THE BIOLOGY BEHIND PD-1 CHECKPOINT BLOCKADE

ABSTRACT

Programmed death 1 (PD-1) pathway inhibitors have transformed cancer therapy, leading to durable responses in some patients. However, many patients do not benefit from PD-1 blockade therapy, which highlights the critical need to identify new therapeutic targets to complement PD-1 pathway inhibitors. To address this need, we have developed an in vivo clustered regularly interspaced short palindromic repeats (CRISPR)-based screening platform to discover novel regulators of anti-tumor immunity. In this article, I will first discuss the biology of the PD-1 pathway and its role in regulating anti-tumor immunity. Next, I will introduce our innovative CRISPR-based platforms designed for conducting gene screens in mature immune cell lineages and for enabling gene perturbation without stimulating or manipulating immune cells, two approaches that can affect immune cell development and function. In addition, I will illustrate how these platforms facilitate discovery of new targets that can promote anti-tumor immunity and their potential to lead to more effective cancer therapies.

INTRODUCTION

Immunology has offered hope for treating cancer for over a century and is beginning to fulfill this promise. We are discovering methods to activate the immune system to combat cancer. The tumor microenvironment can inhibit anti-tumor immunity in a number of ways (1,2). Tumors evolve to evade elimination by the immune system (3). The immune checkpoint blockade strategy of cancer immunotherapy targets immunoinhibitory pathways that limit anti-tumor immunity. These inhibitory pathways, called “immune checkpoints,” provide crucial inhibitory signals that regulate immune homeostasis and defense, mediate tolerance to prevent autoimmunity, and aid in resolving inflammation to curb immune-mediate tissue damage (4,5). Tumors and microbes exploit these pathways to evade eradication by the immune system.

Immune checkpoint pathways provide physiologic mechanisms to regulate T cell responses. T cell activation is a highly regulated process requiring two signals. The first signal arises from engagement of the T cell receptor (TCR) by peptide-major histocompatibility complexes (MHC), conferring specificity to the response. The second signal, or co-signal, is the costimulatory signal (4,5). Interactions between the CD28 receptor on T cells and its ligands CD80 (B7-1) and CD86 (B7-2) on antigen presenting cells (APCs) provide the major costimulatory signal for initial activation of naïve T cells, promoting their proliferation, cytokine production, and survival (6). In addition to positive second signals, inhibitory second signals are induced early during activation to modulate the T cell activation program (4,5,7,8). Upon T cell activation, coinhibitory “immune checkpoints,” such as programmed death 1 (PD-1), are induced to regulate T cells.

PD-1 Pathway

PD-1 inhibits positive signals from the TCR and CD28 by engaging its ligands PD-L1 (CD274) and PD-L2 (CD273), attenuating T cell proliferation and effector responses, including cytokine production and cytotoxic functions (4,5,8,9) (Figure 1). PD-L1 is widely expressed on a variety of hematopoietic cells, including antigen presenting cells and T cells, as well as on non-hematopoietic cells including vascular endothelial cells, epithelial cells, pancreatic islet cells, and syncytiotrophoblasts in the placenta. In contrast, PD-L2 is primarily expressed on hematopoietic cells, including dendritic cells, B cells, and macrophages, but can also be found on some non-hematopoietic cells, including lung epithelial cells (10,11). The expression of these ligands on non-hematopoietic cells enables the PD-1 pathway to influence T cell responses across a range of tissues beyond lymphoid organs. PD-L1 and PD-L2 also can be expressed by cancer cells, contributing to tumor immune evasion (12).

Fig. 1.

PD-1 inhibits T cell responses. Activation of naïve T cells requires interactions of the TCR with peptide-MHC complexes and a costimulatory signal provided by the CD28:CD80/86 interaction. When PD-1 binds it ligands PD-L1 and PD-L2, this dephosphorylates kinases downstream of the TCR and CD28, attenuating T cell responses.

PD-1 modifies membrane-proximal events in T cells, by promoting dephosphorylation of key signaling molecules, thereby attenuating TCR and CD28 signals. When PD-1 binds to its ligands, its cytoplasmic tail becomes phosphorylated on tyrosines within its immunoreceptor tyrosine-based inhibitory motif (ITIM) and immunoreceptor tyrosine-based switch motif (ITSM). This results in recruitment of phosphatases that dephosphorylate kinases downstream of CD28 and TCR, countering TCR and CD28 signaling (13-17) (Figure 1).

The multifaceted functions of PD-1 inhibitory signals exemplify how immunoinhibitory signals help maintain balance in the immune system. PD-1 plays a key role in regulating the threshold for initial activation of T cells, influencing their T cell differentiation and function, limiting effector T cell responses, as well as promoting T cell tolerance and immune homeostasis. PD-1 inhibitory signals (a) counteract positive signals received through the TCR and costimulatory receptors, thereby dampening immune responses and restoring homeostasis; (b) promote resolution of inflammation, helping to prevent immune-mediated tissue injury, and (c) maintain tolerance, by limiting the activation and function of potentially pathologic self-reactive T cells. Moreover, PD-L1 expressed on non-hematopoietic cells can mitigate immune-mediated tissue damage and shield target organs from potentially pathogenic self-reactive T cells (4,5,8).

During chronic infections, such as chronic HIV, hepatitis B, and hepatitis C, as well as in cancer, continuous antigen exposure leads to T cell exhaustion, a state of diminished effector T cell function, and sustained expression on inhibitory receptors, including PD-1 (18,19). PD-1 plays a critical role in mediating T cell exhaustion, limiting the functionality of exhausted T cells during chronic infection and cancer. The significance of the PD-1 pathway in T cell exhaustion was first discovered in the lymphocytic choriomeningitis (LCMV) mouse model of chronic viral infection (20). In this model, virus-specific CD8+ T cells progressively lose effector functions and fail to control viral infection. During chronic LCMV infection, PD-1 expression is both heightened and persistent, in contrast to acute LCMV infection, where the PD-1 expression initially increases but declines after the virus is cleared and the antigenic stimulus is gone. The high and sustained expression of PD-1 during chronic LCMV infection suggested that PD-1 was inhibiting the function of exhausted T cells. Remarkably, PD-1 pathway blockade during chronic LCMV infection improved the quantity and function of exhausted virus-specific CD8+ T cells, restoring their capacity to proliferate, produce cytokines, kill infected cells, and reduce viral burden. These studies were pioneering in identifying a mechanism of T cell exhaustion and a therapeutic strategy for treating it, demonstrating that PD-1 blockade can alleviate PD-1 inhibition to restore T cell function. The fundamental program of T cell exhaustion is observed in tumors as well as in chronic viral infection (18,19). Studies in cancer models have shown that PD-1 similarly inhibits the functions of exhausted T cells, impairing their ability to control tumor growth (21).

Exhausted CD8+ T cells exist along a spectrum, ranging from a less dysfunctional subset that exhibits slow self-renewal (known as stem-like or progenitor exhausted) and has the capacity to differentiate into an effector-like (transitory) subset to a non-dividing terminally differentiated exhausted T subset (22-27). The draining lymph node (LN) serves as a reservoir for less dysfunctional stem-like T cells (28-30). PD-1 blockade triggers a proliferative burst of the stem-like exhausted CD8+ T cells, causing some to differentiate into effector-like cells (31,32). These effector-like cells traffic from the LN to infected tissues or tumors to provide rapid and robust anti-viral or anti-tumor immunity, resulting in therapeutic benefit (33).

PD-1 pathway blockade has been translated into cancer therapies, with anti-PD-1 or anti-PD-L1 inhibitors revolutionizing cancer treatment (7-9,34-36). Initially approved by the Food and Drug Administration (FDA) in 2014, PD-1 pathway inhibitors have since gained approval for treating over 25 cancer types, and tumor agnostic indications such as tumor mutational burden >10 mutations per megabase and microsatellite instability high (MSI^hi)/mismatch repair deficient (MMRd) tumors. PD-1 pathway blockade can result in durable responses, with some patients experiencing benefit for more than a decade (36). However, effectiveness varies across tumor types, and only a subset of patients responds (7). Thus, it is essential to understand mechanisms of response and resistance to PD-1 pathway inhibitors. This knowledge will help identify and prioritize targets for combination therapy, to extend these benefits to more cancer patients.

Development of an In Vivo CRISPR Screening Approach to Discover New Cancer Immunotherapy Targets and Their Mechanisms of Action

One approach we have taken to uncover new targets for cancer immunotherapy is to develop a robust high-throughput in vivo CRISPR screening platform. This platform is designed to knockout (KO) genes in immune cells to identify gene KOs that enhance anti-tumor immunity (37-39). Transcriptional profiling of the alterations occurring in immune cells during cancer development and progression has resulted in immense catalogs of genes and pathways potentially involved in regulating innate and adaptive immune responses to tumors. Similarly, artificial intelligence approaches are identifying candidate regulators of anti-tumor immunity. However, the functional annotation of these genes and pathways has lagged far behind their description. Consequently, the challenge for understanding innate and adaptive anti-tumor immune responses has shifted from merely characterizing these changes to ascribing specific functions to each gene.

To tackle this challenge, we have built systems for perturbing genes in diverse immune cell types in vivo with a focus on T cells, using CRISPR gene editing (37-39). Most methods for perturbing gene expression in T cells have relied on activating these cells either in vitro or in vivo to achieve efficient transduction of viral vectors that deliver short hair ribonucleic acid (shRNA) or CRISPR constructs (40-43). However, our studies show that such approaches can modify the intrinsic biology of T cells, precluding identification of genes critical for initial T cell activation and differentiation (44). To overcome this limitation, we have devised experimental approaches employing CRISPR/Cas9 technology to delete genes in vivo using unmanipulated immune cells. This strategy aims to identify physiologic regulators of T cell priming in the LN, T cell migration to tissues, and T cell function in tissue microenvironments.

We initially developed the CHimeric IMmune Editing (CHIME) system for conducting CRISPR-based screens and validation studies in immune cells in vivo (38). This system employs the Cas9 protein together with a single chimeric RNA, which includes a customizable deoxyribonucleic acid (DNA)-targeting sequence known as a guide RNA (gRNA) and a trans-activating CRISPR RNA (tracrRNA) scaffold, enabling precise targeting of specific DNA sequences in mammalian cells. The Cas9 and gRNA components are delivered via a lentiviral delivery system. Cas9-expressing hematopoietic stem cells (HSC) are transduced with lentiviral vectors containing the gRNAs and then implanted into irradiated Cas9-expressing recipient mice, which prevents immune rejection of the transferred Cas-expressing cells. These transduced HSCs differentiate into mature immune cells containing the gRNAs and resulting gene KOs. CHIME eliminates the need to directly stimulate or manipulate mature immune cells to deliver gRNAs, two approaches that can alter immune cell development and function. CHIME enables KO of genes in immune mature lineages, including naïve lymphocytes, myeloid cells, and dendritic cells. Our studies show that CHIME does not affect T cell development or function. Thus, unlike other methods that rely on activated immune cells, CHIME enables knockout of individual genes or libraries of genes in mature immune cells without impacting their development or function.

Identification of Ptpn2 as a Cancer Immunotherapy Target That Regulates CD8+ T Cell Exhaustion

We first used CHIME to achieve constitutive KO of individual genes in T cells, and to execute an in vivo CRISPR screen in the LCMV chronic infection model, aimed at identifying genes that regulate LCMV-specific CD8+ T cell responses during T cell exhaustion (38). To investigate LCMV-specific T cells in vivo, we transduced HSCs from Cas9-expressing P14 TCR transgenic mice (which recognize LCMV CD8 epitope GP33-41) with lentiviral vectors containing gRNAs targeting potential regulators. We included the non-immunogenic fluorescent protein Vex as a marker to assess transduction efficiency. Bone marrow chimeras (BMC) generated from the transduced HSCs developed a hematopoietic compartment, including naïve P14 TCR-transgenic T cells, where gRNA expression and gene deletion took place. We transferred perturbed naïve P14 T cells to Cas9-expressing recipient mice that were subsequently infected with LCMV clone 13 to determine how KO of a single gene or pool of genes affected LCMV-specific T cell responses.

In this in vivo screen, we evaluated a curated set of gRNAs targeting genes involved in regulation of TCR and cytokine signaling, coinhibitory or coinhibitory signaling and metabolism, using a pooled CRISPR approach (38). We determined which gene perturbations were enriched (more abundant) or depleted (less abundant) compared to the initial population (Figure 2). Enrichment would indicate KO of a negative regulator, whereas depletion would indicate KO of positive regulator of CD8+ T cells. The gRNAs targeting the phosphatase Ptpn2 were enriched, identifying Ptpn2 as a potential negative regulator of LCMV-specific CD8+ T cell responses.

Fig. 2.

Strategy for in vivo T cell screens. T cell screens are typically performed by transferring a population of KO T cells to a virus-infected or tumor-bearing host. Here the black, dark gray, and light gray cells have three different gRNAs targeting different genes. The cells undergo selective pressure in response to a tumor that they recognize. This leads to changes in relative abundance of the black, dark gray, and light gray cells. In this case, the dark gray cells have increased, the light gray cells have decreased, and the black cells have stayed the same. This suggests that the gene targeted in the dark gray cells is a negative regulator of T cell responses, the gene targeted in the light gray cells is a positive regulator of T cell responses, and the gene targeted in the black cells has no effect.

To validate Ptpn2 as a cell intrinsic regulator of T cell function, we performed an in vivo competitive co-transfer assay to compare Ptpn2 gRNA and control gRNA-containing P14 CD8+ T cells in the same microenvironment. This approach controls for variations in viral antigen load and inflammatory signals across different mice. We co-transferred a 1:1 ratio of congenically marked Ptpn2 gRNA and control gRNA-containing P14 CD8+ T cells into recipient mice that were subsequently infected with LCMV clone 13. We then analyzed the ratio of control and Ptpn2 gRNA containing T cells in the spleen eight days later. Ptpn2 gRNAs efficiently deleted (~80%) Ptpn2. P14 CD8+ T cells with Ptpn2 gRNAs significantly outcompeted those with control gRNAs, confirming Ptpn2 as a cell intrinsic negative regulator of LCMV-specific CD8+ T cells during chronic LCMV infection (38). Further studies revealed that Ptpn2 deletion increased the proliferation and cytotoxic function of exhausted T cells in the LCMV chronic infection model, primarily through enhanced interferon-a signaling (45).

In complementary studies, we explored the role of Ptpn2 in regulating anti-tumor immunity. We determined that Ptpn2 intrinsically restricts anti-tumor CD8+ T cell responses (45). In a competitive co-transfer assay, Ptpn2-deleted OT-1 TCR transgenic CD8+ T cells, which recognize ovalbumin (Ova) CD8 epitope (257-262), significantly outcompeted control CD8+ T cells in the tumor and expressed more Granzyme B. Transfer of Ptpn2-deleted OT1 CD8+ T cells to mice subsequently implanted with B16-Ova tumor cells markedly attenuated tumor growth compared to mice receiving control OT-1 CD8+ T cells and resulted in clearance of 25% of the tumors. These results demonstrated the therapeutic potential of Ptpn2 deletion to enhance anti-tumor immunity.

We also examined the effect of Ptpn2 deletion in all hematopoietic cells on tumor growth control to model therapeutic targeting of Ptpn2, given its broad expression on hematopoietic cells and its regulation of T, B, and myeloid cells. BMCs with Ptpn2 deletion in hematopoietic cells completely eradicated MC38 tumors, whereas control chimeras showed progressive tumor growth (45). Finally, we tested if Ptpn2 deletion in hematopoietic cells could control B16 melanoma tumors, known to be anti-PD-1 resistant. When we combined Ptpn2-deleted BMCs with anti-PD-1 and GVAX therapeutic vaccination, we observed enhanced tumor growth control and survival, with approximately 25% of mice clearing their tumors in contrast control chimeras which exhibited progressive tumor growth (45). These studies underscore the therapeutic potential of Ptpn2 inhibition as a cancer combination therapy with PD-1 pathway inhibitors.

Ptpn2 is an especially appealing target for cancer immunotherapy because its expression in both tumor cells and T cells limits anti-tumor immunity and effectiveness of anti-PD-1. Deleting Ptpn2 in tumor cells enhances IFN-g signaling in the tumor cells, which in turn increases MHC class I expression and tumor antigen presentation to T cells, while also making tumor cells more susceptible to IFN-g mediated apoptosis (46). Thus, inhibiting Ptpn2 could have dual anti-tumor effects by directly impacting tumor cells and increasing anti-tumor immune cell functions. This potential led to development of a Ptpn1/2 small molecule inhibitor by AbbVie, currently being assessed in Phase 1 clinical trials, both as a stand-alone therapy and in combination with anti-PD-1 for solid tumors (47).

Creation of Novel Methods for Mechanistic Evaluation of Candidate Genes

We have engineered four modifications of CHIME to investigate gene function in hematopoietic cells under context-specific conditions in vivo (37). These systems allow for simultaneous deletion of two genes, inducible deletion of individual genes, sequential deletion of gene pairs, or gene deletion in specific immune lineages (e.g., CD8+ T cells). These methods are faster and more modular, compared to generation of knockout mice which is time consuming.

Combinatorial CHIME (C-CHIME) employs a lentiviral vector enabling knockout of two genes at once in hematopoietic cells. We have achieved efficient deletion of two genes in T cells with this method. Combinatorial knockout of genes is valuable for assessing epistasis, performing anchored screens, and determining synergistic interactions.

The inducible CHIME platform (I-CHIME) employs a Frt/FlpO approach for inducible gene deletion. I-CHIME provides temporal control over gene expression, enabling studies of genes that affect immune development and their roles during specific phases of the immune response, (e.g., during the effector phase, contraction to memory, and memory maintenance). We have observed controlled and efficient deletion with I-CHIME. By pairing the inducible and combinatory systems, we created inducible combinatorial CHIME (icCHIME), which enables concurrent knockout of two genes—one inducibly and the other constitutively. This system facilitates more nuanced mechanistic studies and studies of gene pairs that are synthetic lethal during development.

Lineage-specific CHIME (L-CHIME) uses a loxP gRNA vector together with Cre mouse strains to enable lineage-specific gene deletion. For example, using E8iCre, which deletes genes specifically in CD8+ T cells, we were able to evaluate gene function selectively in CD8+ T cells with no off-target editing in other lineages. This method can be used to assess the effects of gene knockout in polyclonal T cell and non-T cell immune populations.

Although CHIME enables evaluation of gene function across hematopoietic cell lineages, we have refined a complementary method using nucleofection to more quickly investigate individual candidate genes in hematopoietic cells by creating gene-edited bone marrow chimeras (37). This method involves nucleofecting c-Kit+ bone marrow stem cells either with control or targeting gRNA-Cas9 ribonucleoprotein complexes, which are then implanted into irradiated recipient mice. One advantage of this nucleofection strategy is its ability to delete genes in an existing KO mouse strain, circumventing the lengthy process of crossing-breeding mouse strains. However, due to its transient delivery nature without genomic integration, nucleofection is not suitable for sequential gene knockouts or pooled screening approaches. In contrast, the Combinatorial CHIME (C-CHIME) and Sequential CHIME (S-CHIME) systems leverage lentiviral integration of the gRNAs, facilitating both sequential gene KO and in vivo pooled screens.

These two methods similarly complement each other for T cell studies. Nucleofection allows for deletion in naïve T cells more rapidly than CHIME, but without genomic integration of the gRNA. Nucleofection is useful for faster analyses of the edited naïve T cells, making it beneficial for comparisons of gene KO and control T cells with a fixed TCR specificity in adoptive transfer studies. Conversely, the CHIME lentivirus-based system enables experiments in intact animals without the need for cell transfers and is useful for analyzing polyclonal T cell responses. Together, these nucleofection and lentiviral methods serve as a robust toolkit for deleting individual genes or two genes at different times in TCR transgenic T cells, polyclonal T cells, and non-T immune cells.

Developing New CRISPR-Based Platforms for High Throughput Screens

Although CHIME enables screening in naïve CD8+ T cells, it is limited by its throughput capacity and lengthy processing time. We thus explored alternative platforms that could enable higher throughput and faster screens in naïve CD8+ T cells. We developed a system to transduce naïve CD8+ T cells with pseudotyped lentivirus, followed by a seven-day in vitro culture period to allow for expression of the transduction marker, all without triggering T cell activation or differentiation (37,39). The transcriptional profiles of these cells closely resemble those of naïve cells, a similarity we have validated through functional assays.

Additionally, we have established experimental and analytical guidelines for conducting robust in vivo T cell CRISPR screens. These include optimizing the number of genes that can be screened, determining the number of naïve T cells that can be transferred and recovered, and using computational modeling to determine optimal numbers of genes and mice required (39). We also incorporated unique molecular identifiers (UMIs) into the tracrRNA tetraloop, which does not affect function but provides each T cell with a nearly unique genetic barcode, improving statistical power and tracking of T cell clonal dynamics for distinct gene knockouts in vivo (39). Using this framework, we determined the optimal number of naïve OT-1 CD8+ T cells needed for engraftment in the tumor draining LN and tumor in the B16-Ova tumor model and designed a proof-of-concept screen to evaluate genes regulating T cell proliferation in the tumor draining LN and tumor, and their migration between sites. This screen identified site-specific regulators of CD8+ T cell responses, such as lymph node predominant effects of KO of Pdcd1 (encoding PD-1) and Zc3h12a (encoding Regnase) and tumor dominant effects of KO of Ptpn1 and Ptpn2. It also identified regulators of T cell trafficking from the tumor draining LN to the tumor, such as Tbx21, which encodes T-bet. Using these approaches, we are now performing large (~800 gene) CRISPR-based loss-of-function screens in naive CD8+ T cells responding to murine tumors.

CONCLUDING REMARKS

PD-1 pathway blockade has transformed cancer therapy, but only a subset of patients benefits from it. There is an urgent need to systematically identify and prioritize targets for combination therapies to extend these benefits to more cancer patients. To address this unmet need, we have developed a robust in vivo platform for conducting CRISPR screens in immune cells along with a comprehensive toolkit for validating and studying candidate genes. Our framework for in vivo T cell screens provides a paradigm for designing and analyzing immune cell CRISPR screens. We currently are expanding these screening techniques to include a wider range of immune cell types and disease contexts, such as autoimmunity, and are incorporating diverse readouts of gene function, like cytokines and transcriptional regulators. Our CHIME and nucleofection methods enable analyses of individual genes or paired genes at different timepoints in immune cells. Our goal is to discover novel immunotherapy targets and determine their mechanisms of action. Our in vivo screening strategies hold the potential to identify immunotherapy targets for a diverse range of immune-mediated diseases.

DISCUSSION

Mann, St. Louis: I want to thank you for all your work in this field; it’s been truly transformative. I’m a cardiologist so I’m going to come at this in a slightly different way. In studies we’ve done with really simple experimental injury models, we’ve noticed that if you knock out PD-1 the inflammation is sustained, which then segues to loss of T cell tolerance and a kind of autoimmune myocarditis. This is probably no surprise to you. Could we use agonistic pathways to try to tamp down the initial inflammation that then leads to collateral damage? I think they’re doing that with arthritis. Is it crazy to think about doing this in the heart?

Sharpe, Boston: Absolutely not. It is exciting to see how this pathway is now being translated to therapy to engage these inhibitory receptors to treat autoimmune and inflammatory diseases. The first PD-1 agonists were described earlier this year for rheumatoid arthritis, but many others are also under investigation. There are more inhibitory receptors in addition to PD-1, and agonists are being developed for those as well. Hopefully, we can learn how to combine them and then translate this therapy. That’s important not only for autoimmune but also inflammatory diseases. One thing I wasn’t able to discuss is that some cancer patients experience adverse events, and we need to learn how to use these types of approaches to treat these adverse events that are inflammatory or autoimmune-like. Thank you for your question.

Mann, St. Louis: Thank you.

Anderson, Chicago: I am also a recovering cardiologist and want to follow up on this. With regard to the cleverness of in vivo CRISPR screening, can you imagine flipping that around in an autoimmune model, to be searching for pathways that crush autoimmunity and looking at the other side of this? Maybe you’re already doing that. I know that people who have knocked out PD-1 just in heart muscle cells get this terrible cardiomyopathy, revealing an unexpected role for that signaling pathway and in muscle biology. I just wonder if you can turn it around and then sort of screen like you’ve done.

Sharpe, Boston: We’re doing that very thing. We’re developing CRISPR technologies to identify targets not only for cancer but also for autoimmune diseases. In addition to the genes that we identify with our CRISPR screens with cancer, which increase abundance, we see genes that reduce T cell responses. We’re beginning to study some of those to see if they might be regulators of autoimmunity.

Anderson, Chicago: That’s great! Thank you.

Sharpe, Boston: Thanks for that question.

Carethers, San Diego: May I ask one question? Can you fully explain how high tumor-mutational burden generates high PD-1 expression? I haven’t figured that one out yet, but it obviously does this for mismatch repair deficiency tumors.

Sharpe, Boston: There are tumor cells that can express multiple mutations, and there can be some mutations in the PD-L1 gene, but that’s not the whole story with mutations. A variety of different mutations can create this tumor burden, but some mutations have been identified that increase PD-L1 and PD-L2. Margaret Shipp and Philippe Armand discovered for PD-L1 and PD-L2 that there can be overexpression changes and they can be used as a biomarker for identifying Hodgkin’s patients that could respond to anti-PD-1.

Carethers, San Diego: Okay, thank you. One more question.

Ross, St. Louis: I’m a nephrologist in St. Louis. Like most people, we have noticed increasing rates of interstitial nephritis from checkpoint inhibitors. These cases are rising, and we’re kind of concerned. There are probably some other downstream effects similar to what were raised in the last question about overstimulating T cells. What can we do to mitigate the rising rates of interstitial nephritis?

Sharpe, Boston: That’s a very important question. We know some patients will develop these adverse events. In the same way that we’re trying to determine who will benefit from PD-1 checkpoint blockade, a number of ongoing studies are trying to identify patients who are at risk for these adverse events. In some cases, there may be preexisting autoimmunity. In other cases, cross-reactivity between antigens on the tumor and normal cells may contribute to the development of immune-related adverse events. One of the approaches being studied for treating these adverse events is interleukin-6 blockade.

Ross, St. Louis: Thank you.