Identification of Metabolic Bone Disease in Patients with EndogenousHyperthyroidism: Role of Biological Markers of Bone Turnover*
E. Jodar Gimeno,1 M. Mun oz-Torres,1 F. Escobar-Jimenez,1 M. Quesada Charneco,1 J. D. Luna del Castillo,2
N. Olea3
1Service of Endocrinology, University Hospital, 12 de Octubre, Madrid, Spain2Department of Biostatistics, University of Medicine, Granada, Spain3Service of Nuclear Medicine, University Hospital, 12 de Obtubre, Madrid, Spain
Received: 5 May 1997 / Accepted: 5 June 1997
Abstract. Active hyperthyroidism is associated with re-duced bone mass. Nevertheless, not all patients show thesame risk for developing osteoporosis. Our aim was to ana-lyze some clinical and biochemical potential predictors oflow bone mass in hyperthyroid patients. We studied 127consecutive hyperthyroid patients (110 females, 17 males;aged 42 ± 16 years). Bone mineral density (BMD) wasmeasured by dual X-ray absorptiometry (DXA) at lumbarspine (LS; L2–L4) and femoral neck (FN). Data were ex-pressed as g/cm2 and T-score. Patients were placed into twogroups based on recent WHO criteria: Group A, no osteo-porosis (n4 98); and group B, lumbar or femoral osteo-porosis (n4 29). Study protocol included evaluation ofosteoporosis risk factors, anthropometrical variables, thy-roid function, and bone turnover markers. Receiver-oper-ating characteristic (ROC) plots for the precision of bonemarkers and multivariate analysis for the prediction ofBMD and osteoporosis were performed. Group B showedgreater age and proportion of menopausal females; lowerweight, height, and calcium intake; longer duration ofmenopause; and greater levels of total and bone alkalinephosphatase and of urine hydroxyproline. No differences inthyroid function, osteocalcin, tartrate-resistant acid phos-phatase, and type I collagen C-telopeptide (ICTP) werefound. The best predictive model accounted for 46% and62% of the variability of lumbar and femoral BMD respec-tively and correctly classified 89% of the osteoporotic hy-perthyroid patients. No significant difference in ROC plotswas observed. It is concluded that hyperthyroid patientswith lumbar or femoral osteoporosis show a typical clinicaland biochemical profile illustrating that the relationship be-tween BMD and bone markers is better in high turnoverstates. Classical bone turnover markers show high perfor-mance in the evaluation of hyperthyroid bone disease.
Key words: Hyperthyroidism — Bone mineral density —Dual X-ray absorptiometry — Bone turnover markers —Osteoporosis prediction.
Since the classical description made by von Recklinghausen[1], histomorphometric [2] and densitometric [3, 4] studies
have demonstrated the detrimental effects of excess thyroidhormones on the skeleton. Therefore, previous hyperthy-roidism has proven to be a risk factor for osteoporosis [5].This hyperthyroid bone disease is associated with elevatedlevels of classical bone turnover markers [6] and also of thenewer markers, serum bone alkaline phosphatase (BONEALP)and urine excretion of pyridinoline cross-links [7].
Several approaches to the identification of the high-riskpopulation have been used. Genetic, endocrinologic, andlifestyle risk factors for osteoporosis have been identified.Nevertheless, the predictive capacity of such risk factors islow [8]. For this reason, different groups have proposedmathematical models to estimate the rate of bone loss [9,10] and to identify the patients with lower BMD values [11].Linear discriminating analysis is a useful tool for exploringthe predictive value of clinical and biochemical character-istics in identifying the patients with osteoporosis and, toour knowledge, has not been used previously to assess theindividual risk in hyperthyroid patients.
Moreover, new markers of bone metabolism with in-creased specificity, sensitivity, and availability have beendeveloped. SerumBONEALP determined by immunoradio-metric assay (IRMA) and serum type I collagen C-terminaltelopeptide (ICTP) determined by radioimmunoassay (RIA)have been introduced as precise formation and resorptionbone turnover markers respectively. The role of these mark-ers in assessing hyperthyroid bone disease must be evalu-ated in larger populations. Receiver-operating characteristic(ROC) plots are a fundamental evaluation tool in clinicalmedicine. ROC plots provide a pure index of accuracy bydemonstrating the limits of a test’s ability to discriminatebetween alternative states of health over all decision thresh-olds [12]. The aim of the present study was to identify thehyperthyroid subjects with osteoporosis on the basis ofclinical and biological data, to study the variability of lum-bar and femoral BMD explained by these data, and to ex-plore the precision of markers of bone turnover in assessinghyperthyroid bone disease.
Patients and Methods
Patients
We studied 126 consecutive active (elevated serum free T4 and T3levels) or former hyperthyroid patients (normal free T4 and T3levels with or without conventional carbimazole treatment) fromthe endocrinology outpatient clinic of the University of GranadaHospital (aged 42 ± 16 years; range 13–75 years; 17 males, 76premenopausal and 34 menopausal females). The diagnosis of hy-
*Preliminary results partially presented at the World Congress onOsteoporosis in Amsterdam, Holland, May 1996.Correspondence to:E. Jodar at C/ Fobos 7, 2° B, Madrid 28030,Spain
perthyroidism was made based on clinical and biochemical find-ings. Patients with subclinical hyperthyroidism were excluded.The therapy regimen in our center consisted of the prescription ofsubmaximal-maximal doses of carbimazole (30–45 mg/day) dur-ing the first month with following adjustment based on serum freeT4, T3, and TSH levels. All patients were caucasian. Patients whowere receiving or had received during the previous year medica-tion that may have altered phosphorus or calcium metabolism orbone mass were excluded. Postmenopausal women on estrogenreplacement therapy were also excluded. None of the patients stud-ied had a history of hepatic or renal disorders, alcoholism, osteo-porotic fractures, early menopause, or any other major medicalcondition. All of them consumed a normal diet and drank less than40 g of alcohol per day. Patients completed a standard question-naire designed to assess calcium intake (scored 1, low intake [<500
mg/day]; 2, medium intake [500–1000 mg/day]; 3, high intake[>1000 mg/day]), tobacco use (scored 0, no use; 1, <20 cigarettes/day; 2, >20 cigarettes/day), and physical activity (scored 1, littleactivity; 2, moderate activity; 3, very active). All participants wereinformed about the nature of the study and gave their consent toparticipate. The study was approved by our hospital’s ethical com-mittee.
Methods
Age, anthropometrical variables (height, weight, body mass index[BMI], body surface area [BSA]) and clinical data (etiology, du-
Table 1. Comparisons of qualitative data from hyperthyroid patients with osteoporosis(group A) or without osteoporosis (group B)
Group A(n 4 29)
Group B(n 4 98) P
SexFemale 96.0% 84.5% P 4 0.0405Male 4.0% 15.5%
MenopausePremenopausal 34.8% 78.2% P 4 0.0002Postmenopausal 65.2% 21.8%
Table 2. Comparisons of quantitative data from hyperthyroid patients with osteoporosis(group A) or without osteoporosis (group B)
Group A(n 4 29)
Group B(n 4 98) P
Age (years) 52.3 ± 17.7 38.4 ± 13.7 P < 0.0001Height (m) 1.559 ± 0.085 1.610 ± 0.071 P 4 0.0017Weight (kg) 59.6 ± 11.6 65.4 ± 12.2 P 4 0.0236BMI (kg/m2) 24.5 ± 4.6 25.3 ± 4.9 P 4 n.s.BSA (m2) 1.58 ± 0.17 1.68 ± 0.15 P 4 0.0025Duration of HT (months) 31.6 ± 41.4 22.5 ± 32.8 P 4 n.s.Duration of men (years) 15.1 ± 8.8 7.1 ± 4.2 P 4 0.0019Number of recurrencesa 0.8 ± 1.4 0.7 ± 1.5 P 4 n.s.FT4 (ng/dl) 3.39 ± 3.49 2.85 ± 2.37 P 4 n.s.FT3 (pg/ml) 15.13 ± 11.57 12.60 ± 9.67 P 4 n.s.TSH (mU/ml) 5.43 ± 20.14 1.79 ± 5.58 P 4 n.s.LS BMD (g/cm2) 0.735 ± 0.113 0.986 ± 0.114 P < 0.0001LS BMD (T-score) −3.062 ± 0.868 −0.915 ± 1.016 P < 0.0001FN BMD (g/cm2) 0.620 ± 0.074 0.812 ± 0.098 P < 0.0001FN BMD (T-score) −2.769 ± 0.681 −0.963 ± 0.937 P < 0.0001
BMI, body mass index; BSA, body surface area; HT, hyperthyroidism; men, menopause; FT4,serum free T4; FT, serum free T3; BMD, bone mineral density; LS, lumbar spine; FN, femoralneck; n.s., not significanta Patients with Graves’ disease
E. Jodar Gimeno et al.: Predicting Bone Disease in Hyperthyroid Patients 371
ration of disease, duration of anti-thyroid treatment, menstrual sta-tus, physical activity, calcium intake, tobacco use) were assessed.After gelatin-free diet, fasting blood samples were collected todetermine serum levels of calcium (sCa), phosphorus (sP), creati-nine (sCr), thyroid hormones, and bone turnover markers. Fastingand 24-hour urine were collected to determine calcium and hy-droxyproline.
Thyroid Function Tests.Serum TSH was measured by RIA (RIA-gnost hTSH, Gif-Sur-Yvette Cedex, France; reference values (RV)0.5–5mU/ml, lower detection limit 0.03mU/ml), also serum freeT3 (FT3, RV: 3.8–8.3 pg/ml) and serum free T4 (FT4, RV: 0.9–2.0ng/dl) (RIA-mat FT3 and RIA-mat FT4, respectively, Byk-SangtecDiagnostica GmgH/Co. KG, Dietzenbach, Germany).
Bone Turnover Markers.Total alkaline phosphatase (TOTALALP;RV: 100–280 Ul/l) measured by autoanalyzer,BONEALP measuredby IRMA (Tandem-R Ostase, Hybritech Europe, Liege, Belgium;RV: males 12.4 ± 4.36mg/l, females 11.6 ± 4.11mg/ml) andosteocalcin by RIA (BGP, Incstar Corporation, Stillwater, MN;RV: 1.8–6.6 ng/ml) were determined as bone formation markers.Serum activity of tartrate-resistant acid phosphatase measured bykinetic assay (TRAP, RV:#7 Ul/l), serum type I collagen C-telopeptide measured by RIA (ICTP, Orion Diagnostica, Espoo,Finland; RV: 1.8–5.0mg/l) and 24-hour urine hydroxyproline ex-cretion adjusted by BSA and fasting hydroxyproline adjusted byurine creatinine measured by Kirivikko’s method (OHP/d/m2 andOHP/Cr, Organon Teknika Boxtel, The Netherlands; RV: 5–17mg/d/m2 and <0.03, respectively) were assayed as bone resorptionmarkers.
BMD Measurements.Bone mass was measured with a dual energyX-ray absorptiometry (DXA) system (Hologic QDR1000, Wal-theim, MA) at the lumbar spine (L2–L4, LS) and femoral neck(FN). Values were expressed as g/cm2 and as score-T. A total of2552 healthy normal subjects (1331 females and 1221 males)served to establish the mean BMD in the healthy Spanish popu-lation and to calculate T-score for each BMD measurement (num-ber of standard deviations of the patient’s value from the mean ofyoung adults). The characteristics of this reference population havebeen described elsewhere [13]. Thein vivo precision has a long-term coefficient of variation (CV) of 2.1% in the spine and 1.8%in the femoral neck.
Patients were placed into two groups based on the WHO cri-teria for postmenopausal osteoporosis [14]. Those patients as-signed to group A had LS and/or FN osteoporosis (T-score# −2.5;n 4 29), while those in group B had no osteoporosis (n 4 98).
Statistical Methods
For the statistical analysis, mean values and standard deviationswere estimated separately for each group (A and B) and differ-ences were tested by Student’st test for unpaired data andx2 asappropriate. In the multivariate analysis, stepwise linear discrimi-nating analysis in two groups (normals and osteoporotics) andstepwise multiple regression for LS and FN BMD (g/cm2) wereperformed with BMDP software (BMDP Statistical Software Inc.,Los Angeles, CA). ROC plots were generated by parametric meth-ods and compared with CLabROC software (C. E. Metz and H. B.Kronman, The University of Chicago, Chicago, IL).
Results
The general characteristics, thyroid function tests, and BMDmeasurements of the hyperthyroid patients with (group A)and without (group B) lumbar and/or femoral osteoporosisare shown in Tables 1 and 2. Of 127 hyperthyroid patients,29 (22.8%) met lumbar and/or femoral osteoporosis criteria(24/127: 18.9% at LS; 22/127: 17.3% at FN). The hyper-thyroid patients with osteoporosis were older and showed
greater proportion of menopausal females; lower weight,height, BSA, and calcium intake; and longer duration ofpostmenopause. No differences in thyroid function testswere found. These patients were also characterized by glob-al damage of axial bone mass (lumbar and femoral). Amongthe bone turnover markers,TOTALALP, BONEALP, andOHP/d/m2 were significantly elevated in the osteoporoticgroup (Figs. 1 and 2).
The stepwise linear discriminating analysis showed thatthe best predictive model for LS and/or FN osteoporosiscomprised age (P < 0.001), OHP/d/m2 (P < 0.001), BMI (P< 0.001),BONEALP (P < 0.001), and calcium intake oncetransformed into a quantitative variable (P < 0.001). Theequation of the model and the correct classification rate(.89%) are shown in Table 3.
Fig. 1. Bone formation markers in hyperthyroid patients (mean ±SEM). (A) Total alkaline phosphatase (TOTALALP); (B) bone al-kaline phosphatase (BONEALP); (C) osteocalcin (BGP). *P < 0.01;** P < 0.001 versus osteoporotic patients. OP: Lumbar and/orfemoral osteoporosis; NO: no osteoporosis. The gray band denotesreference values.
E. Jodar Gimeno et al.: Predicting Bone Disease in Hyperthyroid Patients372
Although significant differences between osteoporoticand nonosteoporotic patients were found only in ALPs andOPH/d/m2 levels, ROC plots were generated for every boneturnover marker (Fig. 3). No differences were found in areaindex (AI) under ROC plots among ALPs and OHP/d/m2
(AI: TOTALALP 4 0.7473;BONEALP 4 0.7601; OHP/d/m2
4 0.6358;P 4 n.s.). Table 4 shows the accuracy of testsover different decision thresholds.
In the multivariate analysis, the best predictive model forlumbar spine BMD (g/cm2) accounted for more than 46% ofthe variability in vertebral bone mass. The most predictivevariable was an artificial one scored 1 in males, 2 in pre-menopausal females, and 3 in menopausal female. Named‘‘sexmen’’ (tEXP 4 47.91 [1:115] d.f.;P < 0.001), thisartificial variable significantly improved the model. Thesecond predictive variable wasBONEALP (tEXP 4 16.96[2:114] d.f.;P < 0.001), both with inverse relationship withlumbar BMD. The next variable with positive relationshipwas calcium intake (tEXP 4 8.81 (3;113) d.f.;P < 0.001).The last variable included was OHP/d/m2 (tEXP 4 7.37(4;112) d.f.;P < 0.001). The equation of the model was (R4 0.6820; R2 4 0.4651):
6 TOTALALP 0.04 4 Dù 0 → No osteoporosis7 Ca intake 4.28 5 (CCR: 91%)
D < 0 → Osteoporosis(CCR: 82%)Global CCR: 88.9%
TOTALALP, total alkaline phosphatase; OHP/d/m2, 24 h urine hy-droxyproline/body surface; BMI, body mass index;BONEALP,bone alkaline phosphatase
E. Jodar Gimeno et al.: Predicting Bone Disease in Hyperthyroid Patients 373
hyperthyroid patients as stated by other authors using simi-lar methodologies [3, 4, 7, 15, 16]. Moreover, 29/127patients (22.8%) met osteoporosis criteria with 22/127(17.3%) having Z-score# −2 (data not shown). Consider-
ing the mean age of our study population (42 ± 16 years)and the low rate of menopausal females (26.7%), the inten-sity and extent of this hyperthyroid bone disease are greatenough to justify a more precise characterization.
Fig. 3. ROC plots for bone formation markers(A), serum bone resorption markers(B), andurine bone resorption markers(C). TOTALALP:Total alkaline phosphatase;BONEALP: bonealkaline phosphatase; BGP: osteocalcin; TRAP:tartrate-resistant acid phosphatase; ICTP: type Icollagen C-terminal telopeptide; OHP/Cr: fastingurine hydroxyproline/creatinine ratio; OHP/d/m2:24 h urine hydroxyproline/body surface.
E. Jodar Gimeno et al.: Predicting Bone Disease in Hyperthyroid Patients374
As expected from genetic, endocrinologic, and lifestylerisk factors for osteoporosis, osteoporotic hyperthyroid pa-tients showed greater age and proportion of menopausalfemales; lower weight, height, BSA, and calcium intake;and longer duration of postmenopause. The lack of differ-ences in thyroid function tests is not surprising because ofthe lack of correlation between thyroid function and BMDshown in several studies [4, 15]. Moreover, there is an epi-demiological association between osteoporosis and priorhyperthyroidism but not with present active hyperthyroid-ism [17]. On the other hand, BGP, TRAP, and ICTP showedno significant differences between osteoporotics and non-osteoporotics. This may be due to low accuracy (TRAP) orhigh dependence on thyroid hormone levels, especially forICTP, which has been suggested as a good indirect markerof thyroid status [18, 19].
In the ROC plots analysis, our data show similar preci-sion for TOTALALP, BONEALP, and OHP/d/m2 levels in thediagnosis of osteoporosis. This agrees with previous obser-vations by electrophoretic [20] and immunoradiometric [7]assays identifying the bone as the major source of ALPs inthe hyperthyroid state, and highlights the high accuracy ofthe classical bone turnover markers (TOTALALP and OHP)in the hyperthyroid bone disease, probably related to thehigh bone turnover observed in this setting.
Christiansen and coworkers [9, 10] proposed a math-ematical equation to estimate the annual rate of bone loss inpostmenopausal osteoporosis. The tests used to make thisassessment comprised serum alkaline phosphatase, fasting
urine calcium/creatinine ratio, fasting urine hydroxyproline/creatinine ratio, and fat mass calculated from height andweight. Analysis of these parameters with ‘‘cutoff’’ tech-nique and with multiple regression analysis yielded identi-cal conclusions: just 20% of the women were correctlystratified. In the prospective study of Slemenda and cowork-ers, the analysis of predictors of the femoral neck bonemineral density (BMD) showed that the best predictivemodel accounted and correctly classified only 65% of theperimenopausal women whose bone mass was in the lowesttertile [11].
In contrast, by means of discriminating analysis, our datashow that a subset of simple and easily obtainable variables(age, calcium intake, BMI, OHP/d/m2, andBONEALP) cor-rectly classify 89% of our hyperthyroid patients with osteo-porosis. Although such multivariate models must not re-place bone mass measurements, these variables can identifythe patients in which the determination of bone mass iswarranted. The high performance of our model in compari-son with those for menopausal osteoporosis may be relatedto the higher turnover demonstrated in hyperthyroidism orto the higher performance of bone turnover markers in thissituation. The model’s high performance may also be re-lated to the greater importance of calcium intake in hyper-thyroid patients, with a daily calcium loss about four timesgreater than normal controls [21] and with long-term alter-ations in calcium homeostasis [22]. Nevertheless, in con-trast with multiple regression analysis, sex and menopausedid not show any discriminating information. Our data, asdata from Wakasugi et al. [4], do not support a preferentialdeleterious effect of hyperthyroidism on postmenopausalwomen.
As suggested by ROC plots, the most discriminatingsingle variable wasTOTALALP, emphasizing its high per-formance in hyperthyroid bone disease. Just with the inclu-sion in the model of the discriminating information pro-vided by age, OHP/d/m2 (again, a classical bone turnovermarker), BMI, andBONEALP, the information supplied byTOTALALP proved to be worthless and then excluded in thestepwise analysis. Taken together, these data suggest thatTOTALALP may be a reasonable screening tool for bone lossin hyperthyroid patients.
The results of multiple regression analysis were similar,both in variables included and in performance. The signifi-cant improvement in our predictive model once included,‘‘sexmen’’ shows the well-known role of sex and meno-pausal status in the maintenance of bone mass. On the otherhand, the same considerations stated before are valid forcalcium intake and OHP/d/m2. As expected from compari-sons between osteoporotics and non-osteoporotics, thyroidhormones and TSH did not supply any predictive informa-tion. Besides, the high performance ofBONEALP in the lum-bar spine is not surprising sinceBONEALP showed the bestprofile in ROC plots for isolated lumbar spine osteoporosis(data not shown). Finally, in the femoral neck the remainderof the predictive variables were related to the well-knowndeterminant role of weight on bone mass [23], the agingprocess, and the calcium balance.
In conclusion, hyperthyroid patients with lumbar and/orfemoral osteoporosis show an easily available clinical andbiochemical profile with potential utility in the managementof hyperthyroid bone disease. Our findings illustrate that therelationship between BMD and bone markers and that thediagnostic performance of bone markers are greater in highturnover states compared to normal or low turnover states.BONEALP shows high accuracy in assessing lumbar bone
Table 4. Accuracy of total and bone alkaline phosphatase and 24h urine hydroxyproline adjusted by body surface area for the di-agnosis of osteoporosis over different decision thresholds
TOTALALP, total alkaline phosphatase;BONEALP, bone alkalinephosphatase; OHP/d/m2, 24 h urine hydroxyproline/body surface
E. Jodar Gimeno et al.: Predicting Bone Disease in Hyperthyroid Patients 375
mineral density in hyperthyroid patients, whereas ICTP hasno predictive value in hyperthyroid osteopenia, probablybecause of its great dependence on thyroid hormone levels.Classical bone turnover markers (TOTALALP, OHP) showhigh performance in the evaluation of hyperthyroid bonedisease.
Acknowledgments.This study was partially supported by the An-dalusian Regional Government through Project no. 183 ‘‘UnidadMetabolica.’’ The authors are grateful to the anonymous refereesfor their helpful comments.
References
1. von Recklinghausen F (1981) Die fibrose oder deformirendeOstitis, die Osteomalazie und die Osteoplastiche Carzinose, inihren gegenseitigen Beziehungen. In: Reimer G (ed) Fest-schrift Rudolf Virchow. Berlin, pp 1–89
2. Melsen F, Mosekilde L (1977) Morphometric and dynamicstudies of bone changes in hyperthyroidism. Acta Path Micro-biol Scand 85:141–150
3. Campos Pastor MM, Mun˜oz-Torres M, Escobar-Jime´nez F,Ruiz de Almodovar N, Jo´dar E (1993) Bone mass in femaleswith different thyroid disorders: influence of menopausal sta-tus. Bone Miner 21:1–8
4. Wakasugi M, Wakao R, Tawata M, Gan N, Koizumi K, OnayaT (1993) Bone mineral density in patients with hyperthyroid-ism measured by dual X-ray absorptiometry. Clin Endocrinol38:283–286
5. Riggs BL, Melton III LJ (1992) The prevention and treatmentof osteoporosis. New Engl J Med 327:620–627
6. Charles P, Poser JW, Mosekilde L, Jensen FT (1985) Estima-tion of bone turnover evaluated by47Ca-kinetics. Efficiency ofserum bone gamma-carboxyglutamic acid-containing protein,serum alkaline phosphatase, and urinary hydroxyproline ex-cretion. J Clin Invest 76:2254–2258
7. Garnero P, Vassy V, Bertholin A, Riou JP, Delmas PD (1994)Markers of bone turnover in hyperthyroidism and the effectsof the treatment. J Clin Endocrinol Metab 78:955–959
8. Spector TD, Edwards AC, Thompson PW (1992) Use of a riskfactor and dietary questionnaire in predicting bone density andsubsequent bone loss at menopause. Ann Rheum Dis 51:1252–1253
9. Christiansen C, Rodbro P, Riis BJ (1987) Prediction of rapidbone loss in postmenopausal women. Lancet 1:1105–1108
10. Christiansen C, Riis BJ, Rodbro P (1990) Screening procedure
for women at risk of developing postmenopausal osteoporosis.Osteoporosis Int 1:35–40
11. Slemenda CW, Hui SL, Longscope C, Wellman H, JohnstonCC (1990) Predictors of bone mass in perimenopausalwomen: a prospective study of clinical data using photon ab-sorptiometry. Ann Intern Med 112:96–101
12. Zweig MH, Campbell G (1993) Receiver-operating character-istic (ROC) plots: a fundamental tool in clinical medicine.Clin Chem 39:561–577
13. Diaz-Curiel M, Castro N (1992) Bone mineral density in theSpanish population measured by dual X-ray absorptiometry(DEXA). (abstract) Bone Miner, 17(suppl 1):133
14. WHO Study Group (1994) Assessment of fracture risk and itsapplication to screening for postmenopausal osteoporosis: re-port of a WHO study group. WHO technical report series;843, Geneva, Switzerland, p 129
15. Lee MS, Kim SY, Lee MC, Cho BY, Lee HK, Koh CS, MinHK (1990) Negative correlation between the change in bonemineral density and serum osteocalcin in patients with hyper-thyroidism. J Clin Endocrinol Metab 70:766–770
16. Diamond T, Vine J, Smart R, Butler P (1994) Thyrotoxic bonedisease in women: a potentially reversible disorder. Ann In-tern Med 120:8–11
17. Cumings SR, Nevitt MC, Browner WS, Stone K, Fox KM,Ensrud KE, Cauley J, Black D, Vogt TM (1995) Risk factorsfor hip fracture in white women. New Engl J Med 332:767–772
18. Garnero P, Gineyts P, Riou JP, Delmas PD (1994) Assessmentof bone resorption with a new marker of collagen degradationin patients with metabolic bone disease. J Clin EndocrinolMetab 3:780–785
19. Jodar E, Munoz-Torres M, Quesada M, Luna JD, Olea N,Escobar-Jime´nez F (1996) Utility of new markers of boneturnover in hyperthyroid patients (Abstract). Osteoporosis Int29(suppl 1):279
20. Cooper DS, Kaplan MM, Gigway EC, Maloof F, Daniel GH(1979) Alkaline phosphatase isoenzyme patterns in hyperthy-roidism. Ann Intern Med 90:164–168
21. Mosekilde L, Ericksen EF, Charles P (1990) Effects of thyroidhormones on bone and mineral metabolism. Endocrinol MetabClin North Am 19:35–63
22. Langdahl BL, Loft AGT, Ericksen EF, Mosekilde L, CharlesP (1996) Bone mass, bone turnover, body composition, andcalcium homeostasis in former hyperthyroid patients treatedby combined medical therapy. Thyroid 6:161–168
23. Kroger H, Tuppurainen M, Honkanen R, Alhava E, SaarikoskiS (1994) Bone mineral density and risk factors for osteopo-rosis: a population-based study of 1600 perimenopausalwomen. Calcif Tissue Int 55:1–7
E. Jodar Gimeno et al.: Predicting Bone Disease in Hyperthyroid Patients376
