Abstract 120P
Background
Neoadjuvant immunotherapy, as the focus of current research and treatment means for long-term survival, has become one of the options in supporting primary treatment intervention in early NSCLC.
Methods
This was a retrospective analysis of patients with locally resectable NSCLC, who received the neoadjuvant drug pembrolizumab plus chemotherapy and underwent surgical resection. The pathology responses and the PFS and OS in the total sample and subgroups were determined and analyzed. Additionally, artificial intelligence was utilized to incorporate multiple factors for developing a high-performance prediction model.
Results
Of the 61 patients included in the retrospective analysis, 31 (50.82%) patients achieved a pCR, and 38 (62.30%) patients obtained an MPR. For the OS, patients with a pCR were significantly better than the patients with non-pCR (HR, 0.093; P = 0.0227); patients with an MPR performed significantly better than the patients with non-MPR (HR, 0.05357; P = 0.0169). Patients with lymph node metastasis after surgery had significantly worse OS and PFS than those without lymph node metastasis (HR, 0.01607; p = 0.0004; HR, 0.08757; p = 0.0004). The PFS of patients with SCC was better than the patients with non-SCC (HR, 0.3939; p = 0.0340). No significant differences in OS and PFS were found between 2 cycles vs. 3 cycles of neoadjuvant therapy before the surgery; ≤5 cycles vs. >5 cycles of adjuvant therapy post-surgery; TPS of <50% vs. ≥50% (P > 0.05). After model training and optimization, and 5-fold cross-validation, KNC (K-Neighbors Classifier) algorithm was able to predict the pCR with an 85.71% accuracy.
Conclusions
Neoadjuvant immunochemotherapy of pembrolizumab plus chemotherapy for non-small cell lung cancer is safe and tolerable. Both pCR and MPR were closely related to OS and PFS, reflecting the good response of tumor tissues to drug therapy. Lymph node metastasis after surgery was a poor prognostic factor, causing worse OS and PFS. Artificial intelligence constructed a prediction model for assessing treatment efficacy.
Legal entity responsible for the study
The authors.
Funding
Has not received any funding.
Disclosure
All authors have declared no conflicts of interest.