Immunodynamics

Bernard A. Fox

ITAs, by definition, modulate the immune response with systemic and local effects. Assessment of the peripheral blood, may be, but is not always, reflective of the changes within the tumor. Though serial tumor biopsies are therefore of fundamental value in monitoring the effects of novel ITAs, the baseline tumor sample may also portend prognostic significance, as the tumor-immune infiltrate has long been associated with improved outcomes. In 2006, Galon and colleagues, using digital imaging and image quantification software [2], reported a strong and highly significant correlation in colon cancer between increased survival and the presence of immune cell densities (CD3⁺, CD8⁺, Granzyme B⁺, CD45RO⁺ cells) at the invasive margin and center of the tumor. A subsequent study further substantiated these findings in a larger cohort and correlated strong infiltration with disease-free, disease-specific, and overall survival [3]. Importantly, objective assessment of T-cell infiltrates (CD8⁺ and CD45RO⁺ cells), or the immune score was a significantly better prognostic biomarker than tumor-node-metastasis (TNM) staging, recognition of the central role immunity plays in this disease [4]. The observation that immune infiltrates are associated with improved outcomes is not limited to colon cancer. More than 100 publications have reported associations between immune infiltrates and improved outcomes for patients with at least 18 different cancer histologies [5]. Currently the Society for Immunotherapy of Cancer is leading an international effort to validate these findings in a retrospective evaluation performed at 23 centers in 17 countries on tissues from 5000 patients [6].

If applied, determination of the immunoscore is obtained on formalin-fixed paraffin-embedded (FFPE) tissue that has both a portion of invasive margin and tumor center, and the greatest degree of immune infiltrate is selected. Automated immunohistochemistry (IHC) is performed on 2 serial sections with one slide stained with anti-CD45RO antibody and one slide with anti-CD8 antibody [3, 7]. As detailed in the following sections, measurement of the immunodynamic effects within the tumor site extend beyond assessment of the T-cell infiltrate. The additional analysis of the immunoprofile of tumors will likely identify other prognostic markers that may be histology dependent. For example, tumor-infiltrating myeloid cells have been associated with poor prognosis in some cancers [8] as well as markers for antigen-presenting cells, B cells, Tregs, and activation and inhibitory or suppressive molecules [9]. It is believed that the immunoprofile correlates with the mutational status of a patient’s tumor: high mutational status would be expected to result in a strong immunoprofile. A limitation of these assessments is the potential heterogeneity of a given tumor, which could alter immune infiltrates of the primary or metastatic sites. Tumor heterogeneity is widely recognized as a hurdle for cancer immunotherapy [10] and may limit the current strategy for immunoprofiling on a single specimen or biopsy due to the confounding nature of intratumor heterogeneity (i.e., variation within a tumor lesion) and intertumor heterogeneity (i.e., variation between metastatic sites). An alternative approach might be to apply novel imaging methods and reagents with short half-lives that identify specific markers and could provide real-time in vivo imaging of an immunoprofile for all metastatic sites simultaneously.

As detailed in the remainder of our review, assessment of the tumor in addition to prognostic value provides predictive import to response to immunotherapy as exemplified by a 3- to 4-fold higher response rate to PD-1/PD-L1 pathway targeted agents among patients with PD-L1–positive tumors. It is anticipated that immunoprofiling of tumors will become a routine evaluation for predictive biomarkers to guide patient selection for specific agents and combination therapies.

Guido Kroemer and Laurence Zitvogel

The antitumor activity of conventional cancer therapies is dependent, at least in part, on the immune response. However not all therapies induce equivalent immune responses in patients, as the manner of cell death induced may be silent, tolerogenic, or immunogenic [11, 12]. Immunogenic cell death (ICD) inducers including radiation therapy, anthracyclines, and oxaliplatin, as well as unconventional cytotoxic agents (e.g., cardiac glycosides, bortezomib, crizotinib) are endowed with the capacity of stimulating premortem stress responses [13, 14, 15, 16, 17, 18]. ICD [19, 20] generates an endoplasmic reticulum (ER) stress response and the activation of the autophagy machinery, both producing a series of damage-associated molecular pattern molecules (DAMPs) culminating in Cxcl10 release promoting the recruitment of intratumoral Th1-Tc1 cells indispensable for tumor control [21, 22].

Monitoring of ICD requires sampling of the tumor itself, ideally, by an excisional biopsy, a core biopsy, or least preferable, a fine-needle aspirate. Immunodynamic monitoring of ICD relies on recent results indicating that the ER stress response, autophagy, and late apoptosis can all be detected in tumor cells at diagnosis and correlate with immune infiltrates and eventually with patient survival. IHC detection of phosphorylated eIF2α (ER stress-response related-marker) [23], nuclear high mobility group box 1 (HMGB1; late apoptosis-related marker) [24], and light chain 3 beta (LC3-B; autophagosome-related marker) [21, 25] are feasible and reliable on FFPE, allowing determining of a relationship between CD8 and forkhead box P3 (Foxp3) or CD8 and CD68 ratios and response to cytotoxic compounds. As our understanding of ICD deepens, the markers that are measured both by IHC and potentially by gene expression that best measure the postchemotherapy immune response will be refined.

Silvia Formenti

Combining immunotherapy with radiation therapy, similar to immunogenic chemotherapy, has demonstrated clinical activity [26]. Radiation therapy can induce ICD in a dose-dependent manner and enhance ICD of some chemotherapy agents when used in concurrent regimens [26]. In metastatic cancer, combining immunotherapy with radiation therapy to a metastatic site can convert into systemic responders patients who have previously failed to respond to the same immunotherapy. In such setting, the emergence of an abscopal response (“ab-scopus”, i.e., away from the target, outside the radiation field) appears to be an unequivocal marker of immune response [27]. The main rationale for combining tumor radiation therapy with immunotherapy is to convert the irradiated tumor into an in situ, individualized vaccine [28]. T-cell receptor (TCR) repertoire analysis may provide proof of successful vaccination with the emergence of antigenic spread postradiation, detectable by demonstrating the expansion of memory T cells specific to tumor antigens that were not recognized before radiation therapy [29].

The localized nature of radiation therapy offers a unique opportunity to follow the evolution of the irradiated tumor microenvironment by serial biopsies and to determine the relationship between radiation-induced changes and the development of abscopal effects. For example, biopsies obtained after topical imiquimod treatment of basal cell carcinoma in a randomized, placebo-controlled trial, identified 637 genes induced by imiquimod (a toll-like receptor-7 agonist). Four distinct pathways associated with imiquimod-mediated tumor rejection were identified and led to the definition of the immunologic constant of rejection (ICR) [30]. According to the ICR hypothesis, common effector pathways suggestive of an innate immune infiltrate are upregulated in regressing tumors. A similar signature may develop postradiation and serve as a biomarker to predict which patients will generate antitumor immune responses sufficient to achieve abscopal effects.

A second example of the immunodynamic effects of radiation therapy builds upon the combination with anti-CTLA-4 antibodies [26, 27, 28, 31], which requires CD8 T-cell expression of the immune activation marker, NKG2D [32], expressed on NK cells. In patients, blockade of NKG2D is mediated by soluble major histocompatibility complex class I-related chain A (sMICA), which is released by some tumors and reaches high levels in the serum [33]. Therefore, sMICA may be the first biomarker of combination radiation and immunotherapy [34], and as radiation therapy upregulates MICA on the surface of tumor cells, serial biopsies of tumors before and after radiation therapy and ipilimumab are required to assess this immunodynamic endpoint with increased expression of MICA [35].

Cellular therapy, adoptive T cells

Immunodynamics assays of T-cell therapy to date illustrate a correlation of clinical response with robust in vivo expansion of T cells, intratumoral accumulation [43], and long-term persistence of engineered cells, as well as strong and transient elevation in systemic levels of proinflammatory cytokines, notably systemic interleukin 6 (IL-6) and C-reactive protein (CRP), coincident with the peak kinetics of in vivo T-cell expansion [40, 41], as well as cytokine-release–associated toxicity [44].

Don Benson, Lewis Lanier, Jeffrey Miller, and Eric Vivier

NK cells are a population of innate lymphoid cells (ILC) that provide host defense against viruses, bacteria, parasites, and fungus, as well as immune surveillance for cancer. In the peripheral blood of healthy individuals, NK cells comprise between 10–20 % of the lymphocyte population [45]. In humans, NK cells are identified as CD3^–, CD56⁺ lymphocytes [45]. The NK-cell population found in peripheral blood includes immature NK cells, identified as CD3^–, CD56^bright, CD16^– lymphocytes, and mature NK cells, which are CD3^–, CD56⁺, CD16⁺. CD56 and CD16 can also be expressed on subsets of myeloid cells in peripheral blood, which can result in the misidentification of the CD3^–, CD56^–, CD16⁺ NK cell. A more definitive identification of this NK-cell subset can be obtained by costaining for CD7 (CD3^–, CD56^–, CD7⁺, CD16⁺), which is expressed on all NK cells, but not myeloid cells. NK cells express an extensive repertoire of activating and inhibitory receptors, including KIR2DL, KIR3DL, and CD94-NKG2A receptors, which recognize human leukocyte antigen (HLA) class I molecules as ligands and suppress NK-cell activation. Activating or coactivating receptors on NK cells include CD16, a low-affinity receptor for immunoglobulin G (IgG) that is responsible for antibody-dependent cellular cytotoxicity (ADCC); the NKG2D receptor that recognizes stress-induced ligands MICA, major histocompatibility complex class I-related chain B (MICB), and the UL1-6 binding proteins (ULBP1-6, CD226, DNAX accessory molecule-1; DNAM-1), which recognizes CD112 and CD155; CD244 (2B4), which recognizes CD48, and others. NK cells mediate immune protection by release of perforin, granzymes, cytokines, and chemokines, in particular IFN-γ. Upon activation, NK cells can produce abundant amounts of tumor necrosis factor alpha (TNF-α), granulocyte-macrophage colony-stimulating factor (GM-CSF), IL-10, and chemokine (C-C motif) ligands 3 and 4 (CCL3, CCL4) [46, 47].

Based on the premise that autologous NK cell–based therapies are limited by self-tolerance mediated by inhibitory killer-cell immunoglobulin-like receptor (KIR) recognizing residual self–class I major histocompatibility complex (MHC) molecules on tumors, adoptive transfer of haploidentical NK cells with exogenous IL-2 has been used to treat patients with acute myeloid leukemia (AML), non-Hogdkin lymphoma, and ovarian cancer [48, 49, 50]. IL-15 administration may be optimal for stimulating the selective expansion of NK cells and not Tregs. Bispecific killer engagers (BIKEs), which can impart antigen-specific selectivity to NK cells as one created from single-chain Fv specific for CD16 and CD33 (expressed on AML targets) can trigger CD16 on NK cells to kill primary AML and induce cytokine production [51, 52]. The standardization of methods to define and measure NK-cell persistence, expansion, and function after adoptive therapy will facilitate the comparison of different NK-cell products and treatment platforms.

Monoclonal antibodies, tumor-targeting

Michael Kalos, Ignacio Melero, Antoni Ribas, Paul C Tumeh, and Jedd Wolchok

The success of therapeutics that block the PD-1/PD-L1 inhibitory axis has ushered in a new era in oncology, with these agents likely to become the backbone of cancer therapy in a wide range of cancer types. The immune checkpoints, CTLA-4 and PD-1, are cell-surface receptors that upon binding to their ligands, trigger downstream signaling pathways that serve to inhibit T-cell activity [63, 64]. Therapies that target PD-1 or PD-L1 have shown significant clinical activity in patients with advanced melanoma [65, 66, 67, 68], non–small cell lung cancer [69, 70, 71, 72], renal cell cancer [73, 74, 75], Hodgkin lymphoma [76], head and neck cancers [77], and metastatic bladder cancer [78, 79].

Leading this therapeutic class are anti-PD-1 agents pembrolizumab (Keytruda, Merck & Co), and nivolumab (Opdivo, Bristol-Meyers Squibb Co), which the FDA approved for unresectable or metastatic melanoma in September and December 2014, respectively. These build on a previously approved immune-checkpoint inhibitor, anti-CTLA4 (Yervoy, Bristol-Meyers Squibb Co) in 2011 for the same indication. Based on these findings and the expectation that immunotherapy will impact all fields of oncology, Science magazine selected cancer immunotherapy as the breakthrough of 2013 [80].

Critical to understanding how these therapeutic antibodies promote tumor rejection is both the identification of cell types that are altered during blockade and the subsequent mechanistic analysis of how these cell types promote or inhibit tumor regression [81]. Recent investigatory efforts have shown that these antibodies are more clinically effective when preexisting immunity is present in tumors; that is, when tumors have already been identified by the immune system, functionally as a result of local levels of IFN-γ leading to the expression of PD-L1 in the tumor and stroma [75, 82]. Hence, approaches that determine the density, phenotype, and location of immune cell types within the tumor microenvironment and respective PD-L1 expression levels represent one key approach to understanding which cell types and their discrete microenvironments promote or inhibit tumor rejection.

Platforms based on this approach include slide-based, quantitative IHC [5, 82, 83, 84] and quantitative multiplexed IHC [85, 86] on tumor samples, as well as assays that reveal in situ gene expression, including transcriptomic profiles on microarrays performed using laser captured microdissected tissue [87, 88]. As the target of interest, PD-L1 may be expressed in the tumor, the stroma, or both, and hence, spatiotemporal information is required, including the invasive tumor margin, stromal components, tumor center, and perivascular niches. As a consequence, optimization in small samples (such as fine-needle aspirates and small core biopsies) represent significant challenges with this approach [65, 66]. PD-1 expression shows superior AUC values and predictive value when compared to single agent PD-L1. Furthermore, the use of different anti-PD1 and anti-PDL1 primary antibodies and the vast number of detection systems available and used by different labs have made it difficult to harmonize IHC read-outs The cellular sources of PD-1 and PD-L1 must be defined and then systemically investigated according to clinical response. Additionally, multiplexed IHC approaches can be used generate to multiparametric, spatially resolved information and capture spatiotemporal interdependencies that are clinically relevant. the presence of constitutive PD-L1 expressing cancer cells without TILs present in the tumor correlates with non-responsiveness to anti-PD1 therapy. The presence of CD68⁺PD-L1⁺ cells at the invasive margin is significantly associated with the presence of interfacing or neighbouring CD8⁺ and PD-1⁺ cells. Determining the relative presence of PD-1 on CD8⁺ and CD8^– cells in tumors at baseline remains largely unknown but potentially very important in terms of predicting response. Adding another dimension to understanding the preexisting immunity in the tumor microenvironment is the application of TCR next-generation sequencing based on the unambiguously identifiable TCR-β CDR3 region using genomic DNA from tissue samples that can be used to quantify the diversity and repertoire of the T-cell infiltrate in tumor tissues [89].

In serum, absolute lymphocyte count, baseline eosinophil count, CRP, lactate dehydrogenase, and white blood cell count have been shown to correlate with improved survival in patients receiving ipilimumab (anti-CTLA4) [90, 91, 92, 93]. Ongoing studies are investigating the correlative relationships of serum markers with treatment outcome to therapies that block the PD-1/PD-L1 axis with, thus far, the tumor microenvironment expression of PD-L1 demonstrating the tightest relationship with response.

Josh Brody and Aurelien Marabelle

The in situ vaccination (ISV) strategy consists of intratumoral administration of immunostimulatory products to stimulate antitumor immunity. As with other cancer vaccines, ISV presents tumor-associated antigens in an immunogenic context by using the tumor itself as an antigen source. ISV is actually the first cancer immunotherapy paradigm ever tested, as it has been used in clinical practice since the end of the XIX^th century in Europe and in the United States [94]. In the modern era, ISV can be performed with many types of immunostimulatory products, including clinical-grade, live, infectious pathogens such as Bacillus Calmette-Guerin in the treatment of cutaneous metastatic melanoma [95] as well as its peritumoral use in superficial bladder cancer⁷⁰. Toll-like receptor 7 (TLR7) and TLR9 agonists mimicking bacterial nucleic acid have demonstrated their ability to generate antitumoral immunity upon direct administration to vulvar intraepithelial neoplasia [96] and low-grade lymphomas, respectively [97, 98]. ISV using vaccinia viruses genetically modified for preferential infection of cancer cells and expressing GM-CSF have the ability to induce tumor responses and survival benefit in patients with hepatocellular carcinoma [99], and modified herpes virus expressing GM-CSF have generated antitumor immune responses and prolonged disease-free survival in patients with melanoma [100]. Importantly, besides their local, immune-mediated, antitumoral activity, these immunostimulatory products also have the ability to generate a systemic antitumor immune response against distant, noninjected, tumor sites [95, 97, 99].

Nina Bhardwaj, Nora Disis, and Karolina Palucka

There are common immunodynamic elements to most agents studied in cancer vaccines, including dendritic cell (DC)–based approaches. DCs can be exploited for vaccination against cancer through various means including (1) nontargeted peptide or protein and nucleic acids–based vaccines captured by DCs in vivo, (2) vaccines composed of antigens directly coupled to anti-DC antibodies, or (3) vaccines composed of ex vivo–generated DCs that are loaded with antigens. Immunodynamic investigation of cancer vaccines has demonstrated that the absolute number of tumor-specific T cells infused or generated with immunization is critical in obtaining a beneficial clinical outcome [101, 102]. Further, recent studies suggest that functional phenotypic changes in immune system cells, such as the induction of polyfunctional T cells, represent a desired endpoint [103]. Flow cytometric–based methodologies are highly quantitative, are reproducible, and can be standardized, and through the use of multiple intracellular and extracellular markers can provide detailed information about both the phenotype and the activation status of the adaptive immune response elicited with immunomodulation [81]. One of the most commonly used quantitative assays in immune-oncology today is enzyme-linked immunospot (ELISPOT) [104]. ELISPOT can enumerate cellular immunity, is possible to standardize, but provides limited functional information [105]. ELISPOT results are limited by lack of reproducibility and requirement for knowledge of antigens recognized or the availability of autologous tumor or tumor lysates with significant clinical material for analysis.

One newer method that may useful is the detailed analysis of the T-cell repertoire via CDR3 spectratyping strategies or deep sequencing with next generation sequencing technology to assess T-cell diversity [106]. The benefit of repertoire analysis is that the method can be quantitative and does not require an a priori knowledge of a specific antigen or depend on T-cell stimulation ex vivo. Repertoire analysis can be accomplished with less than 1 mL of whole blood. While the analysis of the T-cell repertoire is not directly functional, evolution from polyclonality to monoclonality of specific TCRs would suggest an evolving immune response with treatment [101]. Moreover, the development of multiple monoclonal populations could indicate the development of epitope spreading, which has been shown by multiple groups to be predictive of beneficial clinical outcome after cancer vaccination [107]. Unlike flow cytometry or ELISPOT, TCR spectratyping can be applied to direct analysis of tumors as well.

Site of monitoring the immune response postvaccination remains a critical consideration. One of the key locations to examine immune-cancer interactions is the tumor-draining lymph node (TDLN). It has been demonstrated that significant changes in immune-cell populations arise within TDLNs in breast cancer, specifically in CD4⁺ T cells and CD1a⁺ DCs, and such changes strongly correlated with clinical outcome [108, 109]. Data from the site of vaccination supports a direct correlation with antitumor activity and tumor-specific T-cell responses [108]. Delayed-type hypersensitivity (DTH) skin tests have been used to assess cell-mediated immunity in vivo. During a DTH, an antigen (Ag) is introduced intradermally, and induration and erythema at 48 to 72 h postinjection indicate a positive reaction. Lack of a DTH response to a recall Ag is often regarded as an evidence of anergy.

George Peoples and Jeff Weber

Immunodynamic monitoring of peptide-based vaccines provides an advantage to this vaccine strategy. Given that these peptide vaccines stimulate specific T-cell populations with TCRs specific for the peptide-HLA complexes, then clonal expansion or phenotypic assays may be employed. These assays enumerate vaccine-specific T cells by flow cytometry–based testing using peptide-specific dimer, tetramer, or dextramer reagents. Of course, these clonally expanded T cells must also be shown to be functional in cytokine-release or cytotoxicity-measuring assays. Additionally, peptide-based vaccines lend themselves well to DTH monitoring, as these peptides alone are biologically inactive and will only produce a DTH reaction if a peptide-specific cellular immune response has been induced. Immunodynamic studies have demonstrated that vaccine-elicited T cells are heterogeneous with respect to tumor-killing capacity, and only a small subset of vaccine-elicited T cells are efficient at tumor-cell lysis [110, 111]. This is largely due to differences in functional avidity (also known as recognition efficiency): peptide-specific T cells indistinguishable by tetramer staining may differ by up to 1000-fold in peptide requirement for target lysis [111]. Only high-avidity cytotoxic T lymphocytes (CTLs), which may represent 10 % or less of a vaccine-elicited response, could lyse tumor targets [110, 111]. This can be assayed via a flow-cytometric method for rapid assessment of recognition efficiency and functional capacity of antigen-specific T-cell responses [112].

Cytokines

Kunle Odunsi

The immunoregulatory enzyme IDO catalyzes the rate-limiting step of tryptophan (Trp) degradation along the kynurenine (Kyn) pathway [101]. Both the reduction in local tryptophan concentration and the production of tryptophan metabolites contribute to the immunosuppressive effects of IDO, resulting in multiple negative effects on T lymphocytes notably on proliferation, function, and survival. Tryptophan deprivation also biases the differentiation of naive mouse and human CD4⁺ T cells toward Foxp3-expressing regulatory T-like cells. Moreover, certain tryptophan metabolites activate the aryl hydrocarbon receptor, which has been linked to Treg differentiation [102, 103]. IDO is expressed by activated immune and inflammatory cells in TDLNs and in several human malignancies [118]. IDO1 has been observed to be chronically activated in many cancers, and IDO expression and enzymatic activity correlates strongly with extent of disease and is an independent prognostic factor for reduced overall survival in several malignancies [104, 105].

To assess the impact of IDO inhibition in human clinical trials, a number of correlative immunodynamic endpoints are important to develop [102]. The first is to determine the extent to which the IDO inhibitor alters the Kyn/Trp ratios in the serum and tumor microenvironment of treated patients. The second is to determine immunological endpoints of treatment. Pretreatment and post-treatment biopsies should be analyzed for lymphocyte infiltration by IHC and flow cytometric–based immunophenotyping assays. For IHC, the most critical analyses include IDO1 expression, changes in number, distribution, and phenotype of CD8⁺ and CD4⁺ T cells infiltrating tumor, and changes in patterns of CD4⁺ FoxP3⁺ Treg infiltration. Flow cytometric analyses for the effects of treatment on peripheral blood, tumor lymphocyte numbers, and phenotype (i.e., CD8⁺ and CD4⁺ naïve/effector/central memory subsets, Tregs, and exhaustion markers). Finally, changes in myeloid-derived, suppressor-cell populations can also be assessed by flow cytometry and may be an important endpoint of cytokine treatment [106]. There is currently no widely accepted consensus on how to phenotypically and functionally define this cell population. This represents an area of intense investigation.