Basic mechanisms of Immunotherapy
Making it in a simple Lenguage
Dr. Ada G. Blidner Ins tute of Biology and Experimental Medicine, Buenos Aires, Argen na
[email protected] Who, Where, When? Innate and Adaptive Immune Response
Fast Slow
Recognizes Recognizes Pathogen specific associated an gens paterns Very efficient in Capable of elimina ng elimina ng infec ons infec ons
In the Lymph Ready to nodes, act, in the needs IIR ssues ac va on to
migrate to Interacts ssues and with AIR Dranoff, Nature Reviews, 2004 act Antigen presentation and Priming
Koichi, Nature Reviews, 2012 Huppa, Nature Reviews, 2003 Integrating the Immune Response What are checkpoints good for?
PD-1 PD-1 CD4 CTLA-4 Treg CTLA-4 BTLA LAG3 BTLA LAG3
PD-1
CD8 CTLA-4 HVEM BTLA LAG3 HVEM PD-L1
B VISTA PD-L1
BTLA LAG3 Immune homeostasis CTLA-4
Activation Sustained Activation Immunosuppression
CD4 CD4
Treg
CD8 CD8
CTLA-4 migration CTLA-4 Expression Antigen presentation inhibition Ligand Competition Deactivation of proliferation signals PD-1
PD1 CD4 PD-1 is expressed
PD1 following T cel activation
CD8 PD-L1 is expressed in
PD1 lymphoid and non-lymphoid B PD-L1 tissues. PD-L2 mainly in PD-L2 APCs PD1
Treg PD-L1 PD-L1 is expressed on Endothelial cells, which PD1 inhibit T cell activity in SLO PD-L1
PD-L2 PD-1 is expressed in PD1 Exhausted T cells TFh PD-1
Exhausted T Cell
PD-1/PD-L1 interaction inhibits TCR mediated T cell activation and Proliferation T cell ON/OFF
Co-stimulatory molecules Co-inhibitory molecules
Activation Anergy
Inflammation Homeostasis
Autoimmunity Cancer How do Tumors take advantage of IC?
I.S. wins
The I.S. selects resistant clones
Tumor wins How do Tumors take advantage of IC?
EQUILIBRIUM PHASE PD-1 LAG 3 CD4 PD-1 CTLA-4 BTLA Treg CD8 CTLA-4 PD-1 CTLA-4 PD-L1
PD-L2 MDSCs GM-CSF IL-1B IL-12 PD-L1 IFN-γ PD-L2
IDO
HVEM CTLA-4 blockade broadens T cell repertoire
Robert et al. Page 18 NIH-PA Author Manuscript NIH-PA Author Manuscript
Figure 5. Changes in number of unique TCR V-beta CDR3 sequences related to toxicity during the first 3 months of therapy with tremelimumab. NIH-PA Author Manuscript
Clin Cancer Res. Author manuscript; available in PMC 2015 May 01. Please cite this article in press as: Wei et al., Distinct Cellular Mechanisms Underlie Anti-CTLA-4 and Anti-PD-1 Checkpoint Blockade, Cell (2017), http://dx.doi.org/10.1016/j.cell.2017.07.024
A B
C
Please cite this article in press as: Wei et al., Distinct Cellular Mechanisms Underlie Anti-CTLA-4 and Anti-PD-1 Checkpoint Blockade, Cell (2017), http://dx.doi.org/10.1016/j.cell.2017.07.024
Please cite this article in press as: Wei et al., Distinct Cellular Mechanisms Underlie Anti-CTLA-4 and Anti-PD-1 Checkpoint Blockade, Cell (2017), http://dx.doi.org/10.1016/j.cell.2017.07.024 ABC
Differences between CTLA-4 and PD-1 blockade
ABCD F D
E
Fife & Bluestone T-cell tolerance mediated by CTLA-4 and PD-1 E Á D F Peripheral Tolerance Tissue SpecificG Inhibition Pancreatic Pancreas Lymph Node
Tissue Specific Inhibition
T cell Activation T cell T cell Treg
Activated Immune E T cell Trafficking
Islets (β cells)
DC DC
CTLA-4 - B7-1/B7-2 PD-1 - PD-L1 (legend on next page)
PD-1 Inhibition
CTLA-4 Inhibition B7-1/B7-2 8 Cell 170, 1–14, September 7, 2017 PD-L1/PD-L2 Ligand Concentration Fig. 1. Peripheral tolerance mediated by negative regulators. This figure represents the differential effects for CTLA-4 and PD-1 for limiting T-cell G function during T-cell activation and the maintenance of peripheral tolerance. CTLA-4 and PD-1 are rapidly expressed following T-cell receptor engagement and CD28 co-stimulatory signals. B7-1 and B7-2 are constitutively expressed in the lymph nodes at low levels on antigen-presenting cells and upon activation are highly upregulated. PD-L1 and PD-L2 are constitutively expressed in peripheral tissue sites and upregulated in response to inflammatory signals. The temporal and spatial patterns for ligand expression may help to explain the distinct differences between these two inhibitory molecules. CTLA-4 blockade prevents tolerance induction and therefore may act in the earliest stages of immunity, particularly those within the lymphatic system. PD-1 blockade acts in tissue sites to release anergic T cells and cause autoimmunity,suggesting the inhibitory functions are within the target tissue. Therefore, PD-1 and CTLA-4 can play critical yet non-redundant roles in the maintenance of peripheral tolerance.
administer these types of reagents as therapeutics during balance of positive and negative signals establishing mechan- autoimmune disease progression. isms for T-cell expansion and control. CTLA-4 is important for limiting T-cell activity for cells early during the immune Concluding remarks responses. PD-1, on the other hand, limits T-cell activity in PD-1 and CTLA-4 are critical regulators of T-cell function in peripheral tissues and may be the last chance to prevent T-cell diverse processes ranging from autoimmunity, transplantation destruction of self-tissue resulting in autoimmunity (Fig. 1). tolerance, chronic viral infections, and tumor immunity. In vitro PD-1 and PD-L1 are also expressed on Tregs and may help and in vivo animal models as well as clinical trials have control their suppression of effector T cells. Recent studies demonstrated great therapeutic opportunities by effectively indicate that PD-L1 and PD-L2 can signal bi-directionally by targeting these pathways. T-cell activation is a complicated engaging PD-1 on T cells. PD-L1/PD-L2 signaling may aid in biological process with several checkpoints and fail-safe pro- protection from apoptosis and promote tumor cell growth. The cesses built in to limit aberrant or unfaithful T-cell activation, local environment of cytokines and costimulatory molecule including autoimmunity. T-cell signaling through CD28, expression on professional APCs will be vital for dictating T- CTLA-4, and PD-1 controls complex but critical checkpoints helper outcomes. These pathways offer exciting new therapeu- (legend on next page) for T-cell activation and homeostasis. CD28 signaling induces a tic possibilities to selectively silence destructive T cells or
2 Cell 170178, 1–14, September 7,r 2008 2017 The Authors Journal compilation r 2008 Blackwell Munksgaard Immunological Reviews 224/2008
(legend on next page)
2 Cell 170, 1–14, September 7, 2017 Please cite this article in press as: Thorsson et al., The Immune Landscape of Cancer, Immunity (2018), https://doi.org/10.1016/j.immuni.2018.03.023 Please cite this article in press as: Thorsson et al., The Immune Landscape of Cancer, Immunity (2018), https://doi.org/10.1016/j.immuni.2018.03.023
Please cite this article in press as: Thorsson et al., The Immune Landscape of Cancer, Immunity (2018), https://doi.org/10.1016/j.immuni.2018.03.023 Th1/Th17 responses correlate with good prognosis
Figure 3. Immune Response and Prognostics AB Figure 3. Immune Response and Prognostics AB (A) Overall survival (OS) by immune subtype. (A) Overall survival (OS) by immune subtype. (B) Concordance index (CI) for five characteristic (B) Concordance index (CI) for five characteristic AB Figureimmuneimmune 3. Immune expression expression Response signature and signature Prognostics scores scores (Figure ( 1FigureA) in 1A) in (A) Overallrelation survival to (OS) OS, by immune for immune subtype. subtypes and TCGA tu- (B)relation Concordance to OS, index for (CI) immune for five subtypes characteristic and TCGA tu- immunemormor types. expression types. Red signature denotes Red denotes scores higher (Figure higher and 1A) blue in and lower blue risk, lower risk, relationwithwith an to OS, increase an for increase immune in the subtypes in signature the and signature TCGA score. tu- score. mor(C) types. CI(C) for Red CI T for helper denotes T helper cellhigher scores and cell blue scores in lower relation risk, in relation to OS within to OS within with an increase in the signature score. (C)immune CI forimmune T helper subtypes. cell subtypes. scores in relation to OS within immune(D) Risk(D) subtypes. Risk stratification stratification from elastic from elastic net modeling net modeling of of (D)immune Risk stratification features. from Tumor elastic samples net modeling were of divided into immuneimmune features. Tumor features. samples Tumor were divided samples into were divided into discoverydiscoverydiscovery and and validation validation and sets, validation and sets, an elastic and sets, net an and elastic an elastic net net modelmodelmodel was was optimized was optimized on optimized the discovery on the on set discovery the using discovery set using set using CDimmuneimmune gene gene signatures, signatures, TCR/BCR TCR/BCR richness, and richness, and CDneoantigenimmune counts. gene Kaplan-Meier signatures, plot shows TCR/BCR the richness, and CDneoantigen counts. Kaplan-Meier plot shows the high (red)neoantigen and low (blue) riskcounts. groups from Kaplan-Meier this model plot shows the ashigh applied (red) to andthe validation low (blue) set, prisk < 0.0001 groups (G-rho from this model high (red) and low (blue) risk groups from this model familyas applied of tests, Harrington to the validation and Fleming). set, p < 0.0001 (G-rho (E) Predictionas applied versus outcome to the from validation elastic net model set, p < 0.0001 (G-rho infamily validation of set tests, data (from Harrington D). Top: Patient and outcomesFleming). for(E) each Predictionfamily sample of (black, versus tests, survival; outcomeHarrington red, death) from and plotted elastic Fleming). net model with vertical(E) Prediction jitter, alongthe versus sample’s outcome model pre- from elastic net model The increase in Th17 correlates in validation set data (from D). Top: Patient outcomes dictionin (x validation axis). Middle: set Fractional data (from density D). of Top: the Patient outcomes with a better progonosis outcomesfor each plotted sample against (black, their model survival; predictions. red, death) plotted Confidencewithfor vertical each intervals jitter, sample were along generated (black, the sample’s survival; by boot- red, model death) pre- plotted strappingdictionwith with (x vertical replacement.axis). Middle: jitter, Bottom: along Fractional LOESS the fit sample’s ofdensity of model the pre- the actual outcomes against the model predictions; outcomes plotted against their model predictions. narrowdiction confidence (x bands axis). confirm Middle: good prediction Fractional density of the EFaccuracy.Confidenceoutcomes intervals plotted were against generated their model by boot- predictions. (F)strapping CoxPHConfidence models with of stage replacement. intervals and tumor type were Bottom: (‘‘Tissue’’) generated LOESS fit by of boot- withthe (full actual model) outcomes or without against (reduced the model) model the predictions; validationstrapping set predictions with of replacement. the elastic net model Bottom: LOESS fit of werenarrow compared;the confidence actual the outcomes full model bands significantly confirmagainst out- goodthe model prediction predictions; EFperformedaccuracy. the reduced model in all comparisons narrow confidence bands confirm good prediction (p(F) < 0.001; CoxPH false models discovery of rate stage (FDR) and BH-corrected). tumor type (‘‘Tissue’’) EFSee alsoaccuracy.Figure S3. with (full model) or without (reduced model) the (F) CoxPH models of stage and tumor type (‘‘Tissue’’) validation set predictions of the elastic net model werewith compared; (full model) the full or model without significantly (reduced out-model) the (Figureperformedvalidation 3F) improved the setreduced predictive predictions model accuracy, in of all the comparisons elastic net model highlighting(p < 0.001;were the falsecompared; importance discovery the of rate the full (FDR) immune model BH-corrected). significantly out- TME in determining survival. Lymphocyte Seeperformed also Figure S3 the. reduced model in all comparisons expression signature, high number of unique(p TCR < 0.001; clonotypes, false discovery cytokines rate made (FDR) BH-corrected). by activatedSee also Th1Figure and Th17 S3. cells, and M1 reflecting a balanced immune response. While increased Th17 macrophages most strongly associated with improved OS (Fig- cells generally led to improved OS, Th1 associated with worse ure S3E), while wound(Figure healing, 3F) macrophage improved regulation, predictive and accuracy, OS across most immune subtypes, and Th2 orientation had TGF-b associated with worse OS, recapitulating survival associ- mixed effects (Figure 3C). Tumor types displayed two behaviors ations in immune subtypes.highlighting Within tumor the types, importance the prognostic of the immune relative to immune orientation (Figures 3B, OS; S3B, PFI). In the implications of immuneTME subtypes(Figure in determining seen 3F) in improved univariate survival. analyses predictive Lymphocyte accuracy, first group including SKCM and CESC, activation of immune were largely maintained,expression with C3 correlating signature, with better high OS in number of pathways was generally associated with better outcome, while six tumor types and C4 withhighlighting poor OS in thethree importance cancer types of the immune in the other, the opposite was seen. The relative abundance of (Figure S3F). uniqueTME TCR in determining clonotypes, survival.cytokines Lymphocyte made individual immune cell types had complex associations that by activatedexpression Th1 signature, and Th17 cells, high and number M1 of reflecting a balanceddiffered immune between response. tumor types While (Figures increased S3C and S3D). Th17 These an-macrophagesImmune Response most strongly Correlates associated of Somatic Variation with improved OS (Fig- alyses extend beyond mere determination of lymphocyte pres- The immune infiltrate was relatedunique to measures TCR clonotypes, of DNA damage, cytokines made cells generally led toence improved to suggest OS, testable Th1 properties associated that correlate with worse with patientureincluding S3E), while copy number wound variationby healing, activated (CNV) burdenmacrophage Th1 (both and in terms regulation, Th17 of cells, and and M1 OS across most immuneoutcome in subtypes, different tumor and types Th2 and orientation immune contexts. had TGF-numberassociated of segments with and worse fraction OS, of recapitulating genome alterations), survival associ- reflecting a balanced immune response. While increased Th17 macrophagesb most strongly associated with improved OS (Fig- mixed effects (Figure 3WeC). obtained Tumor and types validated displayed a survival two model behaviors using elastic-netationsaneuploidy, in immune loss of subtypes. heterozygosity Within (LOH), tumor homologous types, recombi- the prognostic cells generally led toCox improved proportional OS, hazards Th1 (CoxPH) associated modeling with with cross-valida- worse urenation S3 deficiencyE), while (HRD), wound and intratumor healing, heterogeneity macrophage (ITH) (Fig- regulation, and relative to immune orientation (Figures 3B, OS; S3B, PFI). In the implications of immune subtypes seen in univariate analyses OS across most immunetion. Low- subtypes, and high-score and tumors Th2 displayed orientation significant had survival TGF-ure 4A).associated LF correlated negatively with worse with CNV OS, segment recapitulating burden, with survival associ- first group includingdifferences SKCM in and the validation CESC, set activation (Figure 3D), of with immune good predictionwerestrongest largelyb correlation maintained, in C6 with and C3 C2, correlating and positively with with better OS in mixedpathways effects was (Figure generallyaccuracy 3C). associated Tumor (Figure types 3E). with Incorporating displayed better outcome, twoimmune behaviors features while intosixations tumoraneuploidy, in types immune LOH, and HRD, subtypes. C4 and with mutation poor Within load, OS particularly tumor in three types, in C3. cancer the types prognostic Cox models fit with tumor type, stage, and tumor type + stage These results suggest a differential effect of multiple smaller, relativein the to other, immune the opposite orientation was (Figures seen. The 3B, relative OS; S3 abundanceB, PFI). In of the (Figureimplications S3F). of immune subtypes seen in univariate analyses firstindividual group including immune cell6 SKCMImmunity types and48, had 1–19, CESC, complex April 17, activation 2018 associations of immune that were largely maintained, with C3 correlating with better OS in pathwaysdiffered between was generally tumor types associated (Figures with S3C better and S3D). outcome, These while an- Immunesix tumor Response types and Correlates C4 with of poor Somatic OS in Variation three cancer types inalyses the other, extend the beyond opposite mere was determination seen. The relative of lymphocyte abundance pres- of The(Figure immune S3 infiltrateF). was related to measures of DNA damage, individualence to suggest immune testable cell types properties had complex that correlate associations with patient that including copy number variation (CNV) burden (both in terms of differedoutcome between in different tumor tumor types types (Figures and immune S3C and contexts. S3D). These an- numberImmune of Response segments and Correlates fraction of Somatic genome alterations), Variation alysesWe extend obtained beyond and validated mere determination a survival model of lymphocyte using elastic-net pres- aneuploidy,The immune loss infiltrate of heterozygosity was related (LOH), to measures homologous of recombi- DNA damage, enceCox to proportional suggest testable hazards properties (CoxPH) modeling that correlate with cross-valida- with patient nationincluding deficiency copy (HRD), number and variation intratumor (CNV) heterogeneity burden (both (ITH) in (Fig- terms of outcometion. Low- in different and high-score tumor typestumors and displayed immune significant contexts. survival urenumber 4A). LF correlated of segments negatively and with fraction CNV segment of genome burden, alterations), with differences in the validation set (Figure 3D), with good prediction strongest correlation in C6 and C2, and positively with We obtained and validated a survival model using elastic-net aneuploidy, loss of heterozygosity (LOH), homologous recombi- accuracy (Figure 3E). Incorporating immune features into aneuploidy, LOH, HRD, and mutation load, particularly in C3. Cox proportional hazards (CoxPH) modeling with cross-valida- nation deficiency (HRD), and intratumor heterogeneity (ITH) (Fig- Cox models fit with tumor type, stage, and tumor type + stage These results suggest a differential effect of multiple smaller, tion. Low- and high-score tumors displayed significant survival ure 4A). LF correlated negatively with CNV segment burden, with differences6 Immunity in48 the, 1–19, validation April 17, set 2018 (Figure 3D), with good prediction strongest correlation in C6 and C2, and positively with accuracy (Figure 3E). Incorporating immune features into aneuploidy, LOH, HRD, and mutation load, particularly in C3. Cox models fit with tumor type, stage, and tumor type + stage These results suggest a differential effect of multiple smaller,
6 Immunity 48, 1–19, April 17, 2018 Immunity Review Personalized Immunotherapy
Can we personalize iARE detection?
IFN-α
Figure 3. Therapies that Might Affect the Cancer-Immunity Cycle The numerous factors that come into play in the Cancer-Immunity Cycle provide a wide range of potential therapeutic targets. This figure highlights examples of some of the therapies currently under preclinical or clinical evaluation. Key highlights include that vaccines can primarily promote cycle step 2, anti-CTLA4 can primarily promote cycle step 3, and anti-PD-L1 or anti-PD-1 antibodies can primarily promote cycle step 7. Although not developed as immunotherapies, chemotherapy, radiation therapy, and targeted therapies can primarily promote cycle step 1, and inhibitors of VEGF can potentially promote T cell infiltration into tumors—cycle step 5. Abbreviations are as follows: GM-CSF, granulocyte macrophage colony-stimulating factor; CARs, chimeric antigen receptors.
Grade 3-4 treatment-related toxicities). Although many of these class, inhibition of VEGF by the anti-VEGF antibody bevacizu- were serious and required treatment, therapy discontinuation, mab appears to be a promising candidate based on hints from or hospitalization, the durable partial and complete responses the preclinical and clinical literature (Motz and Coukos, 2013; in melanoma may warrant this approach in some patients. In Hodi et al., 2010). Similarly, B-Raf inhibitors (vemurafenib) may particular, combination therapy appeared to most dramatically also enhance T cell infiltration into tumors (Liu et al., 2013). Of benefit patients who were less likely to benefit from PD-L1 or course, with increased activity due to combinations comes the PD-1 inhibition alone, because their tumors were PD-L1-nega- increased chance for additive or synergistic toxicity. This further tive (6/13 PD-L1-positive and 9/22 PD-L1-negative patients re- highlights the importance of selecting therapeutic targets that sponded to combination therapy; Wolchok et al., 2013). The are specific to the ability of a tumor to escape immune eradica- addition of a CTLA4-targeted therapy may be completing the tion over targets that may also play an important role in mediating defect in the Cancer-Immunity Cycle for patients who are PD- immune homeostasis and preventing autoimmunity. L1-negative. Further studies of preipilimumab and on ipilimumab treatment tumor samples are warranted to better understand this Concluding Remarks effect. The objective of understanding the inherent immune biology Other combinations that merit serious consideration include related to cancer is to better define strategies to harness the anti-PD-L1 or anti-PD-1 with vaccination, especially if it be- human immune response against cancer to achieve durable re- comes possible to monitor a patient’s T cell profile to distinguish sponses and/or complete eradication of cancer in patients individuals who have generated suboptimal T cell responses to safely. Multiple approaches to cancer therapy exist, and few their cancers (Duraiswamy et al., 2013; Ge et al., 2013). In addi- are as complicated as immune-based therapy. Multiple numbers tion, combinations with agents that will enhance T cell trafficking of systemic factors can effect or contribute to the success or fail- and infiltration into the tumor bed should be explored vigorously, ure of immune therapy and lends to this complexity. Results may because the entry step may be important in some patients. In this be confounded by many currently unmeasured variables,
Immunity 39, July 25, 2013 ª2013 Elsevier Inc. 7 Thanks!
Immunopathology Lab, Ins tute of Biology and Experimental Medicine-CONICET
Dr. Gabriel Rabinovich, Senior Inves gator, Full Professor FCEyN-UBA