Cancer Med. 2026 Jun;15(6):e72001. doi: 10.1002/cam4.72001.
ABSTRACT
PURPOSE: To investigate the concordance and correlation between programmed death-ligand 1 (PD-L1) expression in tumor tissue and circulating tumor cells (CTCs), explore the intrinsic link between epithelial-mesenchymal transition (EMT) and PD-L1 expression in CTCs, and examine the predictive value of FDG metabolic parameters, CTCs and their PD-L1 expression for early tumor response and long-term prognosis in patients with advanced non-small cell lung cancer (NSCLC) receiving immune checkpoint inhibitor (ICI) monotherapy or ICI combined with chemotherapy as first-line treatment.
METHODS: A total of 42 patients with advanced or metastatic NSCLC who received ICI monotherapy or combination chemotherapy as first-line treatment were enrolled as the study population. Pre-treatment peripheral blood CTCs and PD-L1 expression on CTCs were detected. Furthermore, F-18-fluorodeoxyglucose Positron Emission Tomography/Computed Tomography (18F-FDG PET/CT) tumor metabolic parameters, including maximum standardized uptake value (SUVmax), standard uptake value of lean body mass (SUL), metabolic tumor volume (MTV), total lesion glycolysis (TLG), whole-body metabolic tumor volume (WMTV), and whole-body total lesion glycolysis (WTLG) were collected and measured. Tumor early response was monitored using response evaluation criteria in solid tumors (RECIST criteria) at 8 weeks after ICI treatment, and the differences in CTC PD-L1 expression, FDG metabolic parameters, and clinical data between early respongders and non-responders were analyzed. Finally, the enrolled patients were followed up to evaluate progression-free survival/overall survival (PFS/OS) and its associated predictive factors. Bootstrap internal validation (1000 repeated samples) was used to evaluate the robustness of the above prognostic model.
RESULTS: The percentage of PD-L1-positive cells in tumor tissue and CTC represents different immune characteristics of patients, and the correlation between these two percentages was not statistically significant (r = 0.041, p = 0.827). However, an intrinsic connection was identified between the EMT process and PD-L1 expression in CTCs, with PD-L1 expression in CTCs gradually increasing as the EMT process in CTCs changed. Furthermore, tumor proportion score (TPS) demonstrated a significant positive correlation with SUVmax (r = 0.684；p < 0.001) and SUL (r = 0.603；p < 0.01), with SUVmax and SUL increasing as TPS increased. Patients who exhibited an early response after 8 weeks of treatment had significantly higher SUVmax and SUL values, but lower mixed CTCs, than those in the non-response group (p < 0.05). The combination of SUVmax and PD-L1+ mixed CTCs yields a sensitivity of 81% and a specificity of 71% for predicting early response to ICI therapy in patients with NSCLC, but the combination did not result in a substantial enhancement in predictive efficacy. PD-L1+ mesenchymal CTCs (HR = 5.520, 95% CI 1.993-15.291) and WTLG (HR = 4.315, 95% CI 1.864-9.991) were identified as independent predictors of PFS, while PD-L1+ mesenchymal CTCs (HR = 2.880, 95% CI 1.124-7.381) were identified as independent predictors of OS. Bootstrap resampling analysis (B = 1000 iterations) revealed that PD-L1+ mesenchymal CTCs and WTLG achieve an apparent area under the ROC curve (AUC) of 0.822 for predicting PFS, with a bootstrap-corrected mean AUC of 0.821 (95% CI: 0.714-0.917). For OS prediction, PD-L1+ mesenchymal CTCs yielded an apparent AUC of 0.893 and a bootstrap-validated mean AUC of 0.88 (95% CI: 0.655-0.975). Furthermore, the combination of PD-L1+ mesenchymal CTCs and WTLG revealed that patients with WTLG < 1627.4 (g) and PD-L1+ mesenchymal CTCs < 1/5 mL exhibited prolonged PFS and OS (median PFS was 12.5 months; median OS was 18.5 months), while those with WTLG ≥ 1627.4 (g) and PD-L1+ mesenchymal CTCs ≥ 1/5 mL experienced the shortest PFS and OS (median PFS was 1.35 months; median OS was 3.0 months). Patients with PD-L1+ mesenchymal CTCs ≥ 1/5 mL or WTGL ≥ 1627.4 (g) exhibited a PFS and OS that fell between the two groups (median PFS was 6.0 months; median OS was 8.0 months). Stability analysis was conducted on the prognostic model after the combination of PD-L1+ mesenchymal CTCs and WTLG. Bootstrap resampling analysis (B = 1000 iterations) showed that the AUC value of the original ROC curve was 0.821, the mean AUC was 0.823 (95% CI: 0.716-0.914) when predicting PFS. For OS prediction, the AUC value of the original ROC curve was 0.875, the mean AUC was 0.874 (95% CI: 0.751-0.966).
CONCLUSION: EMT is associated with PD-L1 expression in CTCs, with CTCs exhibiting a mesenchymal phenotype tending to have higher PD-L1 expression. PD-L1+ mixed CTCs, SUVmax, and SUL were associated with early response to ICI-based treatment. Preliminarily, PD-L1+ mesenchymal CTCs, as an exploratory biomarker, were significantly associated with PFS and OS and may provide prognostic information complementary to conventional tissue-based PD-L1 assessment and FDG metabolic parameters. However, because PD-L1 expression in primary tumor tissue and CTCs was assessed using different detection methodologies, the observed discordance should be interpreted as potentially reflecting both biological heterogeneity and methodological variation. These findings require further validation in larger cohorts using harmonized detection platforms.
PMID:42267455 | DOI:10.1002/cam4.72001
