The field of bone turnover markers has developed considerably in the past decade. Biochemical monitoring of bone metabolism depends upon measurement of enzymes and proteins released during bone formation and of degradation products produced during bone resorption. Various biochemical markers are now available that allow a specific and sensitive assessment of the rate of bone formation and bone resorption of the skeleton.[1] Although these markers are not currently recommended for use in the diagnosis of osteoporosis, they appear to be useful for the individual monitoring of osteoporotic patients treated with antiresorptive agents.[2] (See the image below.)
A summary list of bone formation markers is as follows:
Serum total alkaline phosphatase
Serum bone–specific alkaline phosphatase
Serum osteocalcin
Serum type 1 procollagen (C-terminal/N-terminal): C1NP or P1NP
A summary list of bone resorption markers is as follows:
Urinary hydroxyproline
Urinary total pyridinoline (PYD)
Urinary free deoxypyridinoline (DPD)
Urinary collagen type 1 cross-linked N-telopeptide (NTX)
Urinary or serum collagen type 1 cross-linked C-telopeptide (CTX)
Bone sialoprotein (BSP)
Tartrate-resistant acid phosphatase 5b
Such markers can also be useful in selected cases to improve the assessment of individual fracture risk when bone mineral density (BMD) measurement by itself does not provide a clear answer. The combined use of BMD measurement and bone markers is likely to improve the assessment of the risk of fractures in those cases.[3, 4, 5] See the Fracture Index WITH known Bone Mineral Density (BMD) calculator.
Diagnosis of osteoporosis is not based on evaluation of bone markers, and BMD assessment is still the criterion standard for evaluation and diagnosis. However, mean values for markers of bone turnover are higher in patients with osteoporosis than in the matched controls. In various studies, the mean urinary excretion of deoxypyridinoline (DPD) is 20-100% higher in patients with osteoporosis than in healthy subjects. In another study, an inverse relationship between the quartile of urinary collagen type 1 cross-linked N-telopeptide (NTX) excretion and mean BMD exists, but these results are not consistent. Moreover, values of healthy subjects and patients with osteoporosis overlap substantially.[6] Therefore, measurement of bone markers is not recommended to make a diagnosis of osteoporosis.
For other discussions on osteoporosis, see the overview topics Osteoporosis and Pediatric Osteoporosis, as well as the articles Osteoporosis in Solid Organ Transplantation and Nonoperative Treatment of Osteoporotic Compression Fractures.
Alkaline phosphatase has been clinically available for several years as a marker for bone metabolism. Serum alkaline phosphatase consists of several dimeric isoforms that originate from various tissues, such as liver, bone, intestine, spleen, kidney, and placenta. In adults with normal liver function, approximately 50% of the total alkaline phosphatase activity arises from the liver and 50% from the bone.
The development of immunoassay-based markers with monoclonal antibodies directed to the bone-specific isoform of alkaline phosphatase has improved specificity and sensitivity. Changes in bone-specific alkaline phosphatase can lag by several weeks. Following the start of antiresorptive therapy, the suppression is observed with the resorption markers as the coupling process is normalized.
Osteocalcin is a small protein (49 amino acids) synthesized by mature osteoblasts, odontoblasts, and hypertrophic chondrocytes. Serum osteocalcin is considered a specific marker of osteoblast function, as its levels correlate with the bone formation rate. However, the peptide is rapidly degraded in the serum, and intact and fragmented segments coexist in the serum. The resulting heterogeneity of the osteocalcin fragments in the serum leads to limitations with the use of this marker. Osteocalcin levels follow a circadian rhythm characterized by a decline during the morning, a low around noon, and a gradual increase to a peak after midnight. Serum osteocalcin levels reportedly vary significantly during the menstrual cycle, with the highest levels observed during the luteal phase.[7]
The major advantages of using osteocalcin as a clinical index of bone turnover are its tissue specificity, its wide availability, and its relatively low within-person variation. In general, serum levels are elevated in patients with diseases characterized by a high bone turnover rate, and the serum levels reflect the expected changes in bone formation following surgical and therapeutic intervention. An exception is found in Paget disease, in which serum alkaline phosphatase is a better predictor of severity of disease than osteocalcin.
Procollagen type 1 contains N- and C-terminal extensions, which are removed by specific proteases during conversion of procollagen to collagen. The extensions are the C- and N-terminal propeptides of procollagen type 1 (P1CP and P1NP). Anti-P1NP antibodies are used to detect the trimeric structure of P1NP by enzyme-linked immunosorbent assay (ELISA) or radioimmunoassay. Measurement of P1NP appears to be a more sensitive marker of bone formation rate in osteoporosis. These assays are being developed for clinical use.
The most useful markers of bone resorption are degradation products derived from the enzymatic hydrolysis of type 1 collagen, particularly peptides related to regions of cross-linking with pyridinoline (PYD). Collagen type 1 represents over 90% of the protein in bone, and it is natural that many bone markers are derived from released collagen fragments. Advances have been made in utilizing noncollagenous proteins and enzymes for serum assays of bone resorption. Their application is becoming increasingly widespread.[8]
Until the early 1990s, urinary hydroxyproline was one of the main bone resorption markers available, but this assay lacked specificity and sensitivity. Hydroxyproline is a component of the bone collagen. During degradation of bone, it is released into the serum and reaches the urine in free and bound forms. Today, serum hydroxyproline is considered a nonspecific marker of bone turnover, since it is derived from the degradation of newly synthesized collagens, from collagens of tissues other than bone, and from diet. Therefore, more specific techniques have replaced urinary hydroxyproline. From a practical standpoint, another major drawback of urinary hydroxyproline was the necessity for dietary restrictions on gelatin intake before applying the test.
Hydroxypyridinium cross-links collagen, PYD, and DPD. The pyridinium compounds, PYD and DPD, are formed during the extracellular maturation of fibrillar collagens and are released upon the degradation of mature collagens (see the images below). The measurement of PYD and DPD is not influenced by degradation of newly synthesized collagens and is independent of dietary sources. While PYD is found in cartilage, bone, ligaments, and vessels, DPD is found in bone and dentin only. The PYD-to-DPD ratio in urine is similar to the ratio of these 2 cross-links in bone, which suggests that both of the cross-links are derived predominantly from bone. PYD and DPD are present in urine as free moieties (40%) or peptide bound (60%). Free forms can be detected by direct immunoassays (free DPD, Pyrilinks-D).
Rather than use the cross-links themselves as markers, several groups have developed assays based on specific antibodies raised against isolated collagen peptides containing cross-links. These fragments detected by radioimmunoassay technique are available for C-telopeptide of type 1 collagen (CTX, CrossLaps) and cross-linked N-terminal telopeptide of type 1 collagen by ELISA technique (NTX, Osteomark).
The monoclonal antibody used for NTX assay is directed against the urinary pool of collagen cross-links derived from a patient with Paget disease. Only β-isomer of CTX is measured in serum CrossLaps assay, while α- and β-isomers of CTX are measured in urine CrossLaps assay. These assays have showed detectable reaction with urine from healthy individuals, as well as large increases associated with elevated turnover.
Currently, 2 categories of C-telopeptide methods exist: the CTX and type I collagen cross-linked C-telopeptide (ICTP). These recognize different segment domains of the C-terminal telopeptide region of the α1 chain of type 1 collagen and respond differently to bone metabolic processes. While CTX responds remarkably to antiresorptive therapies, serum ICTP is insensitive to normal metabolic bone processes, such as osteoporosis, but serum ICTP may be a marker of bone degradation in pathological conditions (eg, bone metastasis, rheumatoid arthritis).
The pyridinium cross-links and the collagen telopeptides involving the cross-linking sites are considered the best indices for the assessment of bone resorption. Circadian rhythm in bone metabolism causes markers to vary by 10-20%. A study by Yavropoulou et al investigated postprandial suppression of bone resorption, which is one of the main contributors to the circadian rhythm of bone turnover markers. The study concluded that the physiologic response to nutrients in terms of the general homeostasis and functional integrity of the skeleton is important, as evidenced by the preservation and augmentation of postprandial suppression of bone resorption in patient with disease affecting bone metabolism.[9]
Urine creatinine varies during the day by 20%. Urine cross-links/creatinine ratios vary by 20-30%. This variability could be corrected by measuring the markers in the morning (first or second void urine) when the levels are the highest. Serum CTX is the most commonly used measurement. Since it has been automated and there is a large amount of data available supporting the use of urine and serum CTX for following antiresorptive therapy.
Relatively few noncollagenous proteins or glycoproteins have sufficient specificity for bone to be considered potential markers. Bone sialoprotein (BSP) is thought to be involved in the mineralization of newly deposited bone matrix and/or the calcification of extraskeletal tissues. BSP is a highly acidic protein with strong affinity for hydroxylapatite crystals. BSP may be a sensitive marker of bone turnover, and clinical data suggest that its serum levels predominantly reflect processes related to bone resorption.
The discovery that the type 5b isotype is specific for bone osteoclasts has facilitated an antibody capture activity assay for tartrate-resistant acid phosphatase 5b as a bone resorption marker but is still under development.
Bone mineral density (BMD) is an important predictor of fracture risk; however, a single measurement indicates only current BMD, not the anticipated rate of bone loss. Patients with a given BMD who are losing bone rapidly have a greater risk of fracture later in life. Nonetheless, patients with low BMD or high marker values would be at risk for osteoporosis and warrant preventive measures with antiresorptive agents. See the Fracture Index WITH known Bone Mineral Density (BMD) calculator.
The relationship between biochemical markers of the bone turnover and the rate of bone loss has been investigated in prospective studies that show conflicting results because of the various technical limitations, such as precision error of the repeated measurements of bone markers and precision error of BMD measurements for rate of bone loss.
Ross and Knowlton, in a study of osteoporotic fractures that monitored older postmenopausal women for 4 years, found that baseline markers of bone resorption were significant predictors of bone loss rate and found that the odds for rapid bone loss increased by 1.8-2.0 times for each standard deviation (SD) increase in the marker. They studied a subset of 200 women in a longitudinal component of the large Hawaii osteoporosis study (HOS), which included over 1100 postmenopausal women.[10] They subdivided those women who had baseline markers into groups with rapid bone loss (>2.2%/y) and those with the slowest loss (< 0.4%/y). The average bone loss was twice as great for those women in the highest quartile compared with women in the lowest quartile. Similar data are available for younger postmenopausal women.
Gutierrez-Buey et al found that women with higher baseline levels of serum bone turnover markers and a lower trabecular bone score had lower BMD during the transition to menopause. The researchers analyzed data from 64 premenopausal women with normal bone density who were enrolled in a prospective, longitudinal study. At 5-year follow-up, 48.4% of the women had normal BMD, 45.8% had low bone mass, and 6.3% had osteoporosis. Those with osteopenia or osteoporosis at follow-up had higher collagen type 1 cross-linked C-telopeptide (CTX) and procollagen type 1 N-terminal propeptide (P1NP) levels at enrollment than the women with normal bone mass.[11]
Analysis of studies investigating whether baseline turnover markers can predict the bone density changes in patients treated with antiresorptive drugs has been controversial. In a subset analysis of a trial with alendronate, the baseline urinary collagen type 1 cross-linked N-telopeptide (NTX) or other parameters did not correlate with subsequent spine or hip BMD.[12] The correlation was weak between baseline NTX and the change in spine BMD after 2 years of 2.5 mg of alendronate.
A study by Tsai et al analyzed the relationship between combined antiresorptive/anabolic therapy and BMD and biochemical markers of bone turnover changes. The study found that early biochemical markers of bone turnover changes predict 2-year BMD gains in the spine in women treated with teriparatide or denosumab. BMD increases at the distal radius were associated with less suppression of bone turnover in women treated with combined teriparatide/denosumab therapy.[13]
Baseline markers of bone turnover provide insight into the state of the remodeling units in the adult skeleton. In younger and older postmenopausal women, resorption and the formation markers do predict, with some degree of confidence, the degree of bone loss without therapeutic intervention. However, the data published suggest that baseline turnover markers do not predict the response to therapy.
Miazgowski et al followed bone turnover markers, BMD, and serum adiponectin in postmenopausal women with newly diagnosed type 2 diabetes. After 12 months of diabetes treatment, there was a significant decrease in body weight, waist circumference, and hemoglobin A1c. Improved diabetic control had no impact on bone turnover markers, which did not change significantly during the study. Bone markers, BMD, and the rate of bone loss fell in the normal range for postmenopausal women with newly diagnosed type 2 diabetes.[14]
The level of bone mass can be assessed with adequate precision by measuring bone mineral density (BMD) using dual-energy x-ray absorptiometry (DXA). However, this measurement does not capture all risk factors for fracture. Bone fragility also depends on the morphology, the architecture, and the remodeling of bone, as well as on the quality (properties) of the bone matrix that cannot be readily assessed. In addition, the risk of fracture is also influenced by muscle function, the propensity to fall, and the ability to adapt to such falls.[15, 16, 17]
With emergence of effective but rather expensive treatments, detecting those women at higher risk for fracture is essential. Several prospective studies have demonstrated a strong association between BMD and the risk of hip, spine, and forearm fractures. However, half of the patients with incident hip fractures have baseline BMD assessed by DXA above the diagnostic threshold of osteoporosis, defined as a T-score of -2.5 SD or more. Clearly, improvement is needed in the identification of patients at risk for fracture.
In the large cohort of elderly women in France (EPIDOS), no significant relationship was found between levels of serum osteocalcin and bone alkaline phosphatase and the risk of hip fracture occurring during a 2-year follow-up. In contrast, 2 prospective studies in younger healthy postmenopausal women (OFELY and HOS) showed a significant positive association between an increased level of bone alkaline phosphatase and the risk of vertebral and nonvertebral fracture. Bone resorption assessed urinary and serum C-telopeptide of type 1 collagen (CTX) or urinary free deoxypyridinoline (DPD) above the normal premenopausal range were consistently associated with about a 2-fold higher risk of hip, vertebral, and other fractures over follow-up periods ranging from 1.8-5 years. The combination of BMD and bone turnover measurement allows the identification of women at a much higher risk for hip fracture.
The potential validity of this approach is illustrated in Table 1, below.
Table 1. Fracture Risk Based on BMD and Biochemical Bone Turnover Markers* (Open Table in a new window)
Population |
Odds Ratio (95% CI) |
Likelihood Ratio |
Probability of Fracture Over 5 y, % |
All women (N = 435) |
... |
... |
12.6 |
Women with low femoral neck BMD (T score < -2.5) |
2.8 (1.4-5.6) |
2.80 |
39 |
Women with high serum CTX (T score >2) |
2.1 (1.2-3.8) |
1.70 |
25 |
Women with high urine DPD (T score >2) |
1.8 (1.0-3.4) |
1.68 |
24 |
Women with low BMD + high CTX |
3.8 (1.9-7.3) |
3.70 |
54 |
Women with low BMD + high free DPD |
2.1 (0.7-6.2) |
3.04 |
45 |
*Adapted from Garnero P, Hausherr E, Chapuy MC. Markers of bone resorption predict hip fracture in elderly women, the EPIDOS Prospective Study. J Bone Miner Res 1996 Oct; 11(10): 1531-8.[18] |
The following recommendations for preventive therapy have been proposed based on the above findings:
Women with osteoporosis (T score < -2.5) should be treated
Women with normal bone density (T score >-1) should not be treated
Consider treating women with osteopenia (-1> T score >-2.5) only if their bone turnover is above normal limits for premenopausal women
Finally, a global diagnostic approach is desirable because the pathogenesis of fragility fractures is multifactorial, including not only the level of BMD but also bone architecture and bone matrix quality, bone turnover, fall-related factors, and muscle function. The combination of diagnostic tests should be validated in prospective studies. In addition, the value of bone turnover markers to predict fracture risk should be explored in large and long-term studies in other populations, such as men, other ethnic groups, and persons who take corticosteroids.[19]
The following guidelines apply to the use of bone markers in prediction of fragility fractures:
High levels of bone resorption markers (above the premenopausal range, that is, the mean +2 SD, T score >2) are associated with an approximately 2-fold increased risk of osteoporotic fractures.
Resorption markers can be used in the assessment of fracture risk in selected patients in whom BMD and clinical risk factors are not sufficient to make a treatment decision.
In persons with osteoporosis, a very high level of the bone turnover marker (T score >3) is suggestive of other metabolic bone disease, including malignancy.
Normal values are reference values established in healthy premenopausal women aged 30-45 years.
The goal of treatment is to reduce the occurrence of fragility fractures, but their incidence is low, and the absence of events during the first years of therapy does not necessarily imply that treatment is effective.[20, 21] Bone mineral density (BMD) assessment with dual-energy x-ray absorptiometry (DXA) is a surrogate marker of treatment efficacy that has been widely used in clinical trials. Given the short-term precision error of 1-1.5% of BMD measurement at the spine and hip, the individual change must be greater than 3-5% to be significant.
With potent bisphosphonates such as alendronate and risedronate, repeating BMD measurement 2 years after initiating therapy will show whether a patient is responding to therapy. Patients who are responding to therapy have a significant increase in BMD at least at the lumbar spine, which is the most responsive site. With treatments such as raloxifene or nasal calcitonin that induce much smaller increases in BMD, DXA is not appropriate to monitor therapy.
With any treatment, DXA may not reveal all responders within the first year of therapy. This relatively low signal-to-noise ratio of this technique does not allow rapid (within months) differentiation of responders from nonresponders. Baseline bone turnover markers are weak predictors of the response to therapy with antiresorptive drugs; however, the change in bone markers is of greater value.
Some women continue to lose bone while receiving antiresorptive therapy. This has been estimated to occur in approximately one third of women receiving estrogen and one sixth of those receiving bisphosphonates. Failure to respond may be due to noncompliance, poor intestinal absorption of drug, other factors contributing to bone loss, or other unidentified factors. Monitoring treatment of osteoporosis with bone markers may have the added advantage of improving compliance.
Most of the effective antiresorptive treatments induce a decrease in bone turnover that reaches a plateau within a few weeks or months, depending on the potency and route of administration of the drug and on the marker. Thus, individual patients can be monitored with bone markers earlier than with DXA. A study indicated that monitoring bisphosphonate therapy with bone marker measurement at baseline and at 3 and 6 months can improve the compliance with therapy by 20% at 1 year. A decrease greater than 50% and 30% in urinary NTX and serum CTX respectively provides evidence of compliance and drug efficacy. (Serum CTX has also gained attention as a possible risk indicator for bisphosphonate-related osteonecrosis of the jaw[22] ; however, its potential value in this capacity remains uncertain.)
In the Multiple Outcomes of Raloxifene Evaluation (MORE) trial, fracture risk with raloxifene therapy correlated better with changes in markers of bone turnover than with improvement in BMD. Further, in the Vertebral Efficacy with Risedronate Therapy (VERT) trial, subjects with a decrease in urine NTX of more than 40% and a decrease of urine CTX of more than 60% showed a greater reduction in fracture risk.
For a 90% specificity to predict a positive BMD response (+3%), cut-offs, expressed as a percentage decrease from baseline, are as follows:
Urine NTX and urine CTX, –45% to –65%
Serum CTX, –35% to –55%
Total or free urinary DPD, –20% to –30%
Osteocalcin and bone alkaline phosphatase, –20% to –40%
On average a change of greater than 30% is significant. In case of an equivocal change in bone markers, a third measurement should be taken 3 months later.
Recommendations for the use of bone markers in monitoring antiresorptive therapy in postmenopausal women with osteoporosis are summarized below.[23, 24]
Types of marker include the following:
Bone resorption - Urine NTX or serum CTX or urine CTX
Bone formation - Bone-specific alkaline phosphatase or osteocalcin (Use 1 marker or 1 resorption and 1 formation marker.)
Timing of sample includes the following:
Serum - Morning (before 9 am) after an overnight fast
Urine - Either first or second morning void, with creatinine correction, after an overnight fast
Intervals of measurement include the following:
Resorption markers - Before starting treatment and 3 or 6 months after treatment has been initiated
Formation markers - Before starting treatment and 6 months after treatment has been initiated