Immunology – TCR therapies

We use Whole Exome Sequencing data from the bulk sequencing of a patient’s FFPE*1 sample (tumor tissue), along with the PBMC*2 sample (normal tissue). The data are processed to identify SVs (Structural Variants), somatic SNVs (single nucleotide variants), short indels (insertions/deletions), and other alterations, using multiple variant calling tools. The identified variants are then carefully examined to eliminate false positives. Only mutations present in a sufficient percentage of tumor cells retain.

*1 Formalin-Fixed Paraffin-Embedded

*2 Peripheral Blood Mononuclear Cell

RNA-seq data from the bulk sequencing of a patient’s FFPE sample (tumor tissue) are processed to obtain expression levels for each gene and identify new chimeric genes created by SVs (Structural Variants).

Variant calling is the process of variants identification based on sequencing data. Different variant callers exist, but they indicate low results correspondence. We perform variant filtering based on our machine learning models to filter out false-positive observations.

Short indels (especially those which affect the open reading frame) may lead to severe distortions of the protein sequences and thus might be a rich source of neoepitopes.

Tumor-Associated Antigens (TAAs) are non-mutated proteins or peptides with elevated expression levels in tumors but also expressed at lower levels in healthy tissues. TAAs were the first targets used in immunotherapies, giving foundations for successful treatments.

Somatic SNVs within a tumor lead to local changes in protein sequences. SNVs being the most common type of genetic alterations are rich sources of possibly immunogenic neoepitopes.

Retrotranspositions are events in which transposons (mobile genomic sequences)  are copied and pasted in new genomic locations. This event may lead to the appearance of new fusion proteins, which might be the source of neoepitopes.

Alternative splicing events, especially intron retention, lead to the appearance of new proteins. The fragments of such proteins not expressed within healthy tissue are another source of neoepitopes.

Structural Variants (SVs) are large genomic alterations that may affect a substantial part of the chromosome. Some of them are likely to cause fusion/chimeric genes. Such genes can bring significant alterations in the resulting protein sequences, which might lead to the appearance of new epitopes.

pMHC immunogenicity is acquired only for the minority of presented pMHC molecules. We can accurately assess the immunogenicity potential of the pMHC ligands with our proprietary AI model. We trained the model on the curated set of neoantigens tested in-vitro for immunogenicity.

Off-target toxicity is one of the biggest challenges in modern cancer vaccine development. It happens when TCR binds antigens presented on a patient’s healthy tissue, thus damaging it. We can identify possible off-targets for all epitopes of interest and eliminate those associated with the high risk of off-target toxicity.

ArdImmune Tox is verified to pinpoint off-targets from the available clinical data.

The pMHC molecule presentation on the cancer cell’s surface is an essential process needed for pMHC-TCR interaction to be possible. With the help of our proprietary AI model, we can identify epitopes and neoepitopes with the highest probability of being presented on the cancer cells by a patient’s MHCs. This model is trained on the large set of data obtained from numerous mass spectrometry experiments.

Immune escape mechanisms, such as HLA LOH*1 or HLA downregulation, are known cancer microevolutionary processes which significantly lower the immunotherapies efficiency. We identify such events and manage the problem by selecting epitopes that are not affected by the immune escape mechanisms.

 *1 Loss of Heterozygosity

The epitope binding to the patient’s MHC allele is necessary for its’ presentation on the cell surface and the MHC-related immunogenicity. We can accurately identify strong MHC binders via using different tools designed for predicting binding affinity strength.

Our process accounts for tumor heterogeneity by prioritization of clonal epitopes, driver genes, and hotspot mutations. It maximizes the cleaning potential of an anti-cancer vaccine composition by targeting only genetic alterations vital for tumor development.

Not all peptide sequences predicted in silico can be synthesized in laboratory conditions. We provide peptide manufacturability predictions to determine which peptides can be difficult or even not possible to manufacture.

We are relying on the model confidence of our machine learning models to present only meaningful and crucial information.

The tumor microenvironment (TME) is the environment around a solid tumor composed of surrounding cells and tissues. These can suppress the efficiency of the immune system or even promote tumor growth and survival. Overcoming the highly unfavorable conditions in the tumor microenvironment is the main challenge for effective T cell-based solid cancer therapies.

A mouse model is a foremost mammalian model used within laboratories for studying multiple aspects of human health and disease. Due to the manipulation of mice genes, such models are becoming more effective and appropriate for research. It allows for accelerating treatments discovery and drives medical innovations.

Recent coronavirus pandemic outbreak has proved once more how important a fast response with new vaccine designs is. Our technology allows designing peptide-based vaccines that are effective against infectious diseases and easily redesignable for novel virus variants detection.

Many therapeutic proteins and peptides had elicited immune system responses that decreased their safety and efficacy. Our technology allows us to suggest amino-acid substitutions that reduce binding strength and prevent the formation of anti-drug antibodies (ADAs). Biotherapeutics prepared in such a way should increase the likelihood of a positive clinical outcome.

Target identification is one of the initial steps of drug discovery and helps to lead to a suitable disease target. We use bioinformatics tools for data mining and analysis of genetic association or expression profiles to enable finding a target that is efficacious, safe, and druggable *1.

*1 It means that the target activity can be modulated by a therapeutic.