKD, they remain associative. In particular, the amelioration of pro-calcific and pro-inflammatory effects observed after removal of the CPP-containing fraction from uraemic serum may be partly due to depletion of other factors given the non-specific nature of centrifugal fractionation. Nonetheless, it does help to establish that such in vitro effects are due to high molecular weight species like CPPs rather than smaller, less dense factors like free phosphate.

In conclusion, this is the first clinical trial in HD patients showing that phosphate binder therapy with high-dose SO lowered serum levels of CPM, CPP-I and CPP-II and was associated with immunomodulatory effects. Serum from SO-treated patients had lower pro-calcific and inflammatory properties in vitro and CPPs were found to mediate these effects. Taken together, these studies further establish CPPs as uraemic toxins and probable mediators of vasculopathy.

FUNDING

The SO interventional study was an investigator-initiated study that was financially supported by Vifor Fresenius Medical Care Renal Pharma (Vifor Pharma), St. Gallen, Switzerland. Vifor Pharma had no role in the study design, data collection, data interpretation or preparation of the manuscript. The secondary analyses of this study were supported by a research grant of the Medical Society of Upper Austria (awarded to U.T.). E.R.S. was supported by a Viertel Charitable Foundation Clinical Investigator grant and Royal Melbourne Hospital Project grant (PG-004-2018).

AUTHORS’ CONTRIBUTIONS

D.C. and E.R.S. developed the research question and supervised the study. U.T., M.C.H., A.P. and D.C. contributed to the conception and design of the SO interventional study and to data acquisition. T.D.H., N.D.T., S.G.H. and E.R.S. were involved in conception of the FLEKSI observational study. U.T., D.C. and E.R.S. acquired study funding, prepared the figures and wrote and revised the original draft. T.D.H., N.D.T., S.G.H., A.P. and E.R.S. contributed reagents and performed experiments. U.T., M.C.H., D.C. and E.R.S. were responsible for data analysis. All authors contributed to the interpretation of data, critically revised the manuscript and approved its final version. All authors accept accountability for the overall work by ensuring that questions pertaining to the accuracy or integrity of any portion of the work are appropriately investigated and resolved.

DATA AVAILABILITY STATEMENT

The data underlying this article will be shared upon reasonable request to the corresponding author.

CONFLICT OF INTEREST STATEMENT

D.C. has received research funding from Vifor Pharma and speaker's honoraria, travel support and research funding from Amgen, Astellas, AstraZeneca, Bayer, Chiesi, Medice, Novartis, Sandoz, Takeda, UCB and Vifor Pharma and holds advisory board positions at Alnylam, Amgen, Astellas, AstraZeneca, Boehringer Ingelheim, Chiesi, Gedeon Richter, Stada, Takeda, UCB and Vifor Pharma. S.G.H. has received research funding and travel support from Amgen and AstraZeneca. A.P. is an inventor of the T₅₀ test and holds a patent on it. He is president of the board of directors, a part-time employee and stockholder of Calciscon AG, Nidau, Switzerland, which commercialises the T₅₀ test. E.R.S has received research funding from the Royal Melbourne Hospital Project Grant, Amgen and Sanofi and holds stock in Calciscon AG. U.T. reports a research grant from the Medical Society of Upper Austria and received travel support from AstraZeneca, Baxter, Sandoz and Vifor Pharma. N.D.T. has received speaker's honoraria from Amgen, Shire and Takeda and travel support from Amgen, Shire, Takeda and Sanofi. M.C.H. and T.D.H. declare no relevant conflicts of interest. The results presented in this article have not been published previously in whole or part, except in abstract format.