Initial steps towards developing a predictive algorithm of disease progression for hidradenitis suppurativa (HS): results from a Cox proportional hazard regression analysis on disease progression among a cohort of 335 Danish patients with HS.

Abstract

Hidradenitis suppurativa (HS) is a chronic inflammatory and scarring disease with a wide spectrum of disease severity. The amount of scarring is proportional to the preceding tissue damage and poses a challenge to patients. Severe HS is most often treatment recalcitrant, but hypothetically avoidable through early biologic treatment. Early prediction of individual risk of disease progression is therefore essential for patient management. To investigate risk factors associated with disease progression and to design an algorithm capable of predicting disease -progression. A prospective cohort study of 335 Hurley III-naïve patients with HS, not treated with biologics, was followed for a median of 2 years. Potential risk factors covered basic demographics, HS anamnestic factors and clinical HS factors collected during physical examination. Two separate Cox proportional hazard regression (CPHR) analyses were conducted. A summated 'progression score' was calculated and used in the predictive algorithm of severe disease. Subsequent bootstrap sampling was used to validate the predictability of the predictive algorithm. The CPHR analysis of Transition to severe disease found that active smoking [hazard ratio (HR) 4.01, 95% confidence interval (CI) 1.71-9.40, P = 0.001]; body mass index (BMI) points > 25 at baseline (each point: HR 1.06, 95% CI 1.02-1.09, P < 0.001); active disease in 2 (HR 4.26, 95% CI 1.23-14.84, P = 0.02) and ≥ 3 areas (HR 6.54, 95% CI 1.89-22.72, P = 0.003) all constituted substantial risk factors. Conversely, the CPHR analysis of Disease progression did not yield results of clinical relevance. A 'progression score' of 3.04 was used as a threshold in the predictive algorithm of Transition to severe disease and achieved the following test specifics: sensitivity = 0.51, specificity = 0.86, positive predictive value = 0.50, negative predictive value = 0.86. We found a disparity between factors increasing the risk of simple Disease progression and those increasing the risk of Transition to severe disease. For the latter, active smoking, BMI points > 25, active disease in 2 or ≥ 3 areas were all shown to be the clinically relevant factors that could be used to construct an algorithm that correctly predicted progression to severe HS in more than half of all instances.