Background/Purpose: Ectopic deposition of liver fat (steatosis) is associated with insulin resistance (IR), cardiovascular disease (CVD), and is a potent risk factor for cirrhosis.  RA patients are at risk for IR and CVD, and cirrhosis risk is a concern with several RA pharmacotherapies.

Methods: RA patients underwent abdominal computed tomography (CT).  Liver images were analyzed using Slice-o-Matic software, with the average radiographic attenuation [in Hounsfield units (HU)] for five (250 pixels each) sites of liver parenchyma, selected away from hepatic vasculature.   Attenuation was corrected using a tissue phantom of known density.  Steatosis was defined as average hepatic attenuation <48 HU (a standard definition that correlates histologically to ≥30% fatty deposition).

Results: A total of 113 RA patients were studied [mean age 64 ± 8 years, 70% female, median RA duration=13 years, median DAS28=3.1].  Mean hepatic attenuation was 54 ± 9 HU and steatosis was detected in 20 (18%).  Steatosis was unrelated to age, gender, or ethnicity.  Those with steatosis reported fewer minutes of exercise and more minutes of daily television watching, had significantly higher IR and serum triglyceride levels, and lower HDL levels.  All measures of adiposity were higher among those with steatosis.  Current alcohol use, and average and maximum number of alcoholic drinks consumed were not significantly associated with steatosis.   Among RA characteristics, liver HU was, on average, 5 HU lower for those seropositive for RF vs. negative (p=0.003).  CRP was strongly inversely correlated with liver HU, but DAS28 was unassociated.  Current prednisone use>5 mg/day was associated with, on average, 7 HU lower liver density compared to no prednisone use, an association that was maintained after adjustment (Fig A).  Among non-biologics, leflunomide was associated with a higher adjusted liver HU (Fig B), an association not modified by concomitant methotrexate use.   Methotrexate use was only significantly associated with lower liver HU among the group with trunk fat above the median (14.7 kg).  Likewise, the magnitude of the associations of RF, prednisone, CRP, leflunomide, and lean mass with liver HU was greater in the group with higher trunk fat.  Biologics were not associated with liver density in any analysis.  Liver HU was not associated with coronary calcium score, ultrasound measures of carotid atherosclerosis, or ankle brachial index.

Conclusion: No prior studies have explored hepatic attenuation in RA.  The findings of this cross-sectional investigation link body composition, systemic inflammation, immunologic characteristics, and pharmacotherapies to the degree of steatosis.  For most associations, the effects were accentuated in the setting of increased adiposity, suggesting a biologic interaction.  Although strongly associated with insulin resistance, steatosis was unassociated with measures of atherosclerosis.