Monitoring of molecular response during immunotherapy by circulating tumor DNA in non-small cell lung cancer

Background: Patients with non-small cell lung cancer (NSCLC) may benefit from immune checkpoint blockade (ICB) targeting PD-1/PD-L1. However, only 20-40% of patients respond to ICB. Circulating tumor DNA (ctDNA) analysis in plasma offers a non-invasive biomarker for treatment monitoring. Previous studies have shown that ctDNA detection can predict clinical response earlier than standard radiological evaluation. Objective: This study aimed to evaluate the detection and dynamics of somatic tumor-specific DNA variants in ctDNA during early treatment cycles, with the goal of identifying molecular response patterns. Methods: Thirty-three stage III-IV NSCLC patients treated with ICB were prospectively included. ctDNA levels were analyzed longitudinally at baseline and up to five timepoints during treatment. Based on initial tumor sequencing, four to fifteen patient-specific somatic variants were selected according to pathogenicity classification and variant allele frequency. These variants were monitored at each timepoint using ultrasensitive sequencing assays (Simsen Diagnostics). A model for ctDNA-based response assessment was developed and compared with clinical outcomes defined by RECIST criteria from CT scans performed every third month after ICB initiation. Results Overall, 78.8% (26/33) of patients showed ctDNA results concordant with CT imaging when applying the molecular response model. Thirteen patients with non-detectable ctDNA at baseline had significantly longer overall survival compared with baseline ctDNA-positive patients, independent of radiological response at 3 months. Six patients experienced early progression, which was detected by ctDNA up to six weeks before clinical evaluation by CT scan. Longitudinal ctDNA patterns varied depending on metastatic site and sampling timepoint. The genomic landscape of non-responders was further explored using complementary methodologies. Conclusion: ctDNA monitoring, based on the selection of informative tumor-specific DNA variants, provides a sensitive tool for evaluating treatment response in NSCLC. ctDNA dynamics correlated well with clinical outcomes and enabled earlier detection of progression or response compared with radiological imaging. Longitudinal ctDNA analysis underscores the importance of frequent sampling and highlights its potential to improve patient management in routine clinical practice.