Perspective: Assessing the Clinical Utility of Serum CTX in Postmenopausal Osteoporosis and Its Use in Predicting Risk of Osteonecrosis of the Jaw

Abstract

Bone turnover markers (BTMs) have become increasingly important in the management of postmenopausal osteoporosis (PMO). In bisphosphonate-treated women with PMO, BTMs can provide early indications of treatment efficacy, are predictors of BMD response and fracture risk reduction, and are potentially useful for monitoring patient compliance. The bone resorption marker serum C-telopeptide cross-link of type 1 collagen (sCTX) has shown high sensitivity and specificity for the detection of increased bone resorption. Recently, sCTX has been singled out as a potential indicator of risk of osteonecrosis of the jaw (ONJ) in patients receiving oral bisphosphonates who require oral surgery. However, whether BTMs are capable of predicting ONJ risk and whether sCTX is usable for this purpose are controversial questions. This article presents an overview of the current literature regarding critical issues affecting the clinical utility of BTMs (including variability and reference ranges) and the current applications of BTMs in PMO management, with a focus on sCTX. Last, the appropriateness of using sCTX to predict ONJ risk in women receiving oral bisphosphonates for PMO is evaluated.

INTRODUCTION

Bone remodeling comprises two coupled processes: bone resorption and bone formation. During bone resorption, osteoclasts secrete hydrochloric acid to dissolve bone mineral and proteolytic enzymes to dissolve bone matrix proteins.(¹) During bone formation, osteoblasts produce and secrete osteoid matrix and mineralize it to form new bone.(²) Biochemical bone turnover markers (BTMs) are released during bone remodeling and can provide a measure of the rate of bone metabolism. They comprise enzymes/proteins secreted by osteoblasts and osteoclasts during remodeling, degradation products formed during resorption, and precursors released during new bone formation.(³) Table 1 lists BTMs commonly used in clinical practice.(^4–13)

Table Table 1.. Description of Bone Turnover Markers Currently Used in Clinical Practice

BTM measurements cannot independently establish a specific bone disease diagnosis. They do reflect metabolic abnormalities such as accelerated bone turnover(¹⁴) and thus allow dynamic assessment of bone remodeling(¹⁵) and monitoring of treatment response in metabolic bone diseases. BTM measurements, unlike many other techniques, are repeatable, inexpensive, and noninvasive.(^3,15)

Postmenopausal osteoporosis (PMO) occurs when menopausal decreases in estrogen production result in increased bone resorption relative to bone formation, leading to a net overall loss of bone mass.(²) This process begins ∼2 yr before menopause and continues for 3–4 yr after menopause.(^16,17) PMO causes deterioration of bone microarchitecture and increased risk of vertebral and nonvertebral fractures. Although BMD measured by DXA is the standard diagnostic test for PMO, BTMs play a growing role in PMO management.

The bone resorption marker serum C-telopeptide cross-link of type 1 collagen (sCTX) is a highly sensitive indicator of increased bone resorption in menopausal women and is useful in follow-up of PMO patients.(^18–21) sCTX decreases dramatically within weeks or months after initiation of bisphosphonate (BP) treatment.(^22,23)

Recently, a new role for sCTX has been postulated-that of predicting the risk of osteonecrosis of the jaw (ONJ) in BP-treated PMO patients who require dental surgery.(^24–27) Whether BTMs can predict ONJ risk and whether sCTX is usable for this purpose are controversial questions.(²⁸) This article will examine the current literature on sCTX use in PMO management to evaluate its appropriateness as a predictor of ONJ risk in women receiving oral BPs for PMO.

LITERATURE ANALYSIS

This analysis included English language publications from 1990 to 2008 listed in Medline or ISI Web of Science. Key search terms included biochemical markers of bone turnover, bone turnover, serum C-telopeptide cross-link of type 1 collagen, sCTX, bisphosphonates, the individual BPs (alendronate, risedronate, ibandronate, zoledronate), postmenopausal osteoporosis, and osteonecrosis of the jaw (ONJ).

CLINICAL UTILITY OF BONE TURNOVER MARKERS

Identification of BTMs and assays suitable for clinical evaluation of PMO has proven challenging. Bone remodeling changes in PMO are more subtle and less rapid than those observed in other metabolic bone diseases.(^7,29) BTMs used in PMO management must be highly sensitive and specific. Clinical BTM assays must be reliable and reproducible, with well-established reference ranges.(³⁰) Available BTM assays include radioimmunoassay, enzyme immunoassay, and high-performance liquid chromatography.(⁷)

sCTX is released during the osteoclast-mediated degradation of type I collagen, which comprises >90% of the organic matrix of bone (Fig. 1).(^5,9,31,32) sCTX is derived from α1-chain C-telopeptide of type 1 collagen that has undergone aging-associated peptide chain rearrangement (β-isomerization).(^5,33–35) Current sCTX assays are ELISAs using antibodies that recognize the amino acid sequence Glu-Lys-Ala-His-Asp-β-Gly-Gly-Arg (also written as EKAHD-β-GGR),(³⁶) referred to as the CrossLaps sequence (Biosciences Diagnostics, Herlev, Denmark), of the α1-chain C-telopeptide.(^18,34,37)

Figure Figure 1. Structure of type I collagen molecules linked by pyridinoline cross-links (pyridinoline [PYD] or deoxypyridinoline [DPD]). The N- and C-telopeptide cross-links of type 1 collagen (NTX and CTX) are released by proteases during osteoclastic bone resorption.(^5,9,32)

Sources of variability:

BTMs show wide biological variability both in healthy subjects and in patients with metabolic bone disease.(^38–41) All sources of variability-biological and analytical, intra- and interassay, intra- and interindividual-must be accounted for to interpret clinical BTM measurements accurately.(⁴²) The total variability (CV_T) in BTM measurements (CV_T² = CV_A² + CV_B²) is the sum of the analytical variability of the assay technique (CV_A) and the natural biological variability of the marker (CV_B).(^7,14,41)

Analytical variability derives from BTM choice, sample processing, and assay selection. Urinary markers must be corrected for urine creatinine content and thus show greater analytical variability than serum markers.(⁴¹) Specimen collection, handling, and storage also affect analytical variability. Some BTMs are sensitive to temperature, UV light exposure, and repeated freeze/thaw cycles(^7,43–45); sample storage temperature generally should not exceed −20°C.(⁷) Handling and storage conditions must be established experimentally for each BTM; sCTX is stable for 3 yr at −20°C, −80°C, and −120°C.(⁴⁶) Standardization and assay automation reduce methodological analytical variability.(⁴) Commercial BTM immunoassays report intra- and interassay coefficients of variability (CV) <10%.(^36,47,48) Automated analyzers improve analytical performance, reduce precision error, and improve intraindividual reproducibility in comparison with manual assays.(^4,13,49) A fully automated sCTX assay(⁴) shows a within-run CV of <4.1%, between-run CV of 5.7%, and long-term intraindividual variability of 9.4%.

The biological (preanalytical) variability of BTM measurements normally exceeds the analytical variability(³⁹) and derives from both controllable and uncontrollable sources.(⁵⁰) Controllable sources include fasting versus nonfasting state,(^51,52) calcium intake, physical exercise, smoking and alcohol use,(⁵³) diurnal and seasonal variations, and phase of the menstrual cycle.(^50,54) Circadian sCTX variation in nonfasting individuals (N = 15) was ±35% in a study by Christgau et al.,(¹⁹) but fasting reduced variation to ±8.8%. To minimize diurnal BTM variability, the preferred collection time for blood and urine samples is after an overnight fast(^52,55) at a consistent time(⁵⁶) in the early morning.(^19,57)

Uncontrollable sources of BTM variability include age, sex, menopausal status, fracture status, disease, immobility, pregnancy, and lactation.(⁵⁸) Some medications, including glucocorticoids, anticonvulsants, hormones, anabolics (e.g., teriparatide), and antiresorptives (e.g., BPs) also influence BTMs.(^50,56) Use of well-defined reference ranges can compensate for many uncontrollable factors.(⁵⁰) Methods for establishing reference ranges (including sample size, baseline characteristics, sample collection methods, etc.) need to be strictly defined and standardized for use in clinical practice.(^56,58) Although published BTM reference ranges are available, laboratory-specific ranges are preferable.(^50,58)

Reference ranges:

No consensus has yet been reached on the most appropriate healthy premenopausal age group to use as a benchmark population for sCTX in women treated for PMO. Studies with different inclusion criteria have reported different reference ranges and advocated different reference age groups (Table 2).(^56,59–61)

Table Table 2.. Reference Ranges for sCTX in Healthy Premenopausal Women

Glover et al.(⁵⁶) enrolled premenopausal women not using oral contraceptives or other bone-active medications and measured their sCTX in a single batch assay. The 95% reference interval was determined by calculating the geometric mean ± 1.96 SD, representing the middle 95% of the population distribution. Women younger than 35 yr had higher BTM levels than women older than 35 yr. Thus, Glover et al. recommended using a premenopausal reference range of 35–45 yr because this age group showed the most BTM stability.(⁵⁶) de Papp et al.(⁶⁰) enrolled healthy premenopausal women, one half of whom used oral contraceptives. The reference range was determined by calculating the geometric mean ± 2 SD, corresponding to the middle 95% of the distribution. Results indicated that oral contraceptive use lowers sCTX measurements. This study's relatively young premenopausal population may also have introduced more BTM variability.(^56,61)

Age-dependent decreases in normal premenopausal BTM levels were reported in the BONTURNO (BONe TUrnover Range of NOrmality) study by Adami et al.,(⁶¹) motivating a recommendation to derive the premenopausal reference range from women 45–50 yr of age. The mid-95% CI was used to determine the reference range, which is lower in BONTURNO than ranges reported among younger women. Standardized populations and methods for determining reference ranges should be established before BTMs can be widely used in clinical practice.

Least significant change:

Clinically meaningful BTM changes in individual patients need to be distinguishable from intra-assay or intrasubject variability.(^14,62) The National Osteoporosis Foundation (NOF) recommends calculating the least significant change (LSC)—the size of BTM difference that represents a clinically meaningful biological change(¹⁴) within an individual over time.(⁶³) The recommended formula for calculating the LSC with a 95% level of confidence is LSC = 1.96 × √2 × √(CVa² +CVi²), where CVa is the analytical variability and CVi is the long-term intraindividual variability.(⁶²) Published LSC and CVi values for sCTX have varied, reflecting interstudy differences in sample sizes, follow-up durations (from 1–2 mo to 4 yr), and assay methods (Table 3).(^4,19,64,65)

Table Table 3.. Published Values of LSC and CVi for sCTX in Women

It has been suggested that excessive long-term intraindividual variability of BTMs may preclude their practical clinical use.(¹⁴) However, serum osteocalcin (OC) and sCTX were found to be stable over a 4-yr follow-up period compared with baseline values in a population of untreated postmenopausal women.(⁶⁵) These results are consistent with low biological variability observed over 4–6 mo with other BTMs.(⁶⁶)

If the LSC and CVi are to be used to identify clinically meaningful and predictive sCTX changes, greater consistency across methods and populations must be attained. The literature contains numerous examples of precision testing and use of the LSC in clinical medicine, especially osteoporosis.(^62,67–70)

Commercially available laboratories:

Inconsistencies among commercial clinical laboratories hinder BTM applications in routine practice. Two studies have reported sufficient differences in BTM results of different commercial laboratories to render their analyses noncomparable.(^30,71) Among European laboratories compared in 2001,(³⁰) BTM results for identical samples differed up to 7.3-fold, primarily because of use of different assay methods. Within-run and longitudinal reproducibility of identical samples differed greatly among six U.S. commercial laboratories; even laboratories using the same commercial assay obtained conflicting results.(⁷¹) Proficiency testing programs and interlaboratory standardization, as well as laboratory-specific reference ranges and CVs, are needed to reduce interlaboratory variation.

BTMs IN THE MANAGEMENT OF PMO

Applications of BTMs in PMO management include identifying patients with high bone turnover at risk for future bone loss and future fracture,(^62,72–74) monitoring and predicting BMD response to antiresorptive therapy,(^75–79) correlating BTM changes to fracture risk reduction,(^80,81) and monitoring patient compliance with therapy.(^31,62,82) The specificity, sensitivity, and pronounced response to therapy of sCTX give it great clinical utility in PMO.(³⁴)

High BTM levels are associated with accelerated postmenopausal bone loss at the forearm, total hip, and femoral neck.(^83–86) sCTX helps identify postmenopausal women at risk for bone loss who may benefit from treatment, although only fragility fractures and BMD are diagnostic for PMO.(⁶⁴) BTMs are more reliable risk markers at the population level than the individual level.(⁸⁷) Elevated BTM levels correlate with risk of future fracture independently of BMD, and the combination of BTMs with BMD and clinical risk factors is a better predictor of fracture risk than any of these parameters alone.(^73,88–90)

BTMs and BP therapy:

Nitrogen-containing BPs, including alendronate, risedronate, ibandronate, and zoledronic acid, effectively reduce the risk of fractures in women with PMO.(^91–94) Significantly reduced BTMs manifest BP effects on bone turnover.(^91,92,94) sCTX responds more dramatically to BP therapy than other BTMs and is considered the most useful BTM for assessing BP efficacy.(^19,34) Relative sCTX reductions of 50–70% after 3 mo of treatment (precision error of 10–20%(³²)) have been reported for several BPs.(^23,95–97)

Reduced fracture incidence is the primary measure of BP efficacy(⁹⁸); however, BTMs are accepted surrogate efficacy markers when used in conjunction with BMD measurements.(⁹⁹) One to 2 yr after BP initiation are needed to detect BMD changes exceeding the LSC (∼3% for a precision error of 1%),(^62,100,101) whereas clinically meaningful reductions in sCTX occur within weeks(²²) or months.(¹⁰²) Thus, insufficient BTM reductions after a suitable period of initial BP therapy may indicate lack of drug absorption or lack of patient compliance.(¹⁰³)

Prediction of BMD response and fracture risk reduction with BP therapy:

Early reductions in BTMs with BP therapy may also predict long-term changes in BMD(^{75–79,104–106}) and fracture risk reduction.(^80,81) These relationships have been explored in groups of patients but, as for fracture risk with baseline pretreatment BTM levels, may not necessarily be applicable to individual patients.

Relationships between BTM changes early in therapy and later changes in BMD and fracture risk have been explored for individual BPs. BTM decreases after 3, 6, and 12 mo of alendronate therapy correlate with long-term (2–4 yr) percentage BMD changes from baseline in the lumbar spine, hip, and total body.(^77–79) With risedronate, uNTX and uCTX reductions at 3–6 mo were significantly associated with 1- and 3-yr reductions in vertebral fracture risk (p < 0.05 for each).(⁸¹) Fracture risk reduction benefits in this study reached a plateau estimated at ∼55–60% uCTX reduction.(⁸¹) Short-term decreases in BTMs (within 3–6 mo of therapy initiation) have also been associated with reduced fracture risk in groups of women receiving antiresorptive treatments for 1–4 yr.(^80,81) The optimal sCTX reduction threshold to obtain fracture risk reduction benefits and the exact relationship (linear versus nonlinear) between changes in BTMs and fracture risk remain subjects of debate.(^107,108)

Monitoring patient compliance with BTMs:

Despite the proven efficacy of BPs to reduce fracture risk and increase BMD, maintaining long-term patient compliance and persistence with BPs has been challenging.(^109–111) It has been suggested that BTMs, the earliest indicators of BP therapy response, may be potentially useful for monitoring compliance and persistence.(^{103,112–114}) However, results of studies examining the usefulness of BTM monitoring on patient persistence have varied. Patients informed of a clinically favorable decrease in BTM levels showed greater adherence or persistence than patients not receiving monitoring in a study of raloxifene.(¹¹²) However, improvements were not found to be dependent on the type of monitoring (either nurse monitoring or BTM measurement monitoring).(¹¹²) The Improving Measurements of Persistence on Actonel Treatment (IMPACT) study evaluated the effects of reinforcement, including BTM follow-up and patient education, on persistence in patients treated with daily risedronate for 1 yr.(¹¹⁴) A decrease in uNTX of >30% was considered a “good” response, a change between −30% and 30% was considered stable, and an increase of >30% was considered “poor.” Patients treated with daily risedronate who were informed of good BTM responses showed improved persistence compared with patients who learned that their BTM levels were stable or poor. It was also noted that patients informed of a poor BTM response were more likely to discontinue treatment than patients who did not receive any form of reinforcement.(¹¹⁴)

PREDICTING BP-ASSOCIATED ONJ RISK: A NEW CLINICAL APPLICATION FOR sCTX?

BTMs have proven useful for monitoring patient response to BP therapy and other clinical applications. Recently, it has been suggested that BTMs may also be useful as a potential indicator of ONJ risk in patients receiving oral BPs who require oral surgery.(²⁴) BP-associated ONJ is a rare but serious condition that has recently raised concerns over the safety of BPs.(¹¹⁵) In 2007, both the American Association of Oral and Maxillofacial Surgeons (AAOMS) and the American Society for Bone and Mineral Research (ASBMR) published guidelines to assist clinicians with the identification and treatment of ONJ.(^116,117)

A confirmed case of BP-associated ONJ is defined as an area of exposed bone in the maxillofacial region that does not heal within 8 wk after identification by a health care provider in a patient with current or previous exposure to BP treatment and no history of radiation therapy to the craniofacial area.(^116,117)

Medical conditions considered in the differential diagnosis of BP-associated ONJ include mucositis, periodontal disease, gingivitis, sinusitis, temporomandibular joint disease, osteoradionecrosis, neuralgia-inducing cavitational osteonecrosis, bone tumors, metastases, periapical pathology caused by carious infection, and infectious osteomyelitis.(^117,118)

BP-associated ONJ should be distinguished from specific oral pathology known to occur in the absence of BPs, such as (1) osteonecrosis associated with periapical pathoses, sinus tracts, purulent periodontal pockets, abscesses, and severe periodontitis,(¹¹⁹) and (2) exposed intraoral bone caused by trauma, osteonecrosis associated with herpes zoster infection, odontogenic infections leading to osteomyelitis, HIV-associated necrotizing ulcerative periodontitis, and spontaneous lingual mandibular sequestration with ulceration.(^{118,120–125})

Risk of developing ONJ may be affected by agent, dose, route, and duration of BP administration, concurrent dental or periodontal disease, the presence of a torus palatinus, comorbid conditions (especially breast cancer and multiple myeloma), and concomitant medications such as glucocorticoids and cancer chemotherapeutics.(^{24,116,117,126–128}) In some BP recipients, the development of ONJ has been associated with dental procedures that manipulate bone or periosteum, as well as with poorly fitting dentures and intraoral trauma.(^129–131)

BP-associated ONJ was first observed in patients treated for cancer, who commonly receive higher and more frequent doses (every 3–4 wk)(¹³²) of potent intravenous BPs than do patients treated for osteoporosis.(^133–135) A few ONJ cases have since been reported in patients receiving oral BPs, primarily alendronate, for treatment of osteoporosis(^24,130,133) and, less frequently, patients treated with oral or intravenous BPs for Paget's disease.(¹³⁶) Patients receiving intravenous pamidronate or intravenous zoledronic acid for cancer-related conditions face higher risk of developing ONJ than patients receiving oral BPs for PMO.(¹³⁷)

The estimated risk of ONJ in patients receiving intravenous BP (pamidronate or zoledronic acid) therapy for oncology indications ranges from 1% to 10%. In contrast, it is 1/10,000–1/100,000 patient-years in patients receiving oral BPs in doses registered for osteoporosis.(¹¹⁷) Confounders such as under-reporting, duplicate reporting, lack of an International Classification of Diseases (ICD-9) code (until the October 2007 introduction of ICD-9-CM code 733.45),(¹³⁸) and varying ONJ definitions have made the true incidence of ONJ difficult to determine.(¹¹⁷) Cases of BP-associated ONJ need to be reviewed and adjudicated according to the ASBMR and AAOMS definition, and the rates of ONJ incidence need to be updated.(^116,117)

In clinical trials using intravenous BPs for nononcology indications, no increased incidence of ONJ has been reported.(^94,139) Furthermore, no cases of ONJ were reported in the large randomized placebo-controlled clinical studies of the oral formulations of alendronate, risedronate, and ibandronate, which included hundreds to thousands of patients receiving oral BP treatment for ≥3 yr (>60,000 patient-years of exposure).(^{91–93,140–142})

No clinical trials to date have studied the relationship between intravenous or oral BP use and the risk of developing ONJ. Instead, information on BP-associated ONJ has been obtained predominantly from published case reports.(^126,137) Additionally, several database studies have attempted to elucidate the association between ONJ risk and the use of intravenous or oral BPs by patients with cancer or osteoporosis. Three database studies have examined the risk of jaw surgery or adverse jaw outcomes (surrogate markers for ONJ) in patients receiving intravenous or oral BPs (Table 4).(^143–145) Studies by Cartsos et al.(¹⁴³) and Zavras and Zhu(¹⁴⁴) reported that treatment with intravenous BPs in cancer and/or osteoporosis patients conferred a greater risk of adverse jaw outcomes. Oral BP use was not significantly associated with increased risk of adverse jaw outcomes in any of these three studies.(^143–145)

Importantly, all of these database studies used surrogate markers of potential ONJ because they antedated the availability of a separate code for drug-induced ONJ (ICD-9-CM 733.45).(^143–145) Therefore, the suspected cases of ONJ were not adjudicated according to the accepted clinical definition of ONJ. Thus, the relationship between oral BPs in PMO regimens and ONJ is less clear than that between intravenous BPs in higher-dose oncology regimens and ONJ.

Table Table 4.. Database Studies That Examined Risk of Jaw Surgery or Adverse Jaw Outcomes in Patients Receiving Intravenous or Oral Bisphosphonates for Treatment of Cancer or Osteoporosis

sCTX as predictor of ONJ risk?

Marx et al.(²⁴) have recently proposed use of the BTM sCTX as a biological marker of increased ONJ risk. These authors monitored the sCTX levels of 30 women (mean age, 64.8 yr) who were current (n = 17) or prior (n = 13) oral BP recipients with bone lesions consistent with meeting the definition of BP-associated ONJ.(²⁴) sCTX assays were performed by Quest Diagnostics (San Juan Capistrano, CA, USA) with an interassay CV of 5.2% and an intra-assay CV of 2.2% at low sCTX levels. However, short- and long-term within-individual CVs and LSC are not presently available (M Caulfield, personal communication, October 3, 2008). In the 17 women taking a BP at the time of ONJ onset, mean fasting sCTX levels were 72.9 pg/ml (range: 30–102 pg/ml). After 6 mo of BP discontinuation, mean fasting sCTX increased to 228.2 pg/ml (range, 162–343 pg/ml). Based on these observations, Marx et al.(²⁴) have suggested that dental surgery should not be undertaken until sCTX is ≥150 pg/ml and that BP therapy be suspended for 4–6 mo, if necessary, to attain this sCTX threshold.

Defining a reference population and calculation of the laboratory least significant change impacts utilization of sCTX for ONJ management:

Since the publication of the recommendations of Marx et al., the use of sCTX for clinical decision making about ONJ risk in patients receiving BP treatment has raised several concerns.(^28,146) Whether a sCTX measurement of 150 pg/ml is a meaningful threshold above which ONJ risk is decreased requires further examination. Practical application of a specific sCTX threshold in individual patients may be hampered by the lack of a standardized postmenopausal sCTX reference range and the potential for interlaboratory variation. The LSC must also be considered in applying BTM threshold values. For an observed change in BTM levels to be considered clinically and statistically significant, it must equal or exceed the LSC. For example, a BP recipient with an sCTX level of 140 pg/ml who needed dental surgery would be advised to discontinue BP therapy for several months, according to the 150 pg/ml threshold defined by Marx et al. Suppose that, after 6 mo, her sCTX increases to 180 pg/ml. The change in sCTX, 40 pg/ml (a change of ∼29%), may or may not be significant, depending on the calculated LSC. Published values for the LSC for sCTX levels have ranged from 27% to 57.4% (Table 3).(^4,19,64) If the results are within the test error range, the clinical interpretation would be that no statistically significant change has occurred within the analytical and biological variability of the test.

It is not unusual for patients diagnosed with osteoporosis in published clinical trials to have sCTX levels <150 pg/ml either before initiation of BP treatment or after 3–6 mo of BP therapy. Mean baseline sCTX levels were reported to be 210 pg/ml in the 1009 patients entered in the placebo-controlled Fracture Intervention Trial (FIT) of alendronate and the subsequent Fracture Intervention Trial Long-term Extension (FLEX).(¹⁴⁷) FLEX enrolled 1009 patients previously enrolled in FIT who had previously received either 5 or 10 mg daily alendronate for a mean of 5 yr.(¹⁴⁷) At the start of FLEX, mean baseline sCTX levels were 110, 100, and 120 pg/ml in the placebo, 5-mg daily, and 10-mg daily groups, respectively. Over a mean of 5 yr of additional alendronate treatment in FLEX, mean sCTX levels remained <150 ng/ml. No cases of ONJ were reported in FLEX; however, reductions in sCTX levels to <150 pg/ml were found to be characteristic of alendronate treatment effect.(¹⁴⁷) Also, sCTX levels that fall within the premenopausal normal range are considered desirable effects of BP therapy.(^148,149) Clinical trials show that continuous BP therapy results in sustained sCTX reductions of 50–70%.(^95,148,150) The lower limits (premenopausal mean − 2 SD) of the sCTX reference ranges reported for healthy premenopausal women are between 40 and 114 pg/ml (Table 2).(^151,152)

Control patients (i.e., patients on BPs but without ONJ) were not examined in the study by Marx et al. Therefore, it is not known how many BP users normally have sCTX levels <150 pg/ml but do not develop ONJ. To determine the percentage of patients whose individual sCTX levels were suppressed below 150 pg/ml while receiving oral BP therapy would require a meta-analysis of individual patient data from the clinical trials of the available oral BPs. Whereas a meta-analysis has not been undertaken for oral BPs, a recent meta-analysis examined ONJ incidence with the use of intravenous zoledronic acid in osteoporosis patients. Data from four clinical studies of zoledronic acid, including the Health Outcomes and Reduced Incidence with Zoledronic Acid Once Yearly (HORIZON) Pivotal Fracture Trial in postmenopausal women, a recurrent fracture prevention study in patients with previous hip fracture, a glucocorticoid-induced osteoporosis study, and a male osteoporosis study, were pooled, and sCTX data from ∼11,000 patients were examined.(¹⁵³) Only two cases of potential ONJ (one in a placebo patient and one in a zoledronic acid-treated patient) were reported and have been previously described.(⁹⁴) The sCTX levels of the placebo patient diagnosed with ONJ were >260 pg/ml. Of the 7714 patients enrolled in the safety population of HORIZON, the percentages of patients with sCTX levels <100 pg/ml were 61%, 31%, 20%, and 16% at 6, 12, 24, and 36 mo, respectively.(¹⁵³) Thus, this analysis reported no association between low sCTX levels (<100 pg/ml) and ONJ.

It is also expected that the sCTX levels of the patients included in the study by Marx et al. will increase after a “drug holiday.” Patients who discontinue oral BP therapy may experience increases in BTMs.(^{147,152,154,155}) BTM levels in patients who received 12 mo of ibandronate therapy returned to baseline values after 12 mo of discontinuation.(¹⁵⁴) Patients discontinuing risedronate therapy after 3 yr of treatment had median BTM levels not significantly different from those of placebo patients 12 mo after discontinuation of treatment.(¹⁵⁶) In women who discontinued alendronate therapy after a mean of 5 yr of treatment as part of FLEX, sCTX levels gradually increased to within −7% of pretreatment levels after 5 yr.(¹⁴⁷) Absolute sCTX levels (geometric mean) rose from 110 to 200 pg/ml after 5 yr; this was similar to FIT baseline levels.(¹⁵⁷) Therefore, the specific time needed after BP discontinuation for sCTX levels to increase above the suggested 150 pg/ml threshold is not clear. Available literature includes only the mean or median levels, and great variability in results may have occurred because not all samples were collected as fasting early morning specimens.(^131–134)

The effect of a drug holiday on fracture risk is an issue of concern.(²⁸) Discontinuation of alendronate therapy for 5 yr, after 5 yr of continuous therapy, did not significantly increase the risk of nonvertebral fractures (RR = 1.00; 95% CI: 0.76–1.32).(¹⁴⁷) However, a subgroup analysis showed that, after 5 yr, patients who discontinued alendronate (FLEX placebo group) experienced increased risk of incident clinical vertebral fractures compared with patients continuing alendronate (RR = 0.45; 95% CI: 0.24–0.85). Although not statistically significant at the 95% CI, incident clinical vertebral fractures were higher in the FLEX placebo group with osteoporosis at baseline or prevalent fracture (RR = 0.57; 95% CI: 0.23–1.40 and RR = 0.47; 95% CI: 0.19–1.1). Also, incident nonvertebral fractures were higher in the placebo group with osteoporosis at baseline (RR = 0.77; 95% CI: 0.50–1.2). A database study by Curtis et al.(¹⁵⁸) also suggested that women who discontinued BP therapy after being compliant (medication possession ratio of 66–100%) for 2 yr had a significantly higher hip fracture incidence after discontinuation of therapy compared with women remaining on therapy (8.43 versus 4.67 per 1000 woman-years; p = 0.016). Thus, for some women, BP discontinuation for long periods (>1 yr) may not be advisable.(¹⁵⁸)

The ability of sCTX to act as a predictive marker of ONJ risk has not been explored in a large clinical trial, and no currently available clinical data support the use of a sCTX threshold as a guide to minimize the risk of BP-associated ONJ in patients receiving oral BP treatment. Additional important scientific questions that need clarification are whether sCTX levels are (1) influenced by local site-specific bone turnover in the jaw caused by ONJ disease activity independent of the effect of BPs and (2) predictive and correlative with ONJ healing. Significant dental literature exists that substantiates the presence of elevated bone resorption markers, including C-telopeptide pyridinoline cross-links, in the gingival crevicular fluid (GCF) of patients with periodontal tissue destruction.(^159–163) Despite these observations, there is insufficient data in the ONJ literature that correlate locally derived BTMs, such as in GCF, with systemic levels. However, in cases of monostotic involvement of bone associated with Paget's disease and fibrous dysplasia, use of very discriminatory systemic biochemical markers of bone remodeling may predict local bone disease and response to therapeutic intervention.(^164–171) Inconsistencies in the suggested correlation between sCTX and prediction of ONJ healing are evident. Not all patients with BP-associated ONJ heal after discontinuation of BPs, and spontaneous cases of non-BP-associated ONJ occur despite, in both cases, having sCTX levels above the 150 pg/ml threshold suggested by Marx et al.(^24,94) Healing of BP-associated ONJ has also been documented to occur without interruption of monthly intravenous BPs administered to cancer patients.(¹²⁷) An interesting hypothesis to explain the lack of overall correlation between the predictive value of sCTX and ONJ healing in selective patients may be related to increased local remodeling of bone associated with extensive underlying ONJ after discontinuation of BP therapy.

The study by Marx et al. raises important questions regarding the potential of sCTX to predict ONJ risk in BP patients. However, current evidence does not support routine clinical use of sCTX for this purpose, and several critical issues should be addressed. It is imperative to account for both analytical and biological variability to achieve meaningful interpretation of BTM measurements. To accomplish this, it is necessary to establish standardized laboratory protocols including assay methods and sample collection methods, determine intra- and interassay CVs and intraindividual variability, calculate the LSC, and establish well-defined reference ranges using standardized criteria. Variability across commercial laboratories also requires consideration. Finally, the validity of sCTX as a clinical predictor of ONJ must be explored in future studies. Because of the relatively low incidence of ONJ among patients taking oral BPs, a clinical trial would be difficult to conduct. However, the definition of ONJ provided by AAOMS and ASBMR and the existence of the ICD-9-CM code 733.45 may facilitate a retrospective database analysis of the relationship between sCTX levels and ONJ incidence.

The predictive capacity of sCTX for ONJ should be explored by receiver operating curve analysis, and the sensitivity, specificity, positive predictive value, and negative predictive value should be identified. These values have been determined for sCTX as a predictor of BMD response to BP therapy(^19,77) but have not been determined for sCTX as a predictor of ONJ.

CONCLUSIONS

BTMs have proven useful for several applications related to PMO. They are independent predictors of future bone loss and fracture risk in untreated patients and have shown utility in predicting BMD response, predicting fracture risk reduction, and monitoring treatment efficacy in treated patients. sCTX is a specific and sensitive marker of bone resorption that can rapidly indicate a patient's response to BP therapy. To attain clinically meaningful information, variability in BTM measurements must be carefully controlled. Thus, issues of variability, lack of a standardized sCTX reference range, and interlaboratory inconsistency may limit the appropriate clinical applications of sCTX. As a predictor of BP-associated ONJ, sCTX has yet to receive rigorous evidence-based validation. Thus, insufficient data currently exist to support using sCTX levels to guide modification of BP therapy to minimize the risk of developing ONJ. Defining the precise reference range and LSC of commercial laboratories for sCTX is a prime area for future research.

Acknowledgements

The authors acknowledge the editorial assistance of Rebecca Jarvis, PhD, from Envision Pharma, in the preparation of this manuscript. Editorial assistance was funded by an independent grant from Roche. The funding source did not review or comment on the manuscript.