Background/Purpose:

Juvenile
Idiopathic Arthritis (JIA) commonly affects the temporomandibular
joints (TMJ) and may cause growth disturbance, functional limitation and facial
deformity. Early diagnosis and management may minimize these complications. History
and clinical examination is of limited value in diagnosing TMJ arthritis. Magnetic
resonance imaging with gadolinium is the most sensitive and specific diagnostic
tool. Most centers provide only qualitative assessment of TMJ synovitis. 
The purpose of this study was to develop a technique to quantify synovial
enhancement and to apply this to MRIs of TMJs affected by JIA and controls in order to
establish a disease threshold, sensitivity and specificity for synovial
enhancement.

Methods:

This
is a case-control study of children (<16 years) who had MRIs with gadolinium
that included the TMJs at Boston Children’s Hospital or Massachusetts General
Hospital from 2006-2015.  Subjects were included in the ‘JIA group’ if
they had a pre-existing diagnosis of JIA and had a subjective assessment of synovitis by the reading radiologist.  The ‘control
group’ included subjects without JIA who had an MRI for other reasons (e.g.
hearing loss or visual disturbance). Coronal slices of a T1-weighted,
gadolinium-enhanced MRI were used to assess the ratio of signal intensity of a
0.2mm² region of interest (ROI) within the TM joint spaces to a
50mm² ROI within the longus-capitus
muscle, which controls for time after gadolinium infusion. Two raters carried out independent evaluations. A receiver
operating characteristic (ROC) curve was created to determine the sensitivity
and specificity of the ratios of enhancement (Figure 1). Independent 2-sample t test (2-tailed)
was performed for each group with unequal variance assumed. Statistical
significance was set at p<0.05. Intra and inter-rater reliability was assessed using
Intra-class Correlation Coefficient (ICC).

Results:

68 subjects
(79% female, mean age 15.3 ± 2.5 yrs) with 112 MRIs
with clear visualization of 187 TMJs were included in the JIA group. The
control group consisted of 141 subjects (56% female, mean age 11.6 ± 3.5 yrs) with 159 MRIs with clear visualization of 311 TMJs.
The mean signal intensity ratio in the JIA group was 3.03 ± 1.4 compared to
1.23 ± 0.16 in the control group (p<0.001). The resulting ICC was >0.8 indicating that raters had a high degree
of agreement.
The ROC analysis indicated a sensitivity of 91.4% and specificity of 98.1% in
detecting synovitis with a signal intensity ratio
cutoff of 1.66. The area under the curve was 0.978 (95% CI 0.964-0.993,
p<0.001), and the maximum Youden J-index was 0.98.

Conclusion:

In conclusion, we
present a reliable method to quantify TMJ synovial enhancement in MRIs with
gadolinium that controls for time after contrast infusion. A signal intensity
ratio of 1.66 discriminates TMJs affected by JIA from unaffected controls with
a sensitivity of 91.4% and specificity of 98.1%.