key: cord-0011639-nrfil3oc
authors: Waldman, Alex D.; Fritz, Jill M.; Lenardo, Michael J.
title: A guide to cancer immunotherapy: from T cell basic science to clinical practice
date: 2020-05-20
journal: Nat Rev Immunol
DOI: 10.1038/s41577-020-0306-5
sha: 0769b67e4e19f64e5774990070f521443d2cdd08
doc_id: 11639
cord_uid: nrfil3oc

The T lymphocyte, especially its capacity for antigen-directed cytotoxicity, has become a central focus for engaging the immune system in the fight against cancer. Basic science discoveries elucidating the molecular and cellular biology of the T cell have led to new strategies in this fight, including checkpoint blockade, adoptive cellular therapy and cancer vaccinology. This area of immunological research has been highly active for the past 50 years and is now enjoying unprecedented bench-to-bedside clinical success. Here, we provide a comprehensive historical and biological perspective regarding the advent and clinical implementation of cancer immunotherapeutics, with an emphasis on the fundamental importance of T lymphocyte regulation. We highlight clinical trials that demonstrate therapeutic efficacy and toxicities associated with each class of drug. Finally, we summarize emerging therapies and emphasize the yet to be elucidated questions and future promise within the field of cancer immunotherapy.

The idea to deploy the immune system as a tool to treat neoplastic disease originated in the nineteenth century 1 . Wilhelm Busch and Friedrich Fehleisen were the first to describe an epidemiological association between immune status and cancer. They noticed spontaneous regression of tumours following the development of erysipelas, a superficial skin infection most commonly caused by Streptococcus pyogenes 1 . Later, William Coley, often called the 'Father of Cancer Immunotherapy' , retro spectively demonstrated that erysipelas was associated with a better outcome in patients with sarcoma 2 . With hopes of prospectively verifying his epidemiological evi dence, Coley treated patients with cancer with extracts of heatinactivated S. pyogenes and Serratia marcescens to boost immunity 3 . This extract, termed 'Coley's tox ins' , possessed potent immunostimulatory properties and achieved favourable responses in various cancers 2 . However, lack of scientific rigour and reproducibility, in concert with the discovery of radiotherapy and chemo therapeutic agents, prevented treatment with 'Coley's toxins' from becoming standard practice 1 .

The concept of cancer immunotherapy resurfaced in the twentieth century and made significant headway with the advent of new technology. In 1909, Paul Ehrlich hypothesized that the human body constantly gen erates neoplastic cells that are eradicated by the immune sys tem 3 . Lewis Thomas and Sir Frank Macfarlane Burnet independently conceived the 'cancer immuno surveillance' hypothesis, stating that tumourassociated neoantigens are recognized and targeted by the immune system to prevent carcinogenesis in a manner similar to graft rejection 1 . Productive immune responses following tumoural adoptive transfer in mice 4 and clinical reports of spontaneous regression of melanoma in patients with concomitant autoimmune disease 5 provided additional evidence supporting this hypothesis, although a unifying mechanism was elusive. The advent of knockout mouse models provided the necessary technology to experi mentally demonstrate a link between immunodeficiency and cancer 6 . Additional molecular and biochemical advances led to the identification of tumourspecific immune responses 7 . This provided unequivocal evi dence that the immune system, in particular T cells (see Box 1 and Fig. 1 ), was capable of waging war on cancer tissue 7 . Cancer immunotherapy has now revolutionized the field of oncology by prolonging survival of patients with rapidly fatal cancers. The number of patients eligi ble for immunebased cancer treatments continues to skyrocket as these therapies position themselves as the first line for many cancer indications. Novel treatment combinations and newly identified druggable targets will only expand the role of immunotherapy in the treatment of cancer in the decades to come.

In this Review, we emphasize the role of T cells in modern cancer immunotherapies and discuss three dif ferent categories of immunotherapeutic approaches to treat cancer: immune checkpoint blockade, an approach that is designed to 'unleash' powerful T cell responses; adoptive cellular therapies, which are based on the infu sion of tumourfighting immune cells into the body; and Neoantigens Antigens not expressed by self-tissues under normal conditions that manifest in the context of pathology; in cancer, these could be altered proteins/peptides encoded by mutated genes.

A mechanism of immune cell inhibition that restrains activation.

cancer vaccines, which can be designed to have either prophylactic or therapeutic activity. Finally, we intro duce some of the emerging targets and approaches in cancer immunotherapy.

Immune checkpoint therapy Several evolutionarily conserved negative regulators of T cell activation act as 'checkpoint molecules' to finetune the immune response and regulate hyperacti vation. Cytotoxic T lymphocyte antigen 4 (CTLA4) and programmed cell death 1 (PD1) are the most potent examples of T cell immune checkpoint molecules. They exert their biological effect at distinct body sites and times during the T cell lifespan 8 . Therefore, they com plement each other functionally and ensure that T cell responses preserve selftolerance while effectively pro tecting the body from pathogens and neoplasia. CTLA4 and PD1 have been successfully targeted by several pio neering research groups as treatments for a wide variety of recalcitrant cancers, research that ultimately earned James P. Allison and Tasuku Honjo the 2018 Nobel Prize in Physiology or Medicine.

After the discovery of T cell costimulation mediated by the surface protein CD28 (Box 1), the search for additional immune regulators led to the identification of CTLA4, a receptor with struc tural and biochemical similarities to CD28, as a new immunoglobulin superfamily member 9,10 . The CTLA4 and CD28 genes are found in the same region of chromo some 2 (2q33.2) and are selectively expressed in the haematopoietic compartment 11 . However, in contrast to the high levels of basal CD28 expression on conven tional T cells, CTLA4 is expressed at a low basal level and is strongly induced following antigen activation. Interestingly, CD4 + CD25 + regulatory T (T reg ) cells, which have an immunosuppressive function, express CTLA4 constitutively. Structurally, both CTLA4 and CD28 form membranebound homodimers comprising an extra cellular immunoglobulinlike domain, a transmembrane region and a cytoplasmic tail capable of recruiting sig nalling proteins and controlling surface expression 10, 12, 13 . The trafficking of CTLA4containing vesicles to the cell surface after activation is controlled by a physical interaction with the lipopolysaccharideresponsive and Box 1 | T cell function, development, activation and fate the 1960s represented a period of enlightenment within the field of immunology because two major subtypes of lymphocytes, B lymphocytes and t lymphocytes, were characterized 264, 265 . this was recognized by the 2019 Lasker award for Basic science, awarded for the pioneering work by Jacques a. F. P. Miller and Max Dale Cooper that defined the key roles of T cells and B cells in adaptive immunity. B cells recognize circulating antigen in its native form and respond by secreting protective anti bodies 266 . By contrast, T cells recognize peptide antigens, derived from proteins degraded intracellularly, that are loaded onto cell surface MHC molecules, a process called antigen presentation. two broad classes of T cells that have distinct effector mechanisms are delineated by the expression of either the CD4 or CD8 coreceptor: CD4 + T cells detect antigen in the context of MHC class ii molecules and orchestrate the adaptive arm of the immune system by producing cytokines with chemo tactic, proinflammatory and immunoprotective properties 267 . at least one CD4 + T cell subclass, CD4 + CD25 + regulatory T cells, dampens the immune response following challenge 268 . CD8 + T cells detect antigen in the context of MHC class i molecules and carry out direct cytotoxic reactions that kill infected or neoplastic cells 269 .

A unique clonespecific cell surface protein complex, the T cell receptor (tCr), specifically recognizes antigens and participates in the developmental selection of T cells that can recognize pathogens but are selftolerant 270 . the tCr complex comprises highly polymorphic single α and βglycoprotein chains (a small T cell population harbours γ and δchains instead) that contain variable and constant regions, akin to immunoglobulins, and a group of nonpolymorphic signalling chains, called CD3 γ, δ, ε and ζ. A vast repertoire of T cell clonotypes with unique specificities is generated through rearrangement of α and βchain gene segments within the genome of each T cell 271 . Following clonotype production, positive and negative thymic selection functions to entrain a 'tolerant' immune system, one that efficiently responds to pathogens or cancer cells but generally ignores or 'tolerates' selftissues as nonimmunogenic 269, 270 .

Antigen stimulation of the TCR is necessary for T cell activation and proliferation, but an additional signal, termed costimulation, is required for phosphorylation events crucial for early signal transduction 272 . The non polymorphic surface protein CD28 and its family members are the most potent costimulatory receptors on T cells, as elegantly demonstrated by the synergism of antiCD28 stimulatory antibodies and TCR engagement on T cell activation and proliferation 273, 274 . Additional evidence was provided by studies demonstrating the efficient inhibition of T cell acti vation and proliferation by inhibitory antiCD28 antibodies [275] [276] [277] [278] . The ligands for CD28, B71 and B72, are expressed on antigenpresenting cells and are upregulated when these cells encounter microorganisms that activate Tolllike receptors or other pathogen sensors 279, 280 . inhibitory molecules, including cytotoxic T lymphocyteassociated protein 4 (CtLa4) and programmed cell death 1 (PD1), are induced during immune responses and represent a 'checkpoint' to dampen T cell hyperactivation 281 (see Fig. 2 ). the polymorphic tCr signals through a complex of three sets of dimeric CD3 chains, ε-δ, γ-δ and ζ-ζ 282 . the intracellular portions of the CD3 chains contain immunoreceptor tyrosinebased activation motifs that are phosphorylated by lymphocytespecific protein kinase (LCK), a srC family kinase 283 . at rest, the surface signalling protein CD45 exhibits phosphatase activity that blocks LCK function 284 . Following activation, CD45 removes an inhibitory phosphate on LCK, permitting phosphorylation of ζ chainassociated protein kinase 70 (ZAP70), a SYK kinase family member that binds to immunoreceptor tyrosinebased activation motifs in the CD3 ζchain and recruits the linker for activation of T cells (LAT) and phospholipase Cγ1 (PLCγ) 285 . With ample costimulation, downstream signalling affects calcium release, the activation of the GTPase RAS and transcriptional reprogramming essential for activated T cell function 286 .

Following activation, circulating naive T cells have three major fates in the periphery (Fig. 1) . First, the effector T cell population can contract through apoptosis as the immune response resolves (cytokine withdrawal) or following repeated highdose stimulation (restimulationinduced cell death) [287] [288] [289] . T cells can also exhibit an exhausted phenotype induced by repeated lowdose and lowaffinity stimulation, as seen in chronic infections and neoplastic processes 88 . Lastly, a subset of these effector cells are involved in longterm immunological memory. Memory T cells are primed to react more vigorously to the same antigen during a subsequent encounter, making them critical mediators of immune recall responses to pathogens and tumours 290 . Leveraging the power of technological advances in molecular biology, recent singlecell RNA sequencing and epigenomic studies have provided additional molecular insight into T cell fates and the corresponding features of immunotherapyresponsive T cells. These studies collectively implicate that complex transcriptomic, epigenomic and clonotypic changes of tumourinfiltrating T cells determine the success of immunotherapy 291-294 . beigelike anchor protein (LRBA) 13 . The sequence simi larity between CTLA4 and CD28 is highest within their extracellular binding domain and they therefore bind to the same ligands, called B71 (also known as CD80) and B72 (also known as CD86), which are expressed by antigen-presenting cells (APCs; Box 1). However, CTLA4 has greater affinity and avidity than CD28 for B7 ligands, representing a key difference in their biology [14] [15] [16] .

With further characterization, it became clear that CD28 and CTLA4 had opposite immunoregulatory functions. For example, soluble CTLA4 was shown to inhibit the proliferation of T cells cocultured with B7expressing APCs because it interfered with the CD28-B7 interaction 14 . T cell receptor (TCR) signalling studies unequivocally demonstrated that CTLA4 inhib its T cell activation and proliferation 12, 17, 18 . The negative tolerogenic role of CTLA4 was also evident in vivo, because Ctla4knockout mice developed a character istic T cellmediated lymphoproliferative autoimmune disease 19 . The absence of Ctla4 was sufficient to cause this phenotype, as treatment with an engineered solu ble version of a CTLA4:Fc fusion protein (CTLA4ig) and genetic crosses to B7deficient mice ameliorated dis ease 20, 21 . The autoimmune lymphoproliferative disorder caused by Ctla4 loss depends on the activity of CD28 because mutation of an LCKbinding carboxyterminal proline motif in the intracellular tail of CD28 abrogates disease in mouse models 22 . Moreover, human patients with CTLA4 haploinsufficiency exhibit similar severe multiorgan lymphocytic infiltration and autoimmunity (CHAI disease) that can be treated with abatacept, an FDAapproved CTLA4Ig 23, 24 .

CTLA4 restrains T cell activation through multiple mechanisms: by directly antagonizing CD28, by com peting for costimulatory ligands, by preventing immune conjugate formation and by recruiting inhibitory effec tors 25 (Fig. 2) . To directly oppose CD28 activity, intracel lular vesicles release CTLA4 at the immunological synapse Antigen-presenting cells (APCs) . immune cells involved in the uptake and processing of antigens to initiate cellular immune responses.

Soluble recombinant human cytotoxic T lymphocyte antigen 4 (CTLA4) fused to the immunoglobulin Fc domain that competes with endogenous CD28 for its ligands.

An interface between interacting lymphocytes and antigen-presenting cells that controls antigen-induced signalling. Fig. 1 | Peripheral T cell fates after antigenic activation. Resting T cells become activated after stimulation by cognate antigen in the context of an antigen-presenting cell and co-stimulatory signals. Activated T cells produce and consume proliferative/survival cytokines, for example, IL-2, IL-4 and IL-7 , and begin to expand in number. If CD4 + CD25 + regulatory T (T reg ) cells are present, they can deprive the cycling T cells of proliferative/survival cytokines, especially IL-2, causing them to undergo apoptosis. Once cells are proliferating rapidly , they have different fates depending on their environment. If they receive acute strong antigenic stimulation, especially if it is encountered repeatedly , the cells will undergo restimulation-induced cell death. By contrast, if they receive chronic weak antigenic stimulation, the cells will survive but become reprogrammed into a specific unresponsive transcriptional state known as 'T cell exhaustion'. Finally , as the antigen and cytokine stimulation diminishes as the immune response wanes, usually once the pathogen has been cleared, cytokine withdrawal can occur passively to contract the expanded population of antigen-specific T cells. A small fraction of cells will be reprogrammed to enter a 'memory' phenotype, and this differentiation step is facilitated by IL-7 and IL-15. Memory T cells will continue to persist in the immune system and form the basis of anamnestic responses. In these regulatory processes, T cell death usually takes the form of apoptosis.

where it associates with the TCR 26 . In the context of the immunological synapse, CTLA4 can also reorganize the cytoskeleton and disturb T cell-APC immune conjugate formation 27 . CTLA4 also mediates the internaliza tion of its ligands, thereby preventing their binding to CD28, which, in turn, reduces iL-2 secretion and T cell proliferation 17, 28, 29 . Lastly, phosphatases, including SH2 domaincontaining tyrosine phosphatase 2 (SHP2) and protein phosphatase 2A (PP2A), are recruited and inter act with the cytoplasmic tail of CTLA4, thereby contrib uting to its negative effect on T cell activation. SHP2 is an inhibitor of phosphorylation of the CD3 ζsubunit of the TCR and also inhibits phosphorylation of the adaptor protein linker of activated T cells (LAT) 30, 31 . PP2A is hypothesized to inhibit extracellular signalregulated kinase (ERK), a kinase that acts as a signalling protein downstream of the TCR 32 . However, there is significant debate about which of the molecules that associate with the cytoplasmic tail of CTLA4 are most important for inhibiting T cell activity. Nevertheless, these inhibitory signals reduce the activation of transcription factors, such as activator protein 1 (AP1), nuclear factorκB (NFκB) and nuclear factor of activated T cells (NFAT), which reprogrammes T cells towards an anergic fate 29, 33 . Beyond its function in activated conventional T cells, CTLA4 expression on T reg cells is essential for the direct and indirect immunosuppressive activity of these cells 34, 35 . In vitro studies showed that CTLA4 was necessary for antiinflammatory cytokine release by T reg cells, which reduces polyclonal activation and proliferation of conventional T cells nearby 36, 37 . This result was confirmed in vivo by adoptive trans fer of CTLA4bearing T reg cells to prevent auto immunity induced by CTLA4deficient T cells that had been transferred to T cell and B celldeficient mice (Rag -/mice) 38, 39 . This treatment effect was nullified by antibodymediated neutralization of CTLA4 (reFS 38, 40, 41 

Before activation, antigen-presenting cells (APCs) load antigen onto MHC molecules to prepare for contact with a T cell that displays a cognate T cell receptor (TCR) while also providing necessary co-stimulatory ligands B7-1 and B7-2. The inhibitory molecule cytotoxic T lymphocyte antigen 4 (CTLA4) is contained within intracellular vesicles in naive T cells, whereas it is constitutively expressed on the cell surface of CD4 + CD25 + regulatory T (T reg ) cells. Both classes of T cells express the co-stimulatory receptor CD28. Early after activation, generally in the lymphoid tissue, T cells are activated when their TCRs bind to their cognate antigen presented by APCs in conjunction with CD28 binding to B7-1/B7-2. Also, the activated T cells begin the process of displaying CTLA4 on the cell surface. T cells within peripheral tissues upregulate PD1 at the mRNA level early after activation. Late after activation, in lymphoid tissue, CTLA4 expressed by activated T cells binds to the B7-1 and B7-2 molecules on APCs, thereby preventing their binding to CD28 and promoting anergy by decreasing the T cell activation state. At the same time, constitutive expression of CTLA4 on T reg cells leads to trans-endocytosis of B7 ligands and interferes with the CD28 co-stimulatory ability of APCs. Late after activation in peripheral tissues, PD1 is further upregulated transcriptionally , leading to greater surface expression of programmed cell death 1 (PD1), which binds to its ligands PDL1 and PDL2, thereby promoting T cell exhaustion at sites of infection or when confronted with neoplasms. Image courtesy of the National Institute of Allergy and Infectious Diseases.

A biological unit that comprises interacting lymphocytes and antigen-presenting cells.

A cytokine essential for lymphocyte activation, proliferation and tolerance.

An intracytoplasmic protein that facilitates molecular interactions and signal transduction.

www.nature.com/nri dendritic cells to induce anergy of conventional T cells in a CTLA4dependent fashion by binding to B7 ligands on APCs, followed by internalizing and degrading them, a process termed transendocytosis 28, 44 .

CTLA4 blockade in cancer. The recognition of CTLA4 as a negative regulator of T cell activation gave rise to the idea that blocking its actions could unleash a therapeutic response of T cells against cancer 45 (Fig. 3 ). James Allison and colleagues first tested this idea and demonstrated that neutralizing antiCTLA4 antibodies enhanced antitumoural immunity in mice against transplanted and established colon carcinoma and fibrosarcoma 46 .

In addition, during rechallenge, animals treated with antiCTLA4 were able to rapidly eliminate tumour cells through immune mechanisms, providing evidence that blocking of CTLA4 induces longlasting immunologi cal memory 46, 47 . Although CTLA4targeted mono therapy was shown to confer benefit in animal models of brain 48 , ovarian 49 , bladder 50 , colon 46 , prostate 47 and soft tissue 46 cancers, less immunogenic cancers, including SM1 mammary carcinoma 51 and B16 melanoma 52 , did not respond as favourably. Furthermore, heterogeneity between cancer models yielded discordant tissuespecific results 45, 53 . In addition, a greater tumour burden corre lated with reduced tumour responses to antiCTLA4 treatment because larger tumours foster a more robust antiinflammatory tumour microenvironment 45, 49 . Despite the mixed success in preclinical studies, mAbs targeting CTLA4 proved effective in clinical trials of melanoma 45 . Ipilimumab, a human IgG1κ antiCTLA4 mAb, gained FDA approval in 2011 for nonresectable stage III/IV melanoma following evi dence that it elicited potent tumour necrosis 54 and conferred a 3.6month shortterm survival benefit 55 . Longterm survival data demonstrated that 22% of patients with advanced melanoma treated with ipili mumab benefited from an additional 3 years or more of life 56 . Additional longterm studies have demonstrated the durability of this survival benefit, indicating the per sistence of antitumoural immunity following CTLA4 blockade 56,57 . Unfortunately, trial results in renal cell carcinoma 58 , nonsmallcell lung cancer 59 , smallcell lung cancer 60 and prostate cancer 61 have yielded less impressive effects than those seen in patients with melanoma. Tremelimumab, an IgG2 isotype form of a CTLA4blocking antibody, has yet to receive FDA approval as it did not increase survival in advanced melanoma 62 . It is hypothesized that effectiveness varies between ipilimumab and tremelimumab owing to dif ferences in binding kinetics and the capacity to mediate antibody-dependent cell-mediated cytotoxicity 63, 64 .

The mechanisms of CTLA4mediated tumour regres sion are pleiotropic but unified by the action of one cell type, the T lymphocyte (Fig. 3 ). T cell responses are nec essary for the therapeutic effects of CTLA4targeted agents because T cell depletion in animal models abolishes tumoricidal activity 65 . Inhibition of CTLA4 enhances T cell clonal responses to tumourassociated neoantigens and a high neoantigen burden portends a favourable response to antiCTLA4 therapy 66, 67 . Apart from boosting effector T cell responses, antiCTLA4 therapy depletes local intratumoural T reg cells through antibodydependent cellmediated cytotoxicity in mouse models and shifts the balance of the tumour microenvironment away from immunosuppression 68, 69 . This phenomenon requires further study in human can cer as current data are inconclusive 70, 71 . The relative role of effector T cells and T reg cells in conferring a clinical benefit has been contested, although specific blocking of CTLA4 in both cell populations can lead to synergistic increases in tumour regression 69 . Overall, current data suggest that the most critical factor in predicting out come is the ratio of effector T cells to T reg cells infiltrating the tumour 45, 49 . PD1/PDL1 biological function. PD1 was first identified in 1992 as a putative mediator of apoptosis, although later evidence suggested a role in restraining immune system hyperactivation, analogous to CTLA4 (reF. 72 ). As a type 1 transmembrane glycoprotein within the immuno globulin superfamily, PD1 exhibits a 20% and 15%

The process by which antibody-based opsonization of target cells promotes their lysis by immune cytotoxic cells.

T cell

Cytotoxic T lymphocyte antigen 4 (CTLA4)-blocking antibodies (α-CTLA4), especially when bound to an Fc receptor (FcR) on an antigen-presenting cell (APC), can promote antibodydependent cellular cytotoxicity (ADCC). CD4 + CD25 + regulatory T (T reg ) cells express higher amounts of CTLA4 than conventional T cells and are therefore more prone to α-CTLA4-induced ADCC than conventional T cells.

In addition, α-CTLA4 can bind to CTLA4 on the surface of the T reg cell and prevent it from counter-regulating the CD28-mediated co-stimulatory pathways that are playing a role in T cell activation. At the same time, α-CTLA4 can also promote T cell responses by blocking CTLA4 on the surface of conventional T cells as they undergo activation. amino acid identity to CTLA4 and CD28, respectively 73 . Human PD1 is expressed on T cells after TCR stimula tion and binds the B7 homologues PDL1 (also known as B7H1) and PDL2 (also known as B7DC), which are present constitutively on APCs and can be induced in nonhaematopoietic tissues by proinflammatory cytokines [74] [75] [76] . In this review, we refer to PD1 and its ligands as the 'PD1 axis' . The predominant role of the PD1 axis in the negative regulation of T cell activation became clear in 1999 when loss of the mouse PD1 ortho logue, Pdcd1, was found to cause autoimmunity in vivo. C57BL/6 mice lacking functional PD1 protein devel oped splenomegaly 77 . Ageing of these animals led to mild T cellmediated lupuslike glomerulonephritis and arthritis that was exacerbated by concurrent lpr muta tions in the Fas gene 78 . Characterization of additional mouse strains showed that Pdcd1 -/mice of the BALB/c strain exhibited cardiac inflammation leading to dilated cardiomyopathy 79 . By comparison, non-obese diabetic Pdcd1 -/mice had accelerated type 1 diabetes mellitus compared with their Pdcd1sufficient counterparts 80 .

The heterogeneous and lateonset autoimmune pheno types of Pdcd1 -/mice were distinct from Ctla4 -/ani mals, demonstrating that the PD1 axis regulates T cell biology differently to CTLA4. Spatially, CTLA4 exerts its regulatory effect predominantly within lymphoid organs, whereas PD1 tends towards tempering T cell activation locally within peripheral tissues 8 . Temporally, PD1 acts later in the course of T cell activation and fate determination. Overall, the PD1 axis plays a unique role in maintaining T cell tolerance to self. PD1 restrains immune responses primarily through inhibitory intracellular signalling in effector T cells and T reg cells 81 . The immunoreceptor tyrosine-based switch motif and the immunoreceptor tyrosine-based inhibitory motif of PD1 are phosphorylated and recruit the phosphatases SHP1 and SHP2, which dephosphorylate, and thereby inactivate, downstream effectors (that is, the CD3 ζsubunit and ZAP70) that are important for early T cell activation 76 and CD28 signalling 82 . Both CTLA4 and PD1 inhibit protein kinase B (PKB; also known as AKT) signalling to reduce glucose uptake and utiliza tion, the former through PP2A and the latter by reducing phosphoinositide 3kinase (PI3K) activity 83 . In contrast to CTLA4, the PD1 axis is essential for controlling the continued activation and proliferation of differentiated effectors; when PD1 engages its ligands, it can induce a state of T cell dysfunction called T cell exhaustion [84] [85] [86] . However, what determines whether PD1 mediates exhaustion or apoptosis in certain contexts is still an active area of research. One model suggests that the interaction between PI3K signalling and the mito chondrial B cell lymphomaextra large (BCLX L ) pro tein is a critical control point at which PD1mediated P13K inhibition reduces BCLX L and promotes apop tosis 25, 83 . Beyond regulating conventional T cells, PDL1 on APCs can control T reg cell differentiation and suppressive activity 87 . Unfortunately, tumour cells can exploit this mechanism by upregulating PD1 ligands to induce T cell exhaustion and generate a tumour microenvironment that facilitates tumour growth and invasion 88 .

Once the PD1 axis was implicated in the negative regulation of T cells, preclin ical work examined whether inhibitors of this pathway could be used for cancer treatment and biomarker discovery. First, overexpression of PDL1 or PDL2 in cancer cell lines was found to constrain the CD8 + T cell cytotoxic antitumour response, whereas tumours were rejected in mice without functional PD1 (reFS 89, 90 ). Second, blockade of PD1 suppressed the growth of transplanted myeloma cells in syngeneic animals 90 . Conversely, transplanted cells overexpressing PDL1 or PDL2 in syngeneic mice allowed for increased tumour colonization, burden and invasiveness 90 . Neutralizing the PD1 axis using mAbs 89, 91 or secreted PD1 extracel lular domains 92 reversed these effects and enhanced T cell cytotoxicity towards tumour cells 90 (Fig. 4) .

Rescuing CD8 + T cell cytotoxicity by PD1 blockade depends on the expression of CD28 as PD1mediated immunomodulation is lost in the context of CTLA4Ig, B7 blockade or CD28 conditionalknockout mice 92 . In addition, reinvigorated T cells in the peripheral blood of patients with lung cancer following PD1 blockade were shown to express CD28 (reF. 93 ). PD1 inhibition not only augments antitumoural immunity but also limits haematogenous seeding of B16 melanoma and CT26 colon carcinoma metastases in mouse models 94 

An albino inbred mouse strain commonly used in immunology research.

An inbred mouse stain with enhanced susceptibility to spontaneous development of type 1 diabetes mellitus.

A conserved amino acid sequence (TxYxx(V/i)) involved in both activation and inhibition of downstream signalling depending on the cell type and biological context.

A conserved amino acid sequence (S/i/V/LxYxxi/V/L) involved in the recruitment of inhibitory phosphatases to dampen downstream signalling.

The progressive loss of effector function due to chronic low-affinity antigen stimulation.

www.nature.com/nri cytolysis and limit metastasis. Apart from a role of PD1 and its ligands in cancer treatment, multiple studies have also shown a negative correlation between human tumour expression of proteins involved in the PD1 axis and prognosis, indicating the utility of these proteins as potential biomarkers [95] [96] [97] . Following preclinical success, mAbs designed to counteract negative immunoregulation by the PD1 axis were developed and efficacy was shown in clinical tri als 98 . Development was initiated by Medarex (ultimately purchased by BristolMyers Squibb) in 2001 (reF. 99 ). In 2010, a phase I trial demonstrated that PD1 blockade was well tolerated and could promote antitumoural responses 100 . In 2014, the humanized and fully human antiPD1 mAbs pembrolizumab and nivolumab (both IgG4) became the first FDAapproved PD1targeted therapeutics for refractory and unresectable mela noma [101] [102] [103] [104] . In a headtohead comparison, pembroli zumab showed better 6month progressionfree survival than ipilimumab and conferred an overall survival ben efit 105, 106 . Clinical trials of nivolumab demonstrated an overall survival of 72.9% at 1 year compared with 42.1% survival in the group of patients treated with the chem otherapeutic dacarbazine 104 . In 2015, pembrolizumab was approved for the treatment of PDL1expressing nonsmallcell lung carcinoma because it provided a 4.3month increase in progressionfree survival com pared with platinumbased chemotherapeutics and was more effective than the chemotherapeutic pacli taxel 107, 108 . Increased PDL1 expression on the target tumour was associated with improved responses to PD1 axis blockade 109 . Additional successful clinical trials expanded the use of pembrolizumab to head and neck squamous cell carcinoma 110 , Hodgkin lymphoma 111 , urothelial carcinoma 112 , gastric/gastrooesophageal junc tion cancer 113 and tissueagnostic carcinoma with a high degree of microsatellite instability 114 . Following approval in tissueagnostic cancers with microsatellite instability, pembrolizumab became the first drug to be approved based on a molecular biomarker rather than by cancer site. However, the immunosuppressive microenviron ment of different tissues makes it hard to predict which patients will benefit 115, 116 . Similar to prembrolizumab, the use of nivolumab has since been extended to renal cell carcinoma 117 , head and neck squamous cell carcinoma 118 , urothelial carcinoma 119 , hepatocellular carcinoma 120 , Hodgkin lymphoma 121 and colorectal cancer with a high degree of microsatellite instability 122 . As was seen with antiCTLA4 therapy, longterm survival analyses demonstrate a longlasting immunemediated survival benefit following PD1 blockade 123 . However, the reason why PD1 blockade has demonstrated broader clinical utility than antiCTLA4 treatment has remained elusive. It is hypothesized that the difference may be because the PD1 axis is frequently coopted by tumours via ligand expression, whereas CTLA4 represents a broader immunoregulatory circuit 74, 124 . PDL1 is also targetable by specific antibodies that have proven effective treatments in multiple forms of cancer. In 2016, the first PDL1targeted humanized mAb, atezolizumab (an IgG4 antibody), was approved for treatment of urothelial carcinoma. An overall response rate of 15% was deemed statistically signifi cant based on historical control data, although responses were dependent on tumour PDL1 expression status 125 . Unfortunately, additional trial data have not demon strated that atezolizumab has clinical efficacy beyond the standard of care in urothelial carcinoma, although it is less toxic than traditional chemotherapy 126 . Indications have since expanded to include the treatment of non smallcell lung carcinoma 127 , triplenegative breast cancer 128 and smallcell lung cancer 129 . Additional antiPDL1 human mAbs, avelumab and durvalumab, entered the market in 2017 (reF. 98 ). Avelumab is used for the treatment of Merkel cell carcinoma 130 , urothe lial carcinoma 131 and advanced renal cell carcinoma 132 . Duvalumab is used for urothelial carcinoma 133 and nonsmallcell lung cancer 134 . Therefore, similar to PD1, blockade of PDL1 has been effective in difficulttotreat forms of cancer.

Blocking a natu rally occurring central immune checkpoint unleashes powerful immune effector mechanisms that may not respect the normal boundaries of immune tolerance to selftissues 135 . Ctla4-and Pdcd1knockout mice provided a glimpse into the spectrum of autoimmune responses that occur in humans during immune check point blockade therapy 19, [77] [78] [79] . Human lossoffunction mutations in CTLA4 and its interacting regulatory pro tein, LRBA, also mirror the immunerelated side effects observed with antiCTLA4 therapy 13, 24 . On the basis of a metaanalysis of trial data sets, immunerelated adverse events are estimated to occur in 15-90% of patients 55 . More severe events requiring intervention are observed in 30% and 15% of patients treated with CTLA4 and PD1 axis inhibitors, respectively 136 . The common immune feature of toxicity is the loss of naive T cells and the accumulation of overactive memory T cells that invade peripheral organs, such as the gastrointestinal tract and lungs, and cause inflammatory damage. Keratinized and nonkeratinized mucosa appear to be the most susceptible, as approximately 68% and 40% of treated patients exhibit pruritis and mucositis, respectively 137, 138 . AntiCTLA4 therapy carries an increased risk of severe autoimmune complications compared with therapies targeting the PD1 axis, as was observed in knockout mice and in clinical studies 19, [77] [78] [79] [80] 139 . In addition, data from doseescalation trials support the claim that antiCTLA4 agents elicit dosedependent responses not seen with therapies targeted at the PD1 axis 107, 139 . Toxicities affecting the gastrointestinal tract and brain are more common with antiCTLA4 therapy, whereas patients treated with PD1 axistargeted therapies are at higher risk of hypothyroidism, hepatoxicity and pneumonitis 137 . However, as the number of indica tions treated with checkpoint blockade increases and more patients are treated, rarer side effects in a wider spectrum of organs and heterogeneous responses have manifested 137 . For example, hyperprogression of disease has been observed in a minority of patients with vari ous tumour types treated with PD1 inhibitors [140] [141] [142] . Most recently, it was shown that the PD1 inhibitor nivolumab can lead to the rapid progression of disease in patients Nature reviews | Immunology with adult T cell leukaemia/lymphoma, providing evi dence for a role of tumourresident T reg cells in the patho genesis of this lymphoma 143 . Multiple immunerelated response criteria have been developed to better catego rize patient responses to checkpoint blockade. In addi tion, these criteria aim to distinguish progression from pseudoprogression, a phenomenon in which patients treated with CTLA4 or PD1 inhibitors experience a period of progression followed by rapid tumour clear ance 144, 145 . Overall, checkpoint blockade leads to auto immune toxicities with a therapyspecific pattern of organ involvement, as predicted by the phenotypes of animals genetically deficient for checkpoint molecules.

Interestingly, preclinical immune checkpoint ther apy studies did not demonstrate major adverse effects in vivo and, thus, were not great predictors of human toxicities 146 . This is thought to be due to the short time frame of these studies and the inbred nature of mouse strains 146 . Recently developed humanized mouse mod els represent a platform that better recapitulates side effects due to checkpoint therapy 146, 147 . Nevertheless, toxicity associated with immune checkpoint blockade is tolerated better than the toxicities associated with traditional chemotherapeutics, making these therapies attractive for quality of life reasons beyond their survival benefit 98, 148 .

Recent research has aimed to improve the sideeffect profiles and clinical response of immune checkpoint blockade through the modification of existing antibod ies and the engineering of novel delivery methods. It was recently shown that abnormal CTLA4 recycling and sub sequent lysosomal degradation was a mechanism that contributes to toxicities and reduced drug effectiveness. Modified pHsensitive antibodies that do not interfere with LRBAmediated CTLA4 recycling were shown to limit adverse events and improve clinical outcomes in established tumours in mouse models, which may ulti mately broaden clinical utility 149, 150 . Additional research has focused on developing biomaterials for the localized administration of checkpoint inhibitors 151 . For example, compared with systemic delivery, transdermal patch delivery of antiPD1 antibodies was better tolerated and unleashed a more robust antitumoural response in a mouse model of melanoma 151 . A broad field of research is currently aimed at discovering novel meth ods to reduce toxicities associated with checkpoint ther apy and to increase clinical benefit in a greater variety of tumours.

Clinical management of drugrelated toxicities is the same for all checkpoint drugs, and toxicities are graded according to the 2009 National Cancer Institute Common Terminology Criteria for Adverse Events severity scale 137, 152 . Mild (grade 1) toxicities are not typically treated. In the setting of grade 2 or 3 adverse events, checkpoint inhibitors are discontinued until symptoms and laboratoryvalue abnormalities resolve. Glucocorticoids are also used to effectively control immune hyperactivity. Infliximab and other immuno suppressive agents can be used when glucocorticoids fail. Lifethreatening (grade 4) toxicities necessitate the com plete discontinuation of therapy and the use of lifesaving measures, as required. Active monitoring of symptoms and laboratory parameters is recommended in order to prevent death due to checkpoint blockade (grade 5).

Current research is aimed at identifying predictive biomarkers for organspecific toxicities due to check point therapy. For example, neutrophil activation, as measured by increased expression of the biliary glycopro tein CEACAM1 and the cell surface glycoprotein CD177, correlates with gastrointestinalrelated side effects in patients treated with ipilimumab 153 . Increases in eosino phil counts and release of the proinflammatory cytokine IL17 are associated with toxicity regardless of the organ affected 154, 155 . Pharmacogenomic profiling (using genetic information to predict responses to drugs) may provide more insight into the relevant genes and pathways medi ating toxicity 137 . Ultimately, the hope is that genetic, bio chemical or metabolic profiling could either prescreen or rapidly detect individuals likely to experience the most severe adverse reactions to checkpoint therapy.

Adoptive T cell (ATC) therapy, in which autologous or allogenic T cells are infused into patients with cancer, has shown considerable promise in recent years. The viability of this type of therapy was first shown by Southam et al. in 1966 , when half of the patients with advanced cancer demonstrated tumour regression follow ing cotransplantation with patientderived leuko cytes and autologous tumour cells 156 . Allogenic haemato poietic stem cell transplants for leukaemia represented the first effective adoptive transfer approach deployed clinically, and clinical improvement was shown to be mediated by a T cell graft versus tumour response 157 .

ATC with tumour-infiltrating lymphocytes. ATC ther apy using tumourinfiltrating lymphocytes (TILs) for the treatment of metastatic melanoma was pioneered at the National Cancer Institute in the late 1980s 158 . Lymphocytes isolated from a cancer biopsy were greatly expanded with IL2 and then reinfused intra venously into the same patient with a large bolus of IL2. The objective response rate was 34%; however, the median duration of response was only 4 months and few patients experienced a complete response 159 . Later studies incorporating lymphodepletion before ATC therapy in 93 patients with metastatic melanoma were more successful, with complete tumour regression in 20 (22%) patients, 19 of whom were still in complete remission 3 years after treatment 160 . The screening and enriching for neoantigenspecific TILs, made possible by highthroughput technologies, recently demon strated promise in a patient with metastatic breast cancer 161 . In addition, knockdown of the gene encod ing cytokineinducible SH2containing protein (Cish), a negative regulator of TCR signalling, was shown to boost the antitumoural response of ATC therapy in mouse models 162 . However, in order for TILbased ATC therapy to elicit durable responses (Fig. 5) , effector T cells with antitumour activity must be present in the tumour, which is not the case for many cancer types 163 . Other innovative approaches to tweak T cell activity and pro liferation may allow for a greater palette of treatments to be developed.

Engineered lymphocytes for ATC. The challenges asso ciated with expanding tumourspecific T cells in vitro led to the development of TCRengineered lymphocytes (Fig. 5) . However, these cells are limited to responding to tumour antigens presented by the MHC (also known as human leukocyte antigen (HLA) in humans) rather than surface antigens on tumour cells 163 . However, synthetic chimeric antigen receptors (CARs) can bypass MHC restriction and direct specific cytotoxicity to a target molecule on the surface of the malignant cell. Isolated T cells from the patient (or allogeneic donor) are genet ically modified to express CARs and then expanded and infused into the patient. This overcomes the problem that tumour cells often downregulate MHC molecules, which leaves the cell unable to present antigen to con ventional T cells 164 . CARs comprise an antigenbinding domain, most often from the variable regions of anti bodies, linked to signalling domains of the TCR and various costimulatory molecules (Fig. 5) . Given the domain modularity of cell surface signalling proteins, mixes and matches of extracellular targeting domains and internal signal transduction domains can be assem bled using protein engineering. This offers many options to tailor CARs to specific tumours. The first generation of CAR T cells relied only on the CD3 ζchain to simu late TCR signalling 165 , but this design was ineffective in clinical trials owing to limited T cell proliferation and cytokine production 166, 167 . Subsequent generations of CAR T cells have been engineered to include domains from CD28, CD40 ligand and other positive regulators of T cell activation to potentiate activation and cyto toxicity in vivo [168] [169] [170] [171] . An engineered singlechain PD1 blocker has also demonstrated similar enhanced effi cacy to secondgeneration CAR T cells with solely a CD28 domain 172 . Even though CAR T cells are typically engineered using retroviral transduction, recent work has used CRISPR-Cas9 technology. CRISPR-Cas9 can be used to edit the TCR germline sequence directly, which could lead to more uniform CAR T cell generation and, ultimately, better efficacy 173 . TILs are then infused into a patient who has undergone lymphodepletion to provide a niche for the transferred TILs to expand, act as effector cells and generate immunological memory. As the T cells were derived from the tumour, it is assumed that a good proportion can recognize tumour-associated antigens (TAAs) or neoantigens. b | The physiological T cell receptor (TCR) complex gains its specificity from polymorphic α-and β-glycoprotein chains that have an antigen-binding portion and a conserved domain that associate with and signal through a group of nonpolymorphic proteins, CD3 γ, δ, ε and ζ. Bioengineering of the TCR α-and β-glycoprotein antigen-binding domain (purple), while preserving the conserved domains (Cα and Cβ), allows for the development and expansion of T lymphocytes with specificity to tumour neoantigens. c | Originally , chimeric antigen receptors (CARs) were composed of an extracellular singlechain fragment of an antibody variable region coupled to a CD3 ζ-signalling domain. Poor expansion and functionality of these first-generation CARs led to the development of second and third-generation CARs containing intracellular modules from co-stimulatory molecules (CD28 and/or 4-1BB) that provide additional signals necessary to fully activate the T cell. Subsequent generations of CAR T cells contain further modifications to improve antitumour efficacy. For example, fourth-generation 'armoured' CAR T cells have been engineered to secrete pro-inflammatory cytokines, such as IL-12, to overcome immunosuppression in the tumour microenvironment. The chimeric cytokine receptor 4αβ, comprising the ectodomain of IL-4Rα fused to the IL-2/IL-15Rβ chain, signals in response to IL-4, an abundant cytokine in numerous tumour types. V H , variable heavy chain; V L , variable light chain.

A limitation to the development of CAR T cell ther apies is the requirement for a distinct tissuerestricted target antigen on the tumour cell surface. For example, CAR T cells designed with specificity for the cell surface molecule CD19, which is expressed by all B cells, have been successful in the treatment of B cell malignancies. The first clinical deployment of secondgeneration CD19specific CAR T cells led to durable responses in chronic lymphocytic leukaemia 174 . Additional clin ical trials of CD19specific secondgeneration CAR T cells in B cell acute lymphoblastic leukaemia (BALL) led to remission in all patients with BALL who were tested 175 . A followup report on patients with BALL enrolled in this clinical trial showed complete remis sion of disease in 44 of 53 (83%) patients with a median followup of 29 months 176 . Similar successes were reported for patients with diffuse large B cell lym phoma 177 , leading to FDA approval for these B cell malignancies in 2017.

The clinical success of CAR T cell therapy for the treatment of BALL and diffuse large B cell lymphoma is due, in part, to targeting the CD19 antigen, an ideal candidate owing to its high expression in certain B cell malignancies and specificity to the B cell lineage. Crossover targeting of normal CD19 + B cells does not hamper therapy or cause severe side effects. However, even as an ideal target, CD19 antigen loss is a com mon cause of treatment failure. CD22 is another anti gen commonly expressed by malignant cells in BALL and has shown promise as a target for CAR T cell therapy in a phase I trial 178 . Other targets, especially tumour neoantigens, are currently being investigated for haematological malignancies that do not express CD19, as well as for solid tumours 179, 180 . B cell matu ration antigen (BCMA)targeted CAR T cell therapy is poised for FDA approval for multiple myeloma in 2020 on the basis of promising preclinical and clinical data 181, 182 . However, owing to reported patient relapses, the investigation of additional target antigens continues. A preclinical study recently identified another target antigen, GPRC5D, with comparable efficacy and toxic ity to BCMAtargeted CAR T cell therapy 183 . Thus far, CAR T cell therapy has only been modestly successful for solid tumours [184] [185] [186] and innovative approaches to improve therapy are underway 179 . A recently identi fied pancancer target, B7H3 (also known as CD276), has demonstrated success in multiple paediatric solid tumour models 187 . In addition to directly acting as cytolytic agents, CAR T cells can also target the unhos pitable tumour microenvironment and revive exhausted T cells 188, 189 . For example, a new generation of 'armoured' CAR T cells engineered to produce IL12 can overcome immunosuppression by T reg cells and myeloid cells in the tumour environment, promote CD8 + T cell cytol ytic activity 190 and enhance myeloid cell recruitment and antigen presentation 191, 192 . Preclinical models using IL12expressing CAR T cells that target the conserved extracellular domain of mucin 16 (MUC16 ecto ) have shown promising results in models of ovarian cancer, a tumour with poor prognosis in advanced stages 193, 194 . A phase I clinical trial is currently in progress for patients with ovarian, fallopian or primary peritoneal cancer 195 .

The efficacy of CAR T cells may also be strengthened through coexpression of a chimeric cytokine recep tor (4αβ) that stimulates proliferation in response to IL4, a cytokine that is usually abundant in the tumour microenvironment. Preliminary studies have shown that this approach works for CAR T cells directed against different tumourassociated antigens (TAAs) 196 and clinical trials are underway in head and neck cancer 197 . In addition, overexpression of the transcription fac tor JUN was shown to confer resistance to CAR T cell exhaustion 198 . Overall, CAR T cells have been success ful for the treatment of B cell malignancies and it will be exciting to continue research on this new treatment modality for intractable types of cancer.

Toxicities can arise from CAR T cell therapy and affect many differ ent organ systems with a range of severity 199 . Patients most commonly experience cytokine release syndrome (CRS) and neurotoxicity 200 . CRS results from the power ful activation and proliferation of CAR T cells in vivo and typically appears quickly after cell transfer. The symptoms are often mild and flulike but can also be severe and lifethreatening, involving hypotension, high fever, capillary leakage, coagulopathy and multi system organ failure. Serious neuro logical events can also occur, such as CAR T cellrelated encephalopathy syndrome, typically characterized by confusion and delirium, but sometimes also associated with sei zures and cerebral oedema 199 . Glucocorticoids are the firstline treatment for milder forms of CRS and CAR T cellrelated encephalopathy syndrome. Tocilizumab, a humanized antiIL6 antibody, is a highly effective secondline treatment for CRS caused by CAR T cell therapy 201 . Other side effects of CD19specific CAR T cell therapy include lymphopenia and hypogamma globulinaemia 202 , which can be effectively managed with intravenous immunoglobulin therapy, similar to the treatment that patients with primary B cell immuno deficiencies receive 203 . The mechanisms behind these side effects are unclear and further research may yield ways to avoid or minimize toxicity. Recent develop ment of a novel murine model of CRS demonstrated that it is not mediated by CAR T cellderived IL6 but rather by recipient macrophages that secrete IL6, IL1 and nitric oxide. Therefore, IL1 blockade repre sents a possible novel intervention in the armamen tarium against CRS 204 . Moreover, a clinical study of lowaffinity CD19specific CAR T cells demonstrated reduced toxicity and enhanced efficacy 205 . Additional efforts to reduce toxicity involve the engineering of CAR T cells with multiple receptor specificities 206 and reducing the halflife of cellular toxicity by using mRNAbased methods that allow for transient recep tor expression 207 or including suicide cassettes that can be activated by exogenous agents to clonally delete the infused cells 208 .

The ATC approach necessitates a patientspecific therapy design, its cost can be prohibitive, patient access to the treatment is limited and manufacturing is challenging. In the United States, the CAR T cell thera pies tisagenlecleucel and axicabtagene ciloleucel have a www.nature.com/nri direct cost of US$475,000 and US$373,000 per patient, respectively 209 . However, these values do not take into account the additional costs associated with treating the severe adverse effects common to CAR T cell therapy, which are estimated to increase drugassociated costs by US$30,000 or more 209 . In comparison with CAR T cell therapy, checkpoint blockade has a price tag of approximately US$12,500 per month 210 . Patient access to CAR T cell therapies also represents a major problem as there are only a few laboratories certified to gener ate CAR T cells and only a few specialized tertiary care centres able to administer this therapy 211 . Lastly, varia bility in the manufacturing of CAR T cells and a lack of standard practices can contribute to heterogeneous outcomes 211 .

Cancer vaccines prompt the immune system to pro tect the body from cancer and fall into two categories, prophylactic and therapeutic. Prophylactic vaccines against hepatitis B and human papillomavirus have been instrumental in reducing the incidence of hepatocellular carcinoma and cervical cancer, respectively 212 . These are classic vaccines used to prevent infection by oncogenic viruses. By contrast, therapeutic vaccines aim to harness the immune system to eliminate diseasecausing cells that are already neoplastic 212 . An early example of this is the use of the bacillus Calmette-Guérin vaccine, com prising attenuated Mycobacterium bovis, which is gener ally used as a prophylactic tuberculosis vaccine but has also been repurposed as a primitive therapeutic vaccine for bladder cancer 213 .

Historically, the discovery of TAAs 214 , which are highly expressed on tumour cells and to a lesser extent on normal tissues, opened the door for further therapeutic vaccinebased approaches. However, as TAAs are often recognized by the immune system as 'self ' , viral antigens and neoantigens that are unique to a malignancy may be more suited as vaccine targets.

Early vaccination approaches in the 1970s were based on autologous tumour vaccines and involved the administration of patientderived tumour cells together with an adjuvant or virus in order to activate polyclonal immune responses to TAAs 215 . For example, autologous tumour cells infected with Newcastle disease virus have been used in one type of cancer vaccine that has demon strated success in preclinical models of metastatic lym phoma and melanoma 216, 217 . Modified Newcastle disease virusbased vaccines have been engineered to express granulocyte-macrophage colonystimulating factor (GMCSF) in attempts to enhance efficacy 218 . Synergism of vaccine approaches with checkpoint blockade agents has also been demonstrated in some preclinical studies of melanoma 46, 219 . Numerous autologous tumour vac cines are being investigated in phase II and phase III tri als but have yet to receive FDA approval. This approach suffers from multiple limitations, most notably the diffi culty in obtaining patientderived tumour cells in certain cancer types 212 . Newer approaches include the devel opment of personalized recombinant cancer vaccines informed by nextgeneration sequencing of genomic DNA from tumours.

Vaccines that elicit responses to tumourderived neoantigens should induce more robust immune responses and cause fewer autoimmunerelated toxic ities than vaccines based on selfderived TAAs, as the T cells that are activated by such a vaccine would not have undergone negative selection during develop ment. These factors, as well as the ability to identify neoantigens through nextgeneration sequencing of genomic DNA from tumours, has shifted the focus to investigating the clinical feasibility of making person alized recombinant vaccines that target neoantigens. However, although a higher mutational burden in the tumour has been shown to correlate with greater immuno genicity and survival after checkpoint blockade 66, 220 , only a small percentage of neoantigens spontaneously generate immune responses in patients with cancer 221 . Sahin and colleagues showed that neoantigens identi fied through nextgeneration sequencing can generate antitumour responses in vivo; in mice that were vac cinated with 50 different neoantigens, 16 were immuno genic 222, 223 . Interestingly, most neoantigens induced cytokine responses from CD4 + T cells rather than CD8 + T cells, suggesting that neoantigens are selected for MHC class II binding 222, 223 . Other preclinical studies demonstrated effective CD4 + and CD8 + T cell responses to neoantigen vaccines in various cancer types [223] [224] [225] [226] [227] . However, recent preclinical work has also highlighted the nonoverlapping role of neoantigen responses mediated by CD4 + and CD8 + T cells 228 .

To design and manufacture a personalized vaccine for clinical use, computerbased algorithms are used to identify which tumourderived peptides could poten tially form a suitable TAA or tumour neoantigen with the patient's MHC alleles (Fig. 6 ). There are several dif ferent strategies to formulate neoantigenbased vac cines, including as synthetic peptides, mRNA, viral and DNA plasmids or antigenloaded dendritic cells, and it is difficult to directly compare how each strategy influ ences immunogenicity 229, 230 . In one trial that tested a multipeptide vaccine that included up to 20 personal neoantigens, 4 of 6 patients with melanoma who entered the study with stage III disease experienced complete responses with no recurrence 25 months post vaccina tion, and the other 2 patients with progressive disease subsequently underwent antiPD1 therapy that resulted in complete tumour regression 231 . Further, of the 97 dif ferent neoantigens that were tested for immunogenicity in this study, 60% elicited CD4 + T cell responses whereas 15% elicited CD8 + T cell responses. Another clinical trial, which tested an RNA vaccine that encoded 10 pep tides representing personalized TAAs in 13 patients with advanced melanoma, achieved similar results 232 .

Although these early cancer vaccine experiments have been promising, challenges remain. An individual tumour can harbour thousands of somatic mutations and pre dicting which neoantigens can elicit strong antitumour responses remains an imperfect art. However, the cur rent methods, consisting of validating mRNA expres sion of the mutation in tumour cells and using software/ databases to predict peptide-MHC binding, have been surprisingly effective in clinical trials to date 229 . How ever, this success has been biased towards MHC class I specific neoantigens as prediction for MHC class II molecules presents unique challenges. For example, the increased diversity of MHC class II molecules and the structural nature of their open binding pocket make discerning a predictable binding motif difficult 233 . Taken together, these differences between MHC classes high light the particular need for new MHC class II predic tion algorithms. Other challenging factors to consider are the time and cost associated with developing bes poke vaccines. Currently, development and production of these vaccines takes approximately 4 months, and, although the downtime can be used to initiate other types of treat ment, shortening the time span to personalized treatment is critical. For rapidly growing or meta static tumours, months might matter. Ongoing efforts to improve design and manufacturing could shorten the production time to several weeks 229 .

Overall, the comprehensive identification of somatic mutations, and the evaluation of peptides derived from these mutations to elicit immune responses, has renewed interest in vaccination strategies for cancer treatment. Even though early clinical trials are promising, extrap olation of these findings could be misleading and advanced clinical trials will ultimately determine the efficacy of personalized vaccine therapy. Nonetheless, cancer vaccines are prototypical 'single patient and sin gle disease' precision medications and would have been in the realm of science fiction just a few decades ago. Further research and technological developments will no doubt lead to greater precision and effectiveness and also provide a better understanding of the mechanisms of antitumoural immune responses.

The molecular diversity of genetic changes that trans form cells in human cancers creates a plethora of diseases involving specific tissue types and cancer mechanisms. Given the exciting advances in cancer immunotherapy, various modifications to current immunotherapeutic approaches are being developed and tested to address the complexity of cancer immunopathogenesis and cancer targetability.

Combination therapies. Following the clinical success of checkpoint blockade monotherapy, combination therapies that couple agents with distinct mechanisms of action have augmented treatment success in various cancers. For example, ipilimumab and nivolumab com bination therapy conferred a significant survival bene fit in patients with metastatic melanoma and advanced renal cell carcinoma, leading to FDA approvals for these conditions 234, 235 . The synergism of antiCTLA4 and anti PD1 therapies is not surprising because CTLA4 and PD1 regulate antitumoural immunity in a complementary manner 8 . Crosstalk between the CTLA4 and PD1 path ways, mediated by CD80 and PDL1 dimerization, pro vides additional insight into the mechanism behind the success of dual therapy 236, 237 . However, as expected, combination checkpoint therapy also increases the risk of medicationinduced toxicities 235 .

Combining radiation therapy with checkpoint block ade is another treatment option for recalcitrant tumours. The immunomodulatory effect of radiotherapy alone represents a doubleedged sword. Mechanistically, radiotherapy increases the diversity of antitumoural T cell responses by exposing novel neoantigens at the same time as blunting the immune response through the induction of PDL1 expression on tumour cells 238 . Therefore, and on the basis of preclinical data, combin ing radiotherapy with blockers of the PD1 axis repre sents an attractive synergistic combination 239 . Patients with metastatic disease may represent a target popula tion for deploying this combination as abscopal responses to radiotherapy are boosted by checkpoint blockade for many tumour types 238, 240 . Overall, dual checkpoint block ade and radiation-checkpoint polytherapy represent promising avenues for synergistic therapeutic responses because these drug combinations display unique and complementary pharmacodynamics.

Research is also directed at newly discovered negative regulators of T cell activation, including lymphocyte activation gene 3 (LAG3), T cell immunoglobulin 3 (TIM3), Vdomain immunoglobulin suppressor of T cell activation (VISTA), B7H3 and T cell immunoreceptor with immunoglobu lin and immunoreceptor tyrosinebased inhibitory motif domains (TIGIT), as adjuvant cancer drugs [241] [242] [243] . LAG3

A phenomenon in which the therapeutic effect of radiation is extended beyond the boundaries of the tissue that was treated Fig. 6 | Personalized vaccine development. Healthy tissue and tumour tissue from a patient with cancer are submitted for DNA sequencing and bioinformatic analyses to identify gene variants that encode peptides that are specific to the tumour (neoantigens). Prediction algorithms are then used to screen for neoantigens that are likely to stably bind to the patient's MHC (also known as HLA in human) molecules and their expression is validated by sequencing tumour mRNA. Multiple predicted neoantigens are then formulated into vaccines, which are administered to the patient together with adjuvants. Post treatment, the patient is regularly monitored for neoantigen-specific immune responses and tumour growth.

www.nature.com/nri is an inhibitory ligand that reduces T cell activation by blocking CD4 contact sites on MHC class II proteins and is expressed on activated T cells and T reg cells. It prevents the overexpansion of the T cell compartment by inducing cell cycle arrest 244 . Like PD1, LAG3 is a marker of T cell exhaustion, which portends a poorer prognosis when expressed on TILs 245 . Multiple strategies of blockade have been developed, including a LAG3:Ig fusion protein and LAG3targeted mAbs 246 . In clinical trials in patients with renal cell carcinoma and pancreatic adenocarcinoma, these drugs did not succeed as monotherapies even though they increased the frequency of tumourspecific T cells 246 . However, when combined with paclitaxel for metastatic breast cancer, 50% of patients treated with LAG3:Ig responded to treatment 247 . Recent research has demonstrated that fibrinogenlike protein 1 (FGL1) acti vates LAG3 independently of binding MHC class II mol ecules and interference with this interaction is essential for unleashing potent antitumoural effects 248 .

TIM3 is another negative regulator of the T cell response. Rather than inhibiting cell cycle progression like LAG3, it regulates apoptosis following galectin 9 bin ding 249 . Its upregulation could represent a mechanism of resistance to antiPD1 therapy, making combination therapy an attractive option to boost the effectiveness of antiPD1 therapy. In addition, TIM3 expression cor relates with poor prognosis in nonsmallcell lung cancer and follicular lymphoma, suggesting a role in can cer progression 250 . Similar to TIM3, VISTA is another molecule shown to be associated with resistance to current checkpoint inhibitors and has demonstrated synergism with antiPD1 therapy in mouse models 251, 252 .

B7H3 represents another targetable negative reg ulator of the T cell response. It is highly expressed in many tumour types, including nonsmallcell lung carci noma, prostate cancer, pancreatic cancer, ovarian cancer and colorectal cancer 241, 243 . Enoblituzumab, a human ized mAb targeting B7H3, was effective at inducing antitumoural responses in a phase I study of patients with various tumour types 253 . Dualaffinity retargeting (DART) proteins that bind to B7H3 and CD3, as well as radioactive iodineconjugated B7H3 mAbs, repre sent additional ways to modulate this pathway and are in earlyphase clinical testing 254, 255 .

Lastly, TIGIT, which contains two immunoreceptor tyrosinebased inhibitory motifs in its intracellular domain and dampens T cell hyperactivation, is being investigated as a checkpoint target. It is more robustly expressed in TILs than in peripheral cells, making it an attractive target owing to its increased specificity com pared with other checkpoint molecules 243 . Preclinical evidence demonstrates that TIGIT blockade augments the effect of preexisting checkpoint inhibitors and reinvigorates tumourspecific exhausted T cells 250, 256 . Currently, blockade of immune checkpoints other than CTLA4 or the PD1 axis have not yet shown major clini cal benefits as single agents but rather may increase the effectiveness of preexisting treatments.

Although the blocking of immune checkpoint mol ecules releases potent antitumoural responses, the stimulation of T cell costimulatory receptors, includ ing inducible costimulator (ICOS), tumour necrosis factor receptor superfamily member 4 (TNFRSF4; also known as CD134), tumour necrosis factor recep tor superfamily member 9 (TNFRSF9; also known as 41BB), glucocorticoidinduced tumour necrosis factor receptor (GITR) and CD27, can also amplify the effect of existing immunotherapies, as shown preclinically and in earlystage clinical studies 168, 170, 171, [241] [242] [243] 257 . ICOS is a member of the CD28 family of costimulatory molecules that mediates contextdependent cytokine responses with an emphasis on T helper 2 (T H 2) cell skewing 258 . ICOS stimulation by vaccines modified to express ICOS ligand exhibited synergism with treatment with CTLA4blocking antibodies preclinically 259 . ICOS upregulation follow ing treatment with currently approved antiCTLA4 and antiPD1 therapies may represent a biomarker of active antitumoural responses because it associates with favourable outcomes 260 .

TNFRSF4 is another costimulatory molecule for which preclinical evidence indicates a role in deploying robust antitumoural responses in sarcoma, melanoma and breast cancer 261, 262 . Data suggest that targeting TNFRSF4 amplifies antiPD1 therapy because TNFRSF4 agonism can upregulate PDL1 expression 263 . In addi tion to synergism with checkpoint blockade, TNFRSF4 upregulation within CAR T cells by transfection repre sents a way to augment tumour cytotoxicity 170 . Agonism of additional TNFR family members, such as TNFRSF9, GITR and CD27, is being tested as adjuvant therapy in phase I/II trials for various tumour types, with prom ising results 243 . Therefore, agonism of positive T cell costimulatory signals, in concert with the existing checkpoint inhibitors or CAR T cells, represents a novel therapeutic avenue to boost antitumoural immunity.

Cancer immunotherapy focused on T cells has emerged as a powerful tool in the armamentarium against can cer. Nevertheless, it took many years of basic science discoveries and subsequent clinical translation to une quivocally demonstrate the power of modulating the immune system to treat cancer. Further research that investigates the regulation of T cells and other immune cells, for example APCs and natural killer cells, may allow us to enhance the power of this approach. In 'dif ficult to treat' tumours, the effect sizes observed in clin ical trials of checkpoint blockade agents, ATC transfer therapies and cancer vaccines have been far higher than the most effective chemotherapeutic agents. Although immunerelated adverse effects are common, these inno vative immunetargeting therapies are better tolerated than traditional chemotherapeutic agents. The burgeon ing field of cancer immunotherapy continues to grow as indications for currently approved therapies expand and the search for novel druggable targets continues. The cancer immunotherapy success stories we have recounted highlight the intrinsic connection between basic science research and clinical practice. They also illustrate how a benchtobedside approach, built upon a solid basic science foundation, can be successful in fighting one of humanity's most dreaded diseases.

Published online xx xx xxxx T helper 2 (T H 2) cell skewing Biasing of CD4 + T helper cells towards a phenotype essential for humoral immunity.

Nature reviews | Immunology

Cancer immunotherapy: a brief review of the history, possibilities, and challenges ahead

Bioimmunoadjuvants for the treatment of neoplastic and infectious disease: Coley's legacy revisited

Cancer immunotherapy: historical perspective of a clinical revolution and emerging preclinical animal models

Intradermal immunization of C3H mice against a sarcoma that originated in an animal of the same line

Spontaneous regression of primary malignant melanomas with regional metastases

Inflammation and immune surveillance in cancer

Spontaneous regression of human melanoma/nonmelanoma skin cancer: association with infiltrating CD4 + T cells

Control of peripheral T-cell tolerance and autoimmunity via the CTLA-4 and PD-1 pathways

This study screens mouse T cell cDNA libraries and identifies CTLA4 as a new immunoglobulin superfamily member expressed within the lymphoid lineage predominantly in activated cells

Human Ig superfamily CTLA-4 gene: chromosomal localization and identity of protein sequence between murine and human CTLA-4 cytoplasmic domains

CTLA-4 and CD28 activated lymphocyte molecules are closely related in both mouse and human as to sequence, message expression, gene structure, and chromosomal location

CTLA-4 can function as a negative regulator of T cell activation

AUTOIMMUNE DISEASE. Patients with LRBA deficiency show CTLA4 loss and immune dysregulation responsive to abatacept therapy

CTLA-4 is a second receptor for the B cell activation antigen B7

Human B7-1 (CD80) and B7-2 (CD86) bind with similar avidities but distinct kinetics to CD28 and CTLA-4 receptors

B7-1 and B7-2 selectively recruit CTLA-4 and CD28 to the immunological synapse

CD28 and CTLA-4 have opposing effects on the response of T cells to stimulation

CTLA-4 engagement inhibits IL-2 accumulation and cell cycle progression upon activation of resting T cells

Lymphoproliferative disorders with early lethality in mice deficient in Ctla-4

CTLA4Ig prevents lymphoproliferation and fatal multiorgan tissue destruction in CTLA-4-deficient mice

this in vivo murine study demonstrates that CD4 + T cells are necessary to cause severe multiorgan autoimmunity in the context of CTLA4 deficiency

B7-1 or B7-2 is required to produce the lymphoproliferative phenotype in mice lacking cytotoxic T lymphocyte

Induction of autoimmune disease in CTLA-4 -/-mice depends on a specific CD28 motif that is required for in vivo costimulation

Abatacept alleviates severe autoimmune symptoms in a patient carrying a de novo variant in CTLA-4

Immune dysregulation in human subjects with heterozygous germline mutations in CTLA4

At the bench: preclinical rationale for CTLA-4 and PD-1 blockade as cancer immunotherapy

Surface cytotoxic T lymphocyteassociated antigen 4 partitions within lipid rafts and relocates to the immunological synapse under conditions of inhibition of T cell activation

Reversal of the TCR stop signal by CTLA-4

This study nominates trans-endocytosis of CD80 and CD86 as a cell-extrinsic mechanism of CTLA4-mediated inhibition of T cell activation

CTLA-4 regulates induction of anergy in vivo

Using TCR transgenic mice, this study provides important in vivo evidence that CTLA4 is essential for T cell tolerance by regulating entrance into the anergic state

Regulation of cytotoxic T lymphocyteassociated molecule-4 by Src kinases

Regulation of T cell receptor signaling by tyrosine phosphatase SYP association with CTLA-4

The CD28 and CTLA-4 receptors associate with the serine/threonine phosphatase PP2A

CTLA4 ligation attenuates AP-1, NFAT and NF-κB activity in activated T cells

Dual function of CTLA-4 in regulatory T cells and conventional T cells to prevent multiorgan autoimmunity

CTLA-4 control over Foxp3 + regulatory T cell function

Immunologic self-tolerance maintained by CD25 + CD4 + regulatory T cells constitutively expressing cytotoxic T lymphocyteassociated antigen 4

This study demonstrates that CTLA4 expression within regulatory T cells is essential for the characteristic suppressive functions of these cells on the immune response

Treg and CTLA-4: two intertwining pathways to immune tolerance

Expression of CTLA-4 and FOXP3 in cis protects from lethal lymphoproliferative disease

Blockade of CTLA-4 on CD4 + CD25 + regulatory T cells abrogates their function in vivo

Cytotoxic T lymphocyte-associated antigen 4 plays an essential role in the function of CD25 + CD4 + regulatory cells that control intestinal inflammation

CD4 + regulatory T cells require CTLA-4 for the maintenance of systemic tolerance

CTLA-4 is required by CD4 + CD25 + T reg to control CD4 + T-cell lymphopenia-induced proliferation

A transendocytosis model of CTLA-4 function predicts its suppressive behavior on regulatory T cells

CTLA-4 blockade in tumor models: an overview of preclinical and translational research

Enhancement of antitumor immunity by CTLA-4 blockade

Impact of tumour microenvironment and Fc receptors on the activity of immunomodulatory antibodies

Remarkably similar CTLA-4 binding properties of therapeutic ipilimumab and tremelimumab antibodies

Immune checkpoint protein inhibition for cancer: preclinical justification for CTLA-4 and PD-1 blockade and new combinations

Genetic basis for clinical response to CTLA-4 blockade in melanoma

Tumor exome analysis reveals neoantigen-specific T-cell reactivity in an ipilimumabresponsive melanoma

TLR1/2 ligand enhances antitumor efficacy of CTLA-4 blockade by increasing intratumoral T reg depletion

Blockade of CTLA-4 on both effector and regulatory T cell compartments contributes to the antitumor activity of anti-CTLA-4 antibodies

Differential control of human T reg and effector T cells in tumor immunity by Fc-engineered anti-CTLA-4 antibody

Anti-CTLA-4 immunotherapy does not deplete FOXP3 + regulatory T cells (T regs ) in human cancers

Induced expression of PD-1, a novel member of the immunoglobulin gene superfamily, upon programmed cell death

Using subtractive hybridization, this study is the first to identify PD1 as a new immunoglobulin superfamily member expressed within the thymus

The B7 family of ligands and its receptors: new pathways for costimulation and inhibition of immune responses

PD-L2 is a second ligand for PD-1 and inhibits T cell activation

The PD-1 pathway in tolerance and autoimmunity

Engagement of the PD-1 immunoinhibitory receptor by a novel B7 family member leads to negative regulation of lymphocyte activation

Immunological studies on PD-1 deficient mice: implication of PD-1 as a negative regulator for B cell responses

Development of lupus-like autoimmune diseases by disruption of the PD-1 gene encoding an ITIM motif-carrying immunoreceptor

Autoimmune dilated cardiomyopathy in PD-1 receptor-deficient mice

Establishment of NOD-Pdcd1 -/-mice as an efficient animal model of type I diabetes

PD-1 and its ligands in tolerance and immunity

This study highlights the central role of CD28 dephosphorylation in mediating the negative regulatory effect of PD1-mediated T cell inhibition

This study demonstrates the complementary ways in which CTLA4 and PD1 inhibit protein kinase B signalling to reduce glucose uptake and utilization

Distinct cellular mechanisms underlie anti-CTLA-4 and anti-PD-1 checkpoint blockade

Restoring function in exhausted CD8 T cells during chronic viral infection

PD-L1/B7H-1 inhibits the effector phase of tumor rejection by T cell receptor (TCR) transgenic CD8 + T cells

PD-L1 regulates the development, maintenance, and function of induced regulatory T cells

Molecular and cellular insights into T cell exhaustion

This preclinical murine study shows that blockade of PD1 or PDL1 could reverse inherent resistance of tumours

Involvement of PD-L1 on tumor cells in the escape from host immune system and tumor immunotherapy by PD-L1 blockade

B7-H1 blockade augments adoptive T-cell immunotherapy for squamous cell carcinoma

Blocking programmed death-1 ligand-PD-1 interactions by local gene therapy results in enhancement of antitumor effect of secondary lymphoid tissue chemokine

Rescue of exhausted CD8 T cells by PD-1-targeted therapies is CD28-dependent

PD-1 blockade inhibits hematogenous spread of poorly immunogenic tumor cells by enhanced recruitment of effector T cells

Programmed cell death 1 ligand 1 and tumor-infiltrating CD8 + T lymphocytes are prognostic factors of human ovarian cancer

PD-1 is expressed by tumor-infiltrating immune cells and is associated with poor outcome for patients with renal cell carcinoma

Tumor B7-H1 is associated with poor prognosis in renal cell carcinoma patients with long-term follow-up

this study shows that expression of PD1 on tumour-infiltrating T cells and PD1 ligands on tumours are poor prognostic markers in various cancers

Immune checkpoint blockade therapy for cancer: an overview of FDA-approved immune checkpoint inhibitors

The development of immunomodulatory monoclonal antibodies as a new therapeutic modality for cancer: the Bristol-Myers Squibb experience

Phase I study of single-agent anti-programmed death-1 (MDX-1106) in refractory solid tumors: safety, clinical activity, pharmacodynamics, and immunologic correlates

Development of PD-1 and PD-L1 inhibitors as a form of cancer immunotherapy: a comprehensive review of registration trials and future considerations

Nivolumab versus chemotherapy in patients with advanced melanoma who progressed after anti-CTLA-4 treatment (CheckMate 037): a randomised, controlled, open-label, phase 3 trial

Adjuvant nivolumab versus ipilimumab in resected stage III or IV melanoma

Nivolumab in previously untreated melanoma without BRAF mutation

This phase III clinical study in advanced melanoma patients shows that PD1 blockade was more efficacious and

Pembrolizumab versus ipilimumab for advanced melanoma: final overall survival results of a multicentre, randomised, open-label phase 3 study (KEYNOTE-006)

Pembrolizumab versus docetaxel for previously treated, PD-L1-positive, advanced non-small-cell lung cancer (KEYNOTE-010): a randomised controlled trial

Pembrolizumab versus chemotherapy for PD-L1-positive non-small-cell lung cancer

Pembrolizumab for the treatment of non-small-cell lung cancer

Pembrolizumab versus methotrexate, docetaxel, or cetuximab for recurrent or metastatic head-and-neck squamous cell carcinoma (KEYNOTE-040): a randomised, open-label, phase 3 study

Pembrolizumab in relapsed/ refractory classical Hodgkin lymphoma: primary end point analysis of the phase 2 KEYNOTE-087 study

Pembrolizumab as second-line therapy for advanced urothelial carcinoma

Safety and efficacy of pembrolizumab monotherapy in patients with previously treated advanced gastric and gastroesophageal junction cancer: phase 2 clinical KEYNOTE-059 trial

Significance and implications of FDA approval of pembrolizumab for biomarker-defined disease

Predicting tumor response to PD-1 blockade

Cancer drugs approved based on biomarkers and not tumor type-FDA approval of pembrolizumab for mismatch repair-deficient solid cancers

Nivolumab versus everolimus in advanced renal-cell carcinoma

Nivolumab for recurrent squamouscell carcinoma of the head and neck

Nivolumab in metastatic urothelial carcinoma after platinum therapy (CheckMate 275): a multicentre, single-arm, phase 2 trial

Nivolumab in patients with advanced hepatocellular carcinoma (CheckMate 040): an open-label, non-comparative, phase 1/2 dose escalation and expansion trial

PD-1 blockade with nivolumab in relapsed or refractory Hodgkin's lymphoma

Nivolumab in patients with metastatic DNA mismatch repair-deficient or microsatellite instability-high colorectal cancer (CheckMate 142): an open-label, multicentre, phase 2 study

Five-year survival and correlates among patients with advanced melanoma, renal cell carcinoma, or non-small cell lung cancer treated with nivolumab

Tumor-associated B7-H1 promotes T-cell apoptosis: a potential mechanism of immune evasion

Atezolizumab in patients with locally advanced and metastatic urothelial carcinoma who have progressed following treatment with platinum-based chemotherapy: a single-arm, multicentre, phase 2 trial

Atezolizumab versus chemotherapy in patients with platinum-treated locally advanced or metastatic urothelial carcinoma (IMvigor211): a multicentre, open-label, phase 3 randomised controlled trial

Atezolizumab versus docetaxel in patients with previously treated non-small-cell lung cancer (OAK): a phase 3, open-label, multicentre randomised controlled trial

Atezolizumab and nab-paclitaxel in advanced triple-negative breast cancer

First-line atezolizumab plus chemotherapy in extensive-stage small-cell lung cancer

Avelumab in patients with chemotherapy-refractory metastatic Merkel cell carcinoma: a multicentre, single-group, open-label, phase 2 trial

Avelumab in metastatic urothelial carcinoma after platinum failure (JAVELIN Solid Tumor): pooled results from two expansion cohorts of an open-label, phase 1 trial

Avelumab plus axitinib versus sunitinib for advanced renal-cell carcinoma

Efficacy and safety of durvalumab in locally advanced or metastatic urothelial carcinoma: updated results from a phase 1/2 open-label study

Durvalumab after chemoradiotherapy in stage III non-small-cell lung cancer

Development of immune checkpoint therapy for cancer

Immune-related adverse events with immune checkpoint blockade: a comprehensive review

This comprehensive review describes the diagnosis of the common immune-related adverse events observed in patients treated with immune checkpoint inhibitors and the subsequent clinical management of these side effects

Immune-related adverse events associated with immune checkpoint blockade

Immune-related adverse events associated with anti-PD-1/PD-L1 treatment for malignancies: a meta-analysis

Hyperprogressive disease is a new pattern of progression in cancer patients treated by anti-PD-1/PD-L1

Hyperprogression during anti-PD-1/PD-L1 therapy in patients with recurrent and/or metastatic head and neck squamous cell carcinoma

Hyperprogressive disease in patients with advanced non-small cell lung cancer treated with PD-1/PD-L1 inhibitors or with single-agent chemotherapy

Rapid progression of adult T-cell leukemia/lymphoma as tumor-infiltrating T regs after PD-1 blockade

Patterns of responses in metastatic NSCLC during PD-1 or PDL-1 inhibitor therapy: comparison of RECIST 1.1, irRECIST and iRECIST criteria

Hyperprogressive disease: recognizing a novel pattern to improve patient management

Improved mouse models to assess tumour immunity and irAEs after combination cancer immunotherapies

A reappraisal of CTLA-4 checkpoint blockade in cancer immunotherapy

Safety and tolerability of PD-1/PD-L1 inhibitors compared with chemotherapy in patients with advanced cancer: a meta-analysis

Hijacking antibody-induced CTLA-4 lysosomal degradation for safer and more effective cancer immunotherapy

Preserving the CTLA-4 checkpoint for safer and more effective cancer immunotherapy

Delivery technologies for cancer immunotherapy

Treatment of the immune-related adverse effects of immune checkpoint inhibitors: a review

Gene expression profiling of whole blood in ipilimumab-treated patients for identification of potential biomarkers of immune-related gastrointestinal adverse events

Evaluation of serum IL-17 levels during ipilimumab therapy: correlation with colitis

Correlation of absolute and relative eosinophil counts with immune-related adverse events in melanoma patients treated with ipilimumab

This clinical study is the first to demonstrate the viability of using patient-derived leukocytes and autologous tumour cells to promote cancer regression

Antileukemic effect of graft-versushost disease in human recipients of allogeneic-marrow grafts

Use of tumor-infiltrating lymphocytes and interleukin-2 in the immunotherapy of patients with metastatic melanoma. A preliminary report

Treatment of patients with metastatic melanoma with autologous tumor-infiltrating lymphocytes and interleukin 2

This study expands on the clinical success of IL-2-expanded tumour-infiltrating lymphocytes in cancer treatment through the addition of lymphodepletion before administration of the cellular therapy

Immune recognition of somatic mutations leading to complete durable regression in metastatic breast cancer

Cish actively silences TCR signaling in CD8 + T cells to maintain tumor tolerance

The urgent need to recover MHC class I in cancers for effective immunotherapy

Expression of chimeric receptor composed of immunoglobulin-derived V regions and T-cell receptor-derived C regions

Chimeric Fv-ζ or Fv-ε receptors are not sufficient to induce activation or cytokine production in peripheral T cells

Antitransgene rejection responses contribute to attenuated persistence of adoptively transferred CD20/CD19-specific chimeric antigen receptor redirected T cells in humans

Human T-lymphocyte cytotoxicity and proliferation directed by a single chimeric TCRζ/CD28 receptor

CD40 ligand-modified chimeric antigen receptor T cells enhance antitumor function by eliciting an endogenous antitumor response

Activation of resting human primary T cells with chimeric receptors: costimulation from CD28, inducible costimulator, CD134, and CD137 in series with signals from the TCR ζ chain

Chimeric receptors with 4-1BB signaling capacity provoke potent cytotoxicity against acute lymphoblastic leukemia

Targeted delivery of a PD-1-blocking scFv by CAR-T cells enhances anti-tumor efficacy in vivo

Targeting a CAR to the TRAC locus with CRISPR/Cas9 enhances tumour rejection

This pilot clinical study demonstrates the feasibility and preliminary efficacy of CD19-directed CAR T cells in chronic lymphocytic leukaemia

CD19-targeted T cells rapidly induce molecular remissions in adults with chemotherapy-refractory acute lymphoblastic leukemia

Long-term follow-up of CD19 CAR therapy in acute lymphoblastic leukemia

Axicabtagene ciloleucel CAR T-cell therapy in refractory large B-cell lymphoma

CD22-targeted CAR T cells induce remission in B-ALL that is naive or resistant to CD19-targeted CAR immunotherapy

Driving CAR T-cells forward

Developing neoantigen-targeted T cell-based treatments for solid tumors

Anti-BCMA CAR T-cell therapy bb2121 in relapsed or refractory multiple myeloma

B cell maturation antigen-specific CAR T cells are clinically active in multiple myeloma

GPRC5D is a target for the immunotherapy of multiple myeloma with rationally designed CAR T cells

Mesothelin-specific chimeric antigen receptor mRNA-engineered T cells induce anti-tumor activity in solid malignancies

A phase I study on adoptive immunotherapy using gene-modified T cells for ovarian cancer

Adoptive transfer of chimeric antigen receptor re-directed cytolytic T lymphocyte clones in patients with neuroblastoma

CAR T cells targeting B7-H3, a pan-cancer antigen, demonstrate potent preclinical activity against pediatric solid tumors and brain tumors

Inhibition of interleukin-10 in the tumor microenvironment can restore mesothelin chimeric antigen receptor T cell activity in pancreatic cancer in vitro

Rewiring T-cell responses to soluble factors with chimeric antigen receptors

Differential effects of IL-12 on T regs and non-T reg T cells: roles of IFN-γ, IL-2 and IL-2R

IL-12 release by engineered T cells expressing chimeric antigen receptors can effectively muster an antigen-independent macrophage response on tumor cells that have shut down tumor antigen expression

IL-12 triggers a programmatic change in dysfunctional myeloid-derived cells within mouse tumors

IL-12 secreting tumor-targeted chimeric antigen receptor T cells eradicate ovarian tumors in vivo

Armored CAR T cells enhance antitumor efficacy and overcome the tumor microenvironment

A phase I clinical trial of adoptive T cell therapy using IL-12 secreting MUC-16 ecto directed chimeric antigen receptors for recurrent ovarian cancer

Selective expansion of chimeric antigen receptor-targeted T-cells with potent effector function using interleukin-4

Design of a phase I clinical trial to evaluate intratumoral delivery of ErbB-targeted chimeric antigen receptor T-cells in locally advanced or recurrent head and neck cancer

c-Jun overexpression in CAR T cells induces exhaustion resistance

This review details the adverse effects associated with CAR T cell therapy and their subsequent clinical management

Chimeric antigen receptor T-cell therapy-assessment and management of toxicities

Cytokine release syndrome

Chimeric antigen receptor-modified T cells for acute lymphoid leukemia

Primary B cell immunodeficiencies: comparisons and contrasts

CAR T cell-induced cytokine release syndrome is mediated by macrophages and abated by IL-1 blockade

Enhanced CAR T cell expansion and prolonged persistence in pediatric patients with ALL treated with a low-affinity CD19 CAR

Trivalent CAR T cells overcome interpatient antigenic variability in glioblastoma

Development of anti-human mesothelin-targeted chimeric antigen receptor messenger RNA-transfected peripheral blood lymphocytes for ovarian cancer therapy

Improving the safety of cell therapy products by suicide gene transfer

Total costs of chimeric antigen receptor T-cell immunotherapy

The era of checkpoint blockade in lung cancer: taking the brakes off the immune system

A guide to manufacturing CAR T cell therapies

Therapeutic cancer vaccines: past, present, and future

Intracavitary Bacillus Calmette-Guerin in the treatment of superficial bladder tumors

This early study identifies a melanoma neoantigen that could elicit a specific cytolytic T cell response

Immunotherapy of established micrometastases with Bacillus Calmette-Guerin tumor cell vaccine

Prevention of metastatic spread by postoperative immunotherapy with virally modified autologous tumor cells. I. Parameters for optimal therapeutic effects

Effective anti-metastatic melanoma vaccination with tumor cells transfected with MHC genes and/or infected with Newcastle disease virus (NDV)

Oncolytic Newcastle disease virus activation of the innate immune response and priming of antitumor adaptive responses in vitro

Tumor vaccines expressing flt3 ligand synergize with ctla-4 blockade to reject preimplanted tumors

Cancer immunology. Mutational landscape determines sensitivity to PD-1 blockade in non-small cell lung cancer

Immunogenicity of somatic mutations in human gastrointestinal cancers

By using next-generation DNA sequencing, this study identifies somatic mutations that create novel immunogenic epitopes that can be successfully targeted by therapeutic vaccination

Mutant MHC class II epitopes drive therapeutic immune responses to cancer

Genomic and bioinformatic profiling of mutational neoepitopes reveals new rules to predict anticancer immunogenicity

Checkpoint blockade cancer immunotherapy targets tumour-specific mutant antigens

Cancer exome analysis reveals a T-cell-dependent mechanism of cancer immunoediting

Predicting immunogenic tumour mutations by combining mass spectrometry and exome sequencing

MHC-II neoantigens shape tumour immunity and response to immunotherapy

Personalized vaccines for cancer immunotherapy

Personalized cancer vaccines: targeting the cancer mutanome

An immunogenic personal neoantigen vaccine for patients with melanoma

Personalized RNA mutanome vaccines mobilize poly-specific therapeutic immunity against cancer

MHC class II epitope predictive algorithms

Nivolumab plus ipilimumab versus sunitinib in advanced renal-cell carcinoma

This long-term analysis of a previous phase III clinical trial demonstrates that nivolumab and ipilimumab combination therapy conferred increased survival compared with ipilimumab alone

PD-L1:CD80 cis-heterodimer triggers the co-stimulatory receptor CD28 while repressing the inhibitory PD-1 and CTLA-4 pathways

Programmed death-1 ligand 1 interacts specifically with the B7-1 costimulatory molecule to inhibit T cell responses

Using immunotherapy to boost the abscopal effect

Radiation and dual checkpoint blockade activate non-redundant immune mechanisms in cancer

Immunologic correlates of the abscopal effect in a patient with melanoma

Emerging targets in cancer immunotherapy

Next generation immune-checkpoints for cancer therapy

Next generation of immune checkpoint therapy in cancer: new developments and challenges

Lymphocyte-activation gene-3, an important immune checkpoint in cancer

LAG-3 confers poor prognosis and its blockade reshapes antitumor response in head and neck squamous cell carcinoma

The future of immune checkpoint therapy

First-line chemoimmunotherapy in metastatic breast carcinoma: combination of paclitaxel and IMP321 (LAG-3Ig) enhances immune responses and antitumor activity

Fibrinogen-like protein 1 is a major immune inhibitory ligand of LAG-3

The Tim-3 ligand galectin-9 negatively regulates T helper type 1 immunity

Lag-3, Tim-3, and TIGIT: co-inhibitory receptors with specialized functions in immune regulation

Immune-checkpoint proteins VISTA and PD-1 nonredundantly regulate murine T-cell responses

VISTA is an inhibitory immune checkpoint that is increased after ipilimumab therapy in patients with prostate cancer

Interim results of an ongoing phase I, dose escalation study of MGA271 (Fc-optimized humanized anti-B7-H3 monoclonal antibody) in patients with refractory B7-H3-expressing neoplasms or neoplasms whose vasculature expresses

Compartmental intrathecal radioimmunotherapy: results for treatment for metastatic CNS neuroblastoma

Phase 1, first-in-human, open label, dose escalation ctudy of MGD009, a humanized B7-H3 × CD3 dual-affinity re-targeting (DART) protein in patients with B7-H3-expressing neoplasms or B7-H3 expressing tumor vasculature

TIGIT and PD-1 impair tumor antigen-specific CD8 + T cells in melanoma patients

CD28 costimulation provided through a CD19-specific chimeric antigen receptor enhances in vivo persistence and antitumor efficacy of adoptively transferred T cells

The CD28-related molecule ICOS is required for effective T cell-dependent immune responses

Engagement of the ICOS pathway markedly enhances efficacy of CTLA-4 blockade in cancer immunotherapy

Tremelimumab in combination with exemestane in patients with advanced breast cancer and treatment-associated modulation of inducible costimulator expression on patient T cells

Therapeutic efficacy of OX-40 receptor antibody depends on tumor immunogenicity and anatomic site of tumor growth

Engagement of the OX-40 receptor in vivo enhances antitumor immunity

Triggering of OX40 (CD134) on CD4 + CD25 + T cells blocks their inhibitory activity: a novel regulatory role for OX40 and its comparison with GITR

Thymus-dependent areas in the lymphoid organs of neonatally thymectomized mice

Cell to cell interaction in the immune response. I. Hemolysin-forming cells in neonatally thymectomized mice reconstituted with thymus or thoracic duct lymphocytes

lymphocytes: how they develop and function

The full spectrum of human naive T cells

How regulatory T cells work

Human T cell development, localization, and function throughout life

The many important facets of T-cell repertoire diversity

T-cell receptor structure and TCR complexes

Clonal expansion versus functional clonal inactivation: a costimulatory signalling pathway determines the outcome of T cell antigen receptor occupancy

A 44 kilodalton cell surface homodimer regulates interleukin 2 production by activated human T lymphocytes

A 45-kDa human T-cell membrane glycoprotein functions in the regulation of cell proliferative responses

Role of the CD28 receptor in T-cell activation

3 antigen). A cell surface molecule with a function in human T cell activation

Monoclonal antibodies identifying a novel T-cell antigen and Ia antigens of human lymphocytes

Human T cell activation. II. A new activation pathway used by a major T cell population via a disulfide-bonded dimer of a 44 kilodalton polypeptide

This study provides evidence that the recently discovered CD28 receptor functioned to augment T cell activation

The effects of TLR activation on T-cell development and differentiation

The B7-CD28 superfamily

CTLA-4 and PD-1 pathways: similarities, differences, and implications of their inhibition

T-cell antigen receptor signal transduction

Function of the Src-family kinases, Lck and Fyn, in T-cell development and activation

The CD45 tyrosine phosphatase: a positive and negative regulator of immune cell function

How does the kinase Lck phosphorylate the T cell receptor? Spatial organization as a regulatory mechanism

Transcriptional mechanisms underlying lymphocyte tolerance

Peripheral T cell tolerance

Mature T lymphocyte apoptosis-immune regulation in a dynamic and unpredictable antigenic environment

Restimulation-induced cell death: new medical and research perspectives

Memory T cell subsets, migration patterns, and tissue residence

Defining T cell states associated with response to checkpoint immunotherapy in melanoma

Single-cell map of diverse immune phenotypes in the breast tumor microenvironment

Epigenetic stability of exhausted T cells limits durability of reinvigoration by PD-1 blockade

Distinct immune cell populations define response to anti-PD-1 monotherapy and anti-PD-1/anti-CTLA-4 combined therapy

as a DPhil co-mentor. They also thank Y. Zhang for invaluable editorial and scientific feedback. Lastly, the authors acknowledge and apologize to all researchers in this field who may have authored elegant studies that were not cited owing to space limitations.

The authors contributed equally to all aspects of the article.

The authors declare no competing interests.

Nature Reviews Immunology thanks J. Oliaro and the other, anonymous, reviewer(s) for their contribution to the peer review of this work.

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.This is a U.S. government work and not under copyright protection in the U.S.; foreign copyright protection may apply 2020 www.nature.com/nri