How Immunotherapy of Cancer May Create Novel Clinical Utilities for HLA Typing

Written by Dr. Attila Bérces (Founder and Chairman, Omixon)
The article was originally published in ASHI Quarterly Q3 2017


While the role of HLA has been demonstrated in over hundred diseases, its clinical utility and impact outside transplantation has been limited. Transplantation still dominates the demand for HLA genotyping.  A new therapeutic paradigm in cancer immunotherapy based on neoantigens has a potential to dramatically change the HLA market. Neoantigen-based therapies may become the dominant driver for HLA typing in the not-so-distant future. HLA labs may need to develop new competencies in addition to cross-matching donors and patients, they may need to learn how to prioritize neoantigens based on their affinity to HLA. I will summarize the important developments in cancer immunotherapy in the context of the future demands for HLA typing. The European Neoantigen Summit held in Amsterdam on April 25 and 26, 2017 was dedicated to this topic and I summarize some relevant presentations.

Tumor-specific neoantigens (TSNA) originate from mutated peptides from cancer cells that differ from the self by somatic cancer mutations. Cancers typically carry between 10s and 1000s of such mutations. This number not only varies among individuals but also among cancer types. [1] Exposure to exogenous carcinogens, such as sunlight, X-ray, asbestos or cigarette smoke typically yield high mutation burden proportional to the level of exposure. Non-smoker lung cancer patients typically carry significantly fewer somatic mutations than smokers. [2] Similarly, the number of somatic mutations in melanoma is proportional to exposure to sunlight. In addition, mismatch-repair deficiency or microsatellite instability also yields a high number of mutations and neoantigens. The chance of a neoantigen fitting to the antigen-binding pocket of the HLA molecule of the patient increases with the number of neoantigens. Consequently high mutation burden is associated with high immunogenicity of the tumor, which is a prerequisite for successful immunotherapy of cancer. If one can identify which somatic mutations yield immunogenic peptides, in principle these peptides can be used to stimulate the immune system as therapeutic vaccines. [3]

Currently there are two FDA-approved therapeutic cancer vaccines: sipuleucel-T (Provenge®) for use in metastatic prostate cancer, and talimogene laherparepvec (T-VEC, or Imlygic®), an oncolytic virus therapy for the treatment of metastatic melanoma. However, the road to these first successful therapies has been paved with the history of repeated failures. [4] One may question why neoantigen-based vaccines have a better chance for therapeutic success. The answer lies in the improved understanding of not only the stimulation of the immune system, but also that of tolerance and the immune escape mechanisms. Cancers often successfully evade the T-cell mediated immune response by downregulating T-cell activity by immune checkpoint signaling molecules.  Our understanding of these signaling molecules has significantly advanced in the past decade. Cytotoxic T-Lymphocyte-Associated protein 4, or CTLA-4, Programmed Death 1 (PD-1) receptor and its ligands PD-L1 and PD-L2 are the most important down-regulating checkpoints and they are successfully pursued as therapeutic targets for cancer immunotherapy. The development of therapeutic antibodies blocking these immune checkpoints represents the most important breakthrough in cancer therapy over the past ten years. These anti-checkpoint therapeutic antibodies, also called checkpoint inhibitors became the fastest growing market segment of cancer therapy. Checkpoint inhibitors are estimated to reach $16B market by 2020.

Checkpoint blockade aims at liberating the immune system from downregulation by a therapeutic antibody. Such therapies were first clinically tested in indications and disease stages where no other therapeutic option was available, typically stage IV metastatic disease. The initial results were dramatic. In one of the earliest studies a decade ago, people with a dire form of melanoma were given a single dose of ipilumumab, an anti-CTLA4 antibody; 22 percent were still alive a decade later.  [5]

However, these therapies come at a significant cost not only in financial terms, but also with severe side effects. Several treatment-related death occurred in the clinical studies of the first CTLA-4 antibody. The PD-1 pathway seems to be therapeutically safer, but complementary and not a replacement for anti-CTLA-4 therapy. However, it is not obvious which patients will benefit from these therapies and who will develop side effects. For these reasons, expanding the use of these antibodies as first line therapies has been a challenge. Bristol Myers Squibb failed to show benefits of their anti-PD-1 antibody over standard-of-care as first-line therapy in NSCLC without patient stratification. In contrast, Merck succeeded in the same indication with an anti-PD1 antibody, which experts consider equivalent, in a stratified population using PD-L1 expression as a biomarker. This experience points out the importance of selecting likely responders based on biomarkers. HLA as a potential biomarker for patient stratification has been tested early in the Ipilimumab registration trial on over 1000 patients. The study did not find any HLA association, which is not surprising considering that melanoma typically has high mutation burden. [6] The large number of mutations yields large variety of neoantigens, of which a few are most likely to be presented efficiently by HLA leading to high immunogenicity.

In addition to PD-L1 expression, therapeutic response has been associated with mutational burden. Metastatic melanoma and lung caner were the first target indications not only because they represent the highest unmet need, but also because they carry the highest mutation burden, which makes them particularly suitable targets. High non-synonymous mutational load, as determined from smoking signature was associated with clinical efficacy of pembrolizumab (anti-PD-1 blocking monoclonal antibody) treatment. [7] Mismatch repair deficiency and microsatellite instability also yield higher mutation burden and are also associated with improved response to checkpoint inhibitors. [8]  This is such an important factor for checkpoint inhibitors, that on May 23, 2017, the U.S. Food and Drug Administration granted its first ever tissue or site-agnostic approval for pembrolizumab (anti-PD-1 antibody) for patients with microsatellite instability-high (MSI-H) or mismatch repair deficient (dMMR) solid tumors.

Expanding the use of checkpoint inhibitors to indications with low mutation burden, especially as first-line therapy is expected to be a difficult challenge. The rate of durable and strong response to the therapy is already modest with the highest mutation burden. Applying them to low mutation burden without stratification could render these therapies economically unaffordable. Identifying responders is critical for these therapies to advance to low mutation burden disease.  The strength of HLA-peptide binding, which determines the immunogenicity of cancer, is a critical factor. For these reasons the determination of the HLA genotype, HLA expression, somatic mutations and neoantigen binding is not only a starting point of neoantigen vaccine therapies, but they can contribute to patient stratification for checkpoint inhibitor therapies. Next generation sequencing of tumor tissue and germline DNA allows efficient determination of nonsynonymous mutations, and thus potential neoantigens. Computational algorithms, like NetMHC [9], can estimate the strength of neoantigen-HLA binding.

Recently, Dhodapkar and Dhodapkar argued that therapeutically targeting shared antigens would be logistically much simpler than determining antigens due to private mutations. [10] In contrast to the random mutations caused by exogenous carcinogens or mismatch repair deficiency, shared antigens have common origin. They include differentiation antigens (Melan-A, gp100), aberrantly expressed tumor-associated antigens (Her2, Muc-1), cancer/germline shared antigens (MAGE family, NY-ESO-1), stemness antigens (SOX2, OCT4), viral oncoproteins (HPV E6), or recurrent somatic mutations (B-Raf V600E).  While these authors are primarily concerned with therapeutic opportunities, the same idea can be expanded for the stratification of highly likely responders to checkpoint inhibitors. This can be relevant for tumor types with lower mutational burden. These antigens will likely be detectable either from tumor tissue or from circulating cell-free DNA. However, the therapeutic response may be HLA restricted, which needs to be determined by the combination of theoretical and empirical methods. The combination of testing for shared antigens and HLA can be a powerful tool in patient stratification for low mutation burden cancers. Similarly to other HLA associations, particular genotypes may be a necessary but not a sufficient condition for a therapeutic response.  While these HLA biomarkers could have a theoretical basis, their utility needs to be tested and validated in clinical studies.

The use of neoantigen in therapeutic vaccines was the topic of the recent Neoantigen Summit, where some of the unpublished advances in the field were presented. I summarize some of the interesting presentations. Andrea van Elsas from Aduro Biotech Europe presented how their company harnesses the power of bacteria to direct the immune response against cancer. Engineered T-cell receptors that recognize neoantigens in an HLA-dependent manner is another fast progressing field. Ziopharm Oncology and Medigene represented at the Summit by Laurence Cooper and Dolores Schendel, respectively, develop such engineered T-cell receptors.

Neoantigen therapies require the prioritization of peptide neoantigens based on HLA-antigen binding, HLA expression and other factors. Morten Nielsen, The Technical University of Denmark, the senior author of netMHC software explained in-silico methods developed by his group for the identification of epotopes.  He warned that current experience is based on pathogen-derived peptides, which may differ from the rules that apply to cancer.  Trevor Clancy, from Oncoimmunity, a start-up from Norway talked about his company’s computational approach to neoantigen selection and prioritization. Erin Newburn from Personalis presented their services for the determination of somatic mutations and neoantigens while Pia Kvistborg, from the Netherlands Cancer Institute talked about identificaiton of neoantigen responses in patients treated with immunotherapies. Johanna Olweus from Oslo University Hospital presented neoantigen determination in the context of predicting outcome for cancer patients receiving immunotherapy. She demonstrated that p-HLA stability measurements could improve prediction of which neoantigens are immunogenic in patients. She also showed how allogenic T-cells can be used to treat cancers. Immatics Biotechnologies  represented by Toni Weinschenk also pursues some allogenic adoptive cellular therapies. These approaches may give new utility for stem cell registry databases and increases demand for HLA typing for the production of off-the shelf engineered TCR products.

Mass spectrometry represents an experimental technique to determine neoantigen-HLA complexes. Michal Bassani-Sternber from Lausanne showed examples of mass spectrometry and bioinformatics determination of neoantigens. Neon Therapeutics represented by Mike Roonez at the Summit, is a start-up company with scientific founders of eminent scientists in immunotherapy of cancer, like James Allison also uses the combination of mass spectrometry, sequencing, some public algorithms in addition to proprietary technologies for p-HLA assay. Karin Jooss presented the approach of Gritstone Oncology. Gritstone delivers personalized synthetic TSNAs in a vaccine base to patients with or without the combination with immune checkpoint blockade.

Sebastian Kreiter from BioNTech presented the most advanced neoantigen-based immune therapy based on synthetic mRNA technologies for pharmacologically optimized protein coding RNA for targeted in vivo delivery. In a collaboration with Genentech, the company tested their IVAC personalized treatment and had 8 of 13 people stayed tumor-free for this ongoing study. BioNTech also offers off-the-shelf products for shared neoantigens.  Agnete Fredriksen from Vaccibody presented the company’s DNA vaccine approach to cancer neoantigen vaccines, which is in late preclinical development.

In addition to the companies mentioned, Caperna, Advaxis, ISA Pharmaceuticals, Agenus, Bavarian Nordic are developing neoantigen-based therapies and are collaborating with established pharma and biotech companies, like Genentech, Amgen, and Merck. These developments may have profound implications for HLA labs. Currently the most important goal of HLA labs is to provide input to minimize transplant rejection. An additional goal of the future may well be to maximize rejection of cancer by the patients’ immune system by selecting immunogenic neoantigens or stratifying patients for checkpoint therapies.


[1] Ton N. Schumacher, Robert D. Schreiber  Neoantigens in cancer immunotherapy, Science, 2015 April 3:348 6230, pp. 69-74.

[2] Govindan R, et al. Genomic landscape of non-small cell lung cancer in smokers and never-smokers. Cell. 2012;150:1121–1134.

[3] Ito A, Kitano S, Kim Y, Inoue M, Fuse M, et al. (2015) Cancer Neoantigens: A Promising Source of Immunogens for Cancer Immunotherapy. J Clin Cell Immunol 6: 322.

[4] Constantin N. Baxevanis & Sonia A. Perez; Cancer vaccines: limited success but the research should remain viable, Expert Review of Vaccines, Pages 677-680, 2016,

[5] Jim Allison commencement address at UC Berkeley.

[6] Wolchok JD1, Weber JS, Hamid O, et al. Cancer Immun. 2010 Oct 20;10:9. Ipilimumab efficacy and safety in patients with advanced melanoma: a retrospective analysis of HLA subtype from four trials.

[7] Rizvi NA, Hellmann MD, Snyder A, et al. Cancer immunology. Mutational landscape determines sensitivity to PD-1 blockade in non-small cell lung cancer. Science 2015; 348:124-8.

[5] D.T. Le et al. PD-1 Blockade in Tumors with Mismatch-Repair Deficiency, N Engl J Med 2015; 372:2509-2520.

[8] Wolchok JD, Weber JS, Hamid O, et al. Cancer Immunity : a Journal of the Academy of Cancer Immunology. 2010;10:9.

[9] Gapped sequence alignment using artificial neural networks: application to the MHC class I system. Andreatta M, Nielsen M, Bioinformatics (2016) Feb 15;32(4):511-7

[10] Dhodapkar and Dhodapkar, Harnessing shared antigens and T-cell receptors in cancer: Opportunities and challenges, PNAS July 19, 2016 vol. 113 no. 29 7944-7945.