Construction and validation of a prognostic model for CRSwNP.

OBJECTIVE

To analyze the clinical features of recurrent chronic rhinosinusitis with nasal polyps (CRSwNP), construct a predictive model for the prognosis of CRSwNP and validate it.

METHODS

Patients with CRSwNP who underwent functional nasal endoscopic surgery (ESS) in our hospital from 01/01/2017 to 30/04/2020 and met the study criteria were followed up regularly. The expression level of periostin in polyp tissues was observed by immunohistochemical staining test. The clinical characteristics of CRSwNP in the relapsed group were compared with those in the un-relapsed group. Subsequently, 144 patients were randomly selected as the training cohort to construct a predictive model and the remaining 63 as the test cohort to confirm the model’s performance. Risk factors for recurrence identified by univariate and multivariate Cox regression were used to create a nomogram.

RESULTS

(1) Preoperative peripheral blood EOS count, EOS%, Lund ⁃ Mackay total score of preoperative sinus CT (Total CT score), ethmoid sinus score (E score), ethmoidal/maxillary sinus density ratio (E/M), and composite expression score (CES) of periostin in polyps in the patients with CRSwNP in the relapsed group were significantly higher than in the patients with CRSwNP in the un-relapsed group, and the differences were statistically significant (all P < 0.05), but the maxillary sinus score (M score) was not statistically different between the two groups (P > 0.05). In addition, the proportion of eosinophilic CRSwNP (eCRSwNP) was significantly higher in the relapsed group than in the un-relapsed group (P < 0.05). (2) ROC analysis showed that preoperative peripheral blood EOS count and EOS% had no predictive value for recurrence of CRSwNP (all P > 0.05), whereas preoperative E score, E/M, total CT score, and periostin had predictive value for recurrence (all P < 0.05). The predictive accuracy of the combination of the above four significant indicators was statistically higher than that of any single indicator (all P < 0.05). (3) E score and periostin were identified as the two risk factors with significant differences by a multivariate Cox proportional hazard model (all p < 0.05), and the 5-year recurrent probability ranged from 0. 2 to 0.9 obtained by nomogram. Decision curve analysis (DCA) results revealed that the combination of periostin and E score showed the best prognostic efficiency of the risk threshold compared to a single factor in both the training cohort and test cohort. Kaplan-Meier (K-M) curve showed patients with high levels of periostin had a higher recurrent probability when compared to those with low levels of periostin.

CONCLUSION

Periostin is a potent novel biomarker for predicting recurrence of CRSwNP and combining it with the E score of sinus CT provides strong predictive efficacy for 5-year postoperative prognosis.