Reliable detection of subtypes of nailfold capillary haemorrhages in childhood-onset systemic lupus erythematosus.

Objectives: In systemic lupus erythematosus (SLE), it is necessary to obtain biomarkers that predict cardiovascular complications due to premature atherosclerosis, which is related to endothelial dysfunction. Nailfold capillary abnormalities might be a biomarker for endothelial dysfunction. In adults and children with SLE, nailfold capillary haemorrhages have shown to be significantly correlated with disease activity. Recently, different subtypes of capillary haemorrhages have been described in childhood-onset SLE (cSLE). The aim of the current study was to assess the inter- and intra-rater reliability of observations of different subtypes of haemorrhages in cSLE patients.

Methods: Five raters blindly evaluated 140 capillaroscopy images from 35 cSLE-patients (diagnosed according to the 2012 SLICC criteria). The images were assessed qualitatively (present or absent) and quantitatively (total number) on four different subtypes of haemorrhages: 1) punctate extravasations, 2) perivascular haemorrhage, 3) large confluent haemorrhage and 4) non-definable. As subgroups 1) and 2) were interpreted as a continuous spectrum, a post-hoc analysis with "merged" (mean) kappa/ICC was additionally calculated as one sub-group.

Results: Qualitative assessment showed a kappa 0.65 (95% CI: 0.60-0.70) for "punctate extravasations and perivascular haemorrhages merged" and a kappa 0.78 (95% CI: 0.72-0.83) for large confluent haemorrhages. For the quantitative assessment, ICC was 0.82 (95% CI: 0.76-0.87) for the "merged groups" and ICC 0.93 (95% CI: 0.91-0.95) for large confluent haemorrhages.

Conclusions: Our study shows that different subtypes of capillary haemorrhages in cSLE-patients could be reliably reproduced by different raters. This confirms our recent observation of perivascular extravasations as a subgroup of capillary haemorrhage in cSLE that might reflect endothelial dysregulation.