ABSTRACT
OBJECTIVES: To identify regions of interest (ROIs) relevant to periarticular osteoporosis in RA with low precision error and sufficient inter-rater reliability and to test diagnostic validity for RA. METHODS: Periarticular BMD was measured using dual-energy X-ray absorptiometry (DXA). Five ROIs were defined around MCP and/or PIP joints II-V, II-IV and mid-metacarpal to mid-phalangeal. They were evaluated for precision using the root mean square coefficient of variation (RMS-CV) and the intra-class correlation coefficient (ICC) for inter-reader reliability. To test validity, established RA patients (n = 25) and early arthritis patients (n = 25) were compared with healthy controls (n = 37) matched on sex, age and menopausal status using paired t-tests, ROC curves and scatterplots. RESULTS: The RMS-CV was 0.45-1.07%. The ICC was 0.99. Mean BMDs of the five ROIs ranged from 0.321 to 0.372 g/cm(2) in established RA, from 0.321 to 0.382 g/cm(2) in early arthritis and from 0.342 to 0.401 g/cm(2) in healthy controls. Mean differences ranged from 0.012 to 0.032 g/cm(2) for established RA and from 0.023 to 0.033 g/cm(2) for early arthritis patients compared with matched controls, with P < 0.05 for ROIs 1-5 in early arthritis and the whole hand in established RA. ROC curves indicated low discriminative power, with an area under the curve (AUC) of 0.61-0.64, and scatterplots showed great overlap between BMD values of patients and controls. CONCLUSIONS: Periarticular BMD measured with DXA seems not to be a useful diagnostic feature due to strong overlap of BMD values between healthy controls, established RA patients and early arthritis patients.