Multidimensional endotypes of chronic rhinosinusitis and their association with treatment outcomes

Abstract

Background: The expression of chronic rhinosinusitis (CRS) is multidimensional. Disease heterogeneity in patients with CRS remains poorly understood. This study aimed to identify endotypes of CRS using cluster analysis by integrating multidimensional characteristics and to explore their association with treatment outcomes.

Methods: A total of 28 clinical variables and 39 mucosal cellular and molecular variables were analyzed using principal component analysis. Cluster analysis was performed on 246 prospectively recruited Chinese CRS patients with at least 1-year postoperative follow-up. Difficult-to-treat CRS was characterized in each generated cluster.

Results: Seven subject clusters were identified. Cluster 1 (13.01%) was comparable to the classic well-defined eosinophilic CRS with polyps, having severe disease and the highest proportion of difficult-to-treat CRS. Patients in cluster 2 (16.26%) and cluster 4 (13.82%) had relatively lower proportions of presence of polyps and presented mild inflammation with moderate proportions of difficult-to-treat cases. Subjects in cluster 2 were highly atopic. Cluster 3 (7.31%) and cluster 6 (21.14%) were characterized by severe or moderate neutrophilic inflammation, respectively, and with elevated levels of IL-8 and high proportions of difficult-to-treat CRS. Cluster 5 (4.07%) was a unique group characterized by the highest levels of IL-10 and lacked difficult-to-treat cases. Cluster 7 (24.39%) demonstrated the lowest symptom severity, a low proportion of difficult-to-treat CRS, and low inflammation load. Finally, we found that difficult-to-treat CRS was associated with distinct clinical features and biomarkers in the different clusters.

Conclusions: Distinct clinicopathobiologic clusters of CRS display differences in clinical response to treatments and characteristics of difficult-to-treat CRS.

Keywords: chronic rhinosinusitis; cluster analysis; difficult-to-treatment; nasal polyps.

Fig 1

A modified heat map indicating the relative degree to which some key variables were present across clusters. The graded color scheme contains 5 shades of blue, representing the 20th (lightest), 40th, 60th, 80th, and 100th (darkest) percentiles of the data for each row. For continuous variables, results are expressed as medians. Categorical variables are summarized using percentage. The protein concentrations of detected molecular markers were normalized to total tissue protein levels. AR, allergic rhinitis; VAS, visual analog scale; CT, computed tomography; HPF, high-power field.

Fig 2

The proportions of difficult-to-treat chronic rhinosinusitis (CRS) differed by subject clusters. Fisher exact test was used to test the differences in proportions across 7 clusters. An overall P value indicates differences among the 7 clusters. An overall P value less than 0.05 indicates that at least one out of the 7 clusters was different from other clusters.