Science and Technology Production
Medicina - BUILDING A PREDICTIVE MODEL BASED ON GLYCOGENE EXPRESSION PROFILES OF MELANOMA PATIENTS FROM TCGA-SKCM PROJECT
Congress
Authorship:
Merlo, Joaquín ; Mahmoud, Yamil ; Veigas, Florencia ; Rabinovich, Gabriel A. ; MARIÑO, KARINA VALERIA ; Girotti, María RominaDate:
2020Publishing House and Editing Place:
MedicinaSummary
Objectives. We aim to study glycoimmune pathways involved in resistance to immune checkpoint blockade (ICB) therapies for melanoma to establish a signature for patient classification.Materials & Methods. Analysis were run using R software v3.6. Deconvolution was performed using MIXTURE tool and the glycoimmune pathways analysis was based on GlycoV4 chip (834 genes).Signature score was calculated as the geometric mean of the expression of genes in each signature. Cluster1 and Cluster2 comparisons was performed using Wilcoxon test.Results. Metastatic tumor biopsies (n=357) were clustered using 78 high-variable glycogenes resulting in a Cluster2 of low Overall Survival (OS), and a Cluster1 of high OS (p<0.05). Next, we characterizedthe tumor microenvironment (TME) using deconvolution tools. Cluster2 showed a lower absolute score (p<0.001) and proportion of cell types associated with immune activation (activated CD8+,CD4+ and M1 macrophages), while showing higher proportion of M2 macrophages and resting CD4+ cells. Cluster2 also showed lower Cytolytic Score, lower TMB and higher Intratumor Heterogeneity(p<0.01). When analyzing gene signatures, Cluster2 showed lower score of apoptosis and interferon-g with higher score of proliferation (p<0.01). We built a predictive model based on a Bayes Naïve classifier (AUC=0.902) using 20 features identified by recursive feature elimination and used it to classify 73 baseline biopsies of a separate cohort (anti-PD1, n=41; Combo, n=32). Patients classified as Cluster2 consistently had lower OS (p<0.01). By performing a Fisher?s test, we found a significant association between Cluster2 and non-responding patients (p<0.05).Conclusion. Dysregulation of 78 glycogenes correlates with distinct profiles of TME and biomarkers in association with OS. These profiles were also found in a cohort of ICB-treated patients. Furtheranalysis is required to validate these findings in order to unveil new mechanisms of resistance to immunotherapy.Key Words
glycomicspatient´s responsemelanomabioinformatics