Introduction After its commercial introduction in 1987, dual energy X-ray absorptiometry (DXA) has become a widely available and clinically useful tool in the evaluation and management of adult bone diseases. More recently, its utilization in the pediatric population has rapidly increased. DXA presents unique challenges to pediatric radiologists, and the aim of this article is to review the technique and interpretation of DXA. Special focus is on the issues unique to pediatric DXA emphasizing the need for an expanded role for the pediatric radiologist in the performance and interpretation of DXA examinations. 1 2 3 4 Areal vs. volumetric bone mineral density As in all radiologic studies, DXA relies on the differential absorption of X-rays to differentiate tissues of different radiographic density. In addition, DXA can quantify (in grams) the bone mineral content (BMC) at various body sites. By selecting regions of interest, a bone area (BA) is selected with units of centimeters squared. (The abbreviation BA for bone area should not be mistaken for bone age.) BMD is measured directly for each pixel in the region of interest (ROI) by comparing the X-ray attenuation of that pixel to a reference standard. It has units of grams per centimeter squared. This value is multiplied by the pixel’s area to derive the BMC of that pixel with units of grams. The areas of all the pixels in the ROI are summed to give BA. Thus, BMD = BMC/BA. 5 1 4 Fig. 1 2 5  Technical aspects of DXA performance 6 2 2 2 7 1 8 9 3 Fig. 2 a b c  Table 1 2 2 2 2 Region Pediatric Adult 2 BMC (g) 2 2 BMC (g) 2 Left arm 201 93 0.465 114 71 0.622 Right arm 196 97 0.497 116 76 0.655 Left ribs 79 40 0.507 75 39 0.511 Right ribs 98 51 0.525 92 48 0.525 Thoracic spine 85 48 0.569 83 48 0.524 Lumbar spine 49 28 0.50 33 20 0.599 Pelvis 179 136 0.760 115 95 0.829 Left leg 353 269 0.76 228 199 0.873 Right leg 338 245 0.724 226 187 0.829 Subtotal 1577 1008 0.639 1081 782 0.723 Head 233 367 1.572 233 367 1.572 Total 1810 1375 0.759 1315 1149 0.874 3 10 11 12 3 13 Fig. 3 Lateral thoracic and lumbar spine image from DXA study for vertebral morphology. Note compressive deformities at T-7 and T-9, and a Schmorl’s node at L1, with otherwise normal vertebral morphology of the lumbar spine  14 Accuracy, precision, and least significant change 15 16 14 17 18 2 15 2 19 21 22 23 DXA in newborns 24 28 26 28 25 26 25 29 not 29 Indications for pediatric DXA 3 14 14 30 31 32 Pediatric DXA interpretation 3 loss 3 3 10 19 21 33 21 34 21 34 37 4 19 20 32 38 39 10 34 10 20 36 39 41 36 42 2 43 44 45 Table 2 C/B/H/A/O GA (L) SA Reference Year Scanner No. of patients (M/F) Age range Ethnicity (C/B/H/A/O) Input Output 25 1992 Hologic 1000 PB 29/28 Newborn  GA + weight + height + SA Lumbar BMD and BMC 25 1992 Hologic 1000 PB 22 total 1–24 months  GA + weight + height + SA Lumbar BMD and BMC 29 1996 Hologic 1000 PB 82/68 GA 27–42 weeks  Weight Total BMD, BMC and BA 35 1991 Hologic 1000 PB 84/134 2–17 years 162/56/0/0/0 Weight + Tanner stage Lumbar BMD 56 1990 Lunar DP-3 184 total 5–11 years  Weight Lumbar BMD 57 1993 Norland XR-26 86/68 5–18 years  Gender + Tanner stage Total BMC and % fat 46 1997 Hologic 1000 PB 142/201 4–19 years (L) 343/0/0/0/0 Gender + Tanner stage Total BMD, BMC and BA 58 59 2002 Lunar DPXL/PED 188/256 4–20 years (L) 444/0/0/0/0 Gender + age Lumbar BMD and apparent BMD or total BMC, % fat and lean body mass       Gender + Tanner stage Lumbar BMD and apparent BMD or total BMD and BMC 62 2002 Hologic 4500 FB 107/124 5–22 years 226/0/0/3/2 Gender + age Total BMC and BA       Gender + height Total BA       Gender + total BMC Total BMD 61 2005 Hologic 4500 FB up to 1948 3–20 years  Gender + age  Lumbar, total hip and total BMD 21 1991 Hologic 1000 PB 109/98 9–18 years 207/0/0/0/0 Gender + Tanner stage or age  Lumbar BMC, BA and BMD, femoral neck BMD 22 33 1996 Hologic 2000 FB 110/124 8–17 years (L) 220/0/0/0/0 Gender + age Lumbar and total BMC and BMD        Femoral neck BMC and BMD 40 a 1999 Hologic 1000 PB 193/230 9–25 years (L) 103/114/103/103 Gender + age + ethnicity Lumbar, femoral neck, and total hip BMD and BMAD        Total BMD and BMC/Ht 10 2004 Hologic 4500 FB 0/422 12–18 years 153/264/0/0/5 Age + weight + ethnicity  Lumbar and femoral neck BMD, femoral neck apparent BMD 60 b 2001 Hologic 2000 PB 0/151 9–14 years (L) 151/0/0/0/0 Breast stage + age  Lumbar, femoral neck, trochanter and forearm BMC and BMD 54 2003 Lunar DPX PB 210/249 3–30 years 459/0/0/0/0 Gender + height or age Total lean body mass and total BMC/lean body mass a http://www.stat-class.stanford.edu/pediatric-bones b http://www.bcm.edu/bodycomplab 2 46 12 47 48 Appendix Bone growth over time and changes in BMD 49 50 1.5 51 52 53 54 55 54 In summary, smaller bones will have lower aBMD than larger bones as measured with DXA, even when the vBMD is the same because of limitations of the areal bone density technique. Attempts to correct for or to circumvent this limitation have been proposed and show promise in improving DXA interpretation and our understanding of bone physiology. Clinical case studies The following case studies are examples of common clinical indications for DXA in pediatrics and illustrate the principles used at Columbus Children’s Hospital in DXA interpretation. 2 35 35 43 2 43 2 35 21 2 43 Final recommendations Pediatric radiologists need to assume a more prominent role in DXA performance, interpretation and research. They require expertise with DXA as a laboratory procedure, a numeric result, and a clinical diagnostic examination. The radiologist must be a “clinical pathologist” closely supervising the acquisition of the DXA study with assessment of quality control data as well as the clinical images. The radiologist must be a “statistician” understanding the principles of least significant change and the relationship of standard deviation scores and percentile rankings. The radiologist must also be a “bone specialist” providing a meaningful context for DXA, translating the numeric value to a clinically useful result. As always, pediatric radiologists must be children’s health advocates by ensuring the appropriate clinical use of DXA as well as actively participating in research efforts.