the relationship between regional bone turnover measured using 18f-fluoride positron emission...

Journal of Clinical Densitometry, vol. 10, no. 1, 46e54, 2007� Copyright 2007 by The International Society for Clinical Densitometry1094-6950/07/10:46e54/$32.00DOI: 10.1016/j.jocd.2006.10.006

Original Article

The Relationship Between Regional Bone Turnover MeasuredUsing 18F-fluoride Positron Emission Tomography

and Changes in BMD Is Equivalent to That Seen for BiochemicalMarkers of Bone Turnover

Michelle L. Frost,*,1 Gary J. R. Cook,2 Glen M. Blake,1

Paul K. Marsden,3 and Ignac Fogelman1

1Osteoporosis Screening & Research Unit, King’s College London School of Medicine, Guy’s Hospital, London;2Department of Nuclear Medicine, Royal Marsden Hospital, Sutton, Surrey; and 3Clinical PET Centre, King’s College

London School of Medicine, St Thomas’ Hospital, London, UK

Abstract

Bone turnover is an important determinant of fracture risk. 18F-fluoride positron emission tomography (18F-PET) al-lows the direct assessment of bone turnover at the clinically important skeletal sites such as the lumbar spine. The aim ofthis study was to determine if the relationship between regional bone turnover measured using 18F-PET and changes inbone mineral density (BMD) is equivalent to that seen for global skeletal measurements of biochemical markers of boneturnover. Forty-three women who had previously had an 18F-PET scan at the lumbar spine, assessment of biochemicalmarkers of bone turnover, and a dual-energy X-ray absorptiometry scan of BMD at the lumbar spine and hip (baselineassessments) were split into 1 of 2 groups: (1) 22 women who commenced treatment for osteoporosis within 2 mo ofhaving the baseline assessments (Treatment group); (2) 21 women who had not taken any treatments for osteoporosissince having the baseline assessments (Untreated group). Sixteen of the women in the Treatment group started risedronatetherapy as part of a prospective study they were participating in, whereas the decision to treat the remaining 6 women wasmade by the subject’s treating physician. Subjects had between 2 and 5 BMD scans over a median follow-up time of 4.1 yrto estimate the annual percentage change in BMD since baseline. The relationship between the tertiles of 18F-PET skeletalkinetic parameter Ki, reflecting regional bone turnover, and annual changes in lumbar spine and hip BMD were comparedto that seen for bone formation (bone-specific alkaline phosphatase, BSALP) and bone resorption (urinary deoxypyridi-noline) markers. Treated women in the highest tertile of both regional (18F-PET) and global (biochemical markers) boneturnover showed the greatest annual percentage increases in lumbar spine BMD. The annual increase in lumbar spineBMD was 1.8%, 2.2%, and 3.2% for women in the lowest, middle, and highest tertile of BSALP, respectively, whichwas similar to that obtained for the regional measurement of Ki of 1.7%, 2.2%, and 2.7% respectively. Untreated womenin the highest tertile of regional and global bone turnover had larger decreases in lumbar spine BMD compared to thosewomen in the lowest tertile, with a 1.4- to 4.8-fold difference in the annual decrease in BMD between the two. Less con-sistent patterns were observed when assessing the relationship between regional and global bone turnover with changes inhip BMD. This study has demonstrated that the relationship between regional bone turnover measured directly at thelumbar spine with changes in BMD is similar to that seen for global skeletal bone turnover using biochemical markers.

Key Words: Biochemical markers; bone mineral density; bone turnover; osteoporosis.

Received 08/04/06; Revised 10/26/06; Accepted 10/26/06.*Address correspondence to Michelle L. Frost, PhD, Osteoporo-

sis Screening & Research Unit, Guy’s Hospital, St Thomas Street,

46

London SE1 9RT, United Kingdom. E-mail: [email protected]

mailto:[email protected]

mailto:[email protected]

Spine Bone Turnover and Changes in Bone Density 47

Introduction

In recent years, it has become clear that the rate of boneremodeling is an important determinant of fracture risk (1).Studies show that bone remodeling is an independent predic-tor of fracture risk (2e4) and changes in remodeling rates cor-relate with fracture risk reductions (5e8). Early changes inbiochemical markers of bone turnover in response to antire-sorptive therapy are associated with long-term changes inbone mineral density (BMD) (9e12). It has also been demon-strated that the magnitude of the decrease in biochemicalmarkers can predict subsequent reductions in fracture risk(2,6,13e15). Treatment-induced changes in markers of boneturnover occur within weeks of commencing therapy, whereaschanges in BMD cannot be readily detected until 1 or 2 yrlater (2,6,13,15). These studies indicate that there is a needto investigate more fully the role of bone turnover in the clin-ical management of osteoporosis.

Currently, the most convenient method for measuring boneremodeling is by using biochemical markers of bone turnover(16,17). Although the most practical choice for measuring therate of bone remodeling, particularly in response to therapy,bone markers reflect global skeletal function and cannot pro-vide information about specific sites in the skeleton. The dy-namic quantification of bone remodeling by tetracyclinelabeling and bone biopsy at the iliac crest is considered thegold standard method but its use is limited due to it beingan invasive technique that is restricted to just 1 skeletal site,the iliac crest, which may not be representative of other sitesof the skeleton (18).

The noninvasive functional imaging technique of 18F-fluoride positron emission tomography (18F-PET) allows thedirect quantitative assessment of bone turnover at specificsites of the skeleton, including the clinically important sitesof the lumbar spine and hip (19e27). The use of 18F-fluorideas a bone-seeking tracer for scanning was first introduced byBlau et al (28) who showed that the uptake of 18F-fluoride bybone is rapid and occurs by chemiabsorption onto hydroxyap-atite (29). It has been demonstrated that the single-passageextraction of 18F-fluoride by bone is close to 100% (30).The use of this 18F-PET has been validated by comparisonwith bone histomorphometric indices of bone turnover.Highly significant correlations have been observed betweenregional skeletal kinetic parameters using 18F-PET andparameters associated with osteoblastic number and activity,including the bone formation and mineral apposition rate(25,26). From our previous work, differences in bone turnoverbetween the trabecular-rich lumbar vertebrae and the humerus(19) and hip (23) have been demonstrated using this tech-nique. A comparison of postmenopausal women with andwithout osteoporosis showed that women with osteoporosishave relatively reduced regional bone formation at the lumbarspine (22). We have also used this technique to directly assessthe effects of antiresorptive therapy on osteoblastic activity atthe lumbar spine (21).

The value of biochemical markers of bone turnover forpredicting BMD response in both observational and treatment

Journal of Clinical Densitometry

studies has been demonstrated (9e12,31e33). It would be ofinterest to examine the relationship between bone turnover,measured directly at the lumbar spine using the novel tech-nique of 18F-PET, with subsequent changes in BMD mea-sured at the same skeletal site. This is of interest as theskeleton is heterogeneous in it’s response to therapy anda global measurement of bone turnover, using biochemicalmarkers, may attenuate or mask what is going on at metabol-ically active sites such as the lumbar spine [34]. In this study,we compared global (biochemical markers) and regional (18F-PET) measurements of bone turnover with subsequentchanges in BMD (over a median follow-up time of 4.1 yr)at the lumbar spine and hip in both treated and untreated post-menopausal women. The aim of this study was to determine ifthe relationship between regional bone turnover measured us-ing 18F-PET and changes in BMD is equivalent to that seenfor biochemical markers of bone turnover. If this is confirmedthen larger studies can be performed to fully investigate therelationship between regional bone turnover and changes inBMD with aging or treatment.

Materials and Methods

Subjects

Eighty-nine postmenopausal women who had previouslytaken part in 1 of 5 studies (19,21,22,35,36) to investigateregional bone turnover measured using 18F-fluoride PETwere invited to take part in this study and have a dualX-ray absorptiometry (DXA) scan to obtain an estimate ofchange in BMD. This group of 89 subjects consisted ofwomen with normal BMD and those with osteopenia or oste-oporosis. Each woman had had an 18F-PET scan of the lum-bar spine, measurements of biochemical markers of boneturnover, and a DXA scan of BMD at the lumbar spine andhip between 3 and 8 yr previously (1998e2003). Blood sam-ples were also taken at the same time as these initial assess-ments for routine chemistry tests including bone, renal,liver, and thyroid profiles and serum levels of parathyroid hor-mone and 25-hydroxy vitamin D. Results of these were withinnormal limits for all subjects. None of the subjects were tak-ing treatments for osteoporosis at the time of these initial as-sessments. Women were excluded from taking part in thepresent follow-up study if they had diseases or took any med-ication known to influence bone metabolism (excluding thosefor the treatment of osteoporosis) since the baseline assess-ments or had started but then discontinued treatment for oste-oporosis. A total of 43 postmenopausal women were suitableand agreed to take part in the follow-up study. These womenhad between 2 and 5 BMD scans over a median follow-uptime of 4.1 yr. The women were divided into 1 of 2 groups:(1) 22 women who commenced treatment for osteoporosisin the first 2 mo after 18F-PET scan and had remained ontreatment since then (Treatment group); (2) 21 women whohad not taken any treatments for osteoporosis since havingthe 18F-PET scan (Untreated group). Seventeen of the womenin group 1 had taken bisphosphonate therapy (alendronate,

Volume 10, 2007

48 Frost et al.

etidronate, or risedronate), 1 had taken raloxifene, 3 had takenhormone replacement therapy, and 1 had taken both didroneland hormone replacement therapy (HRT), at doses recom-mended for the treatment of postmenopausal osteoporosis.Sixteen of the women in the Treatment group started risedro-nate therapy as part of a prospective study they were partici-pating in (21), whereas the decision to treat the remaining 6women was made by the subject’s treating physician. Infor-mation on treatment compliance and adherence was obtainedby questionnaire and interview at the follow-up visit. Subjectswho showed less than 80% adherence to treatment were ex-cluded from taking part. Nineteen of the women in group 1reported taking calcium and vitamin D supplements at dosesof �500 mg calcium and �400 IU vitamin D. The study wasapproved by the local research ethics committee and eachsubject gave written informed consent.

Bone Density Measurements

To estimate the change in BMD since the initial bone turn-over (18F-PET and biochemical markers) and baseline BMDassessments, each subject had a follow-up bone densityscan. Sixty-three percent of the women who took part inthis study had undergone one or more bone density scanssince the baseline assessments because part of their routineclinical care before taking part in this study. These datawere used in addition to that obtained for the final follow-up scan to estimate annual changes in BMD. Each subjecthad between 2 and 6 (median 5 3) lumbar spine (L1eL4)and hip BMD measurements using a QDR4500 (Hologic,Bedford, MA, USA) during a median follow-up time of4.1 yr (range: 2e7 yr). The baseline BMD scan was per-formed within 2 wk of the 18F-PET scan and biochemicalmarker assessments. Regions of interest (ROI) on follow-upscans were carefully matched to the initial scan. Standardquality control procedures were performed daily, which dem-onstrated the stability of the scanner during the study period.Long-term precision assessed in a previous study carried outin our Unit using the same scanner was 1.1%, 2.2%, and 1.3%


for lumbar spine, femoral neck, and total hip BMD, respec-tively (37). Annual rates of change in bone density were cal-culated by obtaining the slope of the regression using linearregression of all the BMD measurements available for eachsubject against time.

Biochemical Markers

A nonfasting measurement of serum bone-specific alkalinephosphatase (BSALP) (Alkphase-B, Metra Biosystems) wasperformed as a marker of bone formation. Fasting second morn-ing void urinary deoxypyridinoline crosslinks (Pyrilinks-D,Metra Biosystems), corrected for creatinine, was measured asa marker of bone resorption. Blood and urine samples weretaken on the day of the 18F-fluoride PET scan and werecollected at the same time of day for each subject. Sampleswere stored at �70�C and then analyzed as 2 batches.

18F-fluoride Positron Emission Tomography

Each subject had a 18F-fluoride PET scan of the lumbarspine after the injection of either 90 or 180 MBq of 18F-ion.Radiotracer production, protocols for scan acquisition andanalysis, the method used to calculate the input functionand the mathematical model used to derive the parametersof interest have been described in detail previously (21,22).PET scans were performed on a Siemens ECAT 951R scan-ner. Subjects were positioned supine, with the midlumbarspine region within the field of view. The ROIs were placedover the vertebral bodies using an automated method thatused a threshold of 50% of the maximum bone activity foreach image set (Fig. 1A). Only the vertebral body was in-cluded in the ROI and disc spaces were excluded from anal-ysis. Slices directly above or below the intervertebral discspaces were not used to avoid any spillover effects. Foreach subject, 1 to 3 complete vertebrae were available withinthe field of view for analysis (Fig. 1B). The ROIs were thenused to produce time-activity curves for individual vertebrae(Fig. 2). To derive the kinetic parameters of interest, the activ-ity of 18F-fluoride in arterial plasma over time is required

Fig. 1. (A) Transaxial 18F-fluoride positron emission tomography (18F-PET) image of the lumbar vertebra showing the regionof interest around the vertebral body (image zoomed in inset), (B) sagittal and coronal view of 18F-PET scan of lumbar spine.

Volume 10, 2007


(Fig. 2). The plasma input function was derived using anoninvasive validated method, which involves measuring18F- counts in a ROI placed over the aorta seen on one ofthe early dynamic frames (35).

Mathematical Model

To derive the 18F-fluoride kinetic parameters of interestthat reflect regional bone turnover, the 3 compartmental tracerkinetic model described by Hawkins et al was used (24). Thismodel consists of a vascular (whole blood) compartment, anextravascular bone compartment, and a bone mineral com-partment (Fig. 3). The rate constant K1 and macroparameterKi are the 2 most important parameters of interest, reflectingbone blood flow and osteoblastic activity, respectively. K1 de-scribes the unidirectional clearance of fluoride from plasma tothe whole of the bone tissue, which is related to regionalblood flow by the equation:

K1 5 E:Q:ð1� PCVÞmL min�1 mL�1 ð1Þ

0

0.02

0.04

0.06

0.08

0.1

0 600 1200 1800 2400 3000 3600Time (secs)

cts/

ml/s

ec

Fig. 2. A representative study showing the arterial plasmainput function (continuous line) and vertebral region of inter-est time-activity curve (triangles) used to derive the kineticparameters of interest reflecting bone turnover. The vertebralvalues have been multiplied by 5 for easier comparison withthe input function.

Plasma BoneECF

BoneMineralR

BC

K1k3

k2 k4

K1 = E . Q . (1-PCV) mL.min-1.mL-1

Ki = K1xK3/(k2+k3) mL.min-1.mL-1

Fig. 3. The 3-compartmental bone kinetic model andequations proposed by Hawkins et al [24] to analyze dynamic18F-fluoride positron emission tomography studies. SeeMethods for description of rate constants K1 to k4. RBC,red blood cells; ECF, extracellular fluid. K15E:Q:ð1� PCVÞðk2 þ k3Þ.


where E is the unidirectional extraction fraction (usually as-sumed to approach 100%), Q is regional blood flow, andPCV is the packed cell volume. The macroparameter Ki rep-resenting the net clearance of fluoride to the bone mineralcompartment can be calculated by:

Ki 5 K1� k3=ðk2þ k3ÞmL min�1 mL�1 ð2Þ

Ki is a function of both K1 that reflects bone blood flow, andthe fraction of the tracer that undergoes specific binding tothe bone mineral ((k3/(k2þ k3)). Ki reflects the number andactivity of osteoblasts in bone and so is an indicator ofthe rate of bone turnover (25,26). K1 and Ki are expressedas clearance rates per unit volume of bone tissue, that is,mL min�1 mL�1.

The same authors have previously carried out a 6-mo studyto compare the reproducibility of biochemical markers andthe 18F-PET technique. Precision (coefficient of variation)was comparable for both methods of assessing bone turnover,averaging approximately 14% (36).

Statistical Analysis

Descriptive statistics are presented as mean� standarddeviation. To calculate the annual rate of change, it was as-sumed that the expected change in BMD was linear withtime for each subject. Linear regression was performed,with BMD as the dependent variable, for each subject. Theannual rate of change was expressed as a percentage ofeach subject’s baseline BMD. The annual percentage changesin BMD, the 18F-PET parameters K1 and Ki and biochemicalmarkers were tested for normality, kurtosis, and skewness andwere all found to be normally distributed. An unpaired Stu-dent’s t-test was used to compare group means. Each of themeasurements of global (biochemical markers) and regional(18F-PET) bone turnover was split into tertiles for each ofthe 2 groups. There were no statistically significant differ-ences in age, years since menopause or body mass index(BMI) for women in each tertiles of bone turnover so thesevariables were not corrected for in all subsequent analysis.Differences in the annual percentage change in BMD be-tween tertiles of bone turnover were assessed using ANOVA.An unpaired Student’s t-test was used to assess the differencein the annual percentage changes in BMD between the lowestand highest tertiles of bone turnover. The relationship be-tween baseline bone turnover and the annual rate of changein BMD was assessed using Pearson correlation coefficient.Backward stepwise linear regression analysis models wereused to examine the relationship between predictor variablesincluding age or years since menopause, baseline global andregional bone turnover, and the annual changes in BMD. Allstatistical analyses were repeated using absolute changesrather than percentage changes in BMD. Results were consid-ered significant at the 5% level ( p� 0.05). All statistical cal-culations were performed using SPSS version 13 (SPSS Inc,Chicago, IL, USA).

Volume 10, 2007

50 Frost et al.

Results

Study groups characteristics at baseline are shown in Table 1.The untreated group were significantly younger than the treatedgroup and had significantly greater BMD at the lumbar spine,femoral neck, and total hip. There were no significant differ-ences in global (biochemical markers) or regional measure-ments (18F-PET, K1 reflecting bone blood flow, and Ki

reflecting bone turnover) of bone turnover between groups.There were no significant differences in baseline mean serumlevels of parathyroid hormone, vitamin D, or creatinine betweengroups. Each subject had between 2 and 5 BMD scans over a me-dian follow-up time of 4.1 yr. The annual percentage changes inBMD at the spine and hip are shown in Figure 4. Annual BMDincreases were 2.2%, 1.4%, and 1.2% at the lumbar spine,femoral neck, and total hip, respectively, in the treated group.BMD decreased by 0.4%, 0.4%, and 0.3% at the lumbar spine,femoral neck, and total hip, respectively, in the untreated group.The changes in BMD were all significantly different from zero(Fig. 4). Twenty of the 22 women in the treated group had stableor positive (�0) annual increases in lumbar spine BMD inresponse to therapy. Of the 21 women in the untreated group,15 had annual changes of lumbar spine BMD �0.

The annual percentage changes in spine and hip BMD inrelation to tertiles of baseline biochemical markers and 18F-PET parameters are shown in Table 2. The annual increasesin BMD in response to antiresorptive therapy were greaterfor women in the highest tertile of global and regional boneturnover compared to those in the middle and lowest tertiles(Table 2A). The difference in changes in BMD between ter-tiles of deoxypyridinoline was statistically significant. Theannual increase in lumbar spine BMD was 1.8%, 2.2%, and3.2% for women in the lowest, middle, and highest tertile

Table 1Study Group Characteristics at Baseline

Characteristic Treated Untreated

Number 22 21Age (yr) 64.4 (6.8) 57.1 (7.7)*Years since menopause 14.6 (8.4) 10.9 (8.6)Height 162.7 (7.2) 162.8 (5.8)Weight 63.8 (9.6) 67.0 (7.2)

Lumbar spine BMD (g/cm2) 0.747 (0.11) 0.946 (0.132)*Femoral neck BMD (g/cm2) 0.649 (0.06) 0.754 (0.09)*Total hip BMD (g/cm2) 0.765 (0.08) 0.878 (0.09)*

BSALP (U/I) 25.8 (6.7) 22.2 (7.1)Deoxypyridinoline (nmol/mm) 6.8 (1.7) 6.9 (1.5)Ki mL min�1 mL�1 0.033 (0.009) 0.035 (0.008)K1 mL min�1 mL�1 0.137 (0.093) 0.113 (0.030)

Abbr: BMD, bone mineral density; BSALP, bone-specific alka-line phosphatase.

*p� 0.05.


of BSALP, respectively (Table 2A). These differences in an-nual increases lumbar spine BMD were similar to those ob-served for the lowest, middle, and highest tertile of theregional measurement Ki of 1.7%, 2.2%, and 2.7%, respec-tively (Table 2A). There was a 1.6- to 5.1-fold difference inannual changes in lumbar spine BMD between the highestand lowest tertiles of bone turnover in the treated group(Table 2A). A less consistent pattern was observed betweenannual changes in hip BMD and tertiles of bone turnover, al-though annual changes in femoral neck BMD were greater forwomen in the highest tertile of BSALP and Ki (Table 2A).

As observed for the treated group, the largest annual changesin spine BMD were seen in untreated women in the highesttertile of regional and global bone turnover (Table 2B). Therewas a 1.4- to 4.8-fold difference in the annual decrease inBMD between those women in the highest and lowest tertilesof both global and regional bone turnover. Changes in femoralneck and total hip BMD tended to be greater in those women inthe highest tertile of regional bone turnover (Ki) compared tothose in the lowest tertile, although these differences did notreach statistical significance (Table 2B).

When the relationships between baseline regional andglobal bone turnover and annual changes in BMD were exam-ined using linear and stepwise regression analysis, there wereno significant correlations observed for either treated oruntreated women.

Similar results for all analyses were obtained when abso-lute changes rather than percentage changes in BMD wereused (data not shown).

Discussion

A direct assessment of bone turnover at the clinically im-portant skeletal sites such as the lumbar spine can be madeusing 18F-PET. In this study, we have examined the

-1

-0.5

0

0.5

1

1.5

2

2.5

3

Lumbar Spine Femoral Neck Total hip

Annu

al c

hang

e BM

D g

/cm

2

Treated Untreated

*

**

* * *

* P = < 0.05 from baseline

Fig. 4. Annual percentage changes in lumbar spine andhipbone mineral density for treated and untreated postmeno-pausal women. *p� 0.05 from baseline.

Volume 10, 2007


Table 2Annual Percentage Changes in Spine and Hip BMD in Relation to Tertiles of Baseline Global and Regional Bone Turnover;

in (A) Women Treated With Antiresorptive Therapy and (B) Untreated Women

Tertiles of bone turnover

1 2 3

(A) Women treated with antiresorptive therapyLumbar spine BMD

BSALP (U/I) 1.82 (0.58) 2.21 (1.87) 3.18 (1.57)Deoxypyridinoline (nmol/mm) 0.60 (1.28) 2.79 (1.69) 3.03 (1.13)Ki mL min�1 mL�1 1.69 (2.24) 2.15 (1.32) 2.68 (1.63)K1 mL min�1 mL�1 1.42 (2.08) 2.42 (1.17) 2.64 (1.89)

Femoral neck BMDBSALP (U/I) 1.29 (1.23) 1.24 (1.78) 1.48 (2.25)Deoxypyridinoline (nmol/mm) 1.17 (1.00) 1.93 (1.97) 0.93 (1.85)Ki mL min�1 mL�1 1.20 (1.09) 1.32 (1.68) 1.60 (2.25)K1 mL min�1 mL�1 1.32 (0.99) 1.28 (1.39) 1.53 (2.52)

Total hip BMDBSALP (U/I) 1.20 (0.89) 1.43 (1.35) 1.15 (2.02)Deoxypyridinoline (nmol/mm) 1.25 (1.18) 1.64 (1.05) 0.72 (1.90)Ki mL min�1 mL�1 1.47 (1.24) 1.20 (1.60) 1.00 (1.47)K1 mL min�1 mL�1 1.41 (1.14) 1.56 (1.53) 0.66 (1.49)

(B) Untreated womenLumbar spine BMD

BSALP (U/I) 0.09 (0.98) �0.85 (0.76) �0.43 (0.29)Deoxypyridinoline (nmol/mm) �0.35 (1.03) �0.40 (0.75) �0.50 (0.56)Ki mL min�1 mL�1 �0.28 (0.49) �0.28 (0.92) �0.69 (0.85)K1 mL min�1 mL�1 �0.26 (0.82) �0.30 (0.40) �0.70 (0.99)

Femoral neck BMDBSALP (U/I) �0.32 (0.34) �0.82 (1.25) �0.06 (0.58)Deoxypyridinoline (nmol/mm) �0.30 (0.40) �0.67 (1.19) �0.17 (0.79)Ki mL min�1 mL�1 �0.26 (0.45) �0.20 (0.71) �0.69 (1.22)K1 mL min�1 mL�1 �0.07 (0.55) �0.78 (1.08) �0.22 (0.64)

Total hip BMDBSALP (U/I) �0.12 (0.65) �0.39 (1.01) �0.41 (0.54)Deoxypyridinoline (nmol/mm) �0.17 (0.62) �0.52 (0.60) �0.23 (0.95)Ki mL min�1 mL�1 �0.17 (0.90) �0.34 (0.52) �0.41 (0.77)K1 mL min�1 mL�1 0.13 (0.65) �0.72 (0.53) �0.26 (0.78)

Abbr: BMD, bone mineral density; BSALP, bone-specific alkaline phosphatase.Note: SD shown in parentheses.

relationship between bone turnover with subsequent changesin BMD using both biochemical markers and 18F-PET. Theprimary objective was to see if the association between re-gional bone turnover measured using 18F-PET with changesin BMD was similar to that observed using biochemicalmarkers. Although the number of subjects in the treated anduntreated groups was small, women with the highest globaland regional bone turnover showed the largest annual changes


in BMD at the lumbar spine. This is the first study to showthat bone turnover measured directly at the lumbar spine is re-lated to changes in BMD at the same skeletal site, and thatthis relationship is similar to that observed for biochemicalmarkers.

Significant moderate correlations between baseline bio-chemical markers or early reductions in markers with changesin BMD after 1 or 2 yr have been observed in postmenopausal

Volume 10, 2007

52 Frost et al.

women treated with HRT (9,32,38e40), bisphosphonates(11,12,14), and teriparatide (41). In this study, women inthe highest tertile of baseline bone turnover who subsequentlycommenced antiresorptive therapy (treated group) had1.6e5.0 fold greater annual rates of change at the lumbarspine compared to those in the lowest tertile (Table 2A). Inthe large randomized Fracture Intervention Trial of alendro-nate, higher levels of formation markers were associatedwith a greater difference in change in spine BMD in osteopo-rotic women treated with alendronate, a finding also observedfor subjects on placebo (14). Results from observationalstudies and from the placebo arm of treatment studies showa less consistent pattern between bone turnover and changesin BMD. As seen for the treated group in this study, the un-treated women in the highest tertile of baseline global and re-gional bone turnover displayed larger annual decreases inlumbar spine BMD (Table 2B). Results from 2 observationalstudies have demonstrated a significant association betweenbone turnover measured by biochemical markers and changesin calcaneal (33) and femoral (31) BMD. In contrast, no sig-nificant correlations were observed between baseline bonemarkers and bone loss in a 4-yr observational study (42).Similar conflicting results have been seen for women in theplacebo arm of treatment studies with significant correlationsseen in some studies (9,38,43) but not others (11,39). Oneplausible explanation for these discrepancies is the fact thatsubjects in the placebo arm are often taking doses of calciumand vitamin D that maintain BMD.

One of the major attractions of 18F-PET is that it allowsa noninvasive measurement of bone turnover at specific sitesof the skeleton, which is not possible using other techniques(19,21,23,26). The skeleton is heterogeneous in terms of itsresponse to aging and to treatments that influence bone re-modeling. Biochemical markers reflect global skeletal func-tion and because 80% of the skeleton is cortical in nature,these markers may attenuate the rate of bone remodeling oc-curring at cancellous skeletal sites such as the lumbar spine.This is reflected in the results obtained in this study withonly 20% of the subjects in the treated and untreated groupsin the same tertile of regional and global bone turnover. In ad-dition, it is unlikely that there is a direct one-to-one relation-ship between the rate of global bone turnover and thebiochemical markers that can be readily measured in serumand urine (1). Regional measurements of bone turnover atsites such as the lumbar spine and hip should provide a greaterinsight into the pathophysiology of osteoporosis and the asso-ciated fractures. Although the numbers in the treated and un-treated group were small, precluding a significant correlationbetween turnover and changes in BMD, women in the highesttertile of bone turnover showed the largest annual gains(treated) or decreases (untreated) in BMD. This observationwas true for both 18F-PET and biochemical marker measure-ments (Table 2). A superior relationship between regionalbone turnover and changes in BMD may have been expectedbecause 18F-PET offers the advantage of allowing site-matched assessments of turnover and BMD. However, thispilot study was not statistically powered to detect significant


differences between techniques. Further larger studies, whichallow an evaluation of the sensitivity and specificity of eachtechnique in predicting significant changes in BMD, are re-quired to fully investigate the relationship between regionalbone turnover and changes in bone density.

The associations between bone turnover and changes inBMD at the hip were less clear than those observed forchanges in lumbar spine BMD (Table 2). However, treatedwomen in the highest tertile of BSALP and Ki did havegreater increases in femoral neck BMD compared to thosein the lowest tertile (Table 2). Larger studies on antiresorptivetherapies tend to report weaker correlations between bio-chemical markers and change in hip BMD than that observedat the lumbar spine BMD (11,12,32,39,41,44). In this study,changes in hip BMD in response to treatment were almosthalf of that seen at the lumbar spine (Fig. 4). The modestchanges in hip BMD combined with the fact that our mea-surements of regional bone turnover were taken at the lumbarspine would probably preclude a closer relationship in thissmall study.

In this study, and other studies investigating the associationbetween changes in BMD and bone turnover, changes inBMD have been used as a surrogate marker for the changein risk of the clinical outcomeefragility fractures. Althoughchanges in BMD have been used in this study and many largertrials as a surrogate marker of fracture risk several of the largerandomized, placebo-controlled trials of bisphosphonates andraloxifene have shown that changes in bone turnover doreflect changes in fracture risk directly, with similar andsometimes better predictive abilities than changes in BMD(6,13e15).

This study had several limitations. The number of subjectsin each group was small and consequently the study did nothave adequate power to detect significant correlations be-tween baseline regional and global bone turnover with subse-quent changes in BMD, particularly as both techniques formeasuring bone turnover are associated with measurementserrors of approximately 15% (17,36). This study only evalu-ated site-matched measurements of turnover and BMD atthe lumbar spine. We have previously shown that 18F-PETcan be used to estimate bone turnover at the hip (23). How-ever, due to the relatively narrow field of view of most PETscanners (10.8 cm) it was only possible to acquire dynamicscans at 1 skeletal site. Further work is required to examinethe relationship between bone turnover at the hip and site-matched changes in BMD. Only baseline bone turnover wasmeasured in this study and it has been suggested that a staticmeasurement of turnover at baseline may not be as predictiveof changes in BMD or fracture risk as a dynamic measure-ment of early changes in bone turnover after 3 or 6 mo oftherapy (11,32,40). The treatment group was inhomogeneousin that subjects did not all take the same type of treatment.Because the mechanism of action and potency in terms ofchanges in BMD can vary for different treatments, a homoge-neous group may have increased the likelihood of observingsignificant differences in changes in BMD between tertilesof bone turnover. Older generation biochemical markers

Volume 10, 2007


were used in this study and therefore other more recentlydeveloped markers that have superior precision may haveyielded slightly different results. However, these limitationswould not have influenced the primary study objective, whichwas to evaluate whether the relationship between regionalbone turnover and changes in BMD were equivalent to thatobserved for global measurements of bone turnover using bio-chemical markers.

In conclusion, 18F-PET offers an attractive technique fornoninvasively assessing regional bone turnover at the clini-cally important sites such as the spine and hip in patientswith metabolic bone disease. The method also proves valu-able in evaluating how bone turnover at these skeletal siteschanges in response to treatment or aging. This study hasdemonstrated that the relationship between regional boneturnover measured directly at the lumbar spine with changesin lumbar spine BMD is similar to that seen for global skeletalbone turnover using biochemical markers. Larger studies arenow required to support these findings and fully investigateregional bone turnover in patients with osteoporosis.

Acknowledgments

The authors would like to thank the radiography and scien-tific staff at the Clinical PET Centre at St Thomas’ Hospitalfor their excellent technical support. This work was fundedin part by an unrestricted grant from Aventis Pharma UK.

References

1. Heaney RP. 2003 Is the paradigm shifting? Bone 33:457e465.2. Garnero P, Hausherr E, Chapuy MC, et al. 1996 Markers of bone

resorption predict hip fracture in elderly women: the EPIDOSprospective study. J Bone Miner Res 11:1531e1538.

3. Johnell O, Oden A, De Laet C, Garnero P, Delmas PD, Kanis JA.2002 Biochemical indices of bone turnover and the assessmentof fracture probability. Osteoporos Int 13:523e526.

4. Melton LJ III, Khosla S, Atkinson EJ, O’Fallon WM, Riggs BL.1997 Relationship of bone turnover to bone density and frac-tures. J Bone Miner Res 12:1083e1091.

5. Cummings SR, Karpf DB, Harris F, et al. 2002 Improvement inspine bone density and reduction in risk of vertebral fracturesduring treatment with antiresorptive drugs. Am J Med 112:281e289.

6. Eastell R, Barton I, Hannon RA. 2003 Relationship of earlychanges in bone resorption to the reduction in fracture riskwith risedronate. J Bone Miner Res 18:1051e1056.

7. Hochberg MC, Greenspan S, Wasnich RD, Miller P,Thompson DE, Ross PD. 2002 Changes in bone density andturnover explain the reduction in incidence of nonvertebral frac-tures that occur during treatment with antiresorptive agents.J Clin Endocrinol Metab 87:1586e1592.

8. Riggs BL, Melton LJ III, O’Fallon WM. 1996 Drug therapy forvertebral fractures in osteoporosis: evidence that decreases inbone turnover and increases in bone mass both determine anti-fracture efficacy. Bone 18:197Se210S.

9. Chesnut CH III, Bell NH, Clark GS, et al. 1997 Hormonereplacement therapy in postmenopausal women: urinary N-telopeptide of type I collagen monitors therapeutic effect and pre-dicts response of bone mineral density. Am J Med 102:29e37.


10. Garnero P, Darte C, Delmas PD. 1999 A model to monitor the effi-cacy of alendronate treatment in women with osteoporosis usinga biochemical marker of bone turnover. Bone 24:603e609.

11. Greenspan SL, Parker RA, Ferguson L, Rosen HN,Maitland-Ramsey L, Karpf DB. 1998 Early changes in biochem-ical markers of bone turnover predict the long-term response toalendronate therapy in representative elderly women: a rando-mised clinical trial. J Bone Miner Res 13:1431e1438.

12. Ravn P, Hosking D, Thompson D, et al. 1999 Monitoring ofalendronate treatment and prediction of effect on bone massby biochemical markers in the Early Postmenopausal Interven-tion Cohort Study. J Clin Endocrinol Metab 84:2363e2368.

13. Buaer DC, Black DM, Garnero P, et al. 2004 Changes in boneturnover and hip, non-spine and vertebral fracture in alendronate-treated women: the fracture intervention trial. J Bone Miner Res19:1250e1258.

14. Bauer DC, Garnero P, Hochberg MC, et al. 2006 Pretreatmentlevels of bone turnover and the antifracture efficacy of alendro-nate: the fracture intervention trial. J Bone Miner Res 21:292e299.

15. Bjarnason NH, Sarkar S, Duong T. 2001 Six and twelve monthchanges in bone turnover are related to reduction in vertebralfracture risk during 3 years of raloxifene treatment in postmen-opausal osteoporosis. Osteoporos Int 12:922e930.

16. Delmas PD, Eastell R, Garnero P, et al., for the Committee ofScientific Advisors of the International Osteoporosis Foundation2000 The use of biochemical markers of bone turnover in oste-oporosis. Osteoporos Int 11(Suppl 6):2e17.

17. Seibel MJ. 2000 Molecular markers of bone turnover: biochem-ical, technical and analytical aspects. Osteoporos Int 11(Suppl6):18e29.

18. Compston JE, Croucher PI. 1991 Histomorphometric assessmentof trabecular bone remodelling in osteoporosis. Bone Miner 14:91e102.

19. Cook GJR, Lodge MA, Blake GM, Marsden PK, Fogelman I. 2000Differences in skeletal kinetics between vertebral and humeral bonemeasured by 18F-fluoride positron emission tomography in postmen-opausal women. J Bone Miner Res 15:763e769.

20. Cook GJR, Blake GM, Marsden PK, Cronin B, Fogelman I.2002 Quantification of skeletal kinetic indices in Paget’s diseaseusing dynamic 18F-fluoride positron emission tomography.J Bone Miner Res 17:854e859.

21. Frost ML, Cook GJ, Blake GM, Marsden PK, Benatar NA,Fogelman I. 2003 A prospective study of risedronate on regionalbone metabolism and blood flow at the lumbar spine measuredby 18F-fluoride positron emission tomography. J Bone MinerRes 18:2215e2222.

22. Frost ML, Fogelman I, Blake GM, Marsden PK, Fogelman I.2004 Dissociation between global markers of bone formationand direct measurement of spinal bone formation in osteoporo-sis. J Bone Miner Res 19:1797e1804.

23. Frost ML, Cook GJR, Blake GM, Marsden PK, Fogelman I.2004 A comparison of regional bone turnover at the spine andhip in postmenopausal women with osteoporosis. OsteoporosInt 15(Suppl 2):40e41.

24. Hawkins RA, Choi Y, Huang S-C, et al. 1992 Evaluation of theskeletal kinetics of fluorine-18-fluoride ion with PET. J NuclMed 33:633e642.

25. Messa C, Goodman WG, Hoh CK, et al. 1993 Bone metabolicactivity measured with positron emission tomography and18F-fluoride ion in renal osteodystophy: correlation with bonehistomorphometry. J Clin Endocrinol Metab 77:949e955.

26. Piert M, Zittel TT, Becker GA, et al. 2001 Assessment of por-cine bone metabolism by dynamic 18F-fluoride PET: correlationwith bone histomorphometry. J Nucl Med 42:1091e1100.

Volume 10, 2007

54 Frost et al.

27. Schiepers C, Nuyts J, Bormans G, et al. 1997 Fluoride kineticsof the axial skeleton measured in vivo with fluorine-18-fluoridePET. J Nucl Med 38:1970e1976.

28. Blau M, Nagler W, Bender MA. 1962 Fluorine-18: a new isotopefor bone scanning. J Nucl Med 3:332e334.

29. Blau M, Ganatra R, Bender MA. 1972 18F-fluoride for boneimaging. Semin Nucl Med 2:31e37.

30. Wootton R, Dore C. 1986 The single-passage extraction of 18F inrabbit bone. Clin Phys Physiol Meas 7:333e343.

31. Dresner-Pollak R, Parker RA, Poku M, Thompson J, Seibel MJ,Greenspan SL. 1996 Biochemical markers of bone turnover re-flect femoral bone loss in elderly women. Calcif Tissue Int 59:328e333.

32. Dresner-Pollak R, Mayer M, Hochner-Celiniker D. 2000 Thedecrease in serum bone-specific alkaline phosphatase predictsbone mineral density response to hormone replacement therapyin early postmenopausal women. Calcif Tissue Int 66:104e107.

33. Ross PD, Knowlton W. 1998 Rapid bone loss is associated withincreased levels of biochemical markers. J Bone Miner Res 13:297e302.

34. Heaney RP, Yates AJ, Santora C II. 1997 Bisphosphonate effectsand the bone remodelling transient. J Bone Miner Res 12:1143e1151.

35. Cook GJR, Lodge MA, Marsden PK, Dynes A, Fogelman I.1999 Non-invasive assessment of skeletal kinetics using fluori-ne-18-fluoride positron emission tomography: evaluation of im-age and population-derived arterial input functions. Eur J NuclMed 26:1424e1429.

36. Frost ML, Cook GJR, Blake GM, Marsden PK, Fogelman I.2006 The precision of 18F-fluoride positron emission tomogra-phy as used to measure regional bone turnover at the spineand hip. Osteoporos Int 17(Suppl 3):429e430.


37. Patel R, Blake GM, Rymer J, Fogelman I. 2000 Long-term pre-cision of DXA scanning assessed over seven years in forty post-menopausal women. Osteoporos Int 11:68e75.

38. Hesley RP, Shepard KA, Jenkins DK, Riggs BL. 1998 Monitor-ing estrogen replacement therapy and identifying rapid bonelosers with an immunoassay for deoxypyridinoline. OsteoporosInt 8:159e164.

39. Marcus R, Holloway L, Wells B, et al. 1999 The relationship ofbiochemical markers of bone turnover to bone density changesin postmenopausal women: results from the Postmenopausal Es-trogen/Progestin Interventions (PEPI) Trial. J Bone Miner Res14:1583e1595.

40. Rosen CJ, Chesnut CH III, Mallinak NJS. 1997 The predictivevalue of biochemical markers of bone turnover for bone mineraldensity in early postmenopausal women treated with hormonereplacement or calcium supplementation. J Clin EndocrinolMetab 82:1904e1910.

41. Chen P, Satterwhite JH, Licata AA, et al. 2005 Early changes inbiochemical markers of bone formation predict BMD responseto teriparatide in postmenopausal women with osteoporosis.J Bone Miner Res 20:962e970.

42. Keen RW, Nguyen T, Sobnack R, Perry LA, Thompson PW,Spector TD. 1996 Can biochemical markers predict bone lossat the hip and spine?: a 4-year prospective study of 141 earlypostmenopausal women. Osteoporos Int 6:399e406.

43. Riis BJ, Overgaard K, Chrsitiansen C. 1995 Biochemicalmarkers of bone turnover to monitor the bone response to post-menopausal hormone replacement therapy. Osteoporos Int 5:276e280.

44. Black DM, Cummings SR, Karpf DB, et al. 1996 Randomisedtrial of effect of alendronate on risk of fracture in women withexisting vertebral fractures. Fracture Intervention Trial ResearchGroup. Lancet 348:1535e1541.

Volume 10, 2007

the relationship between regional bone turnover measured using 18f-fluoride positron emission...

Documents