Industry Program Meeting Report

Immuno-Oncology Biomarkers: Today’s Imperatives for Tomorrow’s Needs. A report from the 32nd Annual Meeting of the Society for Immunotherapy of Cancer, 2017

Introduction

As part of the 32nd annual meeting of the Society for Immunotherapy of Cancer (SITC 2017), leaders from across the field of cancer immunotherapy took part in Immuno-Oncology Biomarkers: Today’s Imperatives for Tomorrow’s Needs, a program focused on next-generation biomarker research. Biomarkers are becoming increasingly recognized as critical components of both prediction of therapeutic efficacy as well as prognosis. For example, tumor PD-L1 positivity enhances anti-PD-1/PD-L1 checkpoint blockade efficacy in the treatment of many cancers, including melanoma and NSCLC. Additionally, the first ever indication for biomarker-based disease was recently awarded to pembrolizumab for the treatment of patients with solid tumors positive for the DNA mismatch repair deficiency/ microsatellite instability-high (dMMR/MSI-H) biomarkers [1]. Based on these successes, there has been a significant increase in clinical trials targeting patients with disease positive for specific biomarkers. To address this ongoing transition in the field, presenters discussed advances concerning biomarker discovery, novel high-throughput methods for patient biomarker testing, ongoing efforts to standardize biomarker cutoffs across drug development programs and clinical trials, regulatory and approval perspectives, and expert opinions on the future of biomarker research and utilization. In this report, we summarize data presented at this meeting to update the field of oncology on how biomarker are becoming critical components of patient care.

The current state of cancer immunotherapy biomarkers. 

Gideon Blumenthal, MD (United States Food and Drug Administration [FDA], Silver Spring, Md) provided regulatory perspective concerning biomarkers, including current definitions, approvals, and future considerations. The FDA recognizes that biomarkers have many potential uses: diagnostic, predictive and prognostic, as well as being used to monitor pharmacodynamic and safety parameters, and susceptibility to adverse events. Knowledge of the context and intended designation of a biomarker are essential for validation and approval of a biomarker-based therapeutic agent. Moving forward, the FDA suggests that biomarker data should be collected and interpreted throughout clinical trials, and that drug development programs should integrate biomarker utilization into their initial hypotheses.

Patient biomarker assessment frequently involves a companion or complementary diagnostic. Companion diagnostics are required for approved therapies that target tumors positive for specific biomarkers, such as trastuzumab (Herceptin, Genentech, San Francisco, CA, USA) which was designed for use in patients with HER2-positive breast cancer [2]. By contrast, complementary diagnostics provide information to facilitate treatment risk-benefit analysis but are not essential for administration of the drug. An example of a complementary diagnostic providing clinical benefit is assessing programmed-death ligand 1 (PD-L1) status of patients with non-small cell lung cancer (NSCLC) enrolled within the CheckMate-057 clinical trial. Patients who received anti-PD-1 (programmed-death receptor 1) nivolumab (Opdivo, Bristol-Myers Squibb, New York, NY, USA) had increased median overall survival (OS) (n = 292, 12.2 mos [95% CI: 9.7 - 15]) compared to patients who received docetaxel (n = 290, 9.4 mos [95% CI: 8.1 - 10.7]). Within the nivolumab cohort, however, patients with tumor PD-L1 expression ≥ 1% (n = 123) had an increased objective response rate (ORR) compared to patients with PD-L1 expression < 1% (n = 108; 31% [95% CI: 23 - 40] vs. 9% [95% CI: 5 - 16], respectively) [3]. Development of novel diagnostics for assessing the status of specific biomarkers has accelerated alongside the design of new therapeutics that target tumors positive for specific biomarkers. Blumenthal indicated that harmonization of biomarker cutoffs for diagnostics will be essential in the future, as introduction of similar diagnostics with differential cutoffs can lead to confusion within the clinic. Harmonization efforts currently underway include a pre-competitive partnership lead by Friends of Cancer Research (Washington D.C., USA), who are working to harmonize methods assessing tumor mutational burden, and the Blood Profiling Atlas in Cancer (BloodPac, River Forest, IL, USA), which is developing standardized techniques to characterize biomarkers for liquid biopsies.

Lastly, biomarkers can also be utilized as a surrogate endpoints during both accelerated and regular regulatory approvals. In accelerated regulatory approval settings, biomarkers used as surrogate endpoints must be ‘reasonably likely’ to predict clinical benefit, quantified utilizing cutoff criteria, as initially hypothesized by investigators and later confirmed through clinical trial results. A biomarker used as a surrogate endpoint for a regular approval, however, must be directly measured and correlated with clinical benefit. Blumenthal provided HIV-RNA as an example of a successful biomarker acting as a surrogate endpoint for antiretroviral therapeutic efficacy.  

Despite their novelty to the immune-oncology landscape, biomarkers are rapidly becoming critical components of drug development programs. Priti Hegde, PhD (Roche-Genentech, South San Francisco, CA, USA) discussed the status of current biomarker utilization and provided a prospective view of where the field may be heading in the future. One of the primary goals of biomarker research is to identify an increased number of assessable markers that can more accurately guide therapeutic selection for enhanced immunotherapeutic response. For example, while anti-PD-1/PD-L1 therapies have demonstrated efficacy against a wide range of malignancies, a significant proportion of the patient population does not respond to therapy.

Optimized biomarker assessment methods that are currently available, including IHC and gene expression analyses, have remained accurate in predicting therapeutic response based on single analyte biomarkers. Given the complexity of tumor immune biology, biomarker research is moving towards studying multiple aspects of tumor biology rather than single markers, as well as building upon existing technologies to ascertain patient biomarker status. Tumor immune infiltration and tumor mutational burden (TMB) are increasingly recognized as predictive biomarkers for immunotherapeutic efficacy. For example, results presented at the 2017 European Society for Medical Oncology (ESMO) annual congress (abstract 1295O) revealed that among biomarker-evaluable patients with NSCLC being treated with anti-PD-L1 atezolizumab (Tecentriq, Genentech, San Francisco, CA, USA) in the phase III OAK trial, blood based TMB with a cutoff ≥ 16 was predictive of both OS and progression-free survival (PFS) benefit in the atezolizumab-treated patients compared to those receiving standard-of-care docetaxel. Importantly, blood TMB predictions were independent of patient PD-L1 status [4]. Dr. Hegde stated that future therapy selection methods will likely require assessment of a wide-range of biomarkers to determine efficacy and/or overcome therapeutic resistance in a given patient via the development of treatment decision algorithms. As biomarker-based drug development matures, the field will likely migrate towards a precision style of medicine, assessing disease based tumor biomarker status rather than tissue histogenesis. Precision medicine is likely to further expand utilization of combination immunotherapies, as immunotherapy-resistant tumors, including tumors with immune excluded or immune desert phenotypes, often convert to an inflamed phenotype when under combination immunotherapeutic pressure.

Dr. Hegde noted that the current trajectory of biomarker research will advance precision medicine techniques. Incorporation of research into clinical utilization, however, will require the development of a high throughput, universal biomarker testing platforms that can readily provide treatment selection recommendations. Towards this goal, Carl Morrison, MD, DVM (OmniSeq Inc, Buffalo, NY) discussed the ‘Immune Report Card’ (IRC) – a comprehensive immune profiling (CIP) platform capable of analyzing immune-related primary (on label) and secondary (off label or clinical trial) biomarker status in patient tumor samples (>10% neoplastic tumor nuclei, <50% necrosis, 10ng DNA per assay, 10 ng RNA per assay, unstained sections) by utilizing a pipeline consisting of multiple commercially available CLIA-certified NYS-approved techniques. Immune Report Card, as a comprehensive immune profiling (CIP) assay, uses multiple technologies to qualitatively measure checkpoint blockade and other anti-cancer immune response markers in formalin fixed paraffin embedded (FFPE) tumor tissue specimens - including PD-L1 pathway activation by FDA-approved immunohistochemistry (IHC), RNA-Seq, copy number/amplification, genomic instability including microsatellite instability (MSI) by PCR and mutational burden (MUB) by DNA-Seq, tumor infiltrating lymphocytes TILS by IHC and RNA-Seq to characterize “hot” versus “cold” tumors, and expression of T-cell receptor signaling (TCRS) genes that measure critical mechanisms of the anti-cancer immune response. The latter is especially important for patients who require secondary combination therapies after monotherapy failure. TMB assessment uses full exon sequencing of 409 genes to provide adequate depth of exome representation. The PD-L1/2 gene amplification is characterized using florescent in situ hybridization (FISH) and subsequently compared to PD-L1 expression using IHC and RNA-seq. IHC for CD3 and CD8 are equally important for comparison to RNA-seq values, but give the added value of the pattern of infiltration as infiltrating, non-infiltrating, or excluded.  Combined IRC results are compiled, analyzed, and presented in a clinical report that provides a molecular immune summary, response prediction to immune markers, and combination immunotherapy recommendations based on completed and in-progress clinical trials. Initial results have shown that IRC has been successful in identifying at least one clinical recommendation in more than 90% of samples, with the remaining 10% lacking recognizable markers indicative of an approved or in-progress therapeutic agent (ie. Immune deserts). Dr. Morrison provided an example of using IRC in practice, in the form of LBA18 from the 2017 ESMO Annual Congress – an assessment of efficacy of anti-LAG-3 (lymphocyte-activation gene 3) BMS-986016 in combination with nivolumab in patients with melanoma. Irrespective of PD-L1 status, ORR in patients receiving BMS-986016 with tumor LAG-3 expression ≥ 1% (n=33) was 18%, significantly increased compared to patients with LAG-3 expression < 1% (n=22; 5%) [5]. To further investigate the observed PD-L1 independency, Dr. Morrison’s group retrospectively analyzed IRC results from melanoma samples provided to OmniSeq (n = 306) to further investigate the relationship between biomarkers and BMS-986016 efficacy. IRC analyses revealed that increased LAG 3 expression correlated with increased CD8, but not PD-L1. These data support the researchers’ findings and suggest that increased BMS-986016 efficacy in LAG 3+ tumors is likely due to increased immune infiltration. IRC analyses also revealed that LAG 3 expression was generally increased in melanoma, ovarian cancer, lung cancer, sarcoma, uterine cancer, head and neck cancer, and breast cancer, thus identifying a potential patient pool for this novel therapy.

Development of liquid biopsy biomarker assessment platforms

The time requirement of current biomarker assessment technologies towards acquiring clinically-relevant results is a limiting factor for widespread clinical utilization. As such, it is imperative that in-development biomarker assays improve upon current timeframes. One method to expedite biomarker analyses is to utilize liquid biopsies of biofluids - such as plasma, serum, CSF, or urine - rather than conventional biopsies of solid tumor tissue, which are invasive, time-consuming to process, and evaluate only the sampled site. Jennifer Jones, MD, PhD (National Cancer Institute, Rockville, Md) discussed the development of nanoFACS and other nanoscale flow cytometric technologies capable of rapidly analyzing tumor-derived exosomes and associated cargo from patient plasma samples. Tumor exosomes are packets with cargo ­­­– including tumor-derived proteins and receptors, RNA, and metabolites – that provides insight into active, ongoing tumor processes potentially targetable by therapeutics. Similarly, immune cell-derived exosomes carry cargo that reflects the state of the immune cells that produce them. Exosomes are expected to provide information that will synergize with assays of other liquid biopsy components - such as circulating tumor cells or DNA - due to the high abundance of exosomes and the multidimensional attributes of exosome subsets. First, exosomes are highly abundant in plasma (>10 billion/mL) and significantly more prevalent than circulating tumor cells (typically <10/mL), allowing for greater detection and analysis efficiency. Second, circulating tumor DNA fragments are unidimensional, while exosome cargo is a multidimensional product that reflects the state of the cells of origin. One complication, however, is that exosomes are comparable in size to circulating lipoproteins, which are also highly abundant and carry extracellular RNA that can create false positives using current fluorescence-activated cell sorting (FACS) technologies.

To alleviate these complications, Jones’ group developed an efficient exosome analysis pipeline combining size-exclusion chromatography for exosome purification and using nanoFACS for analyses, which uses novel flow cytometry techniques that increase size resolution through light scattering to ≤ 100nm (compared to ≥ 400nm in traditional FACS) and through florescence to ≤ 40nm (compared to ≥ 100nm). NanoFACS improves upon traditional FACS by increasing sensitivity and reducing laboratory variability through multiplexed epitope analyses of single-fluorophore labeled components. Data presented using next-generation, high-sensitivity, flow cytometric methods demonstrate that single molecule or single label detection is needed to be able to accurately detect specific markers on exosomes and to enumerate marker-positive exosomes in clinical samples. There is significant interest in further enhancing exosome cargo detection with permeabilization methods that would enable flow cytometric evaluate of cargo inside of the exosomes. Such methods are not currently available, however, and one challenge concerning permeabilization is that exosomes are prone to rupture due to limited volume. Additionally, future methods utilizing patient urine, as opposed to patient plasma, for potential exosome analysis are also under investigation.

As an orthogonal approach, directly assessing the circulating proteome in liquid samples for biomarkers and test development is another promising approach. Circulating proteomic data can provide information about a patient’s disease state, interactions and regulation occurring between tumors, the tumor microenvironment, and the immune system. A group led by Heinrich Roder, D. Phil (Biodesix, Boulder, CO) has developed a deep MALDI ToF mass spectrometry technique that can characterize and measure serum proteins in a high-throughput, highly sensitive, reproducible manner. Deep MALDI ToF implements a way to use many hundreds of thousands of laser shots to increase sensitivity and at the same to reduce noise and CV of observable peaks compared to current mass spectrometry techniques. Interestingly, researchers incorporated modern ‘machine learning techniques’ to associate deep MALDI ToF data with clinical phenotypes directly, opposed to matching results with known biomarkers. For example, proteomic tests provide a characterization of a patient’s serum samples – such as early or late progression – that can later be used for prognostic and predictive tests. Deep MALDI ToF phenotypic results can also be utilized to improve therapeutic selection and prognosis, as well as to identify markers that offer potentially novel drug targets. Deep MALDI ToF-based tests can also be designed to predict response and/or resistance of specific therapeutics, such as immune checkpoint inhibitors or HD-IL2. For example, patients with melanoma who were treated with either nivolumab as a monotherapy or in combination with anti-CTLA-4 ipilimumab underwent testing using the Biodesix platform. In blinded validation these tests were able to accurately predict OS independent of any other known biomarker including PD-L1. Deep MALDI ToF accuracy was further verified, in a similar manner, in patients with NSCLC treated with nivolumab. Interestingly, circulating proteome results analyzed for clinical benefit of nivolumab were unable to discern a benefit from docetaxel in patients with NSCLC, indicating that proteomic tests are predictive for anti-PD-1 checkpoint inhibition versus docetaxel. Proteomic tests can also identify specific cellular processes associated with therapeutic response and resistance. Melanoma patients resistant to both anti-PD-1 and HD-IL2, for example, were associated with elevated complement, acute phase reactants, and wound healing processes, a result that was confirmed by other research groups using orthogonal techniques.

Using next-generation sequencing to improve therapeutic efficacy

Using single analyte biomarkers – such as PD-L1 – for therapeutic selection presents significant spatial, temporal, and heterogeneity limitations. Moving past single analytes and utilizing multiple biomarkers for therapeutic selection is a paradigm shift facing the field of oncology. In this regard, next-generation sequencing techniques are capable of concurrently assessing multiple biomarkers in a single analysis. Tobias Schatton, PharmD, PhD (Brigham and Women’s Hospital, Boston, MA, USA) presented an example utilizing gene expression analyses – including the tumor inflammation signature (TIS) panel developed by Nanostring Technologies, Inc. (Seattle, WA, USA) – to investigate patient response to immune checkpoint inhibitors. The TIS panel measures expression of 18 functional genes and 10 housekeeping genes associated with IFN-ƴ release, T cell exhaustion, T cell/NK cell presence, and antigen-presenting cells. Tumor biopsy samples collected from 183 patients with stage IV melanoma treated with anti-PD-1 therapies (pembrolizumab, n=128; nivolumab, n=55) were analyzed using three gene expression panels representing ~700 genes each and including TIS markers. Biopsy samples simultaneously underwent PD-L1 IHC for assay validation and technical comparison. Consistent with previous studies, gene expression/TIS scores were increased in patients with PR/CR compared to patients with PD/SD. TIS results indicating ICI responsiveness strongly correlated with increased PFS compared to patients identified by TIS to be resistant to ICI therapy. Additionally, a higher proportion of ICI responders were accurately predicted using TIS signatures compared to available IHC methods.

Gene expression/TIS analyses also have potential to stratify responders based on distinct therapeutics targeting the same pathway or molecule, such as pembrolizumab and nivolumab. Recent structural studies suggest pembrolizumab and nivolumab might have differences in PD-1 binding, which could potentially account for differential efficacy towards specific cancers. Intriguingly, gene expression/TIS analyses revealed that pembrolizumab and nivolumab responders had distinguishable gene expression profiles, including in lymphoid signatures. Using this information, Schatton’s group identified a gene expression signature capable of stratifying patients who received pembrolizumab, based on increasing both PFS and OS.  The same signature was not as effective in stratifying patients who received nivolumab. Overall, these data suggest that gene expression analyses incorporating TIS markers offer enticing potential to discover predictive biomarkers for ICI therapies as well as other therapies. Ongoing studies aim to identify unique characteristics of patients who responded to ICI therapy but had low TIS signatures, and patients who did not respond to ICI therapy but had high TIS signatures.

Biomarker-based drug development

Drug development programs have begun developing therapeutics targeted towards specific biomarkers. Pembrolizumab, for example, has gained FDA approval for treating any patient with a tumor positive for the dMMR/MSI-H biomarkers [1]. There remain many biomarkers that can be similarly targeted in a therapeutic regimen. Public T cell receptors (pTCR), or TCRs shared between individuals, are enticing biomarkers for drug development.  Approximately 511 ‘private’ TCRs exist, but only 3-4 million are present in a given individual. Private TCR variability across individuals reduces their usefulness for biomarker-based therapeutics. About 10,000 ‘public’ TCRs, however, are shared between individuals.  Public TCRs are involved in immune responses to chronic viral infections, but their importance in cancer settings has not been characterized.

Hailing Lu, MD, PhD (Immune Design, Seattle, WA, USA) discussed the discovery and utilization of an NY-ESO-1-specific pTCR as a therapeutically targetable biomarker in patients with NY-ESO-1-positive cancers. NY-ESO-1 is an immunogenic tumor antigen commonly expressed in melanoma, ovarian cancer, and sarcoma. Patients with soft tissue sarcoma (n=25) received the lentiviral vector ZVex (Immune Design, Seattle, WA, USA), which targets dendritic cells and promotes NY-ESO-1 expression. NY-ESO-1-expressing dendritic cells subsequently promote anti-NY-ESO-1 T cell responses against NY-ESO-1-positive tumors in treated patients. Median OS in patients who received ZVex has not been reached (95% CI: 12.3 – NR), a superior outcome than that observed in patients receiving standard-of-care pazopanib (12.5 mos), eribulin (13.5 mos), or trabectedin (12.4 mos). Patient tumor samples from the ZVex cohort were then analyzed using next-generation sequencing and three NY-ESO-1-specific pTCR were identified. Identified NY-ESO-1 pTCR were also detectable in both the blood and tumor cells post treatment. Analysis of the five NY-ESO-1-specific pTCR found that they shared similar CD3 sequences but were derived from different Vβ families. Retrospective analyses revealed that pTCR+ patients with NY-ESO-1 positive tumors (n = 26) had improved OS compared to pTCR- patients (n = 29, p = 0.04). Interestingly, while there was some concordance (~70%) between NY-ESO-1 pTCR expression and NY-ESO-1-specific T cell presence as measured by ELISPOT, a number of pTCR+ individuals were ELISPOT negative (15%), suggesting that measurement of pTCR could potentially complement ELISPOT assay in evaluating anti-tumor immunity.

Targeting biomarkers with specific therapeutics will require co-development of companion diagnostics for both therapeutic efficacy and FDA approval. Heather Hirsch, PhD (Jounce Therapeutics, Cambridge, MA, USA) provided an example of simultaneous development of therapeutics and diagnostics, detailing the advancement of the novel anti-inducible T cell co-stimulator (ICOS) antibody JTX-2011, as well as ICOS-specific detection assays. JTX-2011, which is currently being assessed for efficacy within the phase 2 ICONIC clinical trial in patients with advanced solid tumors, is believed to promote anti-tumor activity through two mechanisms: activation of T effector cells, and the selective reduction of intratumoral T reg cells through an ADCC-dependent mechanism. Since pre-clinical experiments suggest that JTX-2011 efficacy is correlated with increased percentage of ICOS-expressing T cells within the tumor, assays to measure ICOS level are being developed as a potential predictive biomarker. Researchers analyzed TCGA databases and found that patients with head and neck, lung, triple-negative breast, stomach, or cervical cancers generally had elevated ICOS RNA levels and thus these indications were prioritized as indications of interest.  As such, the ICONIC trial (NCT02904226) was designed to include a subset of these malignancies, and as a further enrichment strategy each ICONIC cohort includes at least ten patients who were classified as ICOS-high based on an ICOS IHC assay. In regard to clinical companion diagnostics, Hirsch discussed three strategies being utilized to assess ICOS expression: IHC, qPCR, and a Nanostring Technologies-based expression analysis (Nanostring Technologies, Inc., Seattle, WA, USA). A chromogenic IHC assay utilizing an anti-ICOS antibody is currently being developed to stratify ICONIC trial patients.  Preliminary results during assay development suggest ICOS proteins, as measured by IHC, strongly correlates with ICOS RNA expression (p < 0.0005). Towards development of a qPCR-based assay, Hirsch’s group analyzed gene expression across multiple solid tumor indications from TCGA and identified 11 genes highly correlated with ICOS expression. This 11 gene signature is currently being integrated into a Nanostring-based platform and thus far has promising correlation to ICOS RNA measured by qPCR (Spearman’s ρ= -0.85).

Characterizing tumor immune status

Tumor CD8+ T cell infiltration is a vital component of immunotherapeutic response and strongly correlates with survival benefit in patients with colon cancer, bladder cancer, head and neck cancers, NSCLC, and melanoma, among others. Despite these correlations, however, standardized and validated clinical methods to assess tumor immunity are lacking. Fabienne Hermitte, PhD (HalioDx, Marseille, France) discussed the progress of a variety of tumor immune assessment methods. The first described test is a new in vitro diagnostic assay, Immunoscore® (HalioDx, Marseille, France), which combines direct tissue staining, digital pathology, and enumeration of CD3+ and CD8+ cells in both the tumor and the invasive margin. Immunoscore® Colon test (CE-IVD in Europe, CLIA for US) is utilized to predict the risk of relapse in patients with stage 2 or 3 colon cancer by assessing tumor immunity. The design of a prospective study – PROSCORE –  aimed to assess the impact of Immunoscore® Colon testing on adjuvant treatment decisions, was presented. Additionally, Hermitte’s group is currently developing Halioseek® (HalioDx, Marseille, France)  which, reflecting the growing appreciation of PD-L1 as a predictive biomarker for anti-PD-1 response, combines PD-L1 and CD8+ staining. In samples collected from patients with NSCLC, Halioseek® has comparable PD-L1 detection efficacy to other diagnostic tests SP263 (1% cutoff: OA = 97.7% [95%CI: 94.7 – 99.0]) and 22C3 (1% cutoff: OA = 91.2% [95%CI: 86.7 – 94.3]). Collaborative studies are ongoing to assess the added predictive value of CD8 quantification and PD-L1 – CD8 proximity measurements. Finally, in collaboration with Nanostring Technologies, Inc (Seattle, WA, USA), the Immunosign assay allows for multiplexed gene expression analyses characterizing tumor immunity. Dr. Hermitte presented an example of utilizing Immunosign to assess immune signatures in patients with non-Hodgkin lymphoma receiving the CAR T therapy axicabtagene ciloleucel (Yescarta, Kite Pharma, Los Angeles, CA, USA). Gene expression profiles comparisons of pre- and post- axicabtagene ciloleucel treatment biopsies from 14 patients showed profound changes in gene expression within the tumor environment after infusion. Among other markers, CTLA4, GZMA and CCL5 were upregulated genes post-treatment. Researchers hope that over time, compiled Immunosign, Immunoscore® and Halioseek® results will be able to offer insight into potential biomarkers predicting efficacy of immunotherapy treatments.

Immune infiltration and immune marker expression can vary spatially within an individual tumor as well as between metastases. Spatial immune variability can impact biomarker and receptor interactions affecting immune activation. As such, information concerning intratumor immune localization could help optimize therapy selection. Svenja Lippok, PhD (Definiens AG, Munich, Germany) discussed using a multiplex IHC approach in combination with digital image analysis to spatially and functionally characterize patient tumor immune status. This approach measures spatial immune marker expression profiles, as well as potential cellular interactions, using multiplexed bright field IHC capable of assessing both single and double positive cells which are detected by automated image analysis solutions. This standardized approach can be applied to full resections as well as to small, solid biopsies. Tumors are initially classified as inflamed or non-inflamed based on CD8, PD-L1, and FoxP3 IHC detection in epithelial versus non-epithelial regions. After the initial classification, CD68, PD-1, granzyme B, and CD3 IHC results are utilized to further delineate the tumor immune status. Dr. Lippok presented two examples of using spatial assessment to determine immune status of NSCLC tumors. In one, analysis of a NSCLC tumor that displayed increased CD8+ T cell infiltration and tumor PD-L1 expression revealed that despite CD8 and FoxP3 positivity, both cell populations were not in spatial proximity, indicating absence of potential influence on CD8+ T-cells. Based on these spatial considerations, this tumor would be classified as inflamed and would represent a good candidate for PD-1/PD-L1 checkpoint monotherapy. In another example, a NSCLC tumor was classified as ‘immune ignorant’ after displaying low CD8+ T cell infiltration and tumor PD-L1 but increased macrophage infiltration as denoted by the CD68 signals, ultimately suggesting that this patient would not be a good candidate for mono checkpoint immunotherapy. The ability to spatially visualize immune interactions within a tumor is an important advance that can improve understanding of tumor immunity for therapy selection. 

Development and assessment of biomarkers for CAR T therapies

It is becoming widely accepted that T cell classification should be broader than the traditional classification – CD8+, Tregs, etc – and instead have a varied, more fluid, classification based on receptor expression. T cell polyfunctionality assessment is difficult, however, due to the requirement for single cell detection methods. As CAR T cells can express a variety of receptors during production and expansion, characterizing their functionality may help predict efficacy. Sean Mackay (IsoPlexis, Branford, CT, USA) discussed the development of a novel high-throughput, single-cell technology capable of assessing individual T cell signatures. Researchers developed the IsoCode single-cell chip that incorporates over 12,000 microchambers, which integrates an ELISA based antibody barcode array that are used to capture deep functional analysis on ≥40 secreted proteins simultaneously per cell. Thousands of single T cells are incorporated into individual microchambers and incubated for 16 hours to capture the full range of secreted cytokines from each individual cell. After incubation, sandwich ELISA procedures reveal the captured cytokines, and a software and informatics pipeline enables validated analysis for the simple processing of complex images and high-dimensional data. Results from single-cell chip analyses are highly correlated with lower throughput single cell assays, including flow ICS and FLUOROSpot. Using the IsoCode Chip, researchers have captured data to determine the polyfunctional strength index (PSI), or functional strength, of individual T-cells, across thousands of cells at a time. By defining the T-cell functional strength, the PSI metric has predicted complete or partial patient response to pre-infusion axicabtagene ciloleucel with statistical significance. Data from the IsoCode assay that shows that patients with highly functional T-cell subsets in vitro tend to have strong responses to CAR-T therapies in vivo (p = 0.01), and the patients with inactive T-cell subsets have weaker responses. Effector, stimulatory, chemoattractive, regulatory, and inflammatory cytokines contributing to PSI are measured, and non-redundant cytokines (IL-8, IL-5, IL-17α, IFN-ƴ, and MIP-1α) were elevated in axicabtagene ciloleucel responders compared to non-responders. Future studies aim to utilize the IsoCode single cell chip and PSI analyses to characterize signatures that can predict response to a variety of cell based immunotherapies. 

Consolidation of Biomarker Research: Network of Cancer Immune Monitoring and Analysis Centers and Cancer Immunologic Data Commons

Immunotherapies have revolutionized treatment of diverse cancers. In order to benefit more cancer patients from treatment regimens involving immunotherapies, research efforts have been made to develop clinical tools for enabling biomarker-guided immunotherapies and their combinations with other therapies that would achieve the least toxic and the most efficacious, durable clinical outcomes. As a consolidated biomarker research strategy towards accomplishing this goal, establishment of a network of four Cancer Immune Monitoring and Analysis Centers [CIMACs at Dana-Farber Cancer Institute (Boston, MA, USA), Precision Immunology Institute and the Tisch Cancer Institute at Icahn School of Medicine at Mount Sinai (New York, NY, USA), Stanford Cancer Institute (Stanford, CA, USA), and University of Texas MD Anderson Cancer Center (Houston, TX, USA)] and one Cancer Immunologic Data Commons (CIDC at Dana-Farber Cancer Institute) has been supported by the NIH Cancer MoonshotSM Program since September 2017. The Purpose of this CIMACs-CIDC network is to provide the infrastructure for pre-, on- and post-treatment biomarker studies in NCI-supported clinical trials involving cancer immunotherapy. It is aimed to identify multiple immune and tumor marker candidates that can later be validated in large prospective trials to optimize therapies for cancer patients. Technology platforms available at the CIMACs will enable comprehensive analyses of mechanistically-informative, diverse variables, including genomic, phenotypical and immunological correlates, by employing analytically-validated, standardized, and harmonized assays as well as novel specialty assays that will be developed and standardized within the network. The CIDC will coordinate standardized data collection, curation, and annotation for further analyses, and share these data with the immuno-oncology community when appropriate.

Discussion

Biomarker complexity is a strength, not a weakness

A primary goal of many oncologists is to integrate precision medicine techniques into routine practice. The recent FDA approval of pembrolizumab for treating patients with MSI-H/dMMR-positive tumors marks a paradigm shift towards classifying disease based on biomarker status rather than tissue histogenesis [1]. The advent of biomarker-based disease classification makes it necessary to characterize novel biomarkers and improve analytic techniques. Towards this goal, next-generation sequencing and single-cell analyses are dramatically altering biomarker classification and signaling a move away from single molecule markers towards more broad assessments. Other strategies, such as the use of liquid or fresh sample biopsies, will provide more accurate information in relation to active, ongoing tumor processes and help optimize treatment selection. These advances in biomarker assessment are critical to facilitate our understanding of tumor complexity, evidenced by the fact that each metastasis can have a different biomarker profile from the parent tumor. While these improvements will need to withstand the rigor of peer review, they offer potential towards reduction of the discrepancy between the amount of available pre-clinical knowledge and the number of clinical options for patients.

Embracing immunology within oncology

Historically, cancer treatment was primarily limited to chemotherapy and radiation, and tumors were classified based on tissue location and morphology. The success of cancer immunotherapy has challenged existing beliefs and prioritized the understanding of immunological mechanisms for drug development and treatment selection. The challenge for the field will be to integrate immunology and classical tumor biology to benefit patients. For example, despite tumor immunity being increasingly recognized as vital for generating response to cancer immunotherapies, cancers are still classified using the TNM staging protocol, which does not address the immunological features of the tumor or microenvironment. Tumor biology is also critical to immunological understanding and therapeutic selection. Immune silent tumor phenotypes, including tumors with immune cell stromal exclusion, for example, display efficient T cell activation and recruitment, but have tumor-specific attributes that occlude immune cell infiltration. Additionally, acquired immunotherapy resistance is likely a tumor-derived defense mechanism, as immunotherapeutic pressure can select for tumor cells resistant to therapy. In order to advance cancer immunotherapies further into clinical practice, it will be important to assess tumor immunology within the context of classical cancer biology.

Conclusions

The cancer biomarker field is expanding and multi-factorial assessment technologies are becoming increasingly necessary. It is also clear that surveying a wide-array of tumor biomarkers, rather than single analytes such as PD-L1, can better optimize treatment selection for individual patients. Data from this program indicate that researchers are heavily-invested in biomarker discovery, testing, standardization, and incorporation into drug-development programs, with the overall goal of increasing immunotherapeutic efficacy and the number of responders. Lead by leaders in the field of cancer immunotherapy, these data and presentations help strengthen the transition of biomarkers to the forefront of cancer treatment.

References

  1. Pembrolizumab (KEYTRUDA) prescribing information [https://www.merck.com/product/usa/pi_circulars/k/keytruda/keytruda_pi.pdf]
  2. Maximiano S, Magalhaes P, Guerreiro MP, Morgado M: Trastuzumab in the Treatment of Breast Cancer. BioDrugs 2016, 30(2):75-86.
  3. Borghaei H, Paz-Ares L, Horn L, Spigel DR, Steins M, Ready NE, Chow LQ, Vokes EE, Felip E, Holgado E et al: Nivolumab versus Docetaxel in Advanced Nonsquamous Non-Small-Cell Lung Cancer. N Engl J Med 2015, 373(17):1627-1639.
  4. Gandara DR, Kowanetz M, Mok TSK, Rittmeyer A, Fehrenbacher L, Fabrizio D, Otto G, Malboeuf C, Lieber D, Paul SM et al: 1295OBlood-based biomarkers for cancer immunotherapy: Tumor mutational burden in blood (bTMB) is associated with improved atezolizumab (atezo) efficacy in 2L+ NSCLC (POPLAR and OAK). Annals of Oncology 2017, 28(suppl_5):mdx380-mdx380.
  5. Ascierto PA, Bono P, Bhatia S, Melero I, Nyakas MS, Svane IM, Larkin J, Gomez-Roca C, Schadendorf D, Dummer R et al: LBA18Efficacy of BMS-986016, a monoclonal antibody that targets lymphocyte activation gene-3 (LAG-3), in combination with nivolumab in pts with melanoma who progressed during prior anti–PD-1/PD-L1 therapy (mel prior IO) in all-comer and biomarker-enriched populations. Annals of Oncology 2017, 28(suppl_5):mdx440.011-mdx440.011.