Study population
Three hundred participants in the Nor-Hand cohort underwent ultrasound and radiographs of both hands. Among those, 246 participants performed MRI of the dominant hand with gadolinium contrast, and 253 participants performed FOI. One adverse event was reported due to subcutaneous administration of ICG, and the FOI images from this participant were excluded from further analyses. Finally, FOI images from two participants were excluded due to lack of contrast enhancement. In total 221 participants performed both FOI and MRI and were included for further analyses. The majority of participants were women, and a wide range in symptom severity, degree of inflammation and structural damage was observed (Table 1).
Frequency distribution of synovitis according to FOI, MRI and ultrasound
For GS synovitis and PD activity 27 joints were missing due to amputation, trapeziectomy, arthrodesis or unknown reasons. Five joints were missing due to trapeziectomy, arthrodesis or amputation on MRI of the dominant hand. One Phase 1 image, seven Phase 2 images and eight Phase 3 images were excluded from analyses due to difficulties defining phases, i.e. no clear descending of the white from fingertips (phase 1) and white (phase 2) or red (phase 3) pixels persisting in fingertips.
None of the participants demonstrated FOI enhancement of the CMC-1 joint, while 70% of CMC-1 joints demonstrated MRI-defined synovitis, 26% GS synovitis and 19% PD activity (Figure 2). Hence, the CMC-1 joint was not included in further analyses. Only three MCP1 joints showed any FOI enhancement, and MRI was the only modality showing frequent findings in the MCP joints (32% of joints, predominantly grade 1). While MRI and FOI (PVM and phases 2 and 3) detected more synovitis and enhancement in the PIP joints than in the DIP joints, GS synovitis and PD activity and FOI phase 1 demonstrated more activity in the DIP joints.
None of the participants demonstrated extensor tenosynovitis by MRI. Fifty-three participants had flexor tenosynovitis in one or more fingers, and the majority (n = 46) demonstrated grade 1 tenosynovitis adjacent to the MCP joint. Flexor tenosynovitis was not included in further analysis due to its localization on the palmar aspect of the hand and thus not detectable via FOI. When assessing frequency of FOI enhancement in PVM according to VV and KL scores, we found a significant trend for higher proportion of joints with FOI enhancement in joints with severe KL and VV grades (Online supplemental figure 1).
Correlations between FOI, ultrasound and MRI
Good correlations were found between MRI and GS synovitis for all joint groups except in the MCP joints (table 2). Similarly, GS synovitis and PD activity demonstrated good to very good correlations for all joint groups. Overall, the correlations between FOI and MRI were poor to fair, while FOI was poorly correlated with GS synovitis. The strongest correlation with MRI was found for PVM in the PIP joints with Spearman’s rho of 0.32, while the DIP joints had consistently the weakest correlations ranging from 0.00 to 0.14 (Table 2, Figure 3).
Diagnostic performance of FOI measuring synovitis
Using MRI and GS synovitis as reference, FOI Phase 1 demonstrated the highest specificity, with corresponding very low sensitivity (Table 3). FOI PVM and Phase 2 had consistently the highest sensitivities with both MRI and GS synovitis as reference, with values ranging from 48% to 69%. FOI reached high NPV with GS synovitis as reference, suggesting that joints with no FOI enhancement were unlikely to have GS synovitis. However, presence of FOI enhancement did not consistently correspond with presence of GS synovitis, demonstrated by low PPV values. GS synovitis was less prevalent than MRI synovitis, which affected the results considerably. Using MRI instead of ultrasound as reference, FOI demonstrated higher PPV and lower NPV. However, improvement of sensitivity, specificity and AUC was found for FOI when presence of MRI synovitis was increased to grade 2 or more (Online supplement table 1, Online supplement figure 2). The agreement between FOI (enhancement yes/no) and MRI (synovitis yes/no) ranged from 53 to 61% while the same values for ultrasound (synovitis yes/no) ranged from 57 to 89%. Using PD activity as reference, the diagnostic performance of FOI was similar to the results when GS synovitis was used as reference (data not shown).
Results from subgroup analysis
Correlation analyses were repeated for participants with erosive hand OA without consistent improvements in the correlations between FOI, MRI and GS synovitis. Further, the diagnostic performance of FOI measuring synovitis with MRI and ultrasound as reference was similar in erosive hand OA and non-erosive hand OA patients (data not shown).