Deciphering immune dynamics and TEPs-derived circRNAs as predictors of disease progression, treatment response, and prognostic outcomes in GEP-NETs

Introduction: Neuroendocrine neoplasms (NENs) are heterogeneous in terms of primary site, behavior, and response to treatment. The possibility to rely on prognostic circulating biomarkers is an unmet need in NENs. The aim of this study was to quantify peripheral blood mononuclear cell (PBMC) subpopulations in patients with gastro-entero-pancreatic neuroendocrine tumor (GEP-NET) at the time of diagnosis and after first line therapy, evaluating their potential use as prognostic biomarkers. Moreover, the role as potential biomarkers of circRNA sequencing from tumor-educated platelet (TEPs) in NENs was evaluated. Materials and Methods: A prospective observational study was conducted, enrolling 26 consecutive treatment-naïve patients with histologically confirmed GEP-NET. PBMC subpopulations, including lymphocytes, monocytes and natural killer (NK) were studied by flow cytometry at baseline and after 3, 6, and 12 months of therapy (somatostatin analogs or surgery). We also performed a RNAseq analysis of circRNAs obtained from TEPs samples of 10 patients at baseline and after 3 months from therapy and from 5 patients affected by non-malignant endocrinological diseases enrolled as a control group. Results: At baseline, higher levels of circulating CD56bright NK cells were associated with disease progression. A cut-off of 2.46 cells/μL showed a sensitivity of 100% and a specificity of 67% (AUC = 0.86, 95% CI 0.7-1.0, p = 0.004) in differentiating patients with progressive or controlled disease, representing a risk factor for tumor progression (OR = 46.49, 95% CI 2.44– 885.15, p = 0.011). Furthermore, patients with CD56bright NK cell levels above the cut-off experienced significantly reduced progression-free survival (19.57 months; 95% CI 9.03- 30.10 vs 45.64 months; 95% CI 37.48-53.79, p = 0.026). A lower count of non-classical monocytes was associated with the presence of liver metastasis. A cut-off of 9.63% showed a sensitivity of 93% and a specificity of 75% in identifying presence of liver metastasis (AUC = 0.81, 95% CI 0.64-0.99, p = 0.006). Patients with a relative count below this threshold had a significantly higher risk of liver metastasis (OR 57.05, 95% CI 2.65-1128.64, p = 0.010). Prospective analysis revealed a progressive decline in NK cell subsets over time, with a more 3 pronounced reduction in patients with progressive disease, suggesting a tumor-driven immunosuppressive environment. We identified of 252 circRNAs differentially expressed between GEP-NET and controls of which 109 were up-regulated and 143 were down- regulated in NET patients. Further analysis based on an FDR value ≤ 0.05 resulted in the selection of 5 circRNAs all highly significant downregulated. The same analysis at baseline and after therapy in 5 patients revealed an average of 4983 remarkably differentially expressed circRNAs between follow-up and baseline samples of which 2648 up-regulated and 2334 down-regulated, respectively. Applying p ≤ 0.05 and FDR ≤ 0.05 filters, only 3/5 comparisons gave statistically significant results. Conclusions: Circulating CD56bright NK cells could represent a novel non-invasive prognostic biomarker in NET patients. Additionally, patients with liver metastases exhibited a peculiar circulating immune profile, where non-classical monocytes levels could be an early indicator of the presence of liver metastases. The observed changes in NK cell subsets after treatment suggest a restoration toward a more physiological immune profile. The more pronounced alterations in patients with progressive disease may be linked to higher baseline levels of these immune cell subsets, reinforcing their role in tumor-driven immune modulation. Moreover, we identified for the first time a circRNAs signature from TEPs as potential biomarkers for GEP-NETs.