5514 - Discovery of an ImmunoTranscriptomics signature in blood for early colorectal cancer detection

Background

Colorectal Cancer (CRC) is the second leading cause of cancer mortality worldwide. An effective and convenient blood test for early detection of CRC is urgently needed to increase screening compliance and reduce mortality. The development of a new blood test for early CRC detection was initiated that leverage the transcriptome analysis of circulating immune cells (ImmunoTranscriptomics) using artificial intelligence and machine learning tools.

Methods

To identify new transcriptional biomarkers for CRC and adenoma detections, peripheral blood mononuclear cells (PBMC) transcriptome were analyzed by RNA-sequencing of 561 subjects (300 Caucasians and 261 Asians) enrolled in the DGNP-COL-0310 study (Ciarloni et al., 2016), a multi-centers case-control study. The cohort included 189 subjects with CRC, 115 with advanced adenoma (AA), 39 with other types of cancer (OC) as well as 218 individuals without any colorectal lesions (CON). Several univariate and multivariate methods were applied to the discovery set (n = 282) and results were integrated into a ranking system. Top ranked genes were selected for further validation and algorithm development in an independent set (n = 279).

Results

A large panel of differentially expressed genes were identified in non-metastatic CRC (I-II-III) and AA compared to CON and OC with significantly high power of discrimination (P-value = 10^-13). The novel developed data analytics pipeline was used to analyse the transcriptomic data measurements and the diagnostic accuracy of the new gene signature for CRC, through application of Machine Learning (ML) and bootstrap, was 82% sensitivity and 88% specificity and AUC of 90%. The signature was enriched in genes associated with myeloid cells activation, inflammation and hemostasis, suggesting a key role of the innate immunity in the early response to cancer.

Conclusions

Mapping the reaction of the immune system to onset of cancer and disease identification through application ML methods is a new approach to ensure an unbiased, genome-wide, unsupervised gene expression analysis for a highly specific biomarker identification.

Clinical trial identification

Editorial acknowledgement

Legal entity responsible for the study

Novigenix SA.

Funding

Novigenix SA.

Disclosure

S. Hosseinian Ehrensberger: Shareholder / Stockholder / Stock options: Novigenix SA. L. Ciarloni: Shareholder / Stockholder / Stock options: Novigenix. All other authors have declared no conflicts of interest.