In-Accuracy Of Bone Density Measurements By DXA In Patients With Hemoglobinopathies and Iron Overload

Abstract

IntroductionWhen monitoring bone health in patients with hemoglobinapathies, it is unknown if iron in surrounding tissues can lead to inaccuracies in the 2-dimensional assessment by Dual Energy X-ray Absorptiometry (DXA). ObjectiveThe aims of this study were: 1) to determine if the accuracy of lumbar spine assessment by DXA is affected by high liver iron concentration in patients with Sickle Cell Disease (SCD) or Thalassemia (Thal), 2) to test the effect of high tissue iron on vertebral Z-scores using phantoms, 3) to explore the ability to account for potential high-iron content effects when performing DXA examinations. MethodsThis study consisted of a retrospective chart review of data collected by the Children's Hospital & Research Center Oakland, Bone Density Clinic and Iron Measurement Program. Data from both DXA and Super Conducting Quantum Interference Device (SQUID) examinations collected between 2002 and 2013 from were abstracted. Only those patients with a diagnosis of SCD or Thal, who had a DXA and SQUID measurement within the same year were divided into an iron overload group (liver iron concentration (LIC) >3,000 µg Fe/g wet) and low iron (LIC <500 µg Fe/g wet) group. These patients were compared with healthy controls of which only 13 had both DXA and SQUID tests, 34 had DXA only. The 34 healthy controls without a SQUID test were included because it was assumed, based on their health screen that their liver-iron content would not interfere with DXA. In order to explore aim 1, a lumbar spine scan, by DXA, of each subject was re-analyzed to compare the derived areal bone mineral density (aBMD) Z-scores of lumbar vertebrae that are covered by the liver (presumed L1 or L1/L2) with the Z-scores of the lumbar vertebrae not covered by the liver (L3/L4). To explore aim 2, phantoms were designed to mimic the geometry of iron loaded tissues in order to explore the contribution of iron in specific tissues on the accuracy of DXA assessments. Phantoms were constructed using KNOX® brand gelatin and iron(II) sulfate heptahydrate and had concentrations ranging from 3,000 to 7,000 ug Fe/g gelatin. The iron-loaded phantoms were positioned obtusely overlying L1/L2 of the DXA daily quality control phantom to mimic the position of the liver. All data were analyzed by STATA ver.9.2 and were considered significant with a p<0.05. ResultsData from 102 total visits abstracted from 88 subjects [19 SCD (13 F), 24 Thal (12 F), age: 30.1 ± 11.9 years, mean ± SD], and 45 healthy controls (24 F, age: 25.4 ± 11.0 yrs) were analyzed. The SCD and Thal group had an average LIC by SQUID of 4651 ± 2079 µg Fe/g wet tissue and serum ferritin of 5408±2706 ng/mL; while the healthy controls, with both a DXA and a SQUID (n=17), had an average LIC of 251±144. Average aBMD Z-score of the lumbar spine L1-L4 in the Thal group was -2.0 ± 1.1 , the SCD was -2.0 ± 1.6 and the healthy controls: -0.3 ± 0.9. However, when individual vertebrae are analyzed separately, a significant difference was observed between the lumbar spine L1 BMD Z-scores compared to the combined means of L3/L4 Z-scores in the iron loaded population (Table 1). The discrepancy was even greater in subjects with LIC >5000 ug/g wet tissue. These findings were reproduced using heavily iron loaded phantoms.Table1Summary of Vertebrae Analysis by DXA in Controls, Patients and PhantomsSubject groupL1 aBMD Z-ScoreAvgL3/L4 aBMD Z-scoreMean difference L1 vs AvgL3/L4Difference (L1 vs AvgL3/L4)p-valueControls (n=52)-0.29 ± 0.16-0.50 ± 0.170.21 ± 0.08-0.42 ± 0.12<0.001SCD & Thal (n=50)-1.60 ± 0.18-2.21 ± 0.180.63 ± 0.09Controls (n=52)-0.29 ± 0.16-0.50 ± 0.170.21 ± 0.080.70 ± 0.16<0.001LIC >5000 (n=18)-1.43 ± 0.28-2.35 ± 0.260.91 ± 0.15Control Phantom (n=10)0.76 ± 0.08-0.17 ± 0.070.93 ± 0.03-1.0 ± 0.16< 0.001Iron Phantom (n=10)1.70 ± 0.47-0.22 ± 0.101.92 ± 0.15 ConclusionsInitial results for this study show that there is a relationship between liver iron content and lumbar spine aBMD Z-scores when evaluated by DXA. The BMD Z-score for L1 appears to be more significantly affected by the liver iron content then L2, which was unanticipated. When evaluating patients with liver iron content >3,000 ug/g wet tissue, it is important to consider the effects of iron contribution from the liver on the DXA spine scans and delete L1 and/or L2 from the total Z-score prior to making an interpretation. Failing to do so may under diagnose low bone mass in this at risk patient population. Disclosures:No relevant conflicts of interest to declare.