Tear film cytokines as prognostic indicators for predicting early recurrent pterygium. (September 2022)

• Record Type: Journal Article
• Authors: Wan, Qi; Wan, Pengxia; Liu, Weiqin; cheng, Yaqi; Gu, Simin; Shi, Qian; Su, Yaru; Wang, Xiaoran; Liu, Chang; Wang, Zhichong
• Is Part Of: Experimental eye research. Volume 222(2022)
• Publication Date: 2022-09
• Subjects: Tear cytokines -- Recurrence-free survival -- Biomarker -- Nomogram -- Pterygium recurrence; Ophthalmology -- Periodicals; Eye -- Periodicals; Œil -- Périodiques; Ophthalmology; Periodicals; Electronic journals; 612.8405
• DOI: 10.1016/j.exer.2022.109140
• Languages: English
• ISSNs: 0014-4835

Abstract

Abstract: Cytokine profiles in tears have become a noninvasive biomarker for various ocular surface diseases. Therefore, the preoperative profile of cytokines in tear samples of 89 primary pterygium patients were obtained from Zhongshan Ophthalmic Center during 2015–2017. Compared to the tear cytokines in primary groups, the concentrations of IL-8, MMP-1, MMP-9, bFGF and VEGF were generally higher in recurrent pterygium group. The five cytokines were used to build diagnostic models by multiple machine learning algorithms, which can accurately distinguish non-recurrent and recurrent samples of primary pterygium patients. Besides, these cytokines were significantly associated with Recurrent-free survival (RFS) time in pterygium patients and further applied to develop a prognostic model which can estimate the prognosis of pterygium after resection. Afterward, a novel nomogram combined risk score of cytokines related biomarker and clinical characteristics was constructed, which manifested ideal accuracies to predict the 1 and 2 years' probability of pterygium recurrent events. Thus, our finding provides a more simple and accurate prediction for early pterygium recurrence after resection. It also affords a useful tool for ophthalmologists to choose the optimal treatment strategies for pterygium patients. Highlights: Our study constructed a diagnostic model by naive Bayes (NB) algorithm, which could accurately distinguish primary and recurrent pterygium patients. we constructed a prognostic model which can estimate the prognosis of pterygium after resection. We also constructed a novel nomogram which could accurately predict the 1- and 2-year probability of pterygium recurrence events. The 2-year AUC of the risk score was superior to that of each cytokine alone.