LASSO regression based risk prediction model for postoperative control in chronic sinusitis with nasal polyps

Abstract

Objective:To establish a risk prediction model for postoperative control of chronic sinusitis with nasal polyps. Methods:Retrospective analysis was done on the clinical of patients who underwent endoscopic sinus surgery in the Department of Otolaryngology of the First Affiliated Hospital of Soochow University during August 2020 to June 2021. Patients were classified into uncontrolled group（40 cases） and controlled group（104 cases）, based on the European Position Paper on rhinosinusitis and nasal polyps（EPOS 2020）, and the clinical and pathological characteristics of the two groups were compared. The least absolute shrinkage and selection operator（LASSO） regression was used to screen the factors that might affect the prognosis of chronic sinusitis with nasal polyps and multivariate logistic regression was performed. The Receiver operating characteristic curve（ROC） was ploted, the area under curve（AUC） was calculated, and the ability of the prediction model was evaluated using the consistency index（C-index）. Results:A total of 144 patients with CRS with nasal polyps 1 year after operation were enrolled in this study, including 40 patients in the uncontrolled group and 104 patients in the control group（complete control or partial control）. 12 risk factors（allergic rhinitis, allergic dermatitis, olfactory dysfunction, E/M ratio, serum alkaline phosphatase, number of pathological eosinophils, number of pathological lymphocytes, number of plasma cells in pathological tissues, percentage of eosinophils in pathological tissues, stromal edema, basement membrane thickening, and hyperplasia of goblet cells） were found to be associated with postoperative recurrence of chronic sinusitis with nasal polyps. The seven variables（allergic rhinitis, olfactory dysfunction, E/M ratio, pathological eosinophilic percentage, stromal edema, basement membrane thickening, and hyperplasia of goblet cell） were extracted after reduced by LASSO regression. Multivariate logistic regression analysis showed that the 7 variables were risk factors for postoperative recurrence of chronic sinusitis with nasal polyps（P<0.05）. Nomogram prediction model for postoperative recurrence of chronic sinusitis with nasal polyps were established based on the 7 variables above. The verification results of the model showed that the C-index and AUC of the model were 0.937 and 0.937（95%CI 0.901-0.973）, suggesting that the nomogram model had a relatively accurate prediction ability. Conclusion:Combined with the basic clinical data of patients, the prediction model established in this study can facilitate the risk prediction of postoperative control of chronic sinusitis with nasal polyps, and thus help to formulate better therapeutic plans for patients.