Hunting for Cancer Neoantigens Whitepaper/media/informa...the benefits of checkpoint inhibitors...

Hunting for cancer neoantigens: What’s the best approach?

WHITEPAPER

2 / May 2020 © Informa UK Ltd 2020 (Unauthorized photocopying prohibited.)

Most recently, attention in cancer immunotherapy has shifted to neoantigen-based immunotherapies – immunotherapies which target neoantigens, patient-specific peptides found on the surface of tumor cells. These highly personalized immunotherapies promise to unleash the natural power of a patient’s immune system to recognize an individual patient’s tumor antigens and to destroy cancer cells.

Currently, the key limitation to developing successful neoantigen-based immunotherapies is the ability to accurately predict which mutations, out of the vast number that arise in each patient’s cancer cells, are most likely to produce clinically relevant neoantigens.

Here, we conduct a deep dive into the neoantigen prediction market to identify which developers have the best in silico neoantigen prediction platforms and explore how machine learning can be successfully harnessed to improve prediction algorithms. Three key metrics are used to compare prediction platforms, namely prediction capability, clinical trial readiness and system strength and the developers are ranked based on the best publicly available knowledge and information gained from a series of interviews with key opinion leaders. From this analysis it is clear that NEC OncoImmunity’s NEC Immune Profiler is the leading neoantigen prediction platform, followed by Gritstone Oncology’s EDGE.

Tumor cells share a majority of their DNA with healthy cells, but looking deeper, there are areas where tumor DNA is distinct. These differences can result in neoantigens, immunostimulatory peptides presented on the surface of tumor cells, that the immune system is able to recognize as foreign.

Neoantigens first became an attractive immunotherapy target after it was discovered that patients who respond to checkpoint inhibitors have T cell responses to neoantigens. The hope is that neoantigen-based immunotherapies will have the benefits of checkpoint inhibitors without their negative side-effects. Indeed, neoantigen-based immunotherapy is being touted, by many, as the next big breakthrough in cancer treatment.

Executive summary

Introduction

Anna Osborne, Principal ConsultantSandy Tung, Senior Consultant

May 2020 / 3© Informa UK Ltd 2020 (Unauthorized photocopying prohibited.)

Identification of neoantigens presents a challenge due to their nature – tumors typically have hundreds of mutations, but only a small percentage of those mutations result in neoantigens that are present on the cell surface and able to interact with T cells to induce an immune response. Identifying these bona fide neoantigens is a challenging task and developers are racing to build their own unique platforms fit for this purpose.

Approaches to neoantigen discovery can be very varied, but typically a number of key steps are

in involved (Figure 1). The process starts with a routine clinical biopsy followed by next generation sequencing to profile each patient’s tumor to accurately identify the exact DNA mutations that are in the tumor cells but not in healthy cells. The next step is to try to determine whether peptides containing these mutations would be processed, presented and therapeutically relevant and this can be done in silico or using a wet lab approach. In the final step a personalized immunotherapy is manufactured, containing the relevant neoantigens, and administered back to the patient to induce a T cell response1.

Neoantigen identification

Figure 1: Neoantigen discovery process

Patient

Immunotherapy Manufacture

Sample

Neoantigen Selection

DNA and RNA Sequencing

Candidate Neoantigen Identification


The in silico approach uses a series of algorithms to sift through the mutations identified in the sequencing data to predict those that are likely to lead to the generation of neoantigens. The algorithms use previously generated training data to develop their predictions. In contrast, in a wet lab approach, peptides corresponding to each identified mutation from a patient’s tumor are synthesized and tested in a laboratory setting to see if they have the characteristics of a neoantigen.

While proponents of a wet lab approach would

argue that conventional in silico tools employ what amounts to no more than educated guesses, it is unlikely at the present time that undertaking extensive laboratory work on an individual patient basis is commercially feasible given the costs, resources and time involved. In addition, wet lab approaches can only measure pre-existing T-cell responses which miss the huge untapped potential of antigens that have the ability to induce clinically relevant responses but have been ignored by the immune system to date. As such, for the purpose of this report we have focused on in silico approaches to neoantigen identification.

A large number of players have jumped into the neoantigen prediction field and each has its own platform with its own inherent idiosyncrasies, strengths and weaknesses. In this analysis we aim to compare the platforms developed by key players. Most of the work in this field is being carried out by biotech companies who frequently claim to have the best platform based on analysis of publicly available datasets. As clinical trial results are not yet available a different approach is needed to benchmark these players.

The Parker Institute for Cancer Immunotherapy and the Cancer Research Institute launched the Tumor Neoantigen Selection Alliance (TESLA) in 2016. This global bioinformatics consortium includes scientists from the leading neoantigen research groups in academia, non-profit and industry. To assess the strength of their predictive platforms, their proposed neoantigens are evaluated through laboratory testing by TESLA and each participant receives feedback to inform and improve their predictions in the future. The

In silico vs wet lab approach

Comparing in silico approaches to neoantigen prediction

“When you’re faced with a patient, you need to know how to quickly build your vaccine so you can’t really do a wet lab approach. You just don’t have the time to take out the tumor and identify peptides. You don’t really have the time to isolate tumor-infiltrating lymphocytes and look at their specificity. You need to then

rely on in silico approaches based on the sequencing results.”EU KOL


results of this analysis are awaiting publication; however, the only way that companies would agree to participate is by remaining anonymous. While this publication should consolidate our understanding of the characteristics of a good platform, it will not publicly rank the developers against each other2.

Here we aim to achieve this by using three key metrics to compare the prediction platforms, namely prediction capability, clinical trial readiness and system strength. All companies developing in silico approaches to neoantigen prediction with a clinical asset in development were reviewed, as well as all companies in the

TESLA consortium and a number of leading academic groups. Those with sufficient publicly available information were profiled and all ranking were made based on the best publicly available knowledge and information gained from a series of interviews with key opinion leaders.

The key output of the analysis is presented in Figure 2. From this it is clear that biotech companies have more advanced neoantigen prediction platforms compared to service providers and academic groups and that NEC OncoImmunity’s NEC Immune Profiler is the leading neoantigen prediction platform, followed by Gritstone Oncology’s EDGE3–22.

Figure 2: In silico neoantigen prediction platform rankings

*Neon Therapeutics has been retained as a separate entity despite its recent acquisition by BioNTech as the systems are still distinct.

Clin

ical

Tri

al R

eadi

ness

Prediction Capability

Low High

Circle Size = System Strength

Low

High

BGI

OpenVax

Seven Bridges

Achilles Therapeutics+ pVACtools

BioNTech

Agenus

Neon Therapeutics

Medgenome

EpiVax Oncology+ Personalis

Gritstone Oncology

NEC OncoImmunity


The first key metric used in our analysis is prediction capability. It has been estimated that only 1–2% of mutations create immunogenic neoantigens that are clinically relevant and so the key question for neoantigen discovery is which mutations lead to mutated proteins

that are processed into shorter peptides by the proteasome, loaded onto major histocompatibility complex (MHC) molecules, presented on the cell surface and able to interact with T cells to induce an immune response1 (Figure 3).

Conventional prediction approaches only model specific steps involved in the antigen processing pathway and consequently have a poor hit rate. To increase the chances of identifying therapeutically relevant neoantigens it is prudent

to model as many steps involved as possible. As such, in this analysis prediction platforms have been ranked by the extent to which they model the different steps involved in neoantigen formation.

Prediction capability

Figure 3: Key steps involved in neoantigen formation.

“I think probably the biggest challenge and limitation of the field is that we’re training our models on the little parts of the immune system that we think are important, but we’re not actually training predictive

models on all of the other steps involved.”US KOL

Mutated DNA

Protein

Peptides

Proteasome

Presented Neoantigen

Presented Neoantigen

TCR TCR

MHC Class II MHC Class I

CD4+ T cell CD8+ T cell


Candidate neoantigen identificationTo generate a list of candidate neoantigens, mutations in tumor cells must first be identified using variant callers and considerable effort has been put into improving identification approaches. Several developers use more than one variant caller to improve their chances of identifying mutations and NEON Therapeutics and NEC OncoImmunity take this a step further and use machine learning to combine an ensemble of variant callers3,4.

Predicting peptide processingOnce the mutations are confirmed to exist and be expressed, the platforms then generate a list of neoantigen candidates and consider the probability of successful internal processing prior to MHC binding. Internal processing is a key step to predict and multiple publicly available tools are now available to capture the specificity of proteasomes and to predict the cleavage sites that are targeted by different proteases. However, KOLs have been left unimpressed with these approaches and further work is required to enhance these tools. Companies, including Gritstone Oncology and NEC OncoImmunity, have focused on refining these algorithms and it is hoped that these efforts will improve the overall accuracy of their prediction platforms5-8.

Predicting MHC bindingConventionally, platforms have focused on predicting the binding of candidate neoantigens to MHC molecules and algorithmic development and refinement of reference datasets are very active in this area. However, increasingly, there is an understanding that relying too heavily on this step and disregarding other steps, leads to a

high rate of false positives. Thus, while it is agreed that this step is important, researchers are now increasingly looking to model additional steps in the antigen processing pathway.

Predicting MHC Class I and II presentationT cells are involved in both detecting and killing cancer cells, as well as coordinating immune responses and T cells can be classified into two major subsets, CD4+ T cells and CD8+ T cells, based on cell surface expression of CD4 or CD8 glycoprotein. T cells recognize neoantigens presented by the MHC on the surface of tumor cells using T cell receptors (TCRs). There are two classes of MHC molecule, Class I and Class II, that activate CD8+ and CD4+ T cells, respectively and each person expresses a distinct combination of Class I and Class II MHC molecules.

Most studies have focused on the neoantigen reactivity of CD8+ T cells because of their capacity to directly kill tumor cells and because many tumors lack expression of MHC Class II molecules. It is now evident that CD4+ T cell activity can also be elicited in response to tumor mutations although, at present, it is unclear by which mechanism(s) neoantigen-specific CD4+ T cells contribute to tumor control.

In silico prediction platforms initially focused on MHC-I binders and algorithms for predicting MHC-II binders lagged. However, due to increased understanding of the importance of Class II, a number of the developers, specifically, BioNTech, EpiVax, Gritstone Oncology, NEC Oncolmmunity, Neon Therapeutics and Personalis are now including algorithms to model MHC II binding in their platforms4,6,9-11.


Predicting cell surface presentationA key step to predict is whether a peptide-MHC complex is presented on the tumor cell surface and research is increasingly looking to mass spectrometry (MS) to provide data with which to train prediction models for this step. Surface bound peptide-MHC complexes from a tumor or tumor cell-line can be purified by immunoprecipitation and identified using MS. Gritstone Oncology and NEC OncoImmunity are two companies using this approach on a large scale6.

Gritstone Oncology’s EDGE platform has used this approach to analyze over 300 resected tissue specimens, spanning lung, colon, ovarian and gastric cancers from patients of

various ancestries. Each tumor specimen yields thousands of peptide-MHC complexes and the total dataset has now grown to over one million MHC presented peptides. By directly examining peptides that are eluted from peptide-MHC complexes identified in vivo, predictive models can capture features unique to presented peptides and this extensive dataset can be used to train a prediction model5.

A similar approach has been undertaken by NEC OncoImmunity. The NEC Immune Profiler predicts MHC bound peptides presented on the tumor cell surface through an integrated machine learning algorithm trained on vast amounts of MS, binding affinity and other data.

“From the work that’s been published it seems that when people have been vaccinated with neoantigens a CD4+ T cell response is built, even though they originally developed their target for CD8+ T cells, so it seems

that the CD4+ T cells are important.”US KOL

“I think if I was forced to choose to make a vaccine consisting exclusively of Class II or Class I epitopes, I’d go for Class II, the importance of CD4+ T cells that see Class II in anti-tumor immunity is becoming more and more recognized, but, an ideal vaccine, of course, would contain both Class I and Class II antigens.”

EU KOL

“The tools to predict whether potential peptides bind to the MHC of the patient are fairly robust. What we don’t know is whether those peptides are actually really presented or not.”

AU KOL


MHC molecules are encoded by genes of the HLA locus which is highly polymorphic in nature and the resulting MHC molecules have different peptide binding preferences, mainly due to variation at anchor positions within the MHC binding groove. Due to this polymorphism an individual patient’s HLA genes must be sequenced before a platform can model which of the patient’s MHC subtypes each peptide is most likely to be presented by.

HLA specific tools often only evaluate the most common alleles. An additional benefit of using a MS approach is that the approach is agnostic to HLA. A pan MHC antibody can pull down any MHC molecule and so this approach allows researchers to look at a wide range of different HLA alleles.

In addition, as HLA alleles are derived from each other, they can be very closely related and in looking at related HLA alleles, commonalities in their binding specificities can be seen. As such, HLAs can be grouped together in supertypes and pan-specific methods can be used to predict binding to MHC subtypes for which no experimentally determined peptide-binding measurements are available.

Companies such as NEC OncoImmunity and Gritstone Oncology are taking a pan HLA approach and are therefore not limited to the common HLA allele subtypes which should improve the accuracy of their predictive algorithms and make these approaches suitable for patients from a wide variety of ancestries6.

Predicting immunogenicityA key challenge is to predict which neoantigens have the ability to bind a TCR and elicit a T-cell mediated immune response. This is increasingly

considered to be the key limiting step in neoantigen screening and currently this critical step of TCR binding and activation is being largely overlooked.

“If you had to do elutions and predictions and refine algorithms for each HLA allele, I think that would be a humongous task. If you can somehow cluster them together in supertypes, and sometimes you can, that’s a

huge advantage.”EU KOL

“I think the biggest unmet need is to go beyond the ability of a new antigen to bind to a particular MHC and to start to address what is the likelihood that a particular peptide-MHC complex can prime an immune

response in a particular patient, that the patient has the right T-cell receptor repertoire to recognize that particular epitope.”

US KOL


Several prediction platforms, such as those developed by BGI, OpenVax and pVACtools, still disregard this step entirely, but others are starting to tentatively consider it. Research in this area has built upon expertise obtained in the infectious disease field and one approach has been to assess candidate neoantigen similarity to known antigens and similarity to self. For instance, a key attribute of EpiVax’s Ancer system is the JanusMatrix tool which triages out any epitopes that are highly conserved with self-epitopes and therefore unlikely to be recognized by the immune system, as well as epitopes highly conserved with human commensal- or pathogen-derived sequences that may cross-react with T cells raised over the course of an individual’s immune history9.

However, even after taking tolerance mechanisms into account, not all well-presented peptides are strongly immunogenic, suggesting the existence of peptide features that influence T cell recognition independently of peptide processing and presentation. Research in this field is still in its infancy, but differences in non-anchor residues, peptide size and amino acid composition (i.e., aromatic residues) have all been associated with immunogenicity. In addition, peptide-HLA stability,

measured by biochemical assays, has also been proposed as a parameter to discriminate immunogenic from non-immunogenic peptides, although the predictive value of this tool is still controversial.

The more advanced platforms, such as those in development by Agenus, Medgenome, Neon Therapeutics and NEC OncoImmunity, have the ability to integrate immunogenicity data into their algorithms. However, the quality of such algorithms depends heavily on the size of the available datasets, and unlike datasets that describe MHC binding, datasets that describe immunogenicity of neoantigens are still very modest in size4,12,13.

Key to developing a robust database to test immunogenicity will be access to vast quantities of data identifying mutations that lead to immunogenic neoantigens, but also crucially mutations that don’t and often this second and critical piece of data is not made available publicly. A key differentiator of platforms moving forward will be the extent to which they are able to capture and model the immunogenic potential of candidate neoantigens.

“One of the biggest challenges is trying to predict the full three-part peptide MHC TCR recognition. I think if we could computationally model that, that would be really useful and right now we can’t really do it at all.”

US KOL

“What we really lack is a really good database that shows the neoantigens that have been shown to result in tumor rejection, and then ones that don’t.”

US KOL


Overall, whilst all developers included in this analysis openly acknowledge the importance of modeling as many steps in the antigen processing pathway as possible; some developers are at a more advanced stage than others. NEC

OncoImmunity, Neon Therapeutics, Gritstone Oncology and Personalis, in particular, are leading the way, while some of service providers such as BGI and Seven Bridges, as well as the academic groups, are lagging behind.

“People don’t always publish, and don’t always deposit all the various neoepitopes they looked at that didn’t work, but we need this information to make accurate predictions.”

US KOL

“The power of companies is that ultimately they will have these enormous databases of very useful information.”

AU KOL

The second key metric in our analysis is system strength and this metric considers the extent to which developers are building their own proprietary datasets and tools, particularly machine learning (ML) tools, with which to train neoantigen prediction models.

Proprietary dataTo improve the accuracy of prediction models a greater volume of training data is required, particularly for modelling peptide presentation and immunogenicity. Most of the biotech companies included in this analysis have an advantage over the academic groups and service providers in that they have access to, or are building, additional and complementary datasets

to those available in the public domain.

For instance, Achilles Therapeutics’ PELEUS platform has been trained using exclusive commercial access to TRACERx, the largest ever longitudinal study of tumor evolution in lung cancer14. It is the most extensive and highest quality bioinformatics data set of its kind globally and is more than three times larger than the largest publicly available comparator. In addition, companies including Gritstone Oncology and NEC OncoImmunity are building large internal datasets of MS data obtained by eluting peptide-MHC complexes directly from the surface of tumor cells or tumor cell lines6.

System strength


Proprietary toolsAccess to proprietary tools with which to train prediction algorithms, that are more advanced than publicly available offerings, also offer developers an advantage. ML, a form of artificial intelligence that enables a system to learn from data, is playing an increasingly critical role in neoantigen prediction. ML algorithms find natural patterns that cannot be detected any other way and it is an approach that is expected to be used more and more in this field due to vast amounts of data that need to be analyzed.

Several companies have developed their own ML tools to predict certain steps in the antigen processing pathway, for instance BGI have

developed a patented PSSMHCpan algorithm for peptide-MHC binding15,16. However, only a small number of companies have extensively developed their ML approaches. One such company is NEC OncoImmunity and its powerful platform, the NEC Immune Profiler, is able to model all of the steps in the antigen processing pathway using a novel machine learning approach8.

From our analysis, we consider Agenus, Gritstone Oncology, NEC Oncolmmunity and Neon Therapeutics to have platforms with the strongest systems due to their access to large datasets, proprietary tools and machine learning approaches.

“At the last Neoantigen Summit there was a talk from someone at the Parker Institute, who runs TESLA, and he painted a pretty sobering picture. They worked with the various companies building these algorithms

and none of them were good enough. The message was that everyone needs to get back into their wet labs and use that information to really improve prediction algorithms.”

EU KOL

“I think that it is likely that machine learning will play a greater and greater role in neoantigen prediction.”US KOL

“I think companies are increasingly using machine learning at different steps just because of the huge bodies of data that they’re dealing with.”

EU KOL


“Once we start to have clinical data, that will really help, not just in narrowing the field and showing what works, but also the information we get is going to help our understanding of what constitutes a

good epitope.”US KOL

The third key metric used in this analysis is clinical trial readiness and this encompasses not only current clinical progress but also the commercial strengths and weaknesses of the prediction technologies and the companies developing them.

Clinical progressRecently there has been a rapid expansion in the number of early-stage clinical trials using neoantigen vaccine platforms, and as efficacy outcomes are not available, developmental progress must be used as a marker of success. Several players, including BGI, Medgenome, Personalis and Seven Bridges operate as service

providers and are not developing neoantigen-based assets in-house or through partnerships. However, the biotech companies Achilles Therapeutics, Agenus, BioNTech, Gritstone Oncology, NEC OncoImmunity and Neon Therapeutics have all advanced personalized neoantigen therapeutics into early-stage clinical development. Indeed, BioNTech has initiated clinical development of more than one asset. While clinical testing is not in itself a guarantee of success, it is a sign that a developer is heading in the right direction and critically clinical testing will yield data that can be used to refine prediction algorithms.

Commercial focusFor neoantigen-based immunotherapies to be commercially viable, it will be critical to minimize the time it takes to select neoantigens and for the selection process to be as automated as possible. Although all platforms included in this analysis use an in silico approach, the time taken to complete the process of neoantigen prediction and selection varies, and some platforms are less automated than others.

For instance, Medgenome’s approach incorporates a final in vitro validation step and BioNTech hold a tumor board meeting to select the final peptides11,13. From a commercial standpoint any delay is undesirable and expert panels introduce subjective bias. As such, ultimately it would be preferable to develop algorithms that are so good that they don’t need to be validated or vetted at the end.

Clinical trial readiness


As this is an emerging field, most of companies operating in this space have been recently spun out of academia and therefore they have limited experience conducting clinical trials, never mind developing a commercial product. There is the potential that some developers will be held back by gaps in their knowledge and expertise. BioNTech and Gritstone Oncology hope to have circumvented these concerns by partnering with big pharma, specifically Genentech (Roche) and Merck & Co. respectively.

From a commercial standpoint NEC

Oncolmmunity is in a strong position. NEC has considerable experience in IT systems, data handling and data security. In addition, the NEC Immune Profiler has been awarded CE marking, a certification mark that indicates conformity with health, safety, and environmental protection standards8.

From our analysis it is apparent that NEC OncoImmunity, BioNTech and Gritstone Oncology have the most clinically ready platforms, and this will help to future-proof their prediction systems.

“I think one of the huge challenges is the turnaround time, there’s a long waiting period before the patient receives their treatment.”

US KOL

With rapid advances across many relevant fields, we are entering a new era of cancer immunotherapy in which a sophisticated vaccine loaded with patient-specific neoantigens is poised to generate a powerful yet precisely targeted antitumor immune response.

However, predicting and prioritizing neoantigens is a complicated process that involves many computational steps, each with individualized, adjustable parameters and it will be critical to get these steps right in order to see positive clinical outcomes.

Conclusion


1. Schumacher T, et al. (2019) Cancer Neoantigens. Available from: https://www.annualreviews.org/doi/abs/10.1146/annurev-immunol-042617-053402 [Accessed 30 April 2020].

2. Parker Institute for Cancer Immunotherapy (2020) Tumor Neoantigen Selection Alliance (TESLA). Available from: https://www.parkerici.org/research-project/tumor-neoantigen-selection-alliance-tesla/ [Accessed 30 April 2020].

3. Anzar I, et al. (2019) NeoMutate: an ensemble machine learning framework for the prediction of somatic mutations in cancer. Available from: https://doi.org/10.1186/s12920-019-0508-5 [Accessed 29 April 2020].

4. Neon Therapeutics (2018) Form S-1. Available from: https://www.sec.gov/Archives/edgar/data/1694187/000104746918004205/a2235870zs-1.htm [Accessed 29 April 2020].

5. Bulik-Sullivan B, et al. (2018). Deep learning using tumor HLA peptide mass spectrometry datasets improves neoantigen identification. Available from: https://doi.org/10.1038/nbt.4313 [Accessed 29 April 2020].

6. Gritstone Oncology (2018) Form S-1. Available from: https://www.sec.gov/Archives/edgar/data/1656634/000119312518255669/d565416ds1.htm [Accessed 29 April 2020].

7. Gritstone Oncology (2020) Scientific Platform. Available from: https://gritstoneoncology.com/scientific-platform/ [Accessed 29 April 2020].

8. NEC Oncolmmunity (2020) Our Technology. Available from: http://www.oncoimmunity.com/technology/ [Accessed 29 April 2020].

9. Moise L, et al. (2013) The two-faced T cell epitope. Available from: https://doi.org/10.4161/hv.24615 [Accessed 29 April 2020].

10. Personalis (2020) Applying immunopeptidomics and machine learning to improve neoantigen prediction for therapeutic and diagnostic use. Available from: https://www.personalis.com/wp-content/uploads/bsk-pdf-manager/2019/04/AACR_2019_V8.pdf [Accessed 29 April 2020].

11. Sahin U et al. (2017) Personalized RNA mutanome vaccines mobilize poly-specific therapeutic immunity against cancer. Available from: https://doi.org/10.1038/nature23003 [Accessed 29 April 2020].

12. Agenus Bio (2017) AGENUS’ NEXT GENERATION CANCER VACCINE PLATFORMS. Available from: http://agenusbio.com/wp-content/uploads/2016/08/AACR2017.Poster.AgenusVaccinePlatforms.30Mar2017.pdf [Accessed 29 April 2020].

References


13. Medgenome (2018) OncoPeptVAC™ accurately predicts neoantigens from tumors using a novel HLApeptide-TCR-binding algorithm. Available from: https://research.medgenome.com/pdf/whitepaper/OncoPEPT_VAC_WhitePaper_July%202018.pdf [Accessed 29 April 2020].

14. Achilles Therapeutics (2019) Jefferies London Healthcare Conference. Available from: https://achillestx.com/wp-content/uploads/2019/11/Achilles-Jefferies-Final-15.11.19.pdf [Accessed 29 April 2020].

15. BGI (2020) GenoImmune Neoantigen Prediction. Available from: https://www.bgi.com/global/pharma-solutions/translational-oncology/genoimmune-neoantigen-prediction/ [Accessed 29 April 2020].

16. Liu G, et al. (2017) PSSMHCpan: a novel PSSM-based software for predicting class I peptide-HLA binding affinity. Available from: https://doi.org/10.1093/gigascience/gix017 [Accessed 29 April 2020].

17. BioNTech (2019) Pipeline. Available from: https://biontech.de/science/pipeline [Accessed 29 April 2020].

18. EpiVax (2017) Ancer: An Innovative Cancer Neoantigen Prediction System. Available from: https://epivax.com/wp-content/uploads/2019/03/2017-Ancer-An-Innovative-Cancer-Neoantigen-Prediction-System.pdf [Accessed 29 April 2020].

19. Hundal J, et al. (2016) pVAC-Seq: A genome-guided in silico approach to identifying tumor neoantigens. Available from: https://doi.org/10.1186/s13073-016-0264-5 [Accessed 29 April 2020].

20. McGranahan N, et al. (2016) Clonal neoantigens elicit T cell immunoreactivity and sensitivity to immune checkpoint blockade. Available from: https://doi.org/10.1126/science.aaf1490 [Accessed 29 April 2020].

21. Rubinsteyn A, et al. (2018) Computational Pipeline for the PGV-001 Neoantigen Vaccine Trial. Available from: https://doi.org/10.3389/fimmu.2017.01807 [Accessed 29 April 2020].

22. SevenBridges (2020) Neoantigen Discovery Using Next-Generation Sequencing Data. Available from: https://www.sevenbridges.com/blog-neoantigen-discovery-using-next-generation-sequencing-data/ [Accessed 29 April 2020].


Anna OsbornePrincipal Consultant, Informa Pharma Custom Intelligence

Anna forms part of the leadership team at Informa Pharma Custom Intelligence, a team which specializes in pharmaceutical and healthcare consultancy. Anna’s team informs investment, clinical development, and commercial planning decisions across the product lifecycle for top-tier and mid-cap pharmaceutical companies, as well as biotech and medtech companies.

Anna has a keen interest in precision medicine and has led multiple projects in this area including opportunity assessments, market landscapes, patient-based forecasts and indication and product prioritization exercises.

Prior to joining the Custom Intelligence team, Anna was an analyst at Informa Pharmaprojects. Here she expertly curated the drug development database, provided research support to clients and produced custom analytics projects, transforming clinical study data and drug data into actionable insights.Anna’s strong scientific background enables her to fully understand the therapeutic potential of developmental therapies. Before joining Informa, Anna completed the Wellcome Trust PhD training program in Infection and Immunology at Imperial College London and she holds a first-class degree in Molecular Biology from the University of York.

If you have any questions, please call Anna on +44 7733 225 410 or email her at [email protected]

Sandy TungSenior Consultant, Informa Pharma Custom Intelligence

Sandy is a senior consultant at Informa Pharma Custom Intelligence, a team which specializes in pharmaceutical and healthcare consultancy. Sandy has a keen interest in novel therapy development strategy and has conducted multiple projects in areas such as cell and gene therapy, RNAi, and digital therapy.

Prior to joining the Custom Intelligence team, Sandy was a business development associate at Horizon Discovery, where she actively developed business opportunities in the cancer diagnostics and gene editing space. Prior to Horizon Discovery, she was part of the R&D Intelligence team at AstraZeneca where she analyzed pipeline progress and proposed portfolio management strategies.

Sandy holds a MPhil in Bioscience Enterprise from University of Cambridge and a PhD in Biotechnology from National Chung Hsing University.

About the Authors

IPI Supergraphic 1

For use as a background elementon white or ligh-colored backgrounds.Please do not change the color orthe angle of the supergraphic.

Informa’s Pharma intelligence is home of the

world’s leading pharma and healthcare R&D

and business intelligence brands—Datamonitor

Healthcare, Sitetrove, Trialtrove, Pharmaprojects,

Medtrack, Biomedtracker, Scrip, Pink Sheet, In

Vivo. Pharma intelligence’s brands are trusted

to provide over 3000 of the world’s leading

pharmaceutical, contract research organizations

(CRO’s), medical technology, biotechnology

and healthcare service providers, including the

top 10 global pharma and top 10 CRO’s, with

an advantage when making critical R&D and

commercial decisions.

Accurate and timely intelligence about the drug

development pipeline is vital to understanding

the opportunities and risks in today’s

biopharmaceutical marketplace—whether you are

targeting an unmet medical need, investigating

promising new therapies or researching drug

development historical trends and treatment

patterns. If you are providing contract research or

other services in the pharma industry, you need to

stand out. A solid understanding of your potential

clients’ pipelines and competition will help you

leave a lasting impression.

Contact us at [email protected]

United States605 Third AvenueFloor 20-22New YorkNY 10158USA+1 908 547 2200+1 888 670 8900

United KingdomBlue Fin Building110 Southwark StreetLondonSE1 0SUUnited Kingdom+44 20 337 73737

Japan21st Floor, Otemachi Financial City North Tower1-9-5, OtemachiChiyoda-kuTokyo 100-0004+81 3 6273 4260

China23rd Floor16F Nexxus Building41 Connaught RoadHong KongChina+85 239 667 222

AustraliaLevel 424 York StreetSydneyNSW, 2000+61 (0)2 8705 6968

Pharma Intelligence © 2020. All rights reserved. Pharma Intelligence is a trading division of Informa UK Ltd. Registered office: Mortimer House, 37-41 Mortimer Street, London W1T3JH, UK. Registered in England and Wales No 1072954

Hunting for Cancer Neoantigens Whitepaper/media/informa...the benefits of checkpoint inhibitors...

Documents

Transcript of Hunting for Cancer Neoantigens Whitepaper/media/informa...the benefits of checkpoint inhibitors...