JOURNAL OF CLINICAL DENSITOMETRY THE OFFICIAL JOURNAL OF THE INTERNATIONAL SOCIETY FOR CLINICAL DENSITOMETRY EDITOR-IN-CHIEF CLIFFORD J. ROSEN, MD HUMANA PRESS VOLUME 5 NUMBER 4 WINTER 2002 ISSN: 1094 6950
Journal of Clinical Densitometry, vol. 5, no. 4, 343 353, Winter 2002 Copyright 2002 by Humana Press Inc. All rights of any nature whatsoever reserved. 1094-6950/02/5:343 353/$20.00 Original Article Centile Curves for Bone Densitometry Measurements in Healthy Males and Females Ages 5 22 Yr Teresa L. Binkley, MS, 1 Bonny L. Specker, PHD, 1 and Timothy A. Wittig, PHD 2 1 Ethel Austin Martin Program in Human Nutrition and 2 Department of Mathematics and Statistics, South Dakota State University, Brookings, SD Abstract Normative pediatric bone measurement data are necessary for defining osteopenia in children and for identifying factors associated with normal bone growth. The LMS statistical method is used to produce centile curves plotting a growth characteristic as a function of age. The purpose of this study was to provide centile curves of bone measurements using peripheral quantitative computed tomography (pqct) and dual X-ray absorptiometry (DXA) in 231 (107 male) healthy individuals ages 5 22 yr using the LMS method. pqct (Norland XCT 2000; Norland, Ft. Atkinson, WI) was used to image a single slice at the 20% distal tibia. Periosteal circumference, endosteal circumference, and cortical density measurements were used to obtain centile curves. Whole-body DXA (Hologic QDR 4500; Hologic, Bedford, MA) was obtained and scans were analyzed using adult whole-body software for total body bone mineral content (BMC) and total body bone area. pqct measurements showed prepubertal expansion of the tibia that plateaued in females at age 14 and continued in males until age 18. Tibia cortical density increased during the age of puberty more gradually in females than in males. DXA measurement curves showed that total body BMC and total body bone area plateaued in females at approximately age 15, whereas male curves of the same measurements showed a continued increase. Key Words: Reference curve; pediatric bone densitometry; peripheral quantitative computed tomography; dual X-ray absorptiometry. Introduction Osteopenia and osteoporosis are possible complications of several childhood diseases including chronic rheumatic diseases and asthma. The use of bone densitometry to predict osteoporotic risk in these cases would be beneficial. Normative data for children, and in particular reference curves for bone Received 11/28/01; Revised 02/15/02; Accepted 02/25/02. Address correspondence to Teresa L. Binkley, EAM Building, Box 2204, South Dakota State University, Brookings, SD 57007. E-mail: Teresa_Binkley@sdstate.edu 343 measurements using peripheral quantitative computed tomography (pqct) and dual X-ray absorptiometry (DXA), would be useful for physicians treating these patients as well as for research investigators. The purpose of the present study was to provide centile curves of bone measurements using pqct (Norland XCT 2000; Norland, Ft. Atkinson, Wl) and DXA (Hologic QDR 4500; Hologic, Bedford, MA) in healthy males and females ages 5 22 yr. Reference curves can be generated by plotting growth characteristics as a function of age showing the median as well as other centiles using the LMS method (1). Named after the variables used in the procedure, this
344 Binkley et al. Table 1 Mean (SD) Age, Weight, and Height by Gender and Age Group Age group (yr) N Age (yr) Weight (kg) Height (cm) Females 5 6 12 6.3 22.9 (4.2) 119.0 (7.1) 7 8 22 7.8 28.4 (7.3) 128.0 (7.1) 9 10 31 10.1 38.3 (11.8) 143.0 (10.5) 11 12 16 11.8 43.9 (13.1) 151.0 (10.3) 13 14 9 13.9 57.9 (10.4) 165.0 (5.2) 15 16 19 16.2 57.8 (5.7) 167.0 (6.6) 17 18 10 17.7 57.6 (8.9) 164.5 (8.2) 19 22 5 20.4 62.7 (6.9) 166.0 (7.1) Males 5 6 9 6.3 23.5 (3.9) 121.5 (6.2) 7 8 22 8.1 31.7 (5.9) 133.0 (6.3) 9 10 21 9.9 35.8 (7.7) 143.5 (7.3) 11 12 17 11.7 44.9 (10.1) 152.5 (9.7) 13 14 14 13.9 55.4 (8.6) 164.5 (8.3) 15 16 13 16.0 69.6 (12.7) 178.0 (8.5) 17 18 9 17.9 78.0 (17.8) 180.0 (6.4) 19 20 2 19.4 94.0 (18.9) 189.0 (1.4) method uses penalized likelihood to fit three curves the median (M), the coefficient of variation (S), and the skewness as a Box-Cox Power (L) as cubic splines using nonlinear regression techniques. To our knowledge, there are no previously published reference curves for pqct data from the 20% distal tibia site. Although DXA reference curves have been published for this age group (2 4), we are not aware of any that have used the LMS method for data acquired using a fan-beam scanner. Materials and Methods The study protocol was approved by the South Dakota State University Human Subjects Review Board and written informed consent was obtained from participants or participants parents. Subjects included 231 (107 male) healthy males and females ages 5 22 yr (mean: 11.6 yr). Participants described their race as white (n = 226), Asian (n = 3), or Native American (n = 2). Individuals were recruited from a convenience sample taken in a midwest town (population 20,000), a midwest rural elementary school (118 students enrolled in grades K 6), and a midwest rural high school (57 students enrolled in grades 9 12). Materials explaining the study were sent home with the students. Approximately 68% of the elementary students and 55% of the high school students returned the material and participated in the study. Because of privacy issues, no data were collected on students who did not wish to participate. None of the participants had known chronic diseases that affect bone growth or mineralization. Questionnaires, completed by parents of younger children, were used to obtain information on date of birth, ethnicity, and brief health history. Eighteen percent of the participants took vitamin supplements regularly. Of these, 56% stated that the supplements contained vitamin D, 17% stated that they did not contain vitamin D, and 27% were unsure if they contained vitamin D. Mineral supplements were taken regularly by 6%. Of these, 21% were unsure of the amount of calcium in the supplement, and the rest of the group reported supplements containing 50 762 mg of calcium (mean: 379 mg). A modified food frequency questionnaire (5) to target high-calcium foods showed that the mean calcium intake was 1597 (SD = 693) mg/d. Height without shoes was measured to the nearest 0.5 cm, and weight in light clothing was measured by digital scale to the nearest 0.1 kg. Bone density measurements were made using pqct and DXA densitometry. pqct was used to
Centile Curves for Bone Densitometry 345 Fig. 1. pqct tibia periosteal circumference (mm) by age centile curves for females (top) and males (bottom) showing 5th, 50th, and 95th centiles. measure a single slice (no scout view) at the 20% distal tibia site. Our coefficients of variation (CVs) for pqct in children 6 14 yr of age are 1.4, 1.9, and 0.48% for periosteal circumference, cortical area, and cortical density, respectively. The pqct scan has an effective radiation dose of <10 µsv. Settings for pqct acquisition and analysis are described elsewhere (6). Whole-body DXA was obtained on all subjects and analyzed for total body bone mineral content (TBBMC) and total body bone area (TBBA). Adult software was chosen for analysis since the age range of the study was 5 22 yr and pediatric software is suggested only for ages 4 12 yr. The manufacturer s CV for TBBMC and our CV using a
346 Binkley et al. Fig. 2. pqct tibia endosteal circumference (mm) by age centile curves for females (top) and males (bottom) showing 5th, 50th, and 95th centiles. phantom are <1%. Effective dose for the whole-body DXA is <10 µsv. Weight-for-age and height-for-age Z-scores were obtained using EPIINFO (7). All centile curves were generated using LMS software provided by Dr. Tim Cole of the Institute of Child Health, London, UK. Model selection for the best fit centile curve was obtained as suggested previously (1). Results Age, weight, and height are given in Table 1. The mean height-for-age Z-score of the study group was 0.47 for females and 0.80 for males (value > 0 at p < 0.05 for both). The mean weightfor-age Z-score was 0.42 for females and 0.72 for males (value > 0 at p < 0.05 for both). Body mass
Centile Curves for Bone Densitometry 347 Fig. 3. pqct tibia cortical area (mm 2 ) by age centile curves for females (top) and males (bottom) showing 5th, 50th, and 95th centiles. index (BMI) was plotted on gender-specific BMIfor-age percentile charts (8) (not shown) and indicated that 7.4% of the females and 7.7% of the males were above the 95th percentile, while 2.5% of the females and 5.8% of the males were below the 5th percentile. Gender-specific centile curves for tibial pqct measurements by age were generated for periosteal circumference (Fig. 1), endosteal circumference (Fig. 2), cortical area (Fig. 3), and cortical density (Fig. 4). Because of poor scan quality, 37 scans (21 male) were not included. Cortical density by age curve was generated on a subset of participants (76 female, 62 male) with a cortical thickness of >2 mm owing to decreased density readings when the cortical shell is <2 mm thick (6).
348 Binkley et al. Fig. 4. pqct tibia cortical density (mg/cm 3 ) by age centile curves for females (top) and males (bottom) showing 5th, 50th, and 95th centiles. Centile curves for tibial measurements show that periosteal circumference increased gradually and plateaued by age 14 in females. The endosteal circumference increased in females until approx age 9, plateaued until age 15, then decreased until age 17, when a second plateau occurred. The combination of periosteal gain and endocortical contraction in females after the age of 15 resulted in an increase in cortical area until approx age 16. In males, there was no decrease or contraction of the endosteal surface, as noted in the female curve, but cortical area continued to increase in males owing to periosteal
Centile Curves for Bone Densitometry 349 Fig. 5. centiles. DXA TBBA (cm 2 ) by age centile curves for females (top) and males (bottom) showing 5th, 50th, and 95th expansion throughout the ages studied. Unlike the gradual increase in cortical density seen in females, cortical density in males remained constant until approx age 14 and then increased sharply to plateau at age 18. Gender-specific centile curves by age were created for TBBA (Fig. 5) and TBBMC (Fig. 6). Gender-specific centile curves also were obtained for TBBA by height (Fig. 7) and TBBMC by TBBA (Fig. 8). Discussion The centile curve pattern that we generated for tibial cortical density by age is similar to reference data published for the ultradistal radius (9,10). In males,
350 Binkley et al. Fig. 6. centiles. DXA TBBMC (g) by age centile curves for females (top) and males (bottom) showing 5th, 50th, and 95th we noted an increase in cortical density between the age of 14 and 18 that was not seen in females. A similar gender difference has been reported at the ultradistal radius for total bone mineral density (BMD) but not for trabecular BMD (9,10). In females, by age 14 the periosteal circumference curve plateaued, while the endosteal circumference began to contract by age 15 to 16. The contraction of the endosteal surface has been attributed to estrogen effects (11,12). Schiessl et al. (13) hypothesized that increased estrogen secretion at puberty results in females making more bone than they did before puberty relative to the mechanical loads on their bones. The increase in estrogen would decrease the remodeling-dependent bone losses, while modelingdependent additions of bone would continue nor-
Centile Curves for Bone Densitometry 351 Fig. 7. centiles. DXA TBBA (cm 3 ) by height centile curves for females (top) and males (bottom) showing 5th, 50th, and 95th mally. This could be seen as a decrease in the endosteal circumference and an increase in periosteal expansion. Muscle mass also would be increasing, and as the bones become stronger, there would be a reduction in the strains to the remodeling threshold, thus resulting in a plateau in cortical bone area. In males, the periosteal surface continued to increase up to age 20 while the endosteal circumference began to flatten around age 17 but showed no decrease in circumference. This reflects the effect of androgens on periosteal gain (11,12) and explains the increase in cortical area in males throughout the ages studied. These results also can be viewed in terms of the hypothesis proposed by Schiessl et al. (13). Because males would not be exposed to the same levels of estrogen as females, they would not make more bone relative to the mechanical loads on their bones. The decrease in remodeling-dependent
352 Binkley et al. Fig. 8. centiles. DXA TBBMC (g) by TBBA centile curves for females (top) and males (bottom) showing 5th, 50th, and 95th bone loss that is seen with increased estrogen would not be present and endosteal expansion would continue. Muscle mass also would be increasing, and there would be continual strains to the remodeling threshold until the bone was large enough to reduce the strains to the threshold. Normative DXA data on healthy children (2 4) show patterns similar to the TBBMC by the age centile curves that we generated using the LMS method. Centile curves for height by age, TBBA by height, and TBBMC by TBBA are presented. These curves help indicate whether bones are short, narrow, or light for a particular age as suggested by Molgaard et al. (2). Although our study population was taller and heavier than normative growth data published by the Centers for Disease Control, the TBBA by height
Centile Curves for Bone Densitometry 353 and TBBMC by TBBA curves should be applicable to other populations. We have presented pqct centile curves for the 20% distal tibia site and for DXA bone parameters in healthy children. These results may be helpful to researchers and clinicians working with pediatric populations. Acknowledgments We wish to thank the participants, parents, and school personnel who helped organize schedules and worked with us to conduct the study. References 1. Cole TJ, Green PJ. 1992 Smoothing reference centile curves: the LMS method and penalized likelihood. Stat Med 11:1305 1319. 2. Molgaard C, Thomsen BL, Prentice A, Cole TJ, Michaelsen KF. 1999 Whole body bone mineral content in healthy children and adolescents. Arch Dis Child 76:9 15. 3. Zanchetta JR, Plotkin H, Alvarez Filgueira ML. 1995 Bone mass in children: normative values for the 2 20 year old population. Bone 16:393S 399S. 4. Faulkner RA, Bailey DA, Drinkwater DT, McKay HA, Arnold C, Wilkinson AA. 1996 Bone densitometry in Canadian children 8 17 years of age. Calcif Tissue Int 59:344 351. 5. Cummings SR, Block G, McHenry K, Baron RB. 1987 Evaluation of two food frequency methods of measuring dietary calcium intake. Am J Epidemiol 126:796 802. 6. Binkley TL, Specker B. 2000 pqct measurement of bone parameters in young children. J Clin Densitom 3:9 14. 7. Dean AG, Arner TG, Sangam S, et al. 2000 Epi Info 2000, a database and statistics program for public health professionals for use on Windows 95, 98, NT, and 2000 computers. Atlanta: Centers for Disease Control and Prevention. 8. Kuczmarski RJ, Ogden CL, Grummer-Strawn LM. 2000 CDC growth charts: United States. Advance data from vital and health statistics; no. 314. Hyattsville, MD: National Center for Health Statistics. 9. Neu C, Manz F, Rauch F, Merkel A, Schoenau E. 2001 Bone densities and bone size at the distal radius in healthy children and adolescents: a study using peripheral quantitiative computed tomography. Bone 28:227 232. 10. Lettgen B. 1996 Peripheral Quantitative Computed Tomography: Reference Data and Clinical Experiences in Chronic Diseases. In: Pediatric Osteology: New Developments in Diagnostics and Therapy. Schoenau E, ed. Elsevier Science: Essen, Germany, 123 133. 11. Parfitt AM. 1994 The two faces of growth: benefits and risks to bone integrity. Osteoporos Int 4:382 398. 12. Garn SM, 1970 The earlier gain and the later loss of cortical bone. Springfield, IL: Thomas Books. 13. Schiessl H, Frost HM, Jee WSS. 1998 Estrogen and bonemuscle strength and mass relationships. Bone 22:1 6.