OsteoporosInt DOI10.1007/s00198-012-2057-z ORIGINAL ARTICLE Sarcopenia and its relationship with bone mineral density in middle-aged and elderly European men S. Verschueren&E. Gielen&T. W. O’Neill&S. R. Pye& J. E. Adams &K. A. Ward&F. C. Wu &P. Szulc & M.Laurent&F.Claessens&D.Vanderschueren&S.Boonen Received:14February2012/Accepted:12June2012 #InternationalOsteoporosisFoundationandNationalOsteoporosisFoundation2012 Abstract dual-energyX-rayabsorptiometry,fromwhichappendicular Summary The aim of this study was to determine the leanmass(aLM),fatmass(FM)andwhole-body,spineand relationship between reduced muscle mass (sarcopenia) hip BMD were determined. Relative appendicular skeletal a and areal bone mineral density (BMD ) in middle-aged musclemass(RASM)wascalculatedasaLM/height².Mus- a and elderly community-dwelling European men. Men cle strength was assessed in subjects from Leuven. Sarco- with sarcopenia had significantly lower BMD and were penia was defined by RASM at <7.26 kg/m² and by the a more likely to have osteoporosis compared with men recentdefinitionoftheEuropeanWorkingGrouponSarco- without sarcopenia. penia in Older People (RASM at <7.26 kg/m2 plus low Introduction Inmen,therelationshipbetweenreducedmus- muscle function). Linear regression was used to determine cle mass (sarcopenia) and BMD is unclear. This study the associations between aLM, FM, muscle strength and a aimed to determine this relationship in middle-aged and BMD and logistic regression to determine the association a elderlycommunity-dwellingmen. between sarcopenia andosteoporosis. Methods Menaged40–79yearsfromtheManchester(UK) Results Six hundred seventy-nine men with a mean age of and Leuven (Belgium) cohorts of the European Male Age- 59.6(SD010.7),contributeddatatotheanalysis;11.9%were ing Study were invited to attend for assessment including sarcopenic by the conventional definition. After adjustment S.VerschuerenandE.Gielencontributedequallytothemanuscript. S.Verschueren K.A.Ward ResearchGroupforMusculoskeletalRehabilitation, NutritionandBoneHealth,MRCHumanNutritionResearch, DepartmentofRehabilitationSciences,KULeuven, Cambridge,UK Leuven,Belgium F.C.Wu : : E.Gielen M.Laurent S.Boonen(*) AndrologyResearchUnit,ManchesterAcademicHealthScience Centre(MAHSC),UniversityofManchester, GerontologyandGeriatrics,DepartmentofClinical Manchester,UK andExperimentalMedicine,KULeuven, Leuven,Belgium P.Szulc e-mail:[email protected] INSERMUMR1033,UniversityofLyon, Lyon,France : T.W.O’Neill S.R.Pye : M.Laurent F.Claessens ArthritisResearchUKEpidemiologyUnit,ManchesterAcademic LaboratoryofMolecularEndocrinology,DepartmentofCellular HealthScienceCentre(MAHSC),UniversityofManchester, andMolecularMedicine,KULeuven, Manchester,UK Leuven,Belgium : J.E.Adams K.A.Ward D.Vanderschueren ManchesterAcademicHealthScienceCentre(MAHSC) ClinicalandExperimentalEndocrinology,DepartmentofClinical andRadiologyatManchesterRoyalInfirmary, andExperimentalMedicine,KULeuven, Manchester,UK Leuven,Belgium OsteoporosInt for age and centre, aLM, RASM and FM were positively mass (sarcopenia) and BMD and found lower BMD in a a associated with BMD . Men with RASM at <7.26 kg/m² sarcopenicwomen[8,11,17].Inthesestudies,sarcopeniain a had significantly lower BMD compared with those with women has been defined as RASM less than 5.45 kg/m2, a RASM at ≥7.26 kg/m2. In a multivariable model, aLM according to the approach of Baumgartner et al. [18]. Re- was most consistently associated with BMD . Men with cently, however, theEuropeanWorking Group on Sarcope- a sarcopenia were more likely to have osteoporosis com- nia in Older People (EWGSOP) suggested restricting the pared with those with normal RASM (odds ratio03.0; definition of sarcopenia by requiring the presence of an 95 % CI01.6–5.8). additional criterion besides reduced muscle mass, either Conclusions Sarcopenia is associated with low BMD and low musclestrengthor poorphysicalperformance [19]. a osteoporosis in middle-aged and elderly men. Further stud- Inmen,theavailabledatasuggestadifferentrelationship iesarenecessarytoassesswhethermaintainingmusclemass between bone and body composition, although the results contributes toprevent osteoporosis. are inconsistent. Some studies showed that both LM (abso- lute or relative) and FM contributed independently to Keywords Arealbonemineraldensity(BMD ) .Leanmass . BMD ,withapositive[7,20]oranegative[21]correlation a a Musclestrength .Osteoporosis .Relative appendicular between FM and BMD . However, in other studies, only a skeletal muscle mass(RASM),sarcopenia absolute LM [22] or RASM [2] remained independently associated with BMD , with no influence of FM, contrary a to the situation in women. Some studies even showed no Introduction relationshipbetweenLMandBMD afteradjustingforBMI a [3] or when effects of skeletal size were removed by divid- Aprogressivedeclineinbonemineraldensity(BMD),mus- ing BMD by height [23]. Thus, the relative importance of a clemassandmusclestrengtharekeyfeaturesoftheageing LM vs. FM on BMD remains uncertain in men. As in a process. They predispose older individuals to disability, women, muscle strength was found to be a determinant of falls,fracturesandfrailtyandsoposeamajorandincreasing BMD ,independentofweight[16,24],thoughnotindepen- a clinicalandpublichealthburden.Thereisnowconsiderable dentofLM[9].Inmen,theassociationbetweenBMD and a evidence that muscle and bone have common genetic, nu- sarcopeniadefinedaslowRASMhasnotbeenthoroughly tritional, lifestyle and hormonal determinants [1–4]. In ad- studied, and there are no data exploring the relationship dition,muscleandboneinteracttoimpactonbonestrength when using the more stringent EWGSOP definition of [5]. A possible mechanism is the dynamic loading of sarcopenia [19]. muscles,towhichweight-bearingbonesadapt.Thisdynam- The aim of this study was to clarify the relationship ic loading arises from muscle contractions as well as from between muscle and bone in men. More specifically, we the ground impact during weight-bearing activities [6]. Ex- wanted to determine the association between muscle mass, ploringtherelationshipbetweenmuscleandbonemayhelp muscle strength and BMD , as well as the relationship a inthedevelopmentofinterventionsthatwillbenefitmuscu- between sarcopenia and BMD in middle-aged and elderly a loskeletalfunction,withtheaimofreducingadverseclinical European men. Sarcopenia will be defined by low muscle outcomessuch asfallsand fractures. mass alone as well as by the more stringent EWGSOP The evidence for this relationship between muscle and definition. To this end, we used cross-sectional data from boneinageingindividualscomesmostlyfromobservational two centres participating in the European Male Ageing epidemiological studies in women. In postmenopausal Study(EMAS),apopulation-basedstudyofageinginmen. women, almost all studies show that lean mass (LM (kg)) is correlated positively with whole-body and/or regional arealbonemineraldensity(BMD (g/cm2))[7–10].Relative Materials and methods a appendicular skeletal muscle mass (RASM, appendicular LM divided by height squared (kg/m²)) was also found to Subjects contribute significantly to regional BMD [11]. In most [7, a 9] but not all studies [8, 10], fat mass (FM (kg)) was an Men aged 40–79 years were recruited from population additionaldeterminant.Insome,onlyFM[12]orbodymass registers in Manchester (UK) and Leuven (Belgium) for index(BMI)[3],andnotLM,waslinkedwithBMD .There participationinEMAS[25].Subjectswereinvitedtoattend a is some evidence that FM may be more important after the by a letter of invitationwhich included a short postal ques- menopause[13,14].Musclestrengthwasfoundtobeasso- tionnaire. Those who agreed to take part were invited to ciated with BMD in postmenopausal women, independent attend a local clinic for an interviewer-assisted question- a ofweight[15,16]butdependentonLM[9].Severalauthors naire, assessment of physical function, height, weight and have assessed the relationship also between low muscle bonedensitometry.SubjectsinLeuvenhadalsoassessment OsteoporosInt of muscle strength. Ethical approval for the study was demonstrated by the assessor before being performed by obtainedinaccordancewithlocalinstitutionalrequirements the volunteer. Maximum isometric strength was measured in each centre. All subjects provided written informed at different angles (60° and 90°), the highest value of three consent. measurements taken as maximum isometric strength for each angle. Maximum isokinetic strength was measured at Assessments different angular velocities (60°/s and 90°/s) as the highest value of three attempts [30]. To determine the short-term Subjectscompletedapostalquestionnairewhichincludeda reproducibility, duplicate measurements (with a minimum questionaboutcurrentsmokingandsubsequentlyattendeda interval of 1 h) were performed in a random sample of 15 researchclinictocompleteaninterviewer-assistedquestion- subjects.CVwere10.8,16.7,11.3and14.6%forisometric naire and undergo clinical assessments. The interviewer- quadriceps strength at 60°, isometric quadriceps strength at assisted questionnaire included the Physical Activity Scale 90°, isokinetic quadriceps strength at 60°/s and isokinetic for the Elderly (PASE), which combines information on quadricepsstrengthat 90°/s, respectively. leisure,householdandoccupationalactivity[26].Theques- tionnaire also included queries about current prescription Diagnosisofosteoporosis and sarcopenia and non-prescription drugs, by examination of medications and prescriptions brought into the clinic for that purpose. OsteoporosiswasclassifiedasaT-scoreatthefemoralneck, Heightandweightweremeasuredinastandardisedfashion; total hip or lumbar spine of at least 2.5 standard deviations height to the nearest 1 mm using a stadiometer (Leicester (SD)belowthepeakBMD ofayounghealthymalereference a Height Measure, SECA UK Ltd) and body weight to the group. The reference population was the Third National nearest 0.1 kg using an electronic scale (SECA, model no. HealthandNutritionExaminationSurvey[31]. 8801321009, SECA UK Ltd). BMI was calculated as Sarcopenia was defined using two approaches. The first weight in kilogrammes divided by height in square metres. was based on the approach of Baumgartner et al. who Physicalability/dysfunctionwasmeasuredbyusingacom- described sarcopenia as RASM (aLM/height2) below a ponent of the Reuben’s physical performance test (seconds threshold of 7.26 kg/m2 [18]. aLM is the sum of LM of takentowalk50ft)[27]andtheTinettitestforbalanceand arms and legs, measured by DXA. DXA-measured LM is postural stability (seconds taken to go from a sitting to a considered a good indicator of skeletal muscle mass [32]. standingposition) [28]. The second approach was based on the new European consensus definition of the EWGSOP in which a person Bone densitometry andassessmentof muscle strength fulfilling only the criterion of low muscle mass is categor- ised as having pre-sarcopenia, while a person who also has Subjects (N0697) had dual-energy X-ray absorptiometry low muscle strength or low physical performance is cat- (DXA) scans performed on QDR 4500A Discovery scan- egorised as having sarcopenia, and a person with all three ners (Hologic Inc, Bedford, MA, USA), to measure whole- criteria as having severe sarcopenia [19]. Low muscle body,femoralneck,totalhipandlumbarspineBMD ,total strength was defined as grip strength at ≤29 kg if BMI a LM, appendicular LM (aLM) and total FM. All scans and is ≤24, ≤30 kg if BMI is 24.1–28 and ≤32 kg if BMI analyses were performed by trained and certified DXA is >28 [33], and low physical performance as a walking technicians.TheHologicSpinePhantomwasscanneddaily speed of <1 m/s [34]. to monitor the device performance and long-term stability. Devices in Leuven and Manchester were cross-calibrated Analysis with theEuropeanSpinePhantom. Muscle strength testing was performed in Leuven only Subjects taking bone active therapies (corticosteroids, (N0361). Grip strength was evaluated with the Jamar 1 bisphosphonates, calcium and vitamin D, N039) were ex- hand-held dynamometer (TEC Inc., Clifton, NJ). Three cluded from the analysis. No subjects were treated with measurements of maximum strength were taken at both parathyroid hormone (PTH). Descriptive statistics were sides, and the highest value was recorded as maximal grip used to summarise subject characteristics. The association strength (in kilogrammes) [29]. Isometric and isokinetic between RASM, relative FM (total FM/height2 (in kilo- strength were evaluated in the knee extensors of the left grammes per square metre)) and muscle strength on the leg, primarily the quadriceps, to correspond to the side of one hand and BMD (total hip and lumbar spine) on the a proximal femur BMD measurement. Strength was mea- other hand was assessed visually using scatter plots, super- a sured using an isokinetic dynamometer (Cybex II, Lumex imposing linear lines and also locally weighted scatter plot Inc., Ronkonkoma, NY) according to the standardised smooth (LOWESS) curves to examine potential non- procedures provided by the manufacturer. All tests were linearity.Thestrengthoftheassociationswasassessedusing OsteoporosInt linearregression(withBMD asthedependentvariable)and Table1 Subjectcharacteristics a results expressed as β coefficients. In subsequent analyses Variable(N0679) Mean(SD) Percent for ease of interpretation and comparison we standardised the BMD measures into Z-scores. Multivariable linear re- Ageatinterview(years) 59.6(10.7) a gressionwasthenusedtodeterminetheassociationbetween Height(cm) 174.5(7.0) the risk factors (anthropometry, physical performance, cur- Weight(kg) 82.7(13.1) rentsmoking,aLMandtotalFM)andtheoutcome (whole- Bodymassindex(kg/m2) 27.1(3.7) body,femoralneck,totalhipandlumbarspineBMDa)with PASEscore(0–1,100) 208.7(83.8) adjustments made for age and centre. Multivariable linear Tinetti:timetakenfromsittingto 12.5(3.3) regressionwasalsousedtoexaminetheassociationbetween standing(s) muscle strength (quadriceps strength) and BMD with PPT:timetakentowalk50ft(s) 13.7(2.6) a adjustments for age (Leuven cohort only). To examine po- Whole-bodyBMDa(g/cm2) 1.162(0.107) tential non-linear/threshold effects we categorised the risk FemoralneckBMDa(g/cm2) 0.807(0.128) factors into quintiles. In a final model we used stepwise TotalhipBMDa(g/cm2) 1.015(0.142) linear regression including all the potential factors (centre, LumbarspineBMDa(g/cm2) 1.049(0.173) age,height,timetowalk50ft,currentsmoking,aLM,total Appendicularleanmass(kg) 25.2(3.6) FM and isometric quadriceps strength at 90°). Both for- RASMa(kg/m2) 8.2(0.9) wards(startingwithanemptymodel)andbackwards(start- Totalfatmass(kg) 19.9(6.0) ing with the full model) variable selection was employed Relativetotalfatmass(kg/m2) 6.5(1.9) with no difference in results. Only significant (p<0.05) Currentsmoker(yesvs.no) 13.8 factors were retained in the models. Absolute aLM and not Sarcopeniab 11.9 RASMwaschoseninthesemodelstoallowtheinfluenceof Osteoporosisc 8.8 height to be independently examined. Isometric quadriceps Leuvencohort(N0361) strengthat90°waschosentorepresentmusclestrengthasit Isometricquadricepsstrength60°(Nm) 170.8(50.5) appeared to be the most strongly associated with BMD . Isometricquadricepsstrength90°(Nm) 165.0(45.3) a Similarly, of the physical activity and performance meas- Isokineticquadricepsstrength60°/s(Nm) 121.2(44.2) ures, time to walk 50 ft and not PASE score or sit to stand Isokineticquadricepsstrength90°/s(Nm) 105.2(44.0) timewaschosenastimetowalk50ftwasfoundtohavethe Gripstrength(kg) 41.5(8.2) mostconsistentassociationwithBMDa.Forallthestepwise Sarcopeniad 3.7 multivariable models, the variance inflation factor was cal- culated to quantify the severity of any potential multicolli- PPT physical performance test, BMDa areal bone mineral density, RASMrelativeappendicularskeletalmusclemass nearity and consequently weight/BMI and grip strength aAppendicularleanmassdividedbyheightsquared were not included due to multicollinearity. The results of alllinearregressionanalysesareexpressedasβcoefficients bSarcopenia according to the definition of Baumgartner et al. [18]: RASMat<7.26kg/m2 orstandardisedβcoefficientsand95%confidenceintervals cT-score≤−2.5atfemoralneck,totalhip,orlumbarspine (CI). Finally, logistic regression was used to examine the dSarcopeniaaccordingtothedefinitionofEWGSOP[19]:RASMat association between sarcopenia (using the two operational <7.26 kg/m2+low muscle strength (grip strength, ≤29 kg if BMI is definitions)andosteoporosis,withresultsexpressedasodds ≤24;≤30kgifBMIis24.1–28;and≤32kgifBMIis>28[33])orlow ratios(OR)and95%CI.Statisticalanalysiswasperformed physicalperformance(walkingspeed,1.0m/s[34] using STATAversion 9.2(http://www.stata.com). 3.7 % based on the EWGSOP definition (Leuven cohort only).Nineper cent were classified as being osteoporotic. Results Association betweenanthropometry, physicalactivity/ Subjects performance,musclestrengthand BMD a A total of 679 men with a mean age of 59.6 (SD010.7) In bivariate unadjusted analysis, height, weight and BMI yearsandmeanBMIof27.1(SD03.7)kg/m²wereincluded were positively associated with BMD at all sites. Also a in the analysis. Details of the subject characteristics are b shown in Table 1. Mean femoral neck BMDa was 0.807 Fig.1 AssociationbetweentotalhipBMDaandaRASM,crelativetotal (SD00.128) g/cm² and mean lumbar spine BMD 1.049 fatmass,eisometricquadricepsstrength90°,andggripstrength.Asso- a ciation between lumbar spine BMD and b RASM, d relative total fat (SD00.173)g/cm².Twelvepercentofmenweresarcopenic a mass,fisometricquadricepsstrength90°andhgripstrength.Thesolid according to the conventional definition of sarcopenia and linesrepresentthelinearrelationship;thedashedlinesrepresentLOWESS OsteoporosInt 1.5 A β coeff = 0.064 p<0.001 2.0 B β coeff = 0.047 p<0.001 2D (g/cm) 2MD (g/cm) 1.5 Total Hip BM1.0 mbar Spine B 1.0 u L 0.5 0.5 6 7 8 9 10 11 6 7 8 9 10 11 RASM (kg/m2) RASM (kg/m2) 1.5 C β coeff = 0.015 p<0.001 2.0 D β coeff = 0.014 p<0.001 2Hip BMD (g/cm) 1.0 2) Spine BMD (g/cm 1.5 Total mbar 1.0 u L 0.5 0.5 0 5 10 15 0 5 10 15 Relative total fat mass (kg/m2) Relativetotalfatmass(kg/m2) E F 1.5 2.0 β coeff = 0.001 p<0.001 β coeff = 0.0004 p<0.05 2g/cm) 2D (g/cm) 1.5 Hip BMD ( 1.0 Spine BM Total mbar 1.0 u L 0.5 0.5 50 100 150 200 250 300 50 100 150 200 250 300 Isometric quadriceps strength 90o (Nm) Isometric quadriceps strength 90o (Nm) G H 1.5 2.0 β coeff = 0.004 p<0.001 β coeff = 0.0006 p=NS 2g/cm) 2D (g/cm) 1.5 Hip BMD ( 1.0 Spine BM Total mbar 1.0 u L 0.5 0.5 20 30 40 50 60 70 20 30 40 50 60 70 Grip strength (kg) Grip strength (kg) OsteoporosInt higheraLM(bothabsoluteandrelativetoheight)wasasso- 50 ft was associated with lower whole-body, femoral neck ciatedwithhigherBMD atallsites(datanotshown).Total andtotalhipBMD ,whilealongertimetakentogofroma a a hip BMD and lumbar spine BMD increased with increas- sittingtoastandingpositionwaslinkedwithlowerBMD at a a a ingRASM(β00.064and0.047g/cm2perkg/m²respective- all sites (data not shown). Current smoking was associated ly, see Fig. 1a, b). Similarly, higher absolute total FM was with lower BMD atthe total hip(data not shown). a associated with higher BMD at all sites (data not shown) After adjustment for both age and centre, height, weight a and increasing relative total FM with increasing total hip and BMI remained positively associated with BMD at all a and lumbar spine BMD (see Fig. 1c,d). sites (see Table 2). Also higher aLM (both absolute and a In the Leuven cohort only, higher quadriceps strength relative)remainedassociatedwithhigherBMD atallsites; a was associated with higher BMD at all sites, and only the compared with those with RASM of ≥7.26 kg/m2, those a associationbetweenisometricquadricepsstrengthmeasured withRASMof<7.26kg/m2hadsignificantlylowerBMD . a at60°andlumbarspineBMD wasnotsignificant(datanot Higher absolute total FM also remained associated with a shown).Isometric quadricepsstrengthmeasuredat90°was higher BMD at all sites and relative total FM was associ- a positivelyassociatedwithBMD atthetotalhipandlumbar ated positively with BMD at the femoral neck, total hip a a spine (see Fig. 1e, f). Higher grip strength was also associ- and lumbar spine (but not whole body). There was no atedwithhigherBMD atthetotalhip,butnotatthelumbar evidence of threshold effects when any of the anthropo- a spine (see Fig. 1g, h). Allthese associations observed were metric variables were included in the models categorised broadlylinearwithno evidence ofthresholdeffects. into quintiles. Physical activity as measured by PASE score was posi- Intermsofthephysicalperformance/activitymeasures,a tively associated with BMD in the whole body, femoral longer time taken to walk 50 ft remained associated with a neck and total hip. Similarly, a longer time taken to walk lower BMD at whole body, femoral neck and total hip, a Table2 Associationbetweenanthropometry,physicalactivity/performance,musclestrengthandbonedensity:adjustedforageandcentre Independentvariables Dependentvariable Whole-bodyBMD (perSD) FemoralneckBMD TotalhipBMD LumbarspineBMD a a a a (perSD) (perSD) (perSD) Wholecohorta Height(cm) 0.043(0.033,0.054)*** 0.036(0.025,0.047)*** 0.043(0.032,0.054)*** 0.036(0.024,0.047)*** Weight(kg) 0.024(0.018,0.029)*** 0.031(0.026,0.036)*** 0.035(0.030,0.040)*** 0.026(0.021,0.031)*** BMI(kg/m2) 0.051(0.032,0.071)*** 0.089(0.071,0.108)*** 0.098(0.079,0.116)*** 0.069(0.050,0.089)*** Appendicularleanmass(kg) 0.117(0.096,0.137)*** 0.119(0.099,0.139)*** 0.139(0.119,0.159)*** 0.102(0.080,0.123)*** RASM(kg/m2) 0.317(0.235,0.398)*** 0.373(0.293,0.453)*** 0.433(0.353,0.513)*** 0.294(0.209,0.379)*** RASM(kg/m2) ≥7.26 Referent Referent Referent Referent <7.26 −0.560(−0.786,−0.335)*** −0.661(−0.885,−0.437)*** −0.740(−0.968,−0.512)*** −0.593(−0.827,−0.360)*** Totalfatmass(kg) 0.020(0.008,0.032)** 0.041(0.030,0.053)*** 0.049(0.037,0.061)*** 0.034(0.022,0.046)*** Relativetotalfatmass(kg/m2) 0.028(−0.010,0.066) 0.105(0.068,0.143)*** 0.122(0.084,0.160)*** 0.081(0.042,0.120)*** PASEscore/10units 0.009(−0.001,0.019) 0.003(−0.007,0.013) 0.004(−0.007,0.014) 0.003(−0.007,0.014) Timetowalk50ft(s) −0.051(−0.080,−0.021)** −0.034(−0.064,−0.004)* −0.046(−0.076,−0.016)** −0.020(−0.051,0.011) Sittostandtime(s) −0.029(−0.052,−0.006)* −0.015(−0.039,0.008) −0.021(−0.045,0.002) −0.026(−0.049,−0.002)* Currentsmoker(yesvs.no) −0.253(−0.464,−0.042)* −0.252(−0.464,−0.040)* −0.307(−0.523,−0.092)** −0.185(−0.403,0.034) Leuvencohortonlyb Isometricquadricepsstrength60°(per10Nm) 0.039(0.017,0.060)*** 0.046(0.024,0.067)*** 0.050(0.028,0.071)*** 0.021(−0.001,0.044) Isometricquadricepsstrength90°(per10Nm) 0.060(0.037,0.084)*** 0.053(0.029,0.076)*** 0.074(0.051,0.096)*** 0.041(0.016,0.065)** Isokineticquadricepsstrength60°/s(per10Nm) 0.053(0.029,0.077)*** 0.035(0.010,0.060)** 0.051(0.027,0.076)*** 0.043(0.018,0.068)** Isokineticquadricepsstrength90°/s(per10Nm) 0.051(0.027,0.075)*** 0.040(0.015,0.064)** 0.048(0.024,0.073)*** 0.042(0.018,0.067)** Gripstrength(kg) 0.024(0.011,0.037)*** 0.013(−0.0003,0.026) 0.024(0.011,0.037)*** 0.008(−0.005,0.021) Resultsexpressedasβcoefficientsand95%CI BMD arealbonemineraldensity,BMIbodymassindex,RASMrelativeappendicularskeletalmusclemass,PASEPhysicalActivityScaleforthe a Elderly,NmNewtonmeter *p<0.05;**p<0.01;***p<0.001 aAdjustedforageandcentre bAdjustedforage OsteoporosInt while a longer time taken togo from a sitting to a standing significantlystronger(p<0.05)inthoseover60yearsofage positionremainedassociatedwithlowerBMD inthewhole (data not shown). a bodyandlumbarspine.PASEscorehoweverwasnotasso- In a stepwise multivariable model in the Leuven and ciated with BMD at any site after age and centre adjust- Manchester cohort which tested centre, age, height, time to a ment.CurrentsmokingwasassociatedwithlowerBMD at walk 50 ft and current smoking as confounding factors, a thewhole body,femoral neck and total hip. increasing aLM remained associated with higher BMD at a In the Leuven cohort, when examining muscle strength, all sites and total FM was associated with BMD at the a higher isokinetic quadriceps strength remained associated whole-bodyandtotalhipsites(seeTable3).Theeffectsize with higher BMD at all sites. Results were comparable oftotalFMonBMD wassmallincomparisonwiththatof a a for isometric quadriceps strength, though not significant aLMandthedirectionoftheeffectwasnotconsistent,with for the 60° measure and lumbar spine BMD . In contrast, total FM being positively linked with total hip BMD and a a higher grip strength remained only associated with higher negatively with whole-body BMD . Age, centre, time to a whole-body and total hip BMD . Quadriceps strength walk 50 ft and current smoking were retained in some of a explained a larger proportion of the variability in BMD the models. Overall, the significant variables accounted for a comparedwithgripstrength(3–10vs.0–3%,respectively; 12–26 %ofthe variability inBMD . a datanotshown).Therewasnoevidenceofthresholdeffects Similar results were observed in the Leuven and Man- whenanyofthemusclestrengthmeasureswereincludedin chester cohorts individually, though in the Leuven cohort, themodels categorised into quintiles. time to walk 50 ft was not associated with BMD at the a No difference in results was observed after stratification femoral neck and total FM was not associated with any of byage(equalnumbersofmeninfour10-yearagebands— thebonemeasurements (data not shown). 40–49, 50–59, 60–69 and 70–79 years), with broadly sim- In a second stepwise multivariable model in the Leuven ilar associations evident above and below the age of cohort only that also included quadriceps strength, increas- 60 years, though the associations between total FM and ingaLMremainedassociatedwithhigherBMD atallsites a BMD atthefemoralneck,totalhipandlumbarspinewere (see Table 3), and isometric quadriceps strength remained a Table3 Associationbetweenage,leanandfatmass,physicalperformanceandbonedensity:multivariablemodel Dependentvariables Whole-bodyBMD FemoralneckBMD TotalhipBMD LumbarspineBMD a a a a (perSD) (perSD) (perSD) (perSD) Independentvariables Centre:Manchester −0.425(−0.559,−0.291)*** – – 0.158(0.014,301)* Age(years) – – – 0.018(0.011,0.025)*** Height(cm) – – – – Timetowalk50ft(s) −0.034(−0.060,−0.007)* −0.027(−0.054,−0.001)* −0.036(−0.061,−0.008)* – Currentsmoker(yesvs.no) – – −0.210(−0.398,−0.021)* – Appendicularleanmass(kg) 0.130(0.109,0.151)*** 0.121(0.102,0.140)*** 0.118(0.097,0.139)*** 0.100(0.078,0.122)*** Totalfatmass(kg) −0.016(−0.028,−0.003)* – 0.017(0.005,0.029)** – R2forthemodel 0.24 0.21 0.26 0.12 Modelincludingquadricepsstrength:Leuvencohortonly Age(years) – – – 0.018(0.009,0.028)*** Height(cm) – – – – Timetowalk50ft(s) – – – – Currentsmoker(yesvs.no) – – −0.327(−0.570,−0.084)** – Appendicularleanmass(kg) 0.091(0.058,0.124)*** 0.119(0.093,0.145)*** 0.109(0.078,0.140)*** 0.093(0.063,0.122)*** Totalfatmass(kg) – – – – Isometricquadricepsstrength90°(per10Nm) 0.028(0.003,0.052)* – 0.024(0.001,0.048)* – R2forthemodel 0.18 0.20 0.25 0.10 Resultsexpressedasβcoefficientsand95%CI.Stepwiselinearregressionincludingcentre,age,height,timetowalk50ft,currentsmoking, appendicularleanmassandtotalfatmass.IntheLeuvencohortonly,stepwiselinearregressionalsoincludedisometricquadricepsstrength90°and excludedcentre.Variablesremainedinmodelifp<0.05 BMD arealbonemineraldensity,NmNewtonmeter a *p<0.05;**p<0.01;***p<0.001 OsteoporosInt Table4 Theassociationbetweensarcopeniaandosteoporosis likelytohaveosteoporosiscomparedwithmenwithnormal RASM, although the CI were wide as only 14 men were Number Osteoporsisa OR(95%CI) classifiedintothisgroup(OR02.0;95%CI00.4,10.0).No subjects were classified as having severe sarcopenia (low RASM(perSD)b 674 0.7(0.5,0.9)** RASM, low grip strengthand low physicalperformance). Sarcopeniac RASMat≥7.26kg/m2 594 Referent RASMat<7.26kg/m2 80 3.0(1.6,5.8)** Discussion Leuvencohortonlyd Sarcopeniae In this cross-sectional study, both aLM (absolute and rela- Normal 321 Referent tive) and total FM were associated with BMD at all sites, a Pre-sarcopenia 41 3.8(1.6,9.1)** after adjusting for age and centre. Quadriceps strength was Sarcopenia 14 2.0(0.4,10.0) linkedwithBMD atallsites,andgripstrengthwasassoci- a Severesarcopenia 0 – ated with BMD at the whole-body and total hip site. In a a stepwise multivariable model, aLM was the strongest inde- Resultsexpressedasoddsratios(OR)and95%CI pendentdeterminantofBMD atwholebody,femoralneck, a RASMrelativeappendicularskeletalmusclemass totalhipandlumbarspine.Atthewhole-bodyandtotalhip *p<0.05;**p<0.01;***p<0.001 sites, there was an additional independent contribution of aOsteoporosis:T-score≤−2.5atfemoralneck,totalhiporlumbarspine isometricquadricepsstrength,andcurrentsmokingcontrib- bAdjustedforageandcentre utedindependentlytototal hipBMD .Overall,these varia- a cSarcopenia using definition of Baumgartner et al. [18]: RASM at blesaccountedforapproximately10–25%ofthevariability <7.26kg/m² inBMD .aLMexplained20%ofthevariabilityinfemoral dAdjustedforage a neck BMD . When isometric quadriceps strength was not eSarcopeniausingdefinitionofEWGSOP[19]:presarcopenia—RASM a included in the model, physical performance (time to walk at <7.26 kg/m2, sarcopenia—RASM at <7.26 kg/m2+low muscle strength(gripstrength,≤29kgifBMIis≤24;≤30kgifBMIis24.1– 50 feet), total FM and current smoking were independently 28;and≤32kgifBMIis>28[33])orlowphysicalperformance(walking associated with BMD in some of the models in the entire a speed<1.0m/s[34])andseveresarcopenia—allthreecriteria investigatedcohort,whileintheLeuvencohortalone,phys- ical performance and current smoking, but not total FM, associatedwithwhole-bodyandtotalhipBMD .Incontrast, contributedindependentlytoBMD insomeofthemodels. a a total FM was not associated with BMD , nor was time to Several, though not all [3], studies have suggested that a walk 50 ft. Current smoking was associated with total hip LM [7,9, 22]orRASM [2,20] aresignificantly associated BMD . At the lumbar spine, a positive independent associ- withBMD inmen.Inouranalysis,aLMexplained20%of a a ation was present between age and BMD , but age was not the variability in BMD at the femoral neck in midde-aged a a an independent determinant of BMD at the other sites. andelderlymen,whichiscomparablewitharecentstudyin a Overall, these variables accounted for approximately 10– 160 healthy men aged 20 to 72 years, in whom RASM 25%ofthevariabilityinBMD .aLMexplained20%ofthe explained 15 % of the variance in femoral neck BMD [2]. a a variability in femoral neckBMD . Our observation that aLM is an independent contributor to a BMD mayreflectthemechanicalloadingthatmusclecon- a Association betweensarcopenia andosteoporosis tractions and resulting movements place on bone. Alterna- tively,itcouldbeattributedtothefactthatmuscleandbone Sarcopenia (RASM at <7.26 kg/m2) was associated with a have common genetic, nutritional, lifestyle and hormonal 3-fold higher risk of osteoporosis (OR03.0; 95 % CI01.6, determinants operating mainly during growth. 5.8) compared with those with normal RASM after adjust- In a study in men that, in contrast, could not identify an ment for age and centre (see Table 4). Each SD increase in independent effect of aLM on femoral neck and total hip RASM was associated with a 30 % reduction in the likeli- BMD , the authors surmised that the relationship between a hood ofosteoporosis(OR00.7; 95 %CI00.5, 0.9). aLM and BMD was largely mediated by physical activity a Similarly, in the Leuven cohort, men with EWGSOP- [3]. This was previously demonstrated by Walsh et al. in defined pre-sarcopenia (RASM at <7.26 kg/m2) were al- women in whom the relationship between RASM and mostfour times morelikely tohaveosteoporosiscompared BMD disappeared after adjusting for physical activity a with those with normal RASM after adjustment for age (assessedusingtheBaeckePhysicalActivityQuestionnaire) (OR03.8; 95 % CI01.6, 9.1). Those with sarcopenia [35].However,inourstudy,aLMremainedanindependent according to the EWGSOP definition (low RASM and low determinant of BMD when physical performance (time to a grip strength or low physical performance) were twice as walk 50 feet) was included in the multivariate model. OsteoporosInt Physical activity (as measured by PASE) was not related (visceral)fat[39].Thus,anindependentcontributionofFM with any of the bone measurements. The association be- to BMD was not observed in the multivariable model a tween aLM and BMD was also independent of current including isometric quadriceps strength. Yet, FM may con- a smoking. Other authors have suggested that the positive tributetobonemass,secondarytoaromatisationofandrogens relationship between LM and BMD might be attributed to into estrogens, insulin resistance with hyperinsulinemia as a bone or body size, when this factor is not adjusted for [9, well as higher levels of amylin and leptin, all of which are 36]. BMD and LM are influenced by bone/body size. positively associated with obesity [40, 41]. The reason why, a Failing to control for height may then overestimate the despite these obesity-related hormonal changes, we did not relationshipbetweenLMandBMD .Severalauthorsindeed observe anindependentcontributionofFMtoBMD inthis a a showed that the effect of LM on bone diminished when model, might be that testosterone dissociates fat and bone adjusting BMD for body size by dividing it by height or mass in men by respectively decreasing FM and increasing a by using bone mineral apparent density [9, 14, 23]. How- bone mass [42]. The observation that, in contrast with our ever, in our analysis, the relationship between aLM and studyinmen,therelationshipbetweenFMandbonemassis BMD persistedafteradjustingforheight(datanotshown). significant in women supports this concept of a potential a Moreover, according to Khosla et al. the attempt to control dissociationofFMandbonemassbytestosterone[9,23]. forbodysizetendstobiasagainstpotentialeffectsofLMon An additional independent contribution of isometric bone [14]. Finally, Baumgartner et al. supposed that the quadricepsstrength tothevariability ofBMD waspresent a reported association between muscle mass and BMD was at whole body and total hip, but not at femoral neck and a an artifact related to measuring muscle mass as “fat-free lumbar spine. In comparison, Taaffe et al. reported that mass”whichincludesbone,oras“fat-freesoft-tissuemass” musclestrengthcontributedindependentlyfromLMtolimb which includes organ mass. Both are inaccurate parameters BMD in women, but not to femoral neck or whole-body a ofmusclemassandalterthereforetherelationshipbetween BMD in women and not to any site in men [9]. Thus, our a “mucle mass” and BMD : including bone in the measure study is in agreement with others that there may be an a “fat-free mass” falsely strengthens the relationship with independent effect of muscle strength on BMD over and a BMD ,whileincludingorganmassinthemeasure“fat-free abovethat explainedbyLM.Thisadditional effectofmus- a soft-tissue mass” incorrectly attenuates this relationship cle strength may be due to the fact that, although LM and [36]. However, in our study, LM measured by DXA did muscle strength are highly correlated, muscle strength does not include bone mineral ororganmass butonly lean mass not depend solely on LM. This is illustrated by the obser- of botharms and legs. vation that, although loss of LM is accompanied by loss of InadditiontoaLM,agecontributedpositivelytolumbar muscle strength, the age-dependent loss of muscle strength spine BMD , which is probably an artifact related to the is larger than the loss of LM [43]. Yet, as mentioned, the a presence of osteophytosis and/or severe aortic calcification additional effect of muscle strength was not found at the [37]. Smoking was negatively associated with total hip femoral neck and lumbar spine. This may have several BMD ,arelationshipthathasalsobeenobservedbyPluijm explanations. First, finding no additional effect of muscle a etal.[7].WefoundnoindependentcontributionoftotalFM strengthonlumbarspineBMD isnotsurprising,aslumbar a to BMD in the multivariable model including isometric spine BMD is influenced by multiple other factors, e.g. a a quadriceps strength. This is consistent with most studies in osteophytosis that may have confounded the effect of mus- men, in which only LM or RASM was an independent cle strength. Moreover, measuring muscle strength at the determinant of BMD , with no influence of total FM [2, quadricepsandnotatthetrunkmayhavecontributedtothe a 22]. This is in contrast to the situation in postmenopausal factthatnoadditionaleffectofstrengthwasobservedatthe women,inwhomFMusuallywasanadditionalindependent lumbar spine. An alternative explanation is that most of the contributor to BMD [7, 9, 11]. However, when quadriceps effectofmusclestrengthonBMD isexplainedandexpressed a a strength was excluded from the multivariable model, total bytheeffectofLMonBMD ,whiletheadditionalcontribu- a FM was positively linked with total hip BMD and nega- tion of muscle strength to BMD , over and above LM, is a a tivelywithwhole-bodyBMD .Thissuggeststhattheeffect relativelyweak[3,9]. a of FM on total hip BMD is mediated by the dynamic Compared with grip strength, quadriceps strength might a loading of muscles on this weight-bearing bone site. Obese bethestrongerpredictorofBMD sincequadricepsstrength a people need indeed stronger muscles to move their higher was more consistently associated with all BMD sites and a bodyweightandcreatehigherimpactsonbonewhenmov- explained a larger proportion of the variation in BMD . a ing[6].ThenegativelinkbetweentotalFMandwhole-body However, since these results are based on cross-sectional BMD has also been observed by other authors [9, 21, 38] data, more research is needed to understand the relative a and may reflect the increased bone resorption associated contribution of grip strength and quadriceps strength to with the synthesis of inflammatory cytokines in abdominal bone health. OsteoporosInt Based on the definition of sarcopenia of Baumgartner et weight that can be lifted), frail elderly with a mean age of al.(RASMat<7.26kg/m²),12%ofourrandomsampleof 87 years obtained a significant increase in muscle strength, European men between 40 and 79 years were sarcopenic. physical activity and physical performance [50]. Also mus- ThisprevalenceissimilartothatreportedbyBaumgartneret cle mass improved with resistance training in older adults al. (13 %) in non-Hispanic US Caucasian men aged under [49]. An alternative to resistance training is whole-body 70 years [18]. Kyle et al. using a slightly lower cut-off of vibration training. With this therapy, the patient stands on 7.06kg/m²forthedefinitionofsarcopenia,reportedaprev- a platform that generates vertical sinusoidal vibrations. alence of 11 % in healthy Swiss men aged 60 years and These mechanical stimuli activate the muscle spindles, older [44]. With the stricter EWGSOP definition that resulting inthe activationof alpha motor neurons andiniti- requires an additional criterion beside low muscle mass, ate muscle contraction [51]. Similar to resistance training, the prevalence of sarcopenia decreased to 3.7 % in the vibration training has been shown to increase muscle mass Leuven cohort. In literature, the prevalence of sarcopenia and muscle strength inelderlysubjects [52]. varieswidely,from0%inGermansbetween61and83years At this stage, evidence regarding the efficacy of training [45] to 57.6 % in Hispanic US Caucasian men older than on bone loss is inconsistent and further studies are needed. 80 years [18]. It is likely this is due to differences in the A recent Cochrane review about the effectiveness of exer- studypopulation,thereferencegroup,thetechniqueusedto cise in postmenopausal women showed a relatively small, measure muscle mass and the definition of sarcopenia. For but statistically significant effect of physical activity on example,inthesameGermanpopulation,theprevalenceof BMD [53]. Non-weight bearing high force activity such as sarcopenia increased up to 21.8 % when sarcopenia was progressive resistance training was the most effective inter- definedby another measure of muscle mass [45]. vention for femoral neck BMD, while an exercise program Wefoundthatmenwithsarcopenia(RASMat<7.26kg/m²) combining weight bearing exercises and progressive resis- hadsignificantlylowerBMD atallmeasuredsitescompared tance training was most effective for lumbar spine BMD a withthosewithoutsarcopenia.Thesamehasbeenpreviously [53,54].Progressiveresistancetraining wasgenerallyinef- shown in sarcopenic women, with sarcopenia defined as fective for bone adaptations with a load <80 % of 1 RM RASM at <5.45 kg/m2 according to Baumgartner et al. [8, [55]. Also in older men, progressive resistance training 11,17,18].Wealsoobservedthatmenwithsarcopeniawere increased BMD at the hip, but was, contrary to previous more likely to have osteoporosis compared with men with studies in women, not better than walking 30 min three normal RASM. EWGSOP-defined pre-sarcopenia in the times a week [56]. Whole-body vibration had positive Leuvencohort(RASMat<7.26kg/m2[19])wasalsoassoci- effects on BMD in some studies in both genders [52, 54], atedwith a higher riskof osteoporosis.A similar association but a recent meta-analysis failed to observe an important betweenlowRASMandosteoporosiswasfoundbyDiMon- effect, hereby taking into account that the design of whole- acoetal.insarcopenicwomenwithhipfracture,inananalysis body vibration platforms and protocols for their use vary correctedfortimebetweenfractureandDXA,asadecreasein widely [57]. Thus, exercise programs combining strength bothLMandBMD hasbeenobservedafterfracture[46].Men andweightbearingtraining,aswellaswhole-bodyvibration a with EWGSOP-defined sarcopenia (RASM at <7.26 kg/m2 aloneorincombinationwithexercise,mayhelptoincrease andlowgripstrengthorphysicalperformance)weretwiceas oratleastpreventdeclinesinBMD,especiallyinpostmen- likely to have osteoporosis compared with non-sarcopenic opausalwomen,whilemoreresearchisneededinmen[54]. men, although this result was not significant due to lack of Our study had several limitations. This was a cross- power.Toourknowledge,therearenootherstudiesthathave sectional study and so it was not possible to determine the examinedtherelationshipbetween sarcopeniadefinedbythe temporal nature of the observed associations for which EWGSOPdefinitionandBMD orosteoporosisinmen. prospective data are needed. The response rate for partici- a Our observation that aLM determines up to 20 % of the pation in the study in these two centres was 39 %. It is variance in BMD and that RASM at <7.26 kg/m² is asso- possible that those invited, but declined to take part, may a ciated with a higher prevalence of osteoporosis, suggests havedifferedfromthosewhoparticipatedsothattheassess- that an interventional approach with physical training pro- mentsmay beanover-orunderestimateoftheresults from gramsaimedatimprovingmusclemassmaybeimportantto thetotalpopulation.Socautionisneededininterpretationof optimisebonehealth inmiddle-aged and elderly men. the data. However, any such non-response bias would be Numerous studies and meta-analyses have provided evi- unlikely to have influenced the association between bone dencethat,evenintheelderly,progressiveresistancetrain- andmuscleparameters.WeusedLMderivedfromDXAas ingisaneffectiveinterventionforsarcopenia[47–49].With ourestimateofmusclemass.AlthoughDXA-measuredLM, a 10-week training schedule that existed of three times that consists of muscle mass, skin, blood and interstitial a week three series of eight repetitions with a resistance fluid, is assumed to be a good indicator of muscle mass around 80 % of 1 repetition maximum (RM, the maximum [32],theevaluationofLMbyDXAmightunderestimatethe