Development of a Predictive Model for Radiation Pneumonitis Based on Plasma Exosomal miR-200b-5p

Shuwei Zhai¹Yajun Zhu¹Xiaoye Wang¹Qingfeng Zhao²Xu Liang²Shuhao Que²Enhui Dai²Huaiyu Wang²Yuetong Li²Haihua Yang³Wei Feng^2*

• ¹Department of Radiation Oncology, Changzhou jintan first people's hospital, Changzhou, China
• ²Department of Radiation Oncology, The Cancer Hospital of the University of Chinese Academy of Sciences (Zhejiang Cancer Hospital), Hangzhou, China
• ³Department of Radiation Oncology, The Affiliated Taizhou Hospital, Wenzhou Medical University, Taizhou, China

Objective: This study aims to explore the association between plasma exosomal miRNAs and the development of radiation pneumonitis (RP) in non-small cell lung cancer (NSCLC) patients who underwent radiotherapy, and develop a predictive model for symptomatic radiation pneumonitis (SRP) by integrating miRNA expression levels with clinical and dosimetric parameters. Methods: A total of 95 NSCLC patients, who were scheduled to receive definitive radiotherapy, were prospectively enrolled. Plasma exosomes were collected before the radiotherapy, and high-throughput sequencing followed by bioinformatics analysis was performed to identify the candidate miRNAs associated to SRP. Then, the expression levels of these miRNAs were validated using RT-qPCR. Afterwards, a predictive model for SRP was constructed using a nomogram, which combined the miRNA expression data with the clinical and dosimetric factors. Results: Among the 95 patients, 20 (21.10%) patients developed SRP. The high-throughput sequencing revealed 220 differentially expressed miRNAs. Among these miRNAs, 168 miRNAs were upregulated and 52 miRNAs were downregulated in SRP patients (p<0.05). The bioinformatics analysis identified miR-200b-5p as the key candidate miRNA. The univariate and multivariate analyses revealed that lung V5 (OR: 1.264, 95% CI: 1.042-1.532, p=0.018), mean lung dose (MLD; OR: 1.013, 95% CI: 1.004-1.023, p=0.006), and miR-200b-5p expression (OR: 0.144, 95% CI: 0.024-0.877, p=0.032) were the independent risk factors for SRP. The nomogram model that incorporated these factors achieved an area under the receiver operating characteristic curve (AUC) of 0.844, outperforming the individual factors alone (lung V5: 0.748, MLD: 0.760, and miR-200b-5p expression: 0.666). Conclusion: The combination of lung V5 >46.36%, MLD >1120 cGy, and miR-200b-5p expression <0.445 can be used to effectively predict the occurrence of SRP in locally advanced NSCLC patients. This model can aid in the early identification of patients at high risk for RP, allowing for personalized treatment adjustments and improved patient outcomes.