Phosphate binders affect vitamin K concentration by undesired binding, an in vitro study

Chronic kidney disease (CKD) is a major and independent risk factor for cardiovascular disease (CVD) [1]. During CKD disease progression, mortality risk due to CVD rises progressively and accounts for 50% of all deaths in the CKD setting [2]. Vascular calcification (VC), almost universally present in late stage CKD, is an important contributor to CVD and CKD-related mortality [3]. A very recent study supports the assumption that density, besides the volume of calcification is an independent risk factor for dismal outcome [4]. Hyperphosphatemia is associated with VC [5]. Despite the widespread use of phosphate binders to control serumphosphate, evidence from prospective trials showing improved outcome by the use of phosphate binders is lacking. In fact it is unknown if the use of phosphate binders, especially calcium based binders, may actually be harmful [6]. There is some attenuation or delay of progression of VC by the non-calcium-based phosphate binders as compared to the calcium containing binders [7‐9]. However VC frequently progresses despite adequate phosphate concentrations [10]. This has been attributed to increasing calcium load from calcium based binders [11], chronic inflammation [12] and other factors such as vitamin K deficiency [13]. The prevalence of vitamin K deficiency is high in ESRD patients [14]. Adequate levels of vitamin K are mandatory for activation of matrix γ-carboxyglutamate (Gla) protein (MGP) [15]. MGP is secreted by chondrocytes and vascular smooth muscle cells in the arterial medial layer. MGP requires posttranslational modification by carboxylation and phosphorylation to become a functional calcification inhibitor of VC. We hypothesized that phosphate binders inhibit gastrointestinal uptake of vitamin K2 by undesired binding since the binding of fat soluble vitamins has been described previously [16, 17]. If confirmed this could contribute to vitamin K deficiency and as such possibly to the progress of vascular calcification.

All phosphate binders, with the exception of SOH did bind vitamin K in the absence of phosphate. This decline of free K2 concentration was statistically significant for LA and CA/MC. In the presence of phosphate, K2 binding was attenuated for SC and LA, but increased for CC. Table 2 depicts the vitamin K2 concentrations with only vitamin K2 and a phosphate binder in the solution. There is significant binding of vitamin K2 to LA and CA/MC. Vitamin K2 concentrations in a solution containing vitamin K2, phosphate and a binder are shown in Table 3. There is significant binding of vitamin K2 to CA/MC and CC. Vitamin K2 concentrations were also measured in all solutions that did not contain vitamin K2 as a negative control. In none of these samples vitamin K2 was detected. Solutions containing only phosphate with or without a binder only showed negligible vitamin K2 concentration, excluding interference of phosphate or phosphate binders with the vitamin K2 measurement (data not shown).

In this in vitro study we demonstrate that calcium-containing phosphate binders and lanthanum carbonate bind vitamin K2. For LA this binding depends on the absence of phosphate, pointing to competitive binding between phosphate and vitamin K2 for this compound. For calcium carbonate this vitamin K2 binding was statistically significant in the presence of phosphate compared to the solution without phosphate. In the mixture with sevelamer carbonate a nominally lower concentration of K2 was shown as well, but this decline was not statistically significant. Addition of sucroferric- oxyhydroxide did not lead to any decline of vitamin K2 at all, irrespective of presence or absence of phosphate. Our study design precludes a formal quantitative comparison between the different phosphate binders. Several remarkable differences however appear obvious. First, both calcium containing compounds had increased affinity for vitamin K2 in the presence of phosphate. We did not study the chemical explanation for this feature, but possibly the formed calciumphosphate salt itself binds K2 as well. For lanthanum carbonate there appears to be competition for binding between phosphate and vitamin K2. Although we could not find a statistically significant amount of vitamin K2 binding by sevelamer carbonate, where a previous study did [17], we did observe a trend in the same direction. This may be due to difference in concentrations used: 2.5 mg/ml and 4 mg/ml for sevelamer cabonate; and 37 mcg/ml (our study) and 0.5 and 5 ug/ml (Takagi et al.) for vitamin K. Moreover the previous research did not report whether vitamin K1 or K2 was used as substrate [17]. SOH showed no vitamin K2 binding at all. The statistically non-significant higher values of vitamin K2 when using the latter binder, as compared to control is probably the consequence of assay variability. Extrapolating these results to biological systems, including the treatment of hyperphosphatemia in CKD patients can only be done with great caution. However, our data do point to a feature of some commonly used phosphate binders that may have clinical consequences. Vitamin K deficiency results in undercarboxylation of MGP, and as such limit natural defense against ectopic calcification including VC. To some extent, the enigmatic lack of improvement of relevant endpoint in clinical studies, despite effective phosphate lowering potential, could be attributed to a decline in bioavailability of vitamin K from the gastrointestinal tract by phosphate binder therapy. In recent years some studies suggested superior clinical outcome when using sevelamer carbonate as compared to calcium containing binders [20, 21]. Generally this has been ascribed to the calcium loading as a consequence of calcium content in these binders. Our data suggest an additional mechanistic explanation, which is the increased likelihood of vitamin K deficiency. In support of this hypothesis is also a study in which non-hyperphosphatemic CKD patients were treated with either placebo or phosphate binder therapy. Despite a small decline in serum phosphate there was an increase in arterial calcification in the active treatment arm [10]. Our finding that vitamin K2 binding by LA is absent in presence of phosphate does not necessarily imply that in patients no relevant vitamin K binding occurs. It is likely that high gastrointestinal levels of phosphate do not always match local high concentrations of phosphate binders. In addition, dose up titration in case of persistent hyperphosphatemia may increase likelihood of unnoticed aspecific binding of non-phosphorus compounds. This may especially apply in hyperparathyroidism where a substantial amount of phosphate may be bone-derived. Our study has some limitations. We used a fixed pH which does not necessarily match the gastrointestinal tract, where pH may be as low as 1 in the stomach. However, at sites where vitamin K2 uptake occurs pH is not low, so our set-up may be more relevant than at low pH. A single measurement of vitamin K2 was performed at 330 min which was considered a reasonable period to mimic the time needed for the passage to the terminal ileum where the absorption of vitamin K2 starts [22]. Furthermore, only vitamin K2 has been used in our experiments, since vitamin K2 is the preferred cofactor for the carboxylation of MGP [23]. In addition, phosphate and vitamin K concentrations in vivo may differ substantially, and moreover a wide range of additional aspects may be very different as well. In our study we mainly aimed to test the proof of principle of the existence of vitamin K2 binding by contemporary phosphate binders. Obviously, the clinical relevance of our findings cannot be assessed in an in vitro setup. Another limitation, as described, is the fact that no quantitative comparison between phosphate binders can be made.

Sucroferric-oxyhydroxide and sevelamer carbonate were the only binders of the five binders studied that did not bind vitamin K2 in vitro. The clinical significance of this finding requires additional studies.