A methylation-based classifier for neuroendocrine tumour detection using smMIP technology

#4516

Introduction: DNA methylation plays an important role in NET pathogenesis. Distinct methylation profiles can differentiate between tumour subtypes, predict malignancy potential, and provide insights into tumour progression.

Aim(s): We aimed to build a methylation-based classifier model using the most efficient and highest discriminatory single molecule molecular inversion probes (smMIPs) to detect NET presence.

Materials and methods: Through differential methylation analysis of array data (N=279 NET, 1023 healthy tissue, 834 healthy blood, 5786 other tumours), we selected a biomarker panel to distinguish NET from control samples comprising 1152 target CpGs for which we designed smMIPs. We validated our marker panel using the novel, bisulfite-free IMPRESS (Improved Methylation Profiling using Restriction Enzymes and smMIP Sequencing) assay using 33 NET (12 pancreatic, 16 lung, 5 small intestine) and 38 control samples (5 healthy blood, 22 normal adjacent tissue and 11 other tumours). A linear discriminant analysis model was constructed for each smMIP with a cumulative read count > 100. Five-fold cross-validation was carried out and the predictive accuracy of each model was assessed using receiver operating characteristic curves. smMIPs with a cross-validated area under the curve < 0.8 were excluded. The resulting set was combined, and the final classifier model was tuned for the highest overall accuracy.

Conference:

Presenting Author:

Authors: Ibrahim J, Mariën L, Vanpoucke T, Neefs I, Vandenhoeck J,

Keywords: neuroendocrine tumour, DNA methylation, biomarker selection, IMPRESS technology, classifier model,