Skip to main content
Premium Trial:

Request an Annual Quote

NEC OncoImmunity Develops Personalized HLA Typing Tool to Guide Cancer Immunotherapy

Premium

NEW YORK – NEC OncoImmunity has developed a computational human leukocyte antigen typing tool called NeoOncoHLA for guiding personalized cancer immunotherapy.

The company has integrated NeoOncoHLA into its NEC Immune Profiler platform for use in ongoing research projects, with plans to deploy it in future cancer immunotherapy clinical trials. A person's unique HLA genotype can predict response to immune checkpoint inhibitor therapy.

In a paper recently published in HLA Immune Response Genetics, the firm demonstrated that NeoOncoHLA improved the detection of HLA somatic variants in simulated experiments and discovered a novel class I HLA allele.

The algorithm aims to take a more holistic approach to HLA typing by combining database-matching with whole-exome sequencing (WES)-based germline and somatic variant calling, and novel HLA sequence reconstruction. The intent is to enable both a high degree of accuracy and the ability to discover new HLA alleles and tumor-specific variants.

"NeoOncoHLA was designed in [a] manner to leverage the power of novel HLA discovery by integrating abundant WES data, computational mutation identification methods, and ever-increasing HLA sequence reference libraries," Trevor Clancy, CSO of NEC OncoImmunity, said via email.

The algorithm first aligns normal or tumor WES reads, or tumor RNA-seq reads, to the closest-matched HLA allele from the IPD-IMGT/HLA Database and identifies them via an integer linear programing algorithm. The resulting processed alignment files become inputs for multiple variant calling tools, which, in addition to reducing false positives, help facilitate new allele discovery.

In the study, which was conducted in collaboration with Norwegian cancer vaccine biotech Ultimovacs, researchers applied NeoOncoHLA to WES from 10 metastatic melanoma patients. They validated their results using targeted NGS sequencing on patients where at least one candidate HLA germline variant was detected among class I HLA genes, via an Illumina HiSeq 4000 system.

NeoOncoHLA showed 100 percent overlap with targeted NGS in HLA typing, at both the sequence and protein coding levels and gave an overall somatic variant detection accuracy of 91 percent.

NeoOncoHLA analysis also uncovered a novel class I HLA allele in one patient, which was confirmed by both WES and targeted NGS, and named HLA-B*44:02:01:52.

NeoOncoHLA is the latest addition to an active field of HLA typing research. Other methods, such as arcasHLA and Polysolver, seek to accomplish largely the same task through slightly different methods.

Developed at Columbia University, arcasHLA infers HLA types from RNA-seq data, and demonstrated an in silico accuracy rate of 100 percent at two-field resolution for Class I genes and over 99.7 percent for Class II.

One potential limitation of relying on RNA-seq data alone is that there is a greater potential to miss low coverage and/or low expression variants, or those found on a non-transcribed strand.

In contrast, Clancy explained that "NeoOncoHLA can be applied to all types of NGS data."

Polysolver, developed at the Broad Institute, infers alleles for the three major MHC class 1 genes from WES data. The method functions in a broadly similar fashion to NeoOncoHLA, differing mainly in two ways.

First, NeoOncoHLA uses all the alleles available in the IPD-IMGT/HLA Database to align NGS reads, including classical class I, classical class II, and non-classical alleles, compared to only classical class I alleles in Polysolver. Second, Polysolver uses only one variant calling algorithm to detect somatic variants, in contrast to NeoOncoHLA's more stringent approach of seeking consensus calls from multiple variant calling tools.

One slight drawback to the current version of NeoOncoHLA is that despite its ability to infer new alleles, its reliance on known sequence information still limits it largely to testing for classical class I alleles, whose full-length sequences are the most extensively cataloged.

"We hope to benchmark this approach on class II and non-classical alleles as full-length HLA gene sequences become increasingly submitted for these groups into the IPD-IMGT/HLA Database," the authors wrote in their study.

NEC OncoImmunity has incorporated NeoOncoHLA into its Immune Profiler platform for the artificial intelligence-driven prediction of HLA-bound neoantigens presented on the surface of tumor cells. The company is using this platform in research projects and clinical trials, with no plans to commercialize NeoOncoHLA in any other platform, or as a standalone offering.

Commercial HLA typing tools are available from other companies, such as Personalis, which conducts HLA typing through its ImmunoID NeXT platform.

Other companies, such as NGeneBio and Genome Diagnostics, offer CE-IVD-marked HLA typing assays for organ and stem cell transplantation.

NEC OncoImmunity and its collaborator in the study, Ultimovacs, are both members of the Oslo Cancer Cluster, a Norwegian cancer research hub, and have collaborated in several projects over the past few years.

"The primary objective of our collaboration has been to leverage each other's scientific expertise in bioinformatics, immunology, and cancer to gain further insights into treatment outcomes with immunotherapy," Espen Basmo Ellingsen, scientific adviser at Ultimovacs, said via email. "In this context, NeoOncoHLA may be an important tool allowing identification of possible tumor response or resistance mechanisms to immunotherapy in general."

The companies are also collaborating to develop Ultimovacs' candidate cancer vaccine UV1, which is currently being tested in a Phase I/II trial, interim results of which were presented at the annual Society for Immunotherapy of Cancer meeting last year and published in the Journal for ImmunoTherapy of Cancer.