The demise of calcium-based phosphate binders—is this appropriate for children?

Despite national and international guidelines for the management of CKD, many patients have CKD-MBD [10‐13]. This is exemplified by data from the International Pediatric Peritoneal Dialysis Network (IPPN) in 153 children worldwide, in whom the PTH varied from country to country, but overall was over fivefold the upper limit of normal in approximately 50 % of patients. The highest levels were associated with high phosphate and lower calcium levels [14]. This level of PTH is associated with symptomatic bone disease: of nearly 900 children in the IPPN, 1.5 % have active bone pain, 5 % limb deformities, 10 % radiological osteodystropy and 5 % osteopaenia [3]. The latest IPPN data shows that for all ages, 14 % of patients have a serum calcium level below the normal range, 24 % have a serum phosphate level above the normal range, 47 % have a serum PTH level above the normal range; in addition, although 21 % are on sevelamer, 22 % are on a calcium supplement other than a calcium-containing phosphate binder [15]. In a cohort of 249 young Dutch adults with onset of end-stage kidney disease (ESKD) before the age of 14 years, 61 % had severe growth retardation, 37 % had bone disease (defined by at least one of the following conditions: deforming bone abnormalities, chronic pain related to the skeletal system, disabling bone abnormalities, aseptic bone necrosis and atraumatic fractures) and 18 % had disabilities resulting from bone impairment [16].

Hypercalcaemia is not just a result of high calcium intake. Both low and high blood PTH levels and 1,25(OH)D (1,25 hydroxy-vitamin D) can cause hypercalcaemia. Under normal circumstances, after calcium has been absorbed into the blood it can be exchanged as part of bone remodelling, incorporated into bone for growth or excreted in the urine (Fig. 1) [17]. The plasma calcium therefore does not indicate the total body calcium, but is a reflection of calcium movement into and out of the extracellular fluid (ECF). Indeed, just as hypercalcaemia does not necessarily mean that total body calcium is in excess, normocalcaemia does not mean that it is adequate. The ionised calcium in the ECF is crucial for the function of many enzyme systems throughout the body and, in particular, signalling pathways. Its maintenance within a strict, narrow range is therefore essential. A fall in plasma ionised calcium level activates calcium sensing receptors in the parathyroid gland which, in turn, activate PTH production. All of the actions of PTH are to increase the plasma calcium level, which it does by increasing renal tubular calcium reabsorption, increasing production of 1,25(OH)₂D, which in turn increases absorption of calcium from the gut, and by causing an efflux of calcium from bone. Therefore, if calcium influx from the gut is low, PTH stimulation will result in calcium being taken from bone, and if levels rise in the blood, PTH is suppressed and renal tubular reabsorption of calcium is decreased so that calcium is excreted in the urine. Prolonged, uncontrolled PTH secretion will result in downregulation of PTH receptors so that a higher plasma calcium level is needed to suppress PTH, resulting in hypercalcaemia. On the other hand, low PTH levels will result in adynamic bone that is unable to buffer influxes of plasma calcium, also resulting in hypercalaemia (Fig. 2). This bimodal effect is also seen with vitamin D (Fig. 3) [18]. High 1,25(OH)₂D (1,25 dihydroxy-vitamin D; calcitriol) levels stimulate calcium absorption and suppress PTH, whereas low levels decrease calcium absorption and increase PTH, and are also associated with higher C-reactive protein levels.

Recent studies have also suggested that FGF23, a bone-derived protein that was first believed to be mainly involved in phosphate homeostasis, is also involved in calcium regulation. FGF23 increases renal phosphate excretion by downregulating the sodium–phosphate co-transporter in the proximal tubules, thereby, at least in early CKD, increasing phosphorus (P) excretion [19]. It also suppresses 1α-hydroxylase and increases 24-hydroxylase activity, so that there is a reduced production and increased degradation of 1,25(OH)₂D [19]. Recent work has shown that there may be a threshold effect of serum calcium on FGF23 regulation, such that FGF23 is not stimulated when the calcium level is very low (essentially because this would reduce calcitriol and further aggravate hypocalcaemia). Studies in uraemic rats support this finding: in the presence of hypocalcaemia FGF23 production was not increased by low calcitriol or high PTH levels [20]. Interestingly, sevelamer, but not Ca-based binders, have been shown to decrease FGF23 levels in predialysis CKD patients, but this effect could not be attributable to improved phosphate or 1,25 (OH)₂D levels [21], suggesting that a relative hypocalcaemia in these subjects did not allow for FGF23 production irrespective of phosphate or 1,25 (OH)₂D levels. A study in haemodialysis patients has shown that the use of higher dialysate Ca concentrations, as well as the administration of calcitriol and a calcium-based phosphate binder were associated with higher final serum FGF23 levels [22]. In a previous study, we found that in a population of children with well-controlled serum phosphate, high FGF23 levels were seen with a progressive decline in estimated glomerular filtration rate (eGFR) and showed a positive correlation with serum calcium [23]. The complex interactions of calcium, phosphate, vitamin D and FGF23 are shown in Fig. 4.

The growing skeleton has a high demand for calcium, whereas adults may have ‘adynamic’ bones that are unable to incorporate calcium [24]. Clinical measures of bone mineral density (BMD) and bone biopsies have shown defective mineralisation as early as CKD stage 2 [25], and the prevalence increases to 90 % in patients on dialysis [26]. In dialysed children, the mineralisation defect was found to persist after treatment with active vitamin D sterols combined with either calcium-based or calcium-free binders and were associated with a concomitant increase in FGF23 levels [27]. Importantly, on bone biopsies in a cohort of 160 children on peritoneal dialysis, the same group showed that low serum calcium and high PTH levels were associated with defective mineralisation, irrespective of bone turnover [28]. This finding has been further corroborated by dual energy x-ray absorptiometry (DEXA) studies: BMD loss was been attributed to low calcium and high PTH levels and was associated with increased fracture risk [29]. In a recent study in 5- to 21-year-olds with CKD stages 2–5, including patients on dialysis, lower calcium levels were associated with baseline and progression of BMD loss; one standard deviation decrease in BMD was associated with a twofold increase in fracture risk. Adynamic bone disease has a high prevalence, being observed in 40–50 % of adult and almost 30 % of paediatric ESKD patients [30‐32].

One factor that is often forgotten is that the normal range for plasma calcium is highest at birth, falling progressively thereafter, but it does not reach adult values until the age of 4 years. This may not be recognised by clinicians (Fig. 5).

Converging evidence from clinical and experimental studies suggests a role for high serum calcium levels, particularly when associated with high phosphate levels, in the development and progression of vascular calcification. Changes in arterial structure as well as loss of compliance of arteries have been reported as early as in the second decade of life in children on dialysis [5, 33‐39], with approximately 20 % of children on dialysis shown to have coronary artery calcification on computed tomography (CT) scams, increased carotid intima-media thickness (cIMT) and calcified arteries [5, 36, 39]. Evidence of vascular calcification on CT scans is directly related to a high serum calcium × phosphate (Ca–P) product, serum phosphate level, intake of calcium containing phosphate binders and plasma PTH levels [34]. In a postmortem analysis of 120 children with CKD, soft tissue and vascular calcification was associated with the use of active vitamin D and calcium-containing phosphate binders [40]. Progressive increase of cIMT has been associated with higher serum Ca–P product and PTH [5, 41, 42]. In the seminal paper by Goodman et al., not only were the patients with coronary artery calcification older than those without calcification, but they also had a significantly higher mean serum Ca–P level (5.2 vs. 4.5 mmol²/L²) and almost double the intake of calcium from binders (6,456 vs. 3,325 mg/day) [5]. The serum phosphate, Ca–P and PTH levels were very high in both patient groups, and the calcium intake from phosphate binders was significantly above the K/DOQI (National Kidney Foundation Kidney Disease Outcomes Quality Initiative) recommended limit of 1,500 mg/day, suggesting that these patients were at a substantial risk of ectopic calcification [5]. Moreover, transient increases in calcium that occur in relation to dialysis [43] and vitamin D or binder intake may go unrecorded, but these can impact on ectopic calcification, particularly in the setting of high phosphate conditions. In the ‘Treat-to-Goal’ study, the calcium-treated group had significantly more hypercalcaemic episodes than the sevelamer group [8], and the extent of arterial calcification was directly related to the number of episodes of hypercalcaemia during the preceding 6 months [8].

Studies on the role of calcium intake, from diet, binders or dialysate, on vascular calcification have not been performed in children. However, it is important to remember that the purpose of the calcium homeostatic system is to maintain the ECF calcium levels within a very tight range; consequently, serum calcium levels are a poor, and sometimes misleading, marker of total body calcium level. A recent study has shown that elevated serum calcium, phosphate and Ca–P product levels, but not dietary calcium or phosphate consumption, are associated with increased coronary artery calcification [44].

Vascular calcification is an active, highly regulated process and not merely a passive deposition of calcium and phosphate in dead or dying cells [45]. In response to raised extracellular calcium and phosphate levels, vascular smooth muscle cells (VSMCs) undergo specific phenotypic changes, including apoptosis, osteo/chondrocytic differentiation and the release of small membrane-bound bodies called vesicles that form a nidus for the deposition of basic Ca–P in the form of hydroxyapatite nanocrystals [6, 45, 46]. VSMC transformation to an osteo/chondrocytic phenotype is characterised by the upregulation of bone-specific transcription factors and matrix proteins, including Runx2/Cbfa1, osterix and alkaline phosphatase, which in turn lead to accelerated calcification [6, 47‐50]. Raised serum phosphate level is a key factor that triggers osteoblastic differentiation of the VSMC [49].

Using intact arteries from children, we have shown that calcification in the vessel wall begins in pre-dialysis CKD stages 4 and 5 but is significantly greater in dialysis patients, with the calcium load in the vessel wall increasing linearly with time on dialysis and strongly correlated with the mean time-averaged serum Ca–P product [6]. When these vessel rings were cultured in graded concentrations of calcium and phosphate, normal and pre-dialysis vessels were resistant to calcification, but dialysis vessels showed accelerated calcification in high calcium and phosphate media [51]. The synergistic effect of the Ca–P product in driving calcification may be related to hydroxyapatite nanocrystal formation and their lysosomal degradation, leading to very high intracellular calcium levels and apoptotic cell death [52]. Also, Ca–P nanocrystals upregulate bone proteins, including bone morphogenetic protein 2 (BMP-2) and osteopontin [53, 54]. Calcium plays a key role in inducing apoptosis and in the formation and release of hydroxyapatite-laden matrix vesicles: when VSMCs in vitro are incubated in high calcium conditions, apoptotic cell death releases more calcium, which in turn may drive further apoptosis [55]. Apoptotic bodies form a nidus for calcification, and the propensity of these to calcify is markedly increased after calcium and phosphate treatment [56, 57]. Thus, once a nidus of calcification forms in the extracellular matrix of VSMCs, its uptake and phagocytosis will promote further calcification. Importantly, in the presence of a high phosphate level, even a small increase in calcium in the culture medium significantly increases calcification, and experiments have confirmed that calcium is a key mediator of VSMC damage and calcification [51].

In health terms, calcium balance is a measure of the calcium ingested and used for bone remodelling, minus the amount removed in urine and lost in stool. In adults who are not deficient in calcium or vitamin D, about 30 % of dietary calcium is absorbed, and the balance would be expected to be neutral [58]. In contrast, children require a positive calcium balance to allow for growth, particularly during periods of rapid bone accumulation, i.e. infancy and puberty.

In order to try to prevent the deposition in soft tissues of absorbed calcium that might be surplus to need, the K/DOQI nutritional guidelines recommend restricting the upper limit of calcium intake in CKD to <2 × the dietary reference intake, which is less than the upper limit recommended for normal children in those aged less than 8 years and at the upper limit of 2,500 mg for children exceeding that age (Table 2) [72]. The K/DOQI nutritional guidelines also state that ‘the challenge is to ensure that an adequate calcium intake is achieved’, recommending calcium supplementation for infants with CKD, who may not get adequate amounts of calcium if not breast fed, if low-electrolyte infant formulas are required or if fluids are restricted, and for children and adolescents on dialysis in whom a phosphate-restricted diet results in a serious calcium deficit, when typically the dietary calcium intake is <500 mg per day. Calcium-containing phosphate binders may then be the primary source of elemental calcium in the diet.

The need to restrict calcium intake from binders has to be balanced against the potential for inadequate provision of dietary calcium. K/DOQI does acknowledge, however, that ‘There are no data on calcium retention as a function of increased long-term calcium intake in patients with CKD, and it is impossible to accurately assess the actual absorption of calcium derived from binders, which is in large part dependent upon the kind and amount of food present in the stomach with the binder’ [10]. These calcium intake guidelines have recently been approved by the K/DOQI nutrition group [11].

The long-term cardiovascular effects of phosphate binder therapy remain controversial. A recent randomised clinical trial in adults with a GFR of 20–45 ml/min/1.73 m² demonstrated an increase in arterial calcification in those patients receiving calcium, lanthanum and sevelamer binder therapy, but not in the placebo-treated patients [74]. Another study showed progressive arterial calcification in predialysis patients on a low-phosphorus diet alone; progression was less in CaCO₃-treated patients and absent in sevelamer-treated patients [75]. The Independent trial demonstrated improved survival in predialysis patients randomised to sevelamer as compared to those treated with CaCO₃ [9, 76]. In adult dialysis patients, sevelamer was also found to reduce the progression of coronary artery calcifications, but not to be associated with reduced mortality [77]. However, what has clinched the demise of calcium-based phosphate binders in adults is a recent meta-analysis of 11 RCTs, with 2,312 patients on non-calcium-based phosphate binders and 2,310 patients taking calcium-based phosphate binders. There was a 22 % reduction in all-cause mortality in the non-calcium group. A subanalysis of seven RCTs (704 patients) in which vascular calcification had been measured showed that non-calcium based binders resulted in less progression of vascular calcification [78].

Recent studies in adults have shown that a regimen based on a combination of sevelamer + calcium-based binders was capable of effectively managing hyperphosphataemia without hypercalcaemia at reduced financial burden [79]. The average dose of sevelamer was 2.8 g per day, which is substantially lower than the average daily dose (6.5 g) in the study of Chertow et al. [8]. The combination therapy was better tolerated and showed higher patient compliance than CaCO₃ or sevelamer monotherapy [80]. In haemodialysis patients this combination therapy averted both the hypercalcaemia of CaCO₃ and the adverse effects of sevelamer hydrochloride when each was used as monotherapy [81]. Also, provided patients took their prescribed total daily dose of binders, it did not matter if different phosphate binders were taken at the same meal or at separate ones, according to their preference [82].

Only two studies have examined calcium acetate versus sevelamer in children. In an 8-week crossover study in children, sevelamer and calcium acetate were equally effective at reducing serum phosphate levels, but significantly less hypercalcaemia occurred in the sevelamer group [83]. In the second study, bone biopsies suggested that the sevelamer group had reduced bone formation at the 8-month follow-up, but the numbers were too small for comparison [26]. Without evidence, clinical practice is highly variable, with many nephrologists using sevelamer, which is the more expensive option and little is known about its long-term effects in children.

The ideal phosphate binder would be one that has a high affinity for phosphate, is long acting with a small tablet load, has an acceptable taste and texture and is without side effects, is not absorbed, can be administered down a feeding tube and is inexpensive. No phosphate binder that is currently available fulfils all of these criteria. The problem is the increased cost associated with non-calcium-containing binders, which has a substantial effect on prescribing budgets, although this may change when patents expire.