Background Adult large-vessel vasculitides (LVV) are rare conditions, currently classified as two different diseases, Takayasu arteritis (TA) and giant cell arteritis (GCA), on an empirical basis. Insight into phenotypic and pathogenic differences between the two is scarce at best. Arterial involvement, despite being the central disease feature, has been poorly addressed by research. We have developed two novel, imaging-based scores (the arteritis stenosis score [ASS] and arteritis dilation score [ADS]). ASS and ADS define stenotic and aneurysmal disease in a core-set of 17 arteries and respectively represent the sum of stenosis and dilation scores in individual arteries¹.

Objectives To use ASS, ADS and the stenosis and dilation scores of individual arteries to describe a cohort of LVV patients and identify heterogeneity between patients.

Methods The ASS, ADS and individual artery scores have been derived from 110 LVV (81 TA, 29 GCA) patients. Model-based clustering optimising the Bayesian Information Criterion and principal component analysis were performed.

Results Arterial involvement was shown to be differed in GCA and TA: TA has higher ASS than GCA (median, IQR: 20, 11–29 Vs 5, 0–11; p<0.001) and lower ADS (0, 0–5 Vs 6, 0–13; p=0.019). The scatterplot of ASS and ADS revealed incomplete overlap of arterial involvement in GCA and TA. No differences were seen in TA with disease onset before or after 40 yrs. Age at onset, ASS and ADS did not correlate in TA, suggesting stenotic and aneurysmal arterial remodelling are independent. In GCA, ASS and ADS were negatively correlated (ρ=-0.401; p=0.031) and ADS correlated with age at onset (ρ=0.383; p=0.040), suggesting the existence of a biologic “switch” between arterial stenosis and dilation, regulated by age at onset.

We accounted for geographical distribution of lesions by evaluating the scores of individual arteries with correspondence analysis. Arterial involvement was symmetric with tripolar segregation: stenosis in the supra-aortic branches, stenosis in the aorto-abdominal district and arterial dilation (Fig 1A). When patients exhibited the first two components, three different clusters were recognised (Fig 1B), with different ASS, ADS and damage as assessed by the TA damage score (p<0.001 for all tests). 27/29 (93%) of GCA patients were included in cluster 1 and 3, while 77/81 (95%) of TA clustered in 2 and 3. Of interest, density graphs showed (i) a different distribution of arterial involvement in GCA and TA (Fig 1C-D), and (ii) potential for identification of novel disease subsets (2 in GCA and 3 in TA). A comparable distribution was seen in TA with onset before or after 40 yrs (Fig 1E). Lastly, when patients with disease onset after 50 yrs (11 TA, 28 GCA) were studied, a trimodal distribution was observed, suggesting discrete phenotypes of arterial involvement exist, rather than a continuum (Fig 1F).

• Download figure
• Open in new tab
• Download powerpoint

Conclusions Arterial involvement differs in TA and GCA, although some overlap exists. Elderly TA is similar to juvenile TA, while a potential biologic “switch”, yet to be identified, regulating the final outcome of arterial remodelling and influenced by ageing, is present in GCA. Three main patterns of arterial involvement appear to exist. LVVs represent the composition of different discrete subsets rather than a phenotypic continuum.

References

1. : Tombetti et al. EULAR 2015 Poster FRI0258.

References

Disclosure of Interest None declared