Treatment Response Trajectories of Epipharyngeal Abrasive Therapy (EAT) in Chronic Nasopharyngitis: Phenotypic...

Background Treatment responses to epipharyngeal abrasive therapy (EAT) for chronic nasopharyngitis are heterogeneous. Previous analyses of this cohort examined longitudinal changes in EAT and bleeding scores with stratification by tissue type and baseline severity. However, those analyses mainly evaluated overall score changes and baseline-based stratification rather than the timing of individual improvement. The novelty of the present study lies in its focus on temporal response trajectories, specifically when patients improve after EAT. This study classified patients into early responders, delayed responders, and non-responders based on changes in EAT scores during the first three months and examined their association with final outcomes. Methods This single-center retrospective observational study included 546 patients (100.0%) with chronic nasopharyngitis treated with EAT between August 2018 and August 2019. The primary analysis included 238 patients (43.6%) with complete EAT scores at baseline, one month, two months, and three months. Patients were classified into three temporal phenotypes based on ΔEAT, namely early responders, delayed responders, and non-responders. Final outcomes were categorized as complete resolution, partial resolution, dropout, or indeterminate. Logistic regression was performed using complete resolution as the dependent variable, with adjustment for baseline characteristics and treatment intensity. Results Among the 238 patients (43.6%), 103 (43.3%) were early responders, 83 (34.9%) were delayed responders, and 52 (21.8%) were non-responders. Baseline EAT scores differed significantly among the groups (one-way ANOVA, F = 27.04, p < 0.001), as did baseline bleeding scores (one-way ANOVA, F = 4.94, p = 0.008). Final outcome distribution did not differ significantly among the threshold-based response groups (χ² = 0.92, p = 0.921). After adjustment for treatment intensity, early responders were significantly associated with complete resolution compared with non-responders (OR 2.78, 95% CI 1.03-7.45, z = 2.03, p = 0.043). Delayed responders showed a favorable but nonsignificant association (OR 1.72, 95% CI 0.69-4.31, z = 1.16, p = 0.244). Treatment intensity was positively associated with complete resolution but did not differ significantly among the response groups. Bleeding scores decreased in all groups but did not show the clear phenotype separation observed in EAT score trajectories. Conclusions In contrast to analyses focused on overall longitudinal changes and baseline stratification, the present study introduces a trajectory-based framework for evaluating EAT response. Treatment responses may be better understood as distinct temporal phenotypes rather than as a single uniform process. Early improvement within the first month may provide useful prognostic information, whereas delayed improvement suggests that continued treatment may still be beneficial in selected patients. These findings support evaluating EAT not only by final outcomes or overall improvement but also by the timing of improvement. As this was a single-center retrospective observational study, the results should be interpreted as hypothesis-generating rather than causal.