Clinicopathological correlation in erythema induratum

Background - Erythema induratum (EI) is a reactive disorder to mycobacterium tuberculosis infection, a diagnosis not to be missed. Erythema nodosum (EN) is the main clinical differential of EI, but a distinctly different pathological condition that can be difficult to distinguish from EI. Methods – In this retrospective review we assess clinical and histological features of 40 EI cases and 16 EN cases. Six experienced dermatologists blindly diagnosed these cases based on clinical images, thereafter the histology was revealed, and they adjusted their diagnoses accordingly. Fleiss Kappa statistics were applied to determine inter-rater variability. A multi-variate logistic regression model determined the clinical and histological features that contribute most to an accurate diagnosis. Results - After assessing the clinical picture 48.8% of the EI cases and 74% of the EN cases were correctly diagnosed. With added histology results 67.1% EI and 81.2% EN cases were correct. EI cases showed inter-rater variability of 0.478 (pvalue < 0.01) before and 0.469 (p-value < 0.01) after histology was revealed. These features combined in a logistic regression model had a higher diagnostic accuracy than the assessors with regard to EI cases. The model was accurate in 100% and 80% of EI and EN cases respectively. Conclusions - While the study was limited by its retrospective nature and small sample size, valuable features (ulceration, vasculitis and lobular or septal panniculitis) were identified. A biopsy of the lower leg markedly increased the diagnostic accuracy, but there was less concordance between assessors, more research is needed to confirm these results