Clinical significance of predicting the risk of recurrence and metastasis and the benefits of adjuvant chemotherapy in stage Ⅱ-Ⅳa nasopharyngeal carcinoma patients based on MRI radiomics features
10.16066/j.1672-7002.2024.08.001
- VernacularTitle:基于MRI影像组学特征因素预测Ⅱ～Ⅳa期鼻咽癌患者复发转移风险及辅助化疗受益的临床意义
- Author:
Zhiyi DENG
1
;
Yijing YE
;
Dingbo LI
;
Yongjin WU
;
Xianhai ZENG
;
Zaixing WANG
Author Information
1. 深圳市龙岗区耳鼻咽喉医院/深圳市耳鼻咽喉研究所,耳鼻咽喉头颈外科,广东 深圳 518000
- Keywords:
Nasopharyngeal Neoplasms;
Recidivism;
Prognosis;
MRI radiomics;
prediction model;
adjuvant chemotherapy
- From:
Chinese Archives of Otolaryngology-Head and Neck Surgery
2024;31(8):477-484
- CountryChina
- Language:Chinese
-
Abstract:
OBJECTIVE To develop a magnetic resonance(MRI) imaging radiomics and clinical factor model to predict recurrence and metastasis in patients with primary stage Ⅱ-Ⅳa nasopharyngeal carcinoma(NPC) and to validate its predictive effect on adjuvant chemotherapy(AC) outcomes. METHODS A retrospective analysis was performed on 135 patients with stage Ⅱ to Ⅳa NPC diagnosed in Longgang Otolaryngology Hospital of Shenzhen City from February 2018 to October 2021. After receiving standard synchronous radiotherapy and chemotherapy at our hospital,some patients received induction chemotherapy and/or AC based on cisplatin/nedaplatin. The imaging features of enhanced MRI sequences were extracted using PyRadiomics platform. Using the least absolute shrinkage and selection operator(LASSO) algorithm to filter features associated with recurrence or metastasis,a clinical radiomics model(CRM) was constructed by Cox multivariate analysis in a training cohort and validated in a validation cohort. All patients were divided into high-risk and low-risk groups based on the model's median Rad score. Kaplan-Meier survival curves were used to compare 3-year recurrence or metastasis free survival(RMFS) in patients with AC in high-risk group and low risk-group. RESULTS A total of 960 imaging features were extracted. The CRM consists of 9 features(6 imaging features and 3 clinical factors). In the training cohort,the area under the CRM curve(AUC) of 3-year RMFS was 0.867(P<0.001),and the sensitivity and specificity were 90.32％ and 79.66％,respectively. In the validation cohort,the AUC was 0.836(P<0.001) and the sensitivity and specificity were 100.0％ and 71.43％,respectively. The 3-year RMFS in high-risk and low-risk groups was 42.86％(27/63) and 94.44％(68/72)(log rank=50.818,P<0.001),respectively. Among CRM high-risk patients,3-year RMFS was significantly better in patients who received AC than those who did not(log rank=6.204,P=0.013). CONCLUSION CRM based on 3 clinical factors and 6 MRI features provides a non-invasive method for predicting the prognosis of NPC,which may help guide treatment decisions for clinical adjuvant chemotherapy,but further external verification is needed.
