Biochemical markers of bone turnover in children with clinical bone fragility

Introduction

Osteoporosis is defined as a systemic skeletal disease characterized by low bone mass and microarchitectural deterioration of bone, leading to bone fragility and fractures. Once known as a disease of the aging, osteoporosis has now been recognized in significant debilitating conditions affecting pediatric patients starting at very young age. The causes of osteoporosis in children and adolescents are divided into two groups: primary osteoporosis or genetic brittle bone diseases, such as osteogenesis imperfecta (OI) or juvenile idiopathic osteoporosis (JIO), and secondary osteoporosis due to underlying chronic illnesses.

Anti-resorptive or bisphosphonate therapy has been used to treat children and adolescents with a variety of causes of bone fragility to improve bone density and reduce fracture rate. Many studies have demonstrated beneficial outcomes with pamidronate treatment in children with OI [1–3], as well as other bone fragility conditions such as idiopathic juvenile osteoporosis [4], spastic quardriplegic cerebral palsy [5–7], neuromuscular disorders [8], steroids-induced osteoporosis [9], and fibrous dysplasia [10]. Bisphosphonates decrease bone turnover by decreasing bone resorption. They have a direct effect on decreased recruitment and function of osteoclasts and an indirect effect by stimulating osteoblasts to produce an inhibitor of osteoclast formation [11, 12].

Bone metabolism is a dynamic process between bone formation and bone resorption. In adults, bone is continuously being maintained by bone remodeling, which is important for repair of microfractures as a result of wear and tear in response to stress and biomechanical forces. Bone metabolism in children is more complex, and greatly differs from that of adults as it reflects both skeletal growth and modeling. During remodeling, which is seen in adults, bone formation and resorption are tightly coupled. In contrast, bone modeling during growth in children results in new bone formation at a different site than the site of resorption and at a faster rate, leading to alterations in bone shape and bone mass accrual [13]. Interpretation of biochemical markers of bone turnover [bone turnover markers (BTMs)] in children is therefore more challenging than that in adults as biomarkers of bone metabolism represent the combined effects of bone formation, skeletal growth and bone remodeling.

Several biochemical assays are available that measure BTMs. These biochemical assays measure collagen breakdown products and other molecules released from osteoclasts and osteoblasts during the process of bone resorption and formation. Markers that are specific to bone formation include alkaline phosphatase (ALP), and osteocalcin (OC); markers specific to bone resorption include urine pyridinoline (PD), urine deoxypyridinoline (DPD), serum C-terminal cross-linking telopeptide of type I collagen (CTX) and N-terminal cross-linked telopeptide. In normal state of health, half of the ALP activity in serum is derived from bone, where it is produced by osteoblasts during bone formation, so called bone-specific ALP isoenzyme. The rest comes mainly from liver, with some from the intestine. Bone-specific ALP is a widely used bone formation marker in research as it is more specific for bone and therefore, more sensitive in detecting subtle changes in bone formation. Total ALP is considered to be an adequate marker for assessment of metabolic bone disease and it is often used due to convenience and low cost in evaluating and monitoring patients without liver disease. Total ALP is stable, with a relatively long half-life and no significant circadian variations [14]. Serum OC is derived from the most abundant non-collagenous protein in bone and is thought to be secreted exclusively by osteoblasts and thus correlates with osteoblastic activity, bone formation, and mineralization [15]. OC is elevated in high bone turnover states and is reduced in low bone turnover states [15]. PD is present predominantly in articular cartilage and to a lesser extent in bone, whereas DPD is most abundant in mineralized tissues such as bone and dentine [14]. During the process of bone resorption, type I collagen, the most abundant organic component of bone, is fragmented into smaller peptides which are then released into the blood and eventually excreted into the urine.

Several studies have shown that these markers could provide indices of bone turnover in patients with osteoporosis. In adults, biomarkers of bone turnover predict fracture risk and changes in bone mineral density (BMD) [16] and can be used to monitor the effectiveness of bisphosphonate treatment [17]. The evaluation of bone metabolism using such biochemical markers in children and adolescents is complicated by the variation by age, gender, or puberty with rapid growth [13, 15]. The role of biochemical BTMs in the care of pediatric patients with clinical bone fragility is not well established. Published literature contains limited data on the clinical usefulness of these markers in assessing osteoporosis and monitoring the efficacy of anti-resorptive therapy in pediatric patients. The aims of this study were (1) to evaluate associations between the markers of bone formation and bone resorption, (2) to examine the correlations between BTMs (serum ALP, serum OC, urine PD and DPD) and BMD, and (3) to evaluate the effects of bisphosphonate therapy on BTMs in children with bone fragility.

Materials and methods

Results

Clinical data of 115 patients (mean age 9.7±5.8 years, 72 males) with clinical bone fragility and low lumbar BMD were included in the study. Patients were classified as the following diagnoses: OI (41 patients), immobilization from various neuromuscular disorders (57 patients), juvenile osteoporosis (10 patients), and steroid-induced osteoporosis (7 patients). One hundred and two patients received bisphosphonates for bone fragility. Of these, 95 received pamidronate, three zoledronic acid, two residronate and two alendronate. Mean duration of bisphosphonate therapy was 4.0±1.9 years. Mean duration of follow-up was 8.0±2.9 years. Clinical data at baseline of 13 patients who did not receive bisphosphonate treatment were included in the analysis. As shown in Figure 1, the mean lumbar BMD Z-score increased steadily during the first 3 years of bisphosphonate therapy and then plateaued from the 4th to the 7th year.

Figure 1:

Lumbar vertebral BMD Z-scores in children with bone fragility and bisphosphonate therapy.

Lumbar Z-score increased significantly during the first 3 years of bisphosphonate therapy and then plateaued from year 4 to year 7 (as therapy was typically decreased or discontinued). Lumbar Z-score increased significantly during the first 3 years of bisphosphonate therapy and then plateaued from year 4 to year 7. (Numbers of patients were 102 at baseline, 102, 93, 84, 69, 52, 31, and 23 at 1, 2, 3, 4, 5, 6, and 7 years, respectively.)

There were positive correlations between lumbar BMD and age (r=0.6, p<0.001), and also between total BMC and age (r=0.374, p=0.006), demonstrating that lumbar BMD and BMC increase with age, as expected. In contrast, serum ALP, OC and urine PD had significantly negative correlation with age (r=–0.445, p<0.0001; r=–0.282, p=0.032; r=–0.533, p<0.001, respectively).

Associations between the markers of bone formation and bone resorption

There were significant positive correlations between serum OC and each of the following: serum ALP (r=0.31, p<0.015), urine PD (r=0.32, p<0.026), and urine DPD (r=0.48, p<0.001) (Figure 2), indicating that high bone formation and bone resorption markers correlated with high bone turnover state. There was no significant correlation between serum OC and serum parathyroid hormone (PTH).

Figure 2:

Serum osteocalcin in relation to serum alkaline phosphatase, urine pyridinoline and deoxypyridinoline in children with low bone density.

Serum osteocalcin was positively correlated with serum alkaline phosphatase, urine pyridinoline and urine deoxypyridinoline.

Association between BTMs and lumbar BMD

There was a significant association between lumbar BMD and log PD (slope=–0.19, 0.04SE, p<0.001) while controlling for baseline lumbar BMD (slope=0.75, 12SE, p<0.001). Similarly there was a significant correlations between lumbar BMD and log DPD (slope=–0.13, 0.04SE, p<0.001) and log ALP (slope=–0.18, 0.06SE, p=0.002) controlling for baseline lumbar BMD (slope=0.77, 0.12SE, p<0.001 and slope=–0.80, 0.12SE, p<0.001, respectively). Log transformations were used because of the right skewed nature of the BTM distributions, however all three BTM measures were significant (all p<0.005) with the arithmetic scale though the F-values were smaller due to larger variance caused by the tails of the distributions.

There were no significant associations between lumbar BMD Z-score and any of the absolute values of BTMs: serum ALP, serum OC, urine PD, or urine DPD. However, when using OC index (expressed as a ratio of measured OC value to upper limit of normal value for age) for analysis, there was a trend toward significant association between serum OC index and lumbar BMD Z-score (r=0.12, p=0.089) (Figure 3). Other indices of PD (r=–0.135, p=0.097) and DPD (r=–0.048, p=0.477) were also not significantly associated with lumbar BMD Z-score.

Figure 3:

Serum osteocalcin index and lumbar BMD Z-score in children with low bone density.

There was a trend toward significant correlation between serum osteocalcin index and lumbar BMD Z-score. Osteocalcin index was expressed as a ratio of measured osteocalcin value to upper limit of normal value for age. Lines represent linear trend for each follow-up year to illustrate the consistent small upper trend each year. The dotted line was the baseline year.

The effects of bisphosphonate therapy on BTMs

The indices of serum OC, urine PD and DPD over time are depicted in Figure 4. Mean urine PD index was >1.0 at baseline (absolute values being higher than upper limit of normal), indicating high bone turnover in this cohort. Similarly, urine DPD was also at the upper end of normal limit (having index >0.8). On the other hand, mean serum OC index was around half of the upper limit value at baseline. All OC, PD, and DPD indices trended to decrease (only OC was significant) during the first 3 years of bisphosphonate therapy, corresponding to the increasing lumbar BMD Z-scores observed during that time. Urine PD index was the first to increase to or above baseline value by year 4. Similarly, urine DPD index began to increase above baseline value by year 5, and then decreased back to baseline value. Serum OC index remained lower than baseline value for much longer time than PD or DPD index, and started to rise above baseline value by year 6.

Figure 4:

Bone turnover marker indices on bisphosphonate therapy in children with low bone mass.

Bone marker indices at baseline and while on bisphosphonate therapy. Index was expressed as a ratio of measured values and upper limit of normal range for age. Error bars are constructed so that non-overlapping bars are equivalent to an unadjusted p<0.05 (Fisher’s least significant difference test). (n=89, 66, 57, 50, 44, 35, 29, and 23 at baseline, 1, 2, 3, 4, 5, 6, and 7 years, respectively.)

There were no significant differences in bone turnover marker indices among patient classifications based on diagnoses.

Discussion

In this study, we found association between bone formation and bone resorption markers in pediatric patients with metabolic bone disease, as shown by positive correlations between serum OC and ALP, and between serum OC and urine markers for bone resorption. This indicates that high bone formation and resorption markers are both correlated with high bone turnover state. This is similar to studies in healthy children [20, 21], suggesting that the bone markers derive from and reflect the same biological processes of remodeling, modeling and epiphyseal growth, even in the setting of underlying metabolic bone disease, similar to that seen in healthy children.

We also found significantly negative associations between bone resorption markers (urine PD and DPD) and lumbar BMD, in this pediatric cohort with clinical bone fragility. This is similar to other studies in healthy children that showed significant correlations between bone mass and BTMs [20–23]. A study by Jürimäe et al. [21] showed significantly negative associations between BTMs and lumbar spine BMD, which remained significant after adjustment for age and pubertal stages. However, in this study, after controlling for age, we did not find any significant correlation between bone resorption markers and lumbar BMD Z-scores. After the variance accounted for by age was removed, there was no shared variance between BMD and BTMs. This suggests that BTMs do not predict the degree of low BMD, which is in agreement with a study by van der Sluis et al. [24] who found that BTMs cannot predict BMD in children.

It is well-recognized that BTMs in children are influenced by age, gender and pubertal stage, with increasing levels during pubertal growth spurt, paralleling the pattern of the standard height velocity curve, and declining during late puberty to adult levels [20, 25, 26]. This age-related changing pattern of BTMs results in different sets of pediatric normal reference values. Thus, unlike adults, in whom the assessment of BTMs is very straightforward, as the BTMs can be readily compared to baseline values, interpreting BTMs in growing children and adolescents is indeed much more challenging. Comparing the absolute values obtained at various times without taking into account of different age-related normal reference values can lead to misinterpretation of the results. Bisphosphonate therapy suppresses bone turnover, decreasing BTMs. As raw bone marker levels decline during late adolescence, it is impossible for clinicians to differentiate therapy-induced changes from physiologically age-related changes. Therefore, in this study, we utilized the index of bone markers, expressed as a ratio of measured values to upper limit of normal values for age. This gives more meaningful interpretation of the results and better comparison relative to the physiologically changing reference ranges for the longitudinal assessment of BTMs. Calculating standard deviation scores of bone markers have been suggested by Rauchenzauner et al. [27] to evaluate the changes of bone markers. However, this would require a mean value of normal reference with a standard deviation, which is not available to clinicians. Using the index of bone markers as described in our study is a more practical way to track changes of bone markers to monitor the response to bisphosphonate therapy. In addition to age-adjusted values, puberty-adjusted normative values may be more helpful in assessing BMTs in pediatric patients with chronic diseases that may have alteration in their pubertal development.

We observed a decrease in indices of both bone formation and bone resorption markers, with the gain in lumbar BMD Z-score during the first 3 years of active bisphosphonate treatment, indicating suppression of BTMs that occurred during anti-resorptive therapy. These findings are in agreement with what has been well documented in adults [28], as well as many studies in children [1, 29, 30]. Clinical usefulness of BTMs in monitoring therapy has been well recognized in adults, as changes in bone marker levels in response to treatment are more rapid and dynamic than changes in BMD [28]. The absence of change in bone marker levels following the osteoporosis treatment initiation usually indicates non-compliance with oral bisphosphonates therapy, and occasionally a lack of response due to new development of other bone pathologies with high bone turnover such as osteomalacia, primary hyperparathyroidism, hyperthyroidism, bone metastases and multiple myeloma that can be seen in adults [28]. More studies are needed in the pediatric population to see if the biochemical bone markers, repeated at much shorter intervals, can be useful for early assessment of effectiveness of therapy long before changes in bone mass can be ascertained.

We also observed a rise in urine PD and DPD in year 4–5 which coincides with the cessation of bisphosphonate therapy. Rauch et al. observed a steady increase in urinary N-telopeptide but still less than pretreatment baseline values within 2 years after pamidronate discontinuation in patients with OI [31]. BTM levels returned to baseline values rapidly after discontinuation of oral risedronate treatment in adults [32], whereas, after discontinuation of intravenous ibandronate in adults, bone resorption markers returned progressively to the initial level followed by a similar increase in bone formation [33]. By contrast, BTM levels did not return to baseline values even after 3–5 years following cessation of 5 years of oral alendronate treatment [34]. It has been suggested from many adult studies that the increase in bone turnover rate and the parallel decrease in BMD after treatment discontinuation depends on the types or the accumulated dose of bisphosphonates, which also reflects the difference in the degree of inhibition of bone turnover or the difference in the degree of accumulation of drugs in bone matrix [15]. It is unknown whether these observations in adults can be generalized to the pediatric population. Due to the small number of subjects treated with oral bisphosphonates, we could not accurately assess any differences in effects between oral and intravenous agents. There has been no consensus for the dose or how long bisphosphonate should be treated or whether or when the patients should be converted to low dose or oral bisphosphonate therapy in children with clinical bone fragility. More studies are needed to evaluate the clinical utility of BTMs in establishing the optimal dose, determining the duration of bisphosphonate treatment or the timing of retreatment after cessation of therapy in children.

An association between BTMs and fracture risk has also been observed in adults. A greater decrease in BTM levels during the first year of anti-resorptive treatment is associated with greater anti-fracture efficacy over 3 years of treatment [15]. The goal of optimal therapy in adults is that the BTMs should be within the lower end of normal range in adults [15]. A prospective study in pediatric patients to determine the correlations of BTMs with fracture risk would require a large sample size, which may prove to be challenging.

Conclusions

The results of this study showed that there was a consistent relationship between the bone formation and bone resorption markers in children and adolescents with clinical bone fragility, similar to that in healthy children. These biochemical BTMs were correlated with lumbar BMD but not lumbar BMD Z-score. While the BTMs were not reliable predictors of degree of low BMD, the bone markers showed suppression compared to baseline during bisphosphonate therapy. Further studies are needed in children to expand the clinical utility of BTMs in monitoring the response to therapy, assessing anti-fracture efficacy, and in determining the optimal dose or length of bisphosphonate therapy.