|Home | About | Journals | Submit | Contact Us | Français|
With the ageing population in most countries, disorders of bone and mineral metabolism are becoming increasingly relevant to every day clinical practice. Consequently, the interest in, and the need for effective measures to be used in the screening, diagnosis and follow-up of such pathologies have markedly grown. Together with clinical and imaging techniques, biochemical tests play an important role in the assessment and differential diagnosis of metabolic bone disease. These biochemical indices are non-invasive, comparatively inexpensive and, when applied and interpreted correctly, helpful tools in the diagnostic and therapeutic assessment of metabolic bone disease.
This second part of the two part series reviews the current evidence regarding the clinical use of biochemical markers of bone remodelling in the management of osteoporosis.
Part I of this review, published late last year, provided an overview of the basic biochemistry of bone markers, and sources of non-specific variability.1 This second part of the series will review the current evidence regarding the clinical use of biochemical markers of bone remodelling in bone disease. Most of this article will deal with the use of bone markers in osteoporosis, as this is the major clinical application at the time of writing. Other conditions such as Paget’s or metastatic bone disease, in which bone markers may be used clinically, are only briefly mentioned towards the end of this review; these topics have been reviewed recently by Rodman and Windle2 and by Seibel.3
Once somatic growth subsides, the serum and urinary concentrations of most bone markers return to a level much below those seen during normal puberty and growth. This stabilisation usually occurs during the 3rd decade and in healthy men, levels of practically all markers remain more or less unchanged until 70 years of age. After that, a slight increase is usually seen in both formation and resorption markers.4–7 In contrast, menopause is associated with a substantial acceleration in bone turn over, mirrored by a 50–100% increase in both markers of bone formation and resorption.4,5,8–14 In early postmenopausal women, this increase in bone turnover can be attenuated by oral calcium supplementation.15–19 Long-term treatment of women with oestrogen was shown to reduce resorption markers such as deoxypyridinoline (DPD) and N-telopeptide (NTX) to premenopausal levels.9,12,14,16,20–24 A prospective study covering the perimenopausal transition in healthy women suggests that changes in bone turnover occur during the late premenopause with a decrease in bone formation, which only later is followed by a rise in bone resorption.25 It is now widely accepted that the accelerated rate of bone loss seen after the menopause is mainly due to an imbalance in bone turnover and an increase in bone resorption.26,27 Studies employing specific bone markers indicate that bone turnover continues to be increased (and to be associated with bone loss) during late menopause.28–33 In some postmenopausal women,34 but particularly in the very elderly,35–37 this increase in bone turnover is often, but not always found to due to vitamin D and/ or calcium deficiency and secondary hyperparathyroidism.
Osteoporosis is a heterogeneous disease. It is therefore not surprising that in untreated patients with this disorder, rates of bone turnover tend to vary over a wide range. Although most cross-sectional studies show accelerated bone turnover in a certain proportion of postmenopausal osteoporotic women, there is usually broad overlap between diseased and healthy populations.11–13,38–41 In this context, it is important to bear in mind that research studies usually include highly selective patient populations, which may not always represent the population seen in the typical clinical setting. Using a population-based data set, and there fore avoiding this selection bias, we have previously shown that none of the major biochemical markers of bone turnover provide sufficient diagnostic information to be useful in the screening for vertebral osteopenia or osteoporosis.13 However, another population-based study showed that urinary levels of NTX could discriminate between older individuals with normal hip bone density, osteopenia and osteoporosis.42 Again, this association did not hold true for men at the level of the spine.
In retrospective population-based studies, Akesson and co-workers have demonstrated that previous fractures were associated with abnormal bone turnover.28,29,43,44 After adjustment for age and bone mineral density (BMD), women with fractures occurring within six years prior to the study were characterised by lower serum levels of osteocalcin and carboxyterminal procollagen type I propeptide (PICP), but normal rates of bone resorption. In another investigation, the same authors found decreased serum levels of osteocalcin, and elevated urinary concentrations of collagen crosslinks in elderly women at the time of admission for a newly sustained hip fracture.2
Taken together, these data suggest that a long-term imbalance of bone metabolism may lead to increased fragility. Together with the fact that high bone turnover may be sustained for long periods and bone loss may increase with age,44 these findings may provide a rationale for designing more effective intervention strategies. However, other factors such as age (see above), medication,6,18,46–51 immobilisation,32,35 thyroid function,52 co-morbidity35 and the fracture itself40,53,54 itself do influence bone metabolism and therefore need to be considered in the inter pretation of biochemical data and their use in individual patients. Clearly, none of the biochemical markers of bone turnover has proven useful as a single diagnostic index of osteoporosis.
Bone mass, rates of bone loss, and the risk of osteoporotic fractures are interrelated, and both low bone mass and rapid bone loss have been shown to be in de pendent predictors of future fracture risk.5 The rate of bone loss is determined by a number of factors, one of which appears to be the rate of bone remodelling. Earlier observations demonstrated that bone formation and bone resorption increase shortly after natural menopause, a phase that in most women is also associated with significantly accelerated bone loss.4,5,8–14 Similar observations have been made in ovariectomised, pre menopausal women and in castrated men,56,57 indicating that the withdrawal of endogenous sex steroid induces both high bone turnover and rapid bone loss. Conversely, markers of bone metabolism return to premenopausal levels during hormone replacement therapy (HRT).9,12,13,16,20–24 Other biochemical studies suggest that high rates of bone turnover may be sustained well into advanced ages.10,13,31,58,59 However, it is unclear whether this applies to all women.
Most longitudinal studies support the notion that individuals with high rates of bone turnover lose bone at a faster rate than subjects with normal or low bone turnover.16,20,30,60–65 Following a small group of early postmenopausal women, Christiansen and colleagues demonstrated that the combined measurement of serum total alkaline phosphatase, osteocalcin, fasting urinary calcium, hydroxyproline or DPD can predict 60–70% of the variability in bone loss.60,61 These studies also showed that the correlation between baseline markers of bone turnover and the subsequent rate of postmenopausal bone loss is possibly consistent over a period of at least twelve years.55,61 Less optimistic estimates were reported by other groups using different combinations of markers.30,62 For example, a study in elderly women demonstrated that urinary NTX, serum osteocalcin and serum parathyroid hormone together explained only 43% of the variability of bone loss at the hip.30 Markers of bone resorption seemed to be stronger predictors of future bone loss than markers of bone formation, and correlations were stronger in elderly than in younger women.62–65 In a retrospective study of 354 women (mean observation period: 13 years), Ross and Knowlton showed a continuous relationship between the measured levels of various bone turnover markers and the risk of rapid bone loss at the calcaneus: the odds of rapid bone loss (>2.2%/year) doubled for each standard deviation (SD) increase in serum bone specific alkaline phosphatase (BSAP), serum osteocalcin, urinary free pyridinoline (PYD) or DPD.65 In a study of 227 early postmenopausal women treated with calcium alone or HRT plus calcium, Chesnut et al. and Rosen et al. reported that women with high baseline rates of bone resorption were at higher risk of losing bone than women with normal turnover rates.16,20 (Figure 1). Different results were reported by Keen et al., who in a four-year prospective study were unable to detect any correlation between rates of bone turn over and changes in lumbar or hip BMD.66 Other groups argue that due to the high degree of variability in urinary markers of bone turnover, predicting either bone density or changes therein for an individual patient from a single marker measurement may not be possible.62,67 Vestergaard and colleagues showed that serum osteocalcin, BSAP and hydroxyproline are poor predictors of lumbar and hip bone loss in individual perimenopausal women.68
Taken together, there is evidence that rates of bone remodelling are associated with bone loss. However, the strength of this association seems to depend on a number of factors, such as menopausal age, skeletal site and gender. Bone remodelling markers are no substitute for individual bone mass measurements, or for a careful assessment of the patient’s personal and family history.
Bone turnover is an independent predictor of fracture risk. Earlier post-hoc analysis of data from clinical trials suggested that in untreated osteoporotic women, vertebral fracture rates increase as a direct function of either increased bone turn over or of decreased vertebral BMD.69 Thus, at a given level of vertebral BMD, the rate of vertebral fractures increases with the rate of bone turnover. When bone turnover is normal, however, the main determinant of vertebral fractures is vertebral BMD.69
Using the large population-based sample of the Rotterdam study (7983 individuals, 60% women aged 55 years and over), van Daele et al. showed that women with increased urinary DPD levels had an increased risk of hip fracture.70 The relative risk per SD increase in urinary DPD was 3.0 (95% confidence interval (CI) 1.3–8.6). Interestingly, part of this association appeared to be related to disability at baseline. However, when the data were corrected for disability, a relative risk of 1.9 (95% CI 0.6–5.6) remained. This number is very similar to the increase in fracture risk calculated for 1 SD decrease in BMD at the lumbar spine. Later analyses of the same study revealed that low serum osteocalcin concentrations were also associated with an increased risk of hip fracture (odds ratio (OR) : 3.1; 95% CI 1.0–9.2).
In a five-year follow-up of the same population, Weel et al. later showed that an increase in baseline urinary DPD above the premenopausal mean value was associated with an increased future risk of osteoporotic fractures.71 All types of non-vertebral fractures, but especially fractures of the hip (OR 5–6) and the upper humerus (OR 3–5) were associated with urinary levels of DPD above the premenopausal mean, independent of bone mineral density and disability. Fracture risk increased dramatically when elevated rates of bone resorption were combined with low BMD.
Similar results have been published for the French EPIDOS study.72 The relative fracture risks as defined by either BMD or marker measurements were similar (RR ~2) to those reported earlier by van Daele. Again, combined measurements of hip bone density and of bone resorption markers increased the predictive power for hip fractures (RR 5–6). Thus, in elderly women, the relative risk of hip fracture seems to be highest in individuals with both low hip BMD and high rates of bone resorption.
A nested case control study from the same group later suggested that levels of serum under-carboxylated osteocalcin (ucOC), but not of total osteocalcin, were predictive of future hip fractures (OR 2.0; 95% CI: 1.2–3.2).73 These data confirm and extend previous reports, which suggest that increased serum levels of ucOC are predictive of hip fractures in elderly, institutionalised women.74–76 These earlier results, however, may merely indicate an association between poor nutritional status and hip fracture risk among institutionalised subjects, and not a general bio logical mechanism possibly relevant to a more representative sampling of the population. The significance of vitamin K deficiency to the under-carboxylation of osteocalcin had been demonstrated earlier by Price et al.,77 and subsequent clinical studies showed that overt vitamin K deficiency may lead to a disproportionate increase in ucOC in the circulation.78,79 In addition, vitamin K2 levels have been shown to be lower in women with osteoporotic fractures than in healthy individuals.78 Although measurement of ucOC may be useful in providing an integrated assessment of the factors that are responsible for the gamma-carboxylation of osteocalcin, such as vitamins K and D, the underlying biochemical mechanisms by which ucOC could be associated with impaired bone metabolism are, as yet, unknown.
In another prospective study from Sweden, low serum levels of both the carboxy terminal propeptide and telopeptide of type I collagen were associated with an increased risk of hip fracture, independent of age and BMD.43 Thus, increased rates of bone resorption or decreased rates of bone formation seem to be associated with future osteoporotic fractures.
Recently, Meier and colleagues demonstrated in a case-cohort control study of 151 elderly men followed prospectively over 6.3 years that accelerated bone resorption was associated with increased risk of osteoporotic fracture, independent of BMD. Combining measurements of BMD and bone turnover improved fracture prediction in elderly men.80 (Figure 2).
Prospective data from the Australian FREE study of 1112 frail elderly men and women indicate that high bone turnover is also an independent predictor of all cause mortality. This association appeared to be mainly manifested in deaths from cardiovascular causes.33 (Figure 3).
In summary, data from several independent and large prospective studies indicate that in both postmenopausal women and healthy men, increased rates of bone resorption are associated with an increased risk of vertebral and non-vertebral fractures, independent of BMD, age and disability. In the future, markers of bone turnover, in combination with other risk factors for osteoporotic fracture, may be used to define fracture risk and intervention thresholds.
From both a theoretical and clinical point of view, it is conceivable that intervention strategies may differ between patients with accelerated, normal or even abnormally low bone turnover at the time of diagnosis; hence, a patient presenting with high rates of bone resorption may benefit from antiresorptive therapy, whereas in an individual with low bone turnover, a stimulator of bone formation may yield better long-term results. So, are pretreatment bone marker measurements helpful in guiding the selection of therapy for individual patients? Some studies have shown that in osteoporotic patients treated with sub cutaneous calcitonin, increases in lumbar (but not necessarily in hip) BMD were significantly greater in individuals with high than with normal or low base line rates of bone turnover.81–83 Similar results were later reported for short term Alendronate treatment,84 although one report (with an equally small number of subjects) suggests that changes in BMD during treatment with Alendronate are independent of pre-therapeutic bone turnover rates.31
Chesnut et al. and Rosen et al. demonstrated in 227 women treated with either calcium alone or a combination of HRT plus calcium, that individuals within the highest quartile for baseline measures of bone turnover also experienced the greatest gain in BMD after six and twelve months of treatment with HRT and calcium.16,20 In this study, baseline urinary NTX and serum osteocalcin showed the highest predictive values for a change in spinal BMD after one year of either HRT or calcium. In reverse, those women showing a gain in BMD after one year of HRT had significantly higher base line rates of bone resorption (as determined by urinary NTX) than non-responders or subjects losing bone during HRT16 (Figure 1). This observation is in agreement with the hypothesis that changes in bone turnover affects the risk of vertebral fractures only if bone turnover is significantly accelerated, while more subtle changes appear to be without major effect.69 In contrast, Stevenson et al. in a three-year prospective study on the effect of HRT on spine and hip BMD were unable to distinguish between responders and non-responders by means of either base line or follow-up measures of bone turnover.85 Both groups showed the same pretreatment values of bone formation and resorption, and the change in bone markers in response to HRT was identical in the affected and unaffected groups.85
Post-hoc analyses of the Risedronate clinical phase III programs show that the reduction in fracture risk during one and three years of Risedronate treatment is similar in patients with baseline urinary DPD below or above the premenopausal median (ie with normal or accelerated bone resorption).86 (Figure 4). However, the number of patients needing treatment to avoid one fracture during one and three years of treatment with Risedronate is significantly lower in patients with elevated baseline bone turnover as compared to patients with low baseline bone turnover. Thus, although the reduction in overall fracture risk seems to occur independent of baseline bone turnover, patient stratification by pretreatment bone resorption rates seems to make some sense from a pharmacoeconomic point of view.86
A similar post-hoc analysis of the Fracture Intervention Trial (FIT), examining the influence of pretreatment bone turnover on the anti-fracture efficacy of daily alendronate in postmenopausal women found that the non-spine fracture efficacy of alendronate was significantly greater among both osteoporotic and non-osteoporotic women with higher baseline levels of the bone formation marker aminoterminal propeptide of type I collagen (PINP). However, no such association was observed for vertebral fractures, and changes in both BSAP and the carboxyterminal crosslinked teleopeptide (CTX-I) were not associated with fracture outcomes at any site.87 (Figure 5).
Taken together, it remains unclear whether there is a clinically relevant relationship between bone turnover at baseline and the response to antiresorptive treatment. Drugs even of the same class may differ in this respect.
Bisphosphonates, raloxifene, denosumab, strontium ranelate, oestrogens, calcium, calcitonin and teriparatide all improve BMD to varying degrees. In contrast, the effects of these anti-osteoporotic drugs on bone turnover differ greatly: bisphosphonates (Figure 6), oestrogens, denosumab (Figure 7), calcitonin and raloxifene tend to reduce bone resorption and bone formation in a dose dependent manner.88–96 Strontium ranelate, in contrast, has only subtle effects on bone turnover, showing a slight reduction in bone resorption markers and a mild increase in bone formation markers.97 (Figure 8). Finally, teriparatide strongly increases both bone formation and, after a certain gap period, bone resorption markers.98–102 (Figure 9).
Most bisphosphonates, raloxifene, strontium ranelate, oestrogens and teriparatide have also been shown to reduce the risk of osteoporotic fractures. However, the observed reduction in fracture risk is only partly explained by the documented changes in BMD, with the reduction in fracture risk being much greater than predicted from improvements in BMD only.103–106 Hence, it has been estimated that changes in BMD explain only 4% to 28% of the reduction in vertebral fracture risk attributed to antiresorptive treatments.107–109 It is therefore likely that changes in other determinants of bone strength, including the rate of bone turnover and its changes during antiresorptive therapy, may be better predictors of anti-fracture efficacy. In fact, several studies confirmed that short-term reductions in bone turnover were associated with a reduction in vertebral and/or non-vertebral fracture risk in women treated with HRT,69 raloxifene 94,95,111 risedronate,112 alendronate92 and ibandronate.110
The relationship between six and 12 months changes in bone turnover markers and vertebral fracture risk after three years of raloxifene treatment in postmenopausal women clearly favours bone turnover markers as the better predictor of outcome.111 A decrease of 9.3 pg/mL in serum osteocalcin after one year of raloxifene treatment was associated with an OR for new vertebral fractures after 3 years of 0.69 (CI 0.54–0.88, p=0.003). Similarly, for a decrease of 5.91 μg/L in BSAP the OR was 0.75 (CI 0.62–0.92, p=0.005). Importantly, these relationships remained after adjustment for baseline vertebral fracture status and BMD. Two subsequent analyses including postmenopausal women with osteoporosis from
the same cohort (MORE trial) extended and confirmed these results showing that both, one year percentage changes in serum PINP and osteocalcin are able to predict the reduction in vertebral fracture risk after three years of treatment.94,95
Two studies have investigated the change in bone turnover markers and fracture risk in bisphosphonate treated postmenopausal women.92,112 Post-hoc analyses of data from the VERT studies including postmenopausal women with at least one vertebral fracture demonstrated that reductions in urinary CTX-I (by 60%) and NTX-I (by 51%) at three to six months of risedronate treatment were significantly associated with the reduction in vertebral and non-vertebral fracture risk after three years.112 The change in bone resorption markers explained 50–60% of the risedronate-related fracture risk reduction for both vertebral and non-vertebral fractures. Bauer et al. reported that in alendronate-treated women, greater reductions in bone turnover were associated with fewer osteoporotic fractures.92 In their study, each SD reduction in the change in BSAP at one year was associated with fewer spine (OR 0.74; CI: 0.63, 0.87), non-spine (relative hazard (RH) 0.89; CI: 0.78, 1.00) and hip fractures (RH 0.61; CI: 0.46, 0.78). Furthermore, alendronate-treated women with at least a 30% reduction in BSAP had a lower risk of non-spine (RH 0.72; CI: 0.55, 0.92) and hip fractures (RH 0.26; CI: 0.08, 0.83) relative to those with reductions <30%. Again, this effect was at least as strong as the anti-fracture effect observed with one year change in BMD.2
In summary, changes in bone turnover during raloxifene and bisphosphonate therapy seem to be related to subsequent fracture risk with a far greater effect on fracture reduction than has been attributed to treatment-induced changes in BMD. These data suggest that biochemical markers of bone turnover are useful tools to evaluate therapeutic effects after a relatively short period of time, and that serial measurements of bone markers may help to decide whether or not a patient responds to a specific antiresorptive treatment. Whether changes in bone turnover during treatment with agents such as strontium ranelate, denosumab or teriparatide predict fracture outcomes is presently not clear.
Long-term compliance with treatment for osteoporosis is usually poor.113 Several studies reported that up to 50% of postmenopausal women were not adherent to their treatment after one to five years of HRT.114–118 A major cause of non-compliance were unwanted side-effects or fear of side-effects, inconvenience caused by medication, and high drug costs.119.120 Hence, monitoring patients on antiresorptive medication is an eminent part of patient management in order to improve adherence and persistence to therapy, and ultimately treatment effectiveness.
Biochemical markers of bone turnover have been advocated to facilitate follow-up of patients receiving antiresorptive treatments for osteoporosis. As bone turnover markers, in particular indices of bone resorption, decrease rapidly after initiation of treatment within three to six months, they might represent useful surrogate markers for monitoring patient compliance. Only few data, however, are available to support this theoretical approach. Using a decision analysis model, Chapurlat et al.121 compared two strategies of follow-up: a) treatment of a woman without specific monitoring, and b) treatment of this woman with measurement of a serum marker of bone resorption after three months of treatment, with change of treatment if response to treatment as assessed by this marker was not satisfactory. This study suggests that monitoring osteoporotic women with measurements of bone markers early during the treatment course may increase effectiveness of treatment with greater quality adjusted life years than no follow-up. In another study of 75 postmenopausal women treated with raloxifene, Clowes et al. examined whether monitoring (nurse-monitoring or marker-monitoring) enhances adherence and persistence with antiresorptive therapy, and whether presenting information on the biochemical response to therapy provided additional benefit.122 Survival analyses showed that in the group being monitored, cumulative adherence to therapy increased by 57% compared with no monitoring; also, there was a trend for the monitored group to persist with therapy for longer periods of time. However, presentation of results of effects on NTX-I levels did not improve compliance to therapy compared with nurse-monitoring alone. Nevertheless, results from the IMPACT study in postmenopausal women on risedronate have shown that a reinforcement message based on bone marker response influences persistence with long-term treatment.123 In patients in whom a verbal feedback on the change of urinary NTX-I was provided, one-year persistence was higher than in non-reinforced subjects. Interestingly, the message given to patients with a bone turnover marker response considered “good” was associated with significant improvement in persistence, whereas the information given to those with a poor resorption marker response led to a lower persistence.123 Another large study investigating patient compliance using measurements of urinary CTX-I is under way and should give further evidence whether monitoring osteoporosis treatment using bone turnover markers should be encouraged in clinical practice.124
An abundance of experimental and clinical studies have demonstrated that markers of bone formation and resorption are useful tools in the assessment of the skeletal response to a great variety of influences. For example, markers of bone turnover may reflect changes in bone metabolism induced by oophorectomy,57,125 hyperparathyroidism,126,127 Paget’s disease,128 physical exercise,129 immobilisation,32,130 alcoholism, 131 smoking,132 vitamin D deficiency, 33,35,37,133 chronic inflammatory bowel disease,134,135 chronic starvation,136 thyroid disorders,52,137 as well as the pharmacological effects of glucocorticosteroids,48,139,140 androgens,6,7,141 gonadotropin-releasing hormone agonists,142 warfarin,143 growth hormone or insulin-like growth factors.144 Bone turnover markers may in useful in the diagnosis and management of certain of the above conditions, but in most cases have not been rigorously examined.
The situation is somewhat different in cancer: as bone metastases profoundly perturb normal bone remodelling,145 biochemical markers of bone turnover have been shown to reflect these tumour-induced changes in bone remodelling and may therefore be useful in the diagnosis, follow-up and prognosis of patients with malignant (bone) disease.146,147 (Figure 10). Most markers of bone turn over, particularly those of bone resorption, are elevated in patients with established bone metastases (recently reviewed3). While this may indicate a role of bone markers as diagnostic tools in cancer patients, available evidence does not provide any final conclusions as to the accuracy and validity of the presently used markers in the early diagnosis of bone metastases.
Markers of bone resorption respond promptly and profoundly to bisphosphonate and anti-neoplastic therapy, and this response appears to be associated with a favourable clinical outcome in patients with bone metastases.148–149 Recent evidence indicates that the aim of bisphosphonate therapy should be to normalise increased rates of bone remodelling.150,151 However, it remains unknown whether the use of bone markers in the routine clinical setting has any defined beneficial effects on overall outcome in cancer patients. In particular, no study has addressed the question whether patients with bone metastases should be treated according to their rate of bone turnover, and what the treatment goals are in this respect. While it is unlikely that bone turnover markers have sufficient diagnostic or prognostic value to be used in isolation, the combination of these markers with other diagnostic techniques may be the way forward to improve the clinical assessment of patients with bone seeking cancers.
Although the above mentioned studies represent only a small selection of the available literature, they all demonstrate that markers of bone turnover are extremely helpful tools in evaluating the physiology and pathophysiology of bone metabolism, and in elucidating the pathogenesis of bone disease.
Competing Interests: None declared.