Relative perfusion index: An objective, quantitative and noninvasive method for evaluating the severity of keloids

Keloids are the result of abnormal wound healing, and they differ from the normal skin of the patient in the level of blood perfusion and the degrees of inflammation, hypoxia, regeneration of vessels, and expression of sensory receptors. However, there is no objective assessment method to accurately characterize the severity of keloids.

The purpose of this study was to evaluate the perfusion levels of keloids and the expression levels of various internal cytokines, including hypoxia-induced factor-1α (HIF-1α), vascular endothelial growth factor (VEGF), interleukin-17 (IL-17), HT_2A receptor subtype (5-HT_2A R), and H1R, in keloids and nonadjacent normal skin and to propose a laser speckle contrast imaging (LSCI)-based relative perfusion index (RPI), through which keloids can be divided into five grades to objectively characterize their severity.

This population-based cross-sectional study included 70 untreated keloid patients who each had only one keloid on the chest. LSCI was used to measure the area of each patient's keloid ( $$ ) and the perfusion level of each patient's keloid ( $$ ) and normal skin ( $$ ). The Vancouver Scar Scale (VSS) and Visual Analog Scale (VAS) for pain and pruritus were also used to assess each keloid. Immunohistochemistry and Western blot were used to detect the expression levels of various internal cytokines in keloids and normal skin. We compared the perfusion and expression levels of intrinsic cytokines between keloids and normal skin. We established the RPI to grade the severity of keloids and applied different methods to test the utility of the RPI.

The mean perfusion level of keloids was significantly higher than that of normal skin (p < 0.001). The expression levels of HIF-1α, VEGF, IL-17, 5-HT_2A R, and H1R in keloids were significantly higher than those in normal skin (p < 0.05). RPI was defined as: $$ The severity of keloids could be divided into five grades based on RPI. The RPI had a higher correlation with the pain-VAS, pruritus-VAS, and the expression levels of internal cytokines in keloids than blood perfusion levels and the VSS. T-SNE (t-distributed stochastic neighbor embedding) was also used to verify the clinical discriminatory abilities of this RPI model.

The proposed RPI based on LSCI showed the highest accuracy, unlike the VSS and assessment of perfusion, and can be utilized as a reliable, objective, quantitative, and noninvasive tool to evaluate the severity of keloids.