Figure. Correlation in annualized BMD change rate (g/cm2/y) for DBMD1-2 vsDBMD2-3

398 Abstracts

than the general population. Rheumatoid arthritis (RA) is a chronic inflamma-tory rheumatic disease and a frequent cause of secondary osteoporosis inducedby the chronic inflammatory condition and a long-term glucocorticoid therapy(GC). The aim of this study is to evaluate the influence of GC on the trabecularbone score (TBS), BMD and TBS dynamics during one year in patientswith RA.Methods: 134 examined women with RA (age 52.5�12.8 years; height 162.6�6.4cm, weight 68.2�13.7 kg, duration of disease 9.1�7.5 years, duration of postmeno-pausal period 7.6�7.2 years) were divided into three groups: first group, G1e37 pa-tients who did not use GC; second group, G2e 50 patients who used GC in a dose ofO5 mg of prednisolone for more than 3 years; third group, G3 e 47 patients whotook GC only at the exacerbated stage for!6 months. All the patients had been tak-ingmethotrexate as a basic treatment. BMD of total body, PA lumbar spine, proximalfemur and forearm were measured using the DXA method (Prodigy, GEHC Lunar,Madison,WI, USA) and PA spine TBSwas assessed bymeans of TBS iNsight� soft-ware package installed on our DXA machine (Med-Imaps, Pessac, France). Evalua-tion of TBS dynamics in the patients of G1 and G2 groups during the year wasconducted on the background of ongoing therapy which included doses of GC (forthe patients of second group) and/or without any osteotropic treatment.Results: The 3 groups did not differ as to age, basic anthropometric parameters,duration of disease and duration of postmenopausal period in these groups. TBSin G2 was significantly lower compared to G1 (TBSL1-L4: 1.147�0.168 vs.1.250�0.135; t5-3.07; p50.003), and G3 compared to G1 (TBS L1-L4:1.274�0.138; t53.95; p50.0002). However, there were no differences in BMDof PA spine and hip among groups. Only forearm BMD in the second groupwas significantly lower compared to the first one (0.583�0.176 g/cm2 vs.0.675�0.229 g/cm2; t5-2.18; p50.032). Spine TBS decreased by 1.4% afterone year for G1 and by 5.8% for G2.Conclusion: For patients who are GC users, TBS, but not BMD, reflects bone mi-croarchitecture deterioration which is an indicator for those patients to of a highervertebrae and nonvertebral risk of fracture. TBS is a determinant of bone state andmust be monitored during the long-term treatment with GC.Disclosure of Interest: None Declared

O04

WHY RATE OF BMD LOSS DOES NOT PREDICTFRACTURE RISK: THE ‘‘YOGI BERRA’’ PRINCIPLE

W. D. Leslie1,*, S. R. Majumdar2, S. N. Morin3, L. M. Lix1; 1University ofManitoba, Winnipeg, 2University of Alberta, Edmonton, 3McGill University,Montreal, Canada

Aims: Intuitively, rapid BMD loss should predict fractures independent of one’scurrent BMD. However, in 4498 untreated older women undergoing clinical moni-toring we showed that rate of BMD loss estimated from two BMD tests (mean in-terval 3 years) was not a significant risk factor for major osteoporotic fractures(MOF) [1]. To explain this paradox, we hypothesized that rate of BMD loss mightbe too variable to be useful for fracture risk assessment. The current analysis testedwhether past BMD loss predicts future BMD loss.Methods: We identified women age 50+ years in the Manitoba BMD Registry withthree femoral neck BMD tests performed on the same scanner (Prodigy, GE/Lunar)and a minimum of one year between scans. We excluded women with prior prescrip-tions for medications that might affect BMD: systemic estrogen, osteoporosis treat-ment, glucocorticoids or aromatase inhibitors. Rates of BMD loss between the 1st and2nd visits (DBMD1-2) and between the 2nd and 3rd visits (DBMD2-3) were expressedas: (a) annualized loss rate (g/cm2 per year), (b) tertiles of annualized loss rate, and(c) categories of loss (none, significant increase, significant decrease) based on site-specific least significant change (LSC). Pearson correlation coefficients and analysisof variance (ANOVA) were applied to the data.Results: 542 women mean (SD) age at baseline 62 (8) years had 3 femoral neckBMD measurements separated by mean 3.5 years (DBMD1-2) and 3.4 years(DBMD2-3) of which 540 had 3 total hip measurements and 446 had 3 total spinemeasurements. Femoral neck BMD loss was -0.5% (1.6%) per year for DBMD1-2and -0.6% (1.5%) per year for DBMD2-3, with loss exceeding the LSC in 27% and17%, respectively. There was no significant correlation between DBMD1-2 andDBMD2-3 for total hip (r50.01, p50.74) or total spine (r5-0.01, p50.77), whilea weak negative correlation was seen for the femoral neck (r5-0.11, p50.01) (seeFigure). Whether rapid BMD loss was defined as tertile of annualized loss or ac-cording to LSC, DBMD1-2 was not significantly associated with DBMD2-3. AMonte Carlo simulation (N55000) using inputs from the cohort showed theimportance of BMD precision error, and the competition between correlation inBMD loss rates and ‘‘regression to the mean’’. Using short-term precision errorsfrom our program, the predicted correlations between DBMD0-1 and DBMD1-2were r50.08 for femoral neck, r50.08 for total spine and r50.39 for total hipBMD (explained variance !20%). To explain 50% of the variation of DBMD2-

Journal of Clinical Densitometry: Assessment & Management of Muscu

3 from DBMD0-1 required BMD precision error below 0.008 g/cm2 or an intervalbetween BMD tests greater than 5 years.

Conclusion: The weak correlation between past and future BMD loss supportsprevious findings that assessment of BMD change rate is unlikely to be a clinicallyuseful measure. In the words of Yogi Berra, ‘‘It’s tough to make predictions, espe-cially about the future.’’References:1. Leslie WD et al, J Clin Endocrinol Metab 2012;97:1211.Acknowledgement: HIPC file no. 2003/2004-26Disclosure of Interest: None Declared

O05

ABDOMINAL ADIPOSE TISSUE DISTRIBUTION ANDINSULIN RESISTANCE IN ADOLESCENTS e PENN STATECHILD COHORT (PSCC) STUDY

M. D. Sawyer1,*, E. O. Bixler2, F. He1, S. Rodriguez-Colon1,J. Fernandez-Mendoza2, D. Liao1; 1Public Health Sciences, 2Psychiatry, PennState College of Medicine, Hershey, United States

Aims: To investigate the association between abdominal adipose tissue distribu-tion and insulin resistance in a population-based sample of adolescentsMethods: We used available data from 400 adolescents who completed the followup examinations in the PSCC study. We used whole body low-dose DXA to assessand calculate the following abdominal and body fat distribution measures: android/gynoid fat ratio, android/whole body fat ratio, gynoid/whole body fat ratio, visceralfat area, and subcutaneous fat area in the abdominal region. Insulin resistance wasassessed by using homeostatic model assessment of insulin resistance (HOMA-IR).HOMA-IR was log-transformed (log-HOMA) to achieve normal distribution andsubsequently used in the data analysis. We used linear regression models to assessthe association between DXA measures and log-HOMA.Results: The mean (SD) age of the study population was 16.88 (2.19) yrs., with54% male and 77% white. There was no significant difference in age and race dis-tribution between male and female. The mean (SD) of log-HOMA was -0.22(0.79). The overall and gender-specific distribution of DXAvariables are presentedin Table 1. Regression coefficients on log-HOMA per one SD increase in abdom-inal adipose tissue measures are shown in Table 2. In summary, higher abdominaladipose tissue, from both traditional android/gynoid fat and newly developedvisceral area specific measures, is associated with higher HOMA.

Table 1. The distribution of DXA measures by gender.

Mean SD Median IQR 95th percentile

loskeletal Health

Volum

Android/Gynoidfat ratio

All

0.36 0.11 0.34 0.13 0.58

Male

0.38 0.11 0.35 0.13 0.60 Female 0.33 0.10 0.32 0.13 0.52

Android/whole bodyfat ratio

All

0.06 0.02 0.06 0.02 0.09

Male

0.06 0.02 0.06 0.02 0.09 Female 0.06 0.01 0.06 0.02 0.09

Gynoid/whole bodyfat ratio

All

0.18 0.02 0.18 0.03 0.22

Male

0.17 0.02 0.17 0.03 0.20 Female 0.19 0.02 0.19 0.03 0.23

Visceral fat area

All 60.33 39.59 47.79 37.74 140.33 Male 62.36 36.65 48.65 35.54 133.87 Female 57.96 42.77 43.67 45.18 149.62

Subcutaneous fat area

All 217.45 157.70 177.28 211.75 519.08 Male 156.33 138.81 95.91 161.19 458.13 Female 289.22 148.34 256.28 198.99 602.67

e 17, 2014