Markers of Bone Turnover for the Prediction of Fracture Risk and Monitoring of Osteoporosis Treatment: A Need for International Reference Standards

The International Osteoporosis Foundation (IOF) and the International Federation of Clinical Chemistry and Laboratory Medicine (IFCC) formed a committee of experts to examine the efficacy of bone turnover markers (BTMs) on fracture risk prediction, to determine the magnitude of change of BTMs following treatment to apportion their predictive use in fracture risk reduction and to formulate a research strategy if there was insufficient information to make clear recommendations on their use in clinical practice.¹

The discussion document summarised the evidence that osteoporosis is a costly and major health problem worldwide. A number of agents are now available for the treatment of osteoporosis. Current strategies to identify individuals at highest fracture risk and monitor therapeutic response focus on the use of bone mineral density (BMD) measurement. Unfortunately, BMD has several limitations including the fact that not all individuals with low BMD values will sustain osteoporotic fracture and the change in BMD with most therapies underestimates the degree of fracture reduction. BTMs have been studied in a number of clinical trials and may offer clinical potential to assess fracture risk independent of BMD and to monitor therapeutic response independent of the change in BMD.

BTMs are easy to collect in samples of blood or urine, are non-invasive and provide data that could be complementary to BMD. However, limitations of BTMs include the inherent biological variability and in selected cases, multiple methods of analysis are available. Biological variability was separately considered as constituting both uncontrollable variables such as age, gender, menopausal status, disease and controllable sources of variation such as time of day of collection, fasting status, exercise, time of menstrual cycle and season. The importance of each biological and physical variable was summarised and the magnitude and direction of the effect of each biological variable on each BTM were stated.

Recent national guidelines for a number of countries, including Australia, were summarised to include the role of BTMs in the diagnosis and monitoring of osteoporosis treatments. The differences in each separate recommendation reflect the probable differing opinions of each organisation and their members regarding the utility of BTMs in the diagnosis and management of osteoporosis.

The committee conducted a comprehensive search of the english published literature using a previous systematic review of data up to 2001, MEDLINE pre-2000 to verify the former review and PubMed databases between 2000 and 2010. To prevent exclusion of any relevant studies in the review, key recent review articles were separately identified and all additional references from these sources were included. Only prospective cohort studies using the assessment of BTMs prior to fracture were included in the assessment of fracture risk. Cross-sectional, case-controlled, animal and preclinical studies were excluded, as were prospective studies that did not provide separate gender or hip fracture data. Only studies providing individual BTM and fracture outcome data were included in studies examining the performance of BTMs in therapeutic monitoring. Studies using mean changes and not individual changes were consequently excluded.

Each BTM currently available was then summarily described according to whether the marker measured bone resorption or bone formation. The availability of manual or automated methods of analysis, specificity of each marker and sources of variability were summarised in a helpful table. The plausibility of BTM use and how markers could be predictive of skeletal fragility was discussed including the role of remodelling rate in helping to explain the magnitude of ‘stress risers’, ‘plate perforation’ rates and the degree of bone mineralisation. All of these concepts help explain at the micro-architectural level why modest changes in BMD may poorly predict fracture event rates. In contrast, BTMs may reflect micro-architectural changes by explaining reduction in trabecular plate perforations or stress reactions at sites of low remodelling rates and so provide a rationale explanation why BTMs could be independent predictors of fracture risk.²

All but four studies in postmenopausal women showed that one or more BTMs were significantly associated with fracture risk and all studies were conveniently summarised in a comprehensive table. BTMs in women may predict fracture risk independent of BMD and the predictive value may be additive in women with low BMD. Three of only four studies in men also demonstrated that one or more BTMs were associated with fracture risk. However, a number of limitations were noted including the large number of different BTMs used in some studies with the number of observations not being handled statistically appropriately so that false positive associations could occur, the heterogeneity of fracture outcomes (e.g. hip, spine or all fractures) and the different approach to analysis and interpretation of the data. The inconsistency of outcome even for the same BTMs in different studies was noted as well as analytical limitations. Consequently, BTMs independently of BMD predict fracture outcome in some but not all studies.

Expression of risk was critically evaluated noting inconsistency in reported studies. Consequently, lowest and highest tertile, quartile or quintile of BTM may be used in some studies to compare risk. Alternatively, BTM values above or below arbitrary thresholds were examined in some studies. Some studies examined fracture risk as determined by the standard deviation increase in BTM value. The committee also determined that it was preferable that the risk of fracture in the cohort should be reported and be preferably comparable to population risk in order to allow an easier comparison between studies.

Future developments in fracture risk assessment were also discussed with the potential inclusion of BTMs in the fracture risk assessment tool ‘FRAX’ which currently uses age, sex, height, weight, prior fragility fracture and clinical risk factors to help determine future fracture risk. BTMs are not currently included in the FRAX algorithm as there is a scarcity of population-based prospective data with any single analyte.

The second discussion section was devoted to the use of BTMs in monitoring osteoporosis treatment in early drug development as well as an adjunct to routine patient care. The potential use of BTMs to help determine dose, frequency of medication as well as provide an early confirmation of therapeutic effect was stated. The change in BTM according to whether an anabolic agent or an anti-resorptive drug is used as well as the pattern of response was discussed. Route of drug administration has been noted to be important with more rapid reduction in BTM with intravenous compared with oral bisphosphonate therapy. The quantum of effect of each of the currently licensed drugs for use in the treatment of osteoporosis on BTM was conveniently summarised with, for example, the reduction in urine creatinine-adjusted amino-terminal crosslinking telopeptide of type I collagen (NTX) being 60% following denosumab subcutaneous treatment, 50% following oral alendronate or conjugated equine oestrogens, 40% following oral risedronate, 25% following raloxifene and 10% following strontium ranelate treatment.’

Limitation of data include the fact that not all BTMs respond to the same degree to the same antiresorptive agent and clearly different osteoporosis treatments result in different effects on BTMs. The clinical utility of repeat measurement was also discussed with the need to appreciate biological as well as analytical variation and to control for modifiable variables. Intra-individual variability data for each BTM, as expressed in coefficients of variation, was also provided in a table and least significant change (LSC) for each analyte was explained. Valid means of improving the LSC were explained.

Several studies show that the larger the reduction in BTMs following anti-resorptive therapy, the greater the fracture reduction with treatment. Only two studies examined the magnitude of reduction of BTM and represented this as a percentage of the total fracture reduction. In both studies the magnitude of fracture reduction was more accurately reflected in the magnitude of reduction of BTMs than the increase in BMD.

Limitations of the data include that only five of numerous clinical trials have been appropriately analysed according to the committee, only a subset of trial participants had BTM samples measured in such trials and only one study was sufficiently powered to examine fracture outcome. BTMs were not always collected appropriately in all studies according to fasting state, second voided urine samples or before fracture events. Data was not always appropriately normalised and it is often assumed that the BTM subset demonstrated the same reduction in fracture risk as the entire treatment population. The panel recommended future studies should address these controllable variables, BTMs should be measured in all trial participants and the data needs to be analysed in an appropriate statistical fashion.

Other clinical uses of BTMs were briefly discussed. It was acknowledged that BTMs may predict rate of bone loss or therapeutic response and lead to improved adherence to treatment. In addition to the low rates of initiation of appropriate treatment after osteoporotic fracture, poor adherence rates to treatment are recognised as being important and inappropriately low.³ The use of BTMs could help in the identification of secondary causes of osteoporosis, with high rates of bone resorption indicating possible co-existent hyperthyroidism, hyperparathyroidism or multiple myeloma and low rates of bone formation indicating possible cortisol excess.

The final discussion section of the manuscript was devoted to the adoption of appropriate reference analytes and reference standards for future studies. Adoption of reference analytes would allow comparison between studies, help quantify the effects of new therapeutic agents and allow the use of such data in fracture prediction tools such as FRAX. The panel concluded that the reference BTMs should be adequately characterised and defined, need to be bone-specific and useful in fracture prediction as well as treatment monitoring, be widely available, be suitable based on biological and analytical variability, be measurable by automated methods and ideally be present in blood as urine analytes have higher intra-individual variation.

Serum carboxy-terminal crosslinking telopeptide of type I collagen (CTX) was chosen as the bone resorption marker of choice in view that the assay is well characterised, measures an 8 amino acid peptide, has been evaluated in many studies, is widely available as an automated immunoassay or manual ELISA, and the biological and analytical variability is well documented. Serum procollagen type I N propeptide (PINP) was chosen as the reference standard for bone formation, although the peptide antibody target is not well characterised, the immunoassay may recognise the trimeric intact or monomeric forms and the formation of PINP is not entirely specific to bone. However, most PINP probably is produced during bone formation, the assay is widely available as an automated immunoassay or manual radioimmunoassay and the biological and analytical variability is well known. The use of internationally agreed decision limits and target values also necessitate standardisation and the adoption of a reference system with more immediate harmonisation as a short term interim solution proposed.

Discussion

The expert committee’s approach to defining specific BTMs as reference analytes, recommending their universal use in future clinical studies and the adoption of internationally agreed decision limits will help address most of the limitations identified in the use of BTMs currently. Whether BTMs will prove to be useful in the prediction of future fracture events or in the identification of non-responders to anti-osteoporosis treatments awaits further research.

This review is an extremely useful resource summarising the state of art application on the use of BTMs in the diagnosis and treatment of osteoporosis. The initial discussion focuses succinctly but comprehensively on the biological and analytical limitations of the current use of BTMs with the summary of recommendations from differing countries and separate organisations reflecting the lack of consensus in this area.

Subsequent tables summarising the available data regarding the use of BTMs at predicting fracture outcome are helpful and focus the subsequent discussion of study outcomes, highlight the limitations of each study and provide some ideas regarding how to improve future studies to help provide meaningful data.

The adoption of each of the proposed bone formation and bone resorption markers and harmonisation of laboratory reporting is achievable in the short term. The adoption of specific markers by the expert committee may also assist laboratories in deciding which BTM to choose for their local market. However, other issues such as the use of BTMs in the early identification of secondary causes of osteoporosis or the monitoring of other metabolic bone diseases such as Paget’s disease need to be also considered when choosing which BTM to report. The adoption of specific reference markers may accelerate manufacturers’ approaches to harmonisation of results and standardisation of assays, an outcome that will benefit all users of such methods.

The position statement is an extremely useful resource for practicing laboratorians who offer and report BTM results and clinicians who continue to grapple with who to treat and how to determine if a prescribed treatment is likely to be effective in an individual patient at preventing future fracture events.