Our study shows that adding subclavian arteries examination into the Southend Halo Score(3), as proposed in the modified Halo Score(4), does not improve the diagnostic accuracy of the former version for the US scores. However, the modified Halo Score shows better correlation with markers of inflammation in LV-GCA patients.
US has shown high sensitivity and specificity for diagnosing GCA(6–9) and therefore recent EULAR recommendations identified TA and axillary arteries US as the first-line investigation in patients with predominantly cranial GCA(2). In addition, the halo count and Southend Halo Score have recently been proposed to quantify the extent of vascular inflammation by US and correlate with systemic markers of inflammation and risk for ocular ischaemia(3). These novel score systems have also been validated in routine care and showed an excellent diagnostic accuracy for GCA diagnosis(10,11).
Wall swelling at subclavian arteries can be seen by US in LV-GCA patients(12–14), but involvement of subclavian arteries in the absence of vasculitic changes in the axillary arteries is rare and most clinical guidelines include only axillary arteries on the LV examination(2,15). In this context, a novel modified Halo Score has been proposed including the assessment of three vascular territories (TA, axillary and subclavian arteries) instead of two, as the Southend Halo Score may underestimate the burden of inflammation in LV-GCA and Takayasu arteritis(4,16).
To our knowledge, this is the first study specifically designed to compare the diagnostic value of the three published quantitative scores for GCA diagnosis(3,4). According to our findings, the modified Halo Score does not improve the diagnostic accuracy when compared to halo count and Southend Halo Score(Table 2). All scores showed excellent ability to discriminate between patients with and without GCA, as indicated by high AUC in the ROC curve. Thus, the inclusion of the subclavian arteries into the Southend Halo Score increases the burden on the US examination as requires extra time, without improving its diagnostic accuracy.
Overall, all scores showed moderate correlations with markers of inflammation, except for haemoglobin levels(Table 3). However, in the LV-GCA subgroup the modified Halo Score seems to have some advantages, as it showed moderate positive correlations with ESR and platelets and moderate negative correlation with haemoglobin levels(Figure 2). Halo count and Southend Halo Score showed no correlation with laboratory findings in patients with LV involvement. Although grades of the axillary arteries included in the Southend Halo Score are multiplied by 3 in order to equate the inflammation of the TA and LV arteries, the fact that the modified Halo Score is based on the sum of the two higher scores of the 3 scanned regions (TA, axillary or subclavian arteries), seems to be related to a better detection capability of the general burden of inflammation. These findings may be relevant for monitoring purposes. Since the use of Tocilizumab challenged the assessment of activity in GCA due to suppression of the APR, the use of imaging may have a key role in monitoring treatment response(17). According to our data, the modified Halo Score is linked to markers of inflammation at baseline in LV-GCA patients.
Some limitations should be noted. First, our sample size warrants further validation in larger and additional populations. Second, the retrospective design is a prominent limitation, thus TA biopsy was only performed according to clinician criteria. Third, the ultrasonographer was not blinded to clinical data.
In summary, the inclusion of subclavian arteries examination by the modified Halo Score does not improve the diagnostic accuracy of GCA diagnosis over the halo count and Southend Halo Score. However, it correlates better with markers of systemic inflammation in LV-GCA and could be of additional value in monitoring response to therapy. Further studies are necessary to confirm these findings.