Comparison of ultrasound and X-ray absorptiometry bone measurements in a case control study of female rheumatoid arthritis patients and randomly selected subjects in the population

Abstract

To compare quantitative ultrasound (QUS) and dual-energy X-ray absorptiometry (DXA) bone measurements in female rheumatoid arthritis (RA) patients and controls were randomly selected from the population; secondly, to examine disease and demographic factors associated with these bone measurements. In a total of 115 RA patients (mean age 63.0 years) and 115 age- and gender-matched controls demographic and clinical variables were collected and heel QUS parameters [speed of sound (SOS), broadband ultrasound attenuation (BUA) and stiffness index (SI)] as well as DXA bone mineral density (BMD) at spine and hip were measured. The differences in QUS and DXA measurements between RA patients and controls were tested both on a group and on an individual level. Univariate and multivariate statistical tests were applied to explore for associations to the bone measurements. In the RA patients mean disease duration was 16.6 years, erythrocyte sedimentation rate 23.6 mm/h, M-HAQ 1.68, 28-swollen joint count 7.7, 18-deformed joint count 4.5, 50.0% were rheumatoid factor (RF) positive and 44.2% were current users of prednisolone. All bone measurements were reduced in RA patients compared with controls (SOS 1.9%, BUA 9.4%, SI 19.5%, femoral neck BMD 7.4%, total hip BMD 7.5%, spine L2-L4 BMD -3.0%). Only at spine was the BMD reduction not statistically significant ( P=0.21). In the subgroup of never users of prednisolone SOS was decreased by 1.4%, BUA by 3.7%, SI by 11.0, femoral neck BMD by 2.7%, and total hip BMD by 0.6%, whereas for spine L2-L4 BMD was increased by 4.3% and only for SOS and SI was the decrease statistically significant. The QUS discriminated better than DXA between patients and controls on a group level, but this difference in favor of QUS disappeared on an individual level when the measurement errors were taken into account. Age, BMI, RF and deformed joint count, but not corticosteroids, were independently associated with at least one of the QUS and one of the DXA measures; however, the association between disease-related variables was stronger with the QUS bone measures than with the DXA bone measures. The results for the quantitative QUS bone measures seem to mainly reflect bone mass. Disease-related variables in multivariate analysis remained independently associated with all QUS measures even when adjusting for DXA bone measures. Further studies are needed to examine if QUS may reflect other aspects than bone mass and be a potential better predictor for fracture risk in RA and corticosteroid-induced osteoporosis.