Background/Purpose:

Fluorodeoxyglucose (FDG)-positron emission tomography (PET) allows more direct assessment of inflammation than other imaging modalities by portraying metabolic activity. No standards exist for objective quantification of joint activity using PET, and many techniques rely on subjective initial assessments and use of a sample of voxels such as SUVmax or SUVpeak to characterize an entire region. The aim of this study was to compare the accuracy and reliability of novel methods of assessing global synovitis activity in patients with rheumatoid arthritis (RA) and healthy controls.

Methods:

FDG-PET/CT scans were performed on 19 RA patients and compared to 19 asymptomatic age- and sex-matched control subjects using similar acquisition methods. Several methods of quantification were performed. Method A used a traditional subjective approach to only measure ÒactiveÓ joints and an adaptive thresholding algorithm with a lower threshold defined as 40% of the SUVmax in selected joints. Methods B utilized the average SUVmax in control subjects as a lower threshold to exclude only the expected background uptake (separately for each joint type). For both methods, the sum of the partial volume-corrected mean metabolic volume product (cMVPmean) for all ROIs represented the global synovitis score for each subject. RA patients were compared to controls and correlations with clinical indicators of disease activity were assessed among RA patients.

Results:

Method B demonstrated superior inter-reader reliability (r = 0.982) compared to method A (r = 0.636). Patients with RA had much higher PET global activity scores than controls regardless of method (Figure).  Method B correlated strongly with more objective clinical measures of systemic inflammation and disease activity. Method B also generally correlated more strongly with modified disease activity  (M-DAS28) scores that exclude subjective components, compared to traditional disease activity scores (Table).

Conclusion:

For quantification of synovial inflammation by PET/CT, a method of thresholding that eliminates calculated background activity is more accurate and reliable than methods that depend on the assessment of subjectively active joints using focal measurements. Further studies are needed to determine how these methods of quantification perform in longitudinal assessments to assess course of the disease and the effects of various interventions.