Combination of serum cholesterol levels and computed tomography scores can predict degrees of olfactory dysfunction in patients with chronic rhinosinusitis with nasal polyps

Background: Olfactory dysfunction, ranging from mild impairment to complete loss, is a prevalent and bothersome symptom in patients with chronic rhinosinusitis with nasal polyps (CRSwNP). Objective: This study aimed to identify risk factors associated with the severity of olfactory dysfunction in the CRSwNP population, which have been largely overlooked. Methods: A total of 282 patients with CRSwNP were enrolled and categorized into normosmia, hyposmia, and anosmia groups according to their olfactory levels examined by the smell identification test specifically for the Chinese population (CSIT). Patients' demographic characteristics, history of asthma, computed tomography (CT)-based imaging scores, blood cytology, immunoglobulin E, and serum lipid profiles were recorded for all participants. Ordinal logistic regression analysis was used to identify risk factors of gradual olfactory dysfunction, and receiver operating characteristic (ROC) curve analysis was performed to assess the predictive value of those factors for discriminating different degrees of olfactory dysfunction from normosmia. Results: Significant differences were observed in age, prevalence of comorbid asthma, blood eosinophils, serum lipid profiles, and CT-based imaging scores of CRSwNP among the 3 groups stratified by olfactory outcomes ( P < 0.05, respectively). Multivariate ordinal logistic regression analysis identified Lund-Mackay scores, ethmoidal-to-maxillary (E/M) sinus ratio, age, and total cholesterol as independent risk factors for gradual olfactory dysfunction in CRSwNP ( P < 0.01, respectively). Correlation analysis indicated that the 4 identified risk factors were all significantly inversely correlated with the olfactory score measured by CSIT ( P < 0.001, respectively). ROC analysis demonstrated that the combined model based on the 4 risk factors showed strong power to discriminate between anosmic and normosmia (area under the curve [AUC]: 0.893, 95% confidence interval [CI]: 0.848–0.939), and moderate accuracy in classifying hyposmia and normosmia (AUC: 0.716, 95% CI: 0.640–0.792). When compared the AUC between different models, the combined model shows a better prediction ability than E/M ratio, age, and total cholesterol alone ( P < 0.05, respectively). Conclusion: Lund-Mackay score, E/M ratio, age, and serum total cholesterol are independent risk factors for gradual olfactory dysfunction in patients with CRSwNP, which can be used to develop a predictive model for olfactory outcomes.