Introduction

Osteocalcin (OC) is the most abundant non-collagenous protein of bone matrix and is only produced by osteoblasts and odontoblasts under the control of 1,25 dihydroxy-vitamin D and parathyroid hormone (PTH) [1, 2]. γCarboxylation occurs at three glutamine residues in the presence of the cofactor vitamin K, and the carboxylated OC interacts with hydroxyapatite crystals [1]. Undercarboxylated OC (ucOC) has been reported to be associated with a reduced bone mass and a high risk of fracture [3, 4]. OC is released from bone matrix during bone resorption, and increased levels of serum OC are considered to reflect bone resorption, bone formation, or high turnover of bone [1, 5]. OC may play an important role in bone formation, although the function of OC has not yet been fully elucidated [5–7].

Recently, OC, in particular ucOC, has been shown to have a positive effect on both insulin production and insulin sensitivity [8, 9]. Mice lacking OC exhibit decreased β-cell proliferation, glucose intolerance, and insulin resistance [10]. Treatment of wild-type mice with continuously infused recombinant ucOC increases insulin secretion and improves glucose tolerance [11]. Recently, the inhibitory role of decarboxylated OC (ucOC), but not carboxylated OC, in regulating the action of tyrosine phosphatase on insulin secretion and sensitivity has been emphasized [12, 13]. As expected from animal studies, several recent human studies have demonstrated inverse associations between circulating OC, particularly ucOC, and fasting plasma glucose, insulin resistance, and hemoglobin A1C in patients with type 2 diabetes [14–17] and in population-based cohorts [18–20]. However, these human studies have not examined the associations in chronic kidney disease (CKD) with abnormalities of bone metabolism [21]. In patients undergoing maintenance hemodialysis, compared to the remaining population without CKD, several severe abnormalities in bone metabolism (chronic kidney disease-mineral and bone disorder; CKD-MBD), such as secondary hyperparathyroidism, osteomalacia, and adynamic bone disease, have been reported to date [21]. In these states, caused by CKD-MBD, it remains unknown whether ucOC is associated with glucose metabolism.

In the present study, we tested the hypothesis that ucOC levels are associated with indices of glucose metabolism in chronic hemodialysis patients (CKD 5D) with abnormalities of bone metabolism.

Patients and methods

Patients

In 2009, there were 396 maintenance hemodialysis patients at Shirasagi Hospital Kidney Center, who were not admitted to the hospital and had no significant acute illness or malignancy. Of 115 who were diabetic, 96 gave informed consent and were enrolled in the present study. Of the 281 non-diabetic patients, 93 patients who were age- and sex-matched to the above 96 diabetic patients were also enrolled in the present study. All 189 patients (96 diabetics and 93 non-diabetics) gave written informed consent prior to participation in the present study. Underlying kidney diseases of non-diabetic patients were as follows: chronic glomerulonephritis (n = 61), nephrosclerosis (n = 18), polycystic kidney disease (n = 2), other disease (n = 8), and unknown disease (n = 4). None of the patients had a past history of parathyroidectomy or renal transplantation. All patients were free of significant acute illness. They had no past history of fracture or radiographic evidence of vertebral or rib fracture. Past exposure to aluminum was negative in all subjects. The active vitamin D derivatives, calcitriol or alfacalcidol, were given orally to 124 patients. None of the patients received any other medication that might affect calcium or bone metabolism, such as androgen, bisphosphonate, vitamin K or aluminum hydroxide. In the diabetic subgroup, 36 patients were treated with insulin, 38 patients with oral anti-diabetic agents (mainly alpha-glucosidase inhibitors), and 22 patients with diet alone.

Patients underwent hemodialysis 3 times a week in 4–4.5 h sessions using hollow-fiber dialyzer and bicarbonate dialysate containing 3.0 mEq/L calcium. This study was approved by the ethics review committee of the Hospital.

Measurement of serum biochemical parameters of bone metabolism and ucOC

Blood was drawn just before the start of the dialysis session in a non-fasting state. Serum albumin, calcium, phosphorus, and glucose were measured by routine laboratory methods. The serum obtained was stored in aliquots at −20 °C until subsequent assay, with measurements made immediately after thawing. The biochemical parameters for calcium metabolism were determined as described previously [22]. All serum bone markers were measured in the same assay run to avoid inter-assay variance. Serum parathyroid hormone PTH was measured by electrochemiluminescence immunoassay (ECLIA) (Elecsys PTH, Roche Diagnostics GmbH, Mannheim, Germany). The assay detected 5–4,000 pg/ml of intact PTH. The intra- and interassay coefficients of variation were 2.93 % and 3.25 %, respectively. Serum bone alkaline phosphatase (BAP) was measured, as a marker of bone formation, using an enzyme immunoassay (EIA) kit (Alkphase-B; Metra Biosystem, Mountain View, CA) [23, 24]. The assay detected 0.7–140 U/L of BAP. The intra- and interassay coefficients of variation were 2.3 % and 3.1 %, respectively [22]. Serum tartrate-resistant acid phosphatase-5b (TRACP-5b) was measured, as a marker of bone resorption, by fragment absorbed immunocapture enzymatic assay (FAICEA) (Osteolinks TRAP-5b; DS Pharma Biomedical, Osaka, Japan) [25, 26]. The assay detected 10–2,800 mU/dL of TRACP-5b. The intra- and interassay coefficients of variation were 2.15 % and 2.95 %, respectively [27] Serum ucOC concentrations were measured by electrochemiluminescence immunoassay (Picolumi ucOC, Sanko Junyaku, Tokyo, Japan) [17, 28]. The electrochemiluminescence immunoassay for ucOC measures intact ucOC molecule [29]. The assay detected 0.39–50 ng/mL of ucOC [29]. The intra- and interassay coefficients of variation were 2.4 % and 2.9 %, respectively [29]. Glycated albumin was measured by an enzymatic method using the Lucica GA-L kit (Asahi Kasei Pharma Co., Tokyo, Japan). Hemoglobin A1C was measured by routine high performance liquid chromatography, which was standardized according to the Japan Diabetes Society.

Statistical methods

Data are summarized as mean ± standard deviation or median and range as appropriate. Comparisons in mean and median values between the two groups were performed by Student's t test and Mann–Whitney's U test, respectively. Correlations were assessed by non-parametric Spearman's rank correlation test. Multiple linear regression analyses were performed to determine the combined influence of age, gender, hemodialysis duration, body mass index, and intact PTH, BAP, TRACP-5b, and ucOC on the indices of glucose metabolism. Because ucOC, BAP, TRACP-5b, intact PTH, glucose, hemoglobin A1C, and glycated albumin levels exhibited skewed distribution, these variables were entered into multiple linear regression analysis after logarithmic transformation. P values less than 0.05 were considered statistically significant. All calculations were performed on a Windows personal computer using StatView V statistics software (SAS Institute Inc, Cary, NC).

Results

Clinical characteristics of the hemodialysis patients

The clinical characteristics of the diabetic and non-diabetic patients are shown in Table 1. There were no significant differences in age and gender between the diabetic and non-diabetic hemodialysis patients. Hemodialysis duration was significantly shorter for the diabetic patients than for the non-diabetic patients (p < 0.0001), and body weight and body mass index (BMI) was significantly greater in the diabetic patient group (p = 0.0219 and p = 0.0211, respectively). Plasma glucose, hemoglobin A1C, and glycated albumin were significantly higher in the diabetic patient group (p < 0.0001). Intact PTH was significantly lower in the diabetics (p = 0.0188).

Table 1 Baseline characteristics of DM and non-DM patients
Full size table

Relationship between ucOC and indices of bone metabolism

The median values (range) of the serum ucOC, BAP, and TRACP-5b in all patients were 20.0 (2.2–257.4) ng/mL, 18.8 (8.2–69.2) U/L, and 311 (78–1340) mU/dL, respectively. Relationships between ucOC and indices of bone metabolism are shown in Fig. 1. ucOC correlated significantly and positively with BAP and TRACP-5b (ρ = 0.489, p < 0.0001 and ρ = 0.585, p < 0.0001, respectively). Similarly, ucOC correlated significantly and positively with intact PTH (ρ = 0.621, p < 0.0001).

Fig. 1
figure 1

Relationship between serum undercarboxylated osteocalcin (ucOC) and bone alkaline phosphate (BAP), tartrate-resistant acid phosphate-5b (TRACP-5b), or serum intact PTH. Serum ucOC correlated significantly and positively with serum BAP (a), TRACP-5b (b), and intact PTH (c), using non-parametric Spearman's rank correlation test

Full size image

Relationship between ucOC and indices of glucose metabolism

Figure 2 shows ucOC, BAP, and TRACP 5b in non-diabetic and diabetic hemodialysis patients. Serum ucOC levels in the diabetic patients were significantly lower than those of the non-diabetic patients (14.4 (2.7–185.3) vs. 31.5 (2.2–257.4) ng/mL, p < 0.001), although there were no significant differences in serum BAP (18.4 (8.2–69.2) vs. 19.1 (9.1–68.4) U/L, p = 0.5701) and TRACP-5b (291 (87–778) vs. 336 (78–1,340) mU/dL, p = 0.1223) between the diabetic and non-diabetic patients.

Fig. 2
figure 2

Comparison of undercarboxylated osteocalcin (ucOC), bone alkaline phosphate (BAP), and tartrate-resistant acid phosphate-5b (TRACP-5b) between diabetic (DM) and non-diabetic (non-DM) patients. Horizontal bars represent the 10th, 25th, 50th, 75th, and 90th percentile levels. Single asterisk: p < 0.001 vs. non-DM (Mann–Whitney U test)

Full size image

The relationships between serum ucOC and indices of glucose metabolism were examined in all patients (Fig. 3). Serum ucOC correlated significantly and negatively with plasma glucose (ρ = −0.303, p < 0.0001), hemoglobin A1C (ρ = −0.214, p = 0.0036), and glycated albumin (ρ = −0.271, p = 0.0002). However, serum BAP did not correlate significantly with glucose, hemoglobin A1C, or glycated albumin. Serum TRACP-5b did not correlate significantly with hemoglobin A1C or glycated albumin, although it correlated significantly with plasma glucose (ρ = −0.228, p = 0.0018).

Fig. 3
figure 3

Relationship between undercarboxylated osteocalcin (ucOC) and plasma glucose, hemoglobin A1C (HbA1c) or glycated albumin (GA). Serum ucOC correlated significantly and negatively with plasma glucose (a), HbA1c (b), and GA (c), using non-parametric Spearman's rank correlation test. ρ: Spearman's correlation coefficient by rank

Full size image

Multiple regression analysis

Multiple linear regression analyses were performed to investigate whether indices of glucose metabolism were associated with serum log [intact PTH], log [BAP], log [TRACP-5b], and log [ucOC] in each model, respectively, after adjustment for age, gender, hemodialysis duration, BMI (Tables 2, 3, and 4). Log [ucOC] was significantly and independently associated with log [plasma glucose] (β = −0.216, p = 0.0063) after adjustment for age, hemodialysis duration, gender, and BMI (R 2 = 0.090, p = 0.0039), but log [intact PTH], log [BAP], or log [TRACP-5b] was not (Table 2). Log [ucOC] was significantly and independently associated with log [hemoglobin A1C] (β = −0.154, p = 0.0436) after adjustment for the same factors (R 2 = 0.142, p < 0.0001), but log [intact PTH], log [BAP], and log [TRACP-5b] were not (Table 3). Log [ucOC] was also significantly and independently associated with log [glycated albumin] (β = −0.176, p = 0.0226) after adjustment for the same factors (R 2 = 0.127, p = 0.0001), but log [intact PTH], log [BAP] or log [TRACP-5b] were not (Table 4).

Table 2 Multivariate regression analysis for predictors of plasma glucose
Full size table
Table 3 Multivariate regression analysis for predictors of hemoglobin A1C
Full size table
Table 4 Multivariate regression analysis for predictors of glycated albumin
Full size table

Discussion

In the present study of chronic hemodialysis patients, we showed that serum ucOC correlated significantly and positively with other bone metabolism markers (i.e., BAP (a bone formation marker), TRACP-5b (a bone resorption marker), and intact PTH), suggesting that ucOC may be indicative of bone turnover status. We showed that, of the three bone metabolism markers of ucOC, BAP, and TRACP-5b, only ucOC was significantly lower in diabetic hemodialysis patients, compared with non-diabetic hemodialysis patients, and that ucOC correlated significantly and inversely with indices of glucose metabolism. The significant association of ucOC with indices of glucose metabolism was demonstrated to be independent of age, hemodialysis duration, gender and BMI.

Hemodialysis patients have a wide variety of bone disorders, including secondary hyperparathyroidism, osteomalacia, and adynamic bone disease, ranging from high turnover of bone to low turnover of bone [21]. Serum bone markers, such as BAP and TRACP, which are not affected by renal function [1, 30], are significantly and positively associated with bone turnover [21]. In the present study examining hemodialysis patients, ucOC was significantly and positively associated with BAP and TRACP, by non-parametric Spearman's rank correlation test. This finding indicated that increased ucOC could be indicative of the status of increased bone turnover. The finding that ucOC correlated significantly and positively with intact PTH in the present study also indicated that patients with high turnover of bone caused by increased PTH may exhibit bone resorption, likely leading to efflux of bone matrix protein of OC into serum and increased concentrations of serum ucOC in these patients.

Concerning OC in diabetic patients, total OC has been shown to be lower in diabetic patients compared with normal subjects [15, 19, 31]. Im et al. [15] reported a significant reduction in serum total OC among type 2 diabetic postmenopausal women compared with normal glucose and impaired fasting glucose groups. Zhou et al. [16] also reported lower serum total OC levels in subjects with newly diagnosed type 2 diabetes, compared with normal glucose tolerance subjects after adjustment for age, gender, and BMI. Kindblom et al. [19] studied 1,010 elderly men and found that diabetic subjects had 21.7 % lower plasma total OC than non-diabetic subjects. Gerdhem et al. have also shown that serum OC levels, but not BAP, were lower in women with diabetes [32]. In the present study, in which hemodialysis patients with abnormalities of bone metabolism were examined, serum ucOC levels were significantly lower in diabetic hemodialysis patients, compared with non-diabetic hemodialysis patients, which is consistent with the previous findings in patients without chronic kidney disease. Because diabetes has been shown to be associated with low turnover of bone in previous studies [14, 32–34], the lower levels of ucOC in diabetics may be considered to be simply due to an association with the low turnover of bone in diabetics. In the present study, however, BAP and TRACP-5b levels were not significantly different between diabetic and non-diabetic hemodialysis patients, suggesting that potential low turnover of bone in diabetic hemodialysis patients does not explain the difference of ucOC levels between diabetic and non-diabetic hemodialysis patients.

Recently, accumulating emerging evidence has suggested that OC correlates with glucose metabolism in both diabetics and non-diabetics. Inverse relationships between total OC and fasting glucose and/or hemoglobin A1C have been confirmed in elderly subjects [18, 20], postmenopausal women [15], newly-diagnosed type 2 diabetes, and normal glucose tolerance subjects [16]. Among those with diabetes, elevated total OC was associated with lower fasting glucose and hemoglobin A1C in 149 postmenopausal women and 179 men [35]. In older Swedish men with and without diabetes, OC correlated negatively with fasting glucose, fasting insulin, and insulin resistance (HOMA-IR) [19]. Recently, the importance of ucOC, not carboxylated OC, on glucose metabolism has been emphasized [12, 13]. Iki et al. [36] reported that in elderly men without diabetes, serum ucOC correlated inversely with fasting plasma glucose, hemoglobin A1C, and insulin resistance (HOMA-IR). In a recent study in patients with type 2 diabetes, ucOC correlated negatively with fasting plasma glucose and hemoglobin A1C levels in men, even after adjustment for age, duration of diabetes, and body stature [17]. In a study of 199 middle-aged men without overt diabetes, elevated serum ucOC was associated with lower fasting and post-challenge glucose levels and higher insulin sensitivity (higher HOMA-β) [37]. None of the above studies, however, examined patients with chronic kidney disease. In the present study, we showed, for the first time, that there was a significant inverse association of ucOC with glucose metabolism indices in hemodialysis patients with abnormalities of bone metabolism. In relationship to bone metabolism, Kanazawa et al. [17] reported that serum ucOC correlated negatively with fasting plasma glucose and hemoglobin A1C in men with type 2 diabetes and normal renal function, but that serum BAP and urinary N-terminal cross-linked telopeptide of type-I collagen (a bone resorption marker) did not correlate with any parameter related to glucose metabolism. In the present study of hemodialysis patients with abnormalities of bone metabolism, we demonstrated that plasma glucose, hemoglobin A1C, and glycated albumin correlated significantly and negatively with ucOC in multiple regression analyses, but not with BAP, TRACP-5b, or intact PTH.

There are some limitations in this study. First, the sample size was relatively small. Second, the cross-sectional nature of the analysis did not allow for interpretation of causality regarding the relationship between serum ucOC and glycemic control. Third, we measured ucOC using ELISA. As such, further studies should use the hydroxyapatite binding assay to compare OC, ucOC, and the ratio ucOC/OC with glycemic control. Fourth, non-fasting serum glucose is of limited value for glycemic control. Fifth, in the present study, we did not measure bone status, such as bone density or bone quality, which might provide useful additional information on the relationship between ucOC and bone status. However, our cross-sectional study demonstrated a distinct association of ucOC, a possible marker of bone metabolism, with glycemic control, such as hemoglobin A1C, glycated albumin, and non-fasting serum glucose concentration, in hemodialysis patients.

In conclusion, we demonstrated, for the first time, that indices of glucose metabolism in chronic hemodialysis patients with abnormalities of bone metabolism correlated significantly and inversely with ucOC, which, in turn, correlated significantly with bone metabolism markers of BAP and TRACP-5b. These findings suggest that ucOC has a distinct association with glucose metabolism, in addition to being an index of bone markers in hemodialysis patients.