Rationally combining immunotherapies to improve efficacy of immune checkpoint blockade in solid tumors

(1)

Review

Rationally combining immunotherapies to improve ef ﬁcacy of immune checkpoint blockade in solid tumors

Floris Dammeijer

^a,b,¹

, Sai Ping Lau

^b,c,1

, Casper H.J. van Eijck

^c

, Sjoerd H. van der Burg

^d

, Joachim G.J.V. Aerts

^a,b,

*

aDepartmentofPulmonaryMedicine,ErasmusMedicalCenterRotterdam,‘s-Gravendijkwal230,3015CERotterdam,TheNetherlands

bErasmusMCCancerInstitute,ErasmusMedicalCenterRotterdam,‘s-Gravendijkwal230,3015CERotterdam,TheNetherlands

cDepartmentofSurgery,ErasmusMedicalCenterRotterdam,‘s-Gravendijkwal230,3015CERotterdam,TheNetherlands

dDepartmentofClinicalOncology,LeidenUniversityMedicalCenter,P.O.Box9600,2300RCLeiden,TheNetherlands

ARTICLE INFO

Articlehistory:

Received15May2017

Receivedinrevisedform19June2017 Accepted20June2017

Availableonline28June2017

Keywords:

Immunotherapy

Immune-checkpointblockade Tumorimmunology Tumormicroenvironment Personalizedmedicine

ABSTRACT

Withthewidespreadapplicationofimmunecheckpointblockingantibodies(ICBs)forthetreatmentof advancedcancer,immunotherapyhasproventobecapable ofyieldingunparalleledclinicalresults.

However,despitetheinitialsuccessofICB-treatment,stillaminorityofpatientsexperiencedurable responses to ICB therapy. A plethora of mechanisms underlie ICB resistance ranging from low immunogenicity,inadequategenerationorrecruitmentoftumor-speciﬁcTcellsorlocalsuppressionby stromalcellstoacquiredgeneticalterationsleadingtoimmuneescape.Increasingtheresponseratesto ICBsrequiresinsightintothemechanismsunderlyingresistanceandthesubsequentdesignofrational therapeutic combinationson aperpatientbasis.Inthisreview,we aimtoestablishorderintothe mechanismsgoverningprimaryandsecondaryICBresistance,offertherapeuticoptionstocircumvent different modes of resistance and plea for a personalized medicine approach to maximize immunotherapeuticbeneﬁtforallcancerpatients.

Contents

1. Introduction ... 6

2. Currentstateofimmunecheckpointblockade(ICB)inadvancedcancer ... 6

3. MechanismsunderlyingprimaryandsecondaryresistancetoICB ... 7

3.1. PrimaryresistancetoICB ... 7

3.2. SecondaryICBresistance ... 9

4. Therapeuticinterventionsaimedat(re-)sensitizingtumorstoICB ... 9

4.1. ModulatingtheT-cell:novelimmunecheckpointsinvolvingco-inhibition/co-stimulation ... 10

4.2. Chemotherapy,radiotherapyandoncolyticviruses:aimingtore-establishanti-tumorimmunity ... 10

4.3. Cytoreductionbysurgery:an(neo-)adjuvantroleforICBintreatinglocallyadvanceddisease? ... 10

4.4. Immunotherapy:passiveandactiveimmunizationapproachestoinducenovelimmuneresponses ... 11

4.5. Targetingkeyplayersofthetumormicroenvironment–makinganexampleofTAMs ... 11

5. ApersonalizedmedicineapproachtooptimallystratifyandtreatcancerpatientswithICB ... 11

6. Conclusion ... 12

Funding ... 12

Conﬂictofinterest ... 12

References... 12

* Correspondingauthorat:ErasmusMedicalCenterRotterdam,‘s-Gravendijkwal 230,3015CERotterdam,TheNetherlands.

E-mailaddress:j.aerts@erasmusmc.nl(J.G.J.V. Aerts).

1 Sharedﬁrstauthors.

http://dx.doi.org/10.1016/j.cytogfr.2017.06.011

ContentslistsavailableatScienceDirect

Cytokine & Growth Factor Reviews

j o u r n a l h o m e p a g e : w w w . e l s ev i er . c o m / l o c a t e/ c y t o g f r

(2)

1.Introduction

Formanyyears,directingourimmunesystemtotargetcancer wasminimallyeffectiveingeneratingdurableclinicalresponses.T- cellresponsesinducedbyofteninferiorlyformulatedanddesigned vaccineswerenotpowerfulenoughtoovercomethemanybarriers posed by advanced solid tumors [1,2]. However, following the unprecedentedresultsof‘re-invigorating’Tcellsinaproportionof metastatic cancer patients by blocking immune inhibitory checkpoints,tumorimmunotherapyhasregaineditspositionat theforefrontofcancertreatmenttoday[3].Tothisdate,themost studiedandmanipulatedimmunecheckpointsonTcellsarethe receptors T lymphocyte associated antigen 4 (CTLA-4) and programmedcelldeathprotein1 (PD-1). TargetingCTLA-4 and thePD-1-PD-L1-axiswithantagonisticantibodieshasproventobe highlyefﬁcaciousinaproportionofcancerpatients(Fig.1).The ﬁnding that a subgroup of patients has a pre-existing but dysfunctionalanti-tumorimmuneresponsethatcanbetherapeu- tically restored, prompts further investigation into what con- stitutes tumor immunity and precludes response to immunotherapy.

2.Currentstateofimmunecheckpointblockade(ICB)in advancedcancer

Immune checkpoints arereceptorsexpressed byTcells that upon ligationby their respective ligands regulate immune cell effectorfunctionsandproliferationtherebymaintainingtolerance toself-antigensand ensureimmunehomeostasis[4,5].Blocking inhibitorycheckpointsusingantagonisticantibodiesmay‘release the brakes’ on Tcells, including those cells speciﬁc for tumor antigens.

CTLA-4 is upregulated by T cells following recognition of cognateantigenbyantigenpresentingcells (APCs)inthelymph node [6]. The structure of CTLA-4 is nearly identical to the costimulatory receptor CD28 but interacts with much higher afﬁnityfor itsligandsCD80/CD86(B7-1/B7-2)expressedbythe APC[7].IncontrasttoCD28stimulation,CTLA-4hasaninhibitory effect on effector T cells by causing cell cycle arrest [6,7].

Additionally,regulatoryTcells(Tregs)constitutivelyexpresshigh levels of CTLA-4 on theircell surface, furtherfacilitating their immune suppressive potential [8]. Antibodies directedtowards CTLA-4maythereforealsoactbydecreasingTregfrequenciesin blood and tumor via antibody dependent cytotoxicity (ADCC) [9,10].

BesidesCTLA4,activatedTcellsexpressPD-1,andthecoupling ofPD-1toprogramedcelldeathligand1(PD-L1,alsocalledB7-H1) orPD-L2(B7-DC)restrainsT-celleffectorfunctionandprolifera- tion[11].PD-L1isexpressedontumorcells(constitutivelydueto oncogenicsignalingorinresponsetointerferons),myeloidcells includingAPCs,andPD-L2issolelyexpressedbyAPCs[12].Ithas recentlybeenshownthatbothPD-L1onhostmyeloidcellsandon tumorcellsis aprerequisiteforanti-PD-1-therapyefﬁcacy[13].

PD-1 was previouslythoughttoattenuateT-cell receptor(TCR) -signaling but recent insights have ﬁrmly established the inhibitoryroleofPD-1ondownstreamCD28-signallinginT-cells, furtheremphasizingtheimportanceofproper(local)co-stimula- tionforT-cellfunction[14,15].

Thusfar,fourICBsareFDAapproved;anti-CTLA-4(ipilimumab), anti-PD-1 (pembrolizumab and nivolumab) and anti-PD-L1 monoclonal antibodies (atezolizumab). Response rates vary between11and40%dependingontumortypewithPD-1blockade yielding superiorresponses ata more favorable toxicityproﬁle comparedtoCTLA-4inhibition[16–20].Ithasbeensuggestedthat thediscrepancyintoxicitiesbetweenICBscanbeexplainedbythe

Fig.1.Progression-freesurvivalcurvesforchemotherapy,anti-PD-1-andanti-CTLA-4-checkpointblockers;primaryandsecondaryresistancetoimmunecheckpoint blockade(ICB)therapyprecludespatientsfromachievingdurableresponsesandlong-termsurvival.WhenpatientsdonotrespondtoICBsimmediatelyfollowingstartof treatmenttheyexperienceprimaryimmuneresistance.Whenpatientsdorespondinitiallybutrelapseovertime,secondaryresistancetoICB-treatmenthasdeveloped.

PFS-curveshavebeenderivedfromthefollowingclinicaltrialsinvestigatingICB-efﬁccacyinmetastaticmelanoma:Robertetal.NEJM2011,Schachteretal.ASCO#95042016.

(3)

timeofcheckpointengagementintheT-cellresponse.ThePD-1/

PD-L1axishasbeenproposedtooperatelaterduringtheeffector phaseofaTcell,resultinginamoreconﬁnedresponsewhereas CTLA-4actsonthelymphnodeduringT-cellpriming[21].These temperospatialdifferences betweenICBs arebeingexploitedby combining anti-CTLA-4 and anti PD-1/PD-L1 in the clinic.

Combiningipilimumab(anti-CTLA4)and nivolumab(anti-PD-1) inBRAFwild-typemelanomapatientswasefﬁcaciousinreaching itsprimaryendpointofprogression freesurvival[22].Although primaryanalysisshowed asigniﬁcantadvantageofcombination therapyoverbothmonotherapies,recentfollow-updatareporta 2-year survival rate of 64% in the combination treated group comparedto59%survivalin

a

PD-1monotherapytreatedpatients.

Notably, the difference in serious adverse event rate was considerable(58%vs21%)suggestinglimitedclinicalvalueofthis immunotherapycombination[23].Inothersolidtumorsincluding non-small-cell lung cancer (NSCLC) and renal cell carcinoma, responseratesofICBmonotherapyaremoremodestrangingfrom 15 to 20% [24–29]. Reasons underlying this heterogeneity in responseratesshallbefurtheraddressedinthefollowingsections.

DespitethesigniﬁcantprogressthathasbeenmadewithICB across multipletumor types, still much remains to be gained.

Recentinsightsintotumorsfrominitialanddurableresponders andnon-responderstoICBhaveofferednovelinsightsintotumor- immune interactions and the prerequisites for establishing effectiveand durable anti-tumor immunity. A complete under- standingoftheseprocessesisstilllackingbutwithknowledgeof basic(tumor-)immunologicalprinciplesandtheimplementation ofinnovativediagnostics,rationaltherapeuticcombinationscanbe designed to improve ICB response rates in advanced cancer patients.

3.Mechanismsunderlyingprimaryandsecondaryresistanceto ICB

Despite the success of ICBs, only a minority of patients experiencedurableresponses toICBtherapy. The remainderof patients do not respond at all (primary resistance) or initially respondbut relapseover time(secondaryresistance)(Fig.1).A plethoraofmechanismsunderlieICBresistance.Primaryaswellas secondary resistance to ICB results from an intricate interplay betweenimmunecells,otherstromalcells(e.g.cancerassociated ﬁbroblasts(CAF),endothelialcells)andtumorcells,alltogether composing the tumor microenvironment (TME). In general, primary resistance occurs when tumors lack an endogenous adaptiveandfunctionalimmuneinﬁltrate(thisincludesthepre- existence of an irreversibly ‘hyper-exhausted’ T-cell response incapableofrespondingtoICB).Secondaryresistancerecapitulates alltheadaptivemechanismswhich takesplacesubsequentlyto therapeutic pressure resulting in the failure to maintain an effectiveanti-tumorresponse.Ithastobenotedthattheproposed distinctionbetweenprimaryandsecondaryimmuneresistanceis pragmaticandusefulinmostcausesofresistancebutinreality, multipleopposingphenomenamaybeatplayandsome(suchas animmunesuppressiveTME)mayactthroughoutthecourseofICB treatment.

3.1.PrimaryresistancetoICB

Primary resistance to ICB can result fromthe absence of a functional immune response to poorly immunogenic tumor (Fig. 2).Tumors witha high non-synonymous mutational load aremorelikelytodisplayneo-antigensthatcouldbeconsidered foreigntotheimmune system and thus possiblyimmunogenic [30,31].Therefore,itisnotsurprisingthatcancertypeswiththe highestmutationalloadsgenerallyhavehighresponseratestoICB

(melanoma,NSCLC)[32].Also,subtypesoftumorscharacterized by deﬁciencies in mismatch repair genes, as is the case for microsatelliteinstablecoloncancers,respondmarkedlybetterto ICB compared to their microsatellite stable counterparts [33].

However,evenwithinthesametumortype,highmutationalload intumorswasshowntoatleastpartiallypredictresponsetoboth anti-PD1-andCTLA4-inhibitionfurthersupportingtheimportance oftumormutationallandscapeandconcomitantimmunogenicity indeterminingICBefficacy[31,34,35].Butdoesanincreasedneo- antigen load also necessarily lead to enhanced cytolytic T-cell responsesintumors?AseminalstudybyRooneyetal.showsthat increasedneo-antigenload,andinsometumorsthepresenceof viralgenes,wasindeedassociatedwithenhancedcytotoxicT-cell activity [36]. In line with these findings positive correlations betweenanti-CTLA-4therapyefficacyandthepresenceofapre- existingimmuneresponsetogetherwithahighmutational-and neo-antigen loadinmelanomahave beenfound[35]. Asimilar prerequisite for ICB-efficacy was found in melanoma patients where apre-existing CD8+PD1+ T-cellinfiltratein theinvasive tumormarginandcenterpredictedresponsetopembrolizumab (anti-PD-1antibody)treatment[37].Highmutationalloadand/or expression ofneo-antigensalone doesnotseemtofullypredict responsetoICB,andothershaveshownthatexpressionofother antigens such as cancer tests antigens and tumor associated (overexpressed)antigensmayalsocontributetotumorimmuno- genicity [38].Thesedatademonstratethatendogenousimmune reactivitycharacterizedbycytolyticTcellsinthetumorconstitutes abasicrequirementforICBefficacy.

Another major reason for primary ICB resistance is the immune-privileged tumor micro-environment, characterized by thepaucityofinfiltratingtumor-specificTcells.Theexistenceof thisso-called ‘non-inflamed’tumorderivesfromtheinadequate generationorrecruitmentoftumor-specificTcells,orthephysical inabilityofimmunecellstoreachthetumor.Inordertoinducea functional immune response, innate immune recognition and subsequentprimingoftumor-antigenspecificTcellsinthelymph nodeisimperative[39,40].InterrogationofthetheTCGAdatabase byGajewskiandcolleaguestoidentifyfactorsassociatedwithaT- cell inflamed tumor phenotype failed to detect an association between a T-cell inflamed tumor and mutational burden [41].

However,theydidﬁndstrongpositivecorrelationsbetweenT-cell inﬁltrationand presence of DC-related genes emphasizing the importance of DC-mediated anti-tumor immunity over solely tumor antigenicity. In accordance with thesedata, others have found intratumoral DCs to be critical for establishing tumor immunity,withtumorsbeingcapableofactivelysubvertingDC- accumulationorfunctioninvivo[42].Onesuchcauseofimmune ignorance that could be at play is a mutated

b

-catenin/Wnt- signaling pathway in tumor cells, which causes a decrease in chemokinesknowntobecrucialforDC-homingtothetumor[43].

Suchmutationscouldpresentasignificantdownsidetohavinga highmutationalburden,andcouldprovideanexplanationforthe heterogeneityobservedinICBefficacyinhighmutation tumors [41].Interestingly,othermutationsinkeyoncogenicpathwaysare currently beingidentified that impede immune cell-infiltration and/orfunctioninthetumor(e.g.mutationsinPTEN,MYCetc.) [44,45].Re-establishingimmunesurveillancebyskewingmyeloid precursors to the DC-fate, targeting oncogenic pathways or promotingDC-functionmay beessential insensitizing patients toICB.

Moreover, theamountof intratumoral Teffector cells could determinethepotentialofICBtherapytoinducerobustanti-tumor response. T effector cells can be mechanically excluded by a psychicalbarrierconsistingofthickextracellularmatrixproduced by stromal cells (e.g. CAFs) [46].CAFs can alsoexclude Tcells through coating of cancercells withCXC chemokine ligand-12

(4)

(CXCL12)[47].Furthermore,theabnormalvasculatureintheTME expressinghighendothelialFas-ligandpromotesintravascularT cellapoptosis[48].Inaddition,effectorTcellswillneedtoexpress theproperintegrinsinordertobindtothetumorendothelium, egressandexerttheirfunction.Changingtherouteofvaccination wasshowntomodulateintegrinexpressiononT-cellsandimprove homingtothetumortissue[49].

Finally, immune resistance can also be achieved by the preferredattractionofimmuneinhibitorycellstotheTME.Tregs, tumor associated macrophages (TAMs) and myeloid derived suppressor cells (MDSCs) often populate the TME where they exertseveralimmuneinhibitoryproperties,makingitdifﬁcultfor T-cellstosustaintheiranti-tumoreffectorresponses,especiallyin thesettingofICB[50].

Tumors can recruit, induce and expand Tregs capable of suppressing(ICB-induced)anti-tumorTcellsviacompetitionfor key survival factors (CD80/86 co-stimulatory signals, IL-2) and suppressivecytokines(e.g.IL-10,TGF-

b

,IL-35).AsTregsaremuch more potent in binding these survival factors by means of constitutiveCTLA-4 and IL-2-receptor (CD25) expression, CD8+

Tcellsareshortlyoutcompeted.Tregswerefoundtobeinvolvedin limiting

a

PD-1-efﬁcacy as depletion of these cells improved responsestotherapyinseveralsolidtumormousemodels[51].

TAMscontributetoamajoritycancerhallmarksincludingneo- angiogenesis, metastasis, chronic inﬂammation and immune

suppression [52]. Skewing or depleting TAMs could therefore affectmultiplecriticalstepsinoncogenesisandabrogatedifferent modesofimmuneresistance[53].TAMsdisplayanalternatively activated‘M2’-phenotypeknowntobecriticalincontrollingtissue homeostasisandwoundhealing[52].Inthetumor,however,this phenotypeisundesirableasitenablespotentT-cellinhibitionvia cytokines(e.g.IL-10),depletionofkeymetabolites(expressionof arginase,IDO)orbycontactinhibition(e.g.viaPD-L1)[52].This TAM-phenotypeisalsocriticalindeterminingICBefﬁcacyasan innate‘woundhealing’andimmune suppressivegenesignature was foundto optimallypredictnon-responders prior to

a

PD-1 treatment [54]. Recently, Arlauckas et al. identiﬁed another mechanismwherebyTAMscanlimit

a

PD-1therapyefﬁcacy.They foundTAMs tocapture PD-1 targeting antibodies ontheT-cell surfacetherebyconsiderablylimitingthedurationofdrugefﬁcacy [55].

SimilartoTAMs,MDSCscanpotentlyinhibitT-cellfunctionbut theycanalsoindirectlycontributetoanimmunesuppressiveTME bydifferentiatingintoTAMsorskewingthemtoanM2-phenotype [56]. MDSCsare the epitome of chronicand systemic immune modulationbyatumorthatsecretesnumerousmoleculescapable ofskewingmyelopoiesis(e.g.GM-CSF,IL-6,VEGFetc.)[56].

Thepresenceoftheseimmuneinhibitorycellsinmostpatients tumors suggests that a balance exists whereby ICB-responsive anti-tumor Tcells are inequilibrium withimmunesuppressive Fig.2. Differentprocessesunderlyingprimaryand/orsecondaryresistancetocheckpointblockingantibodiesinsolidtumors;Primaryresistancecanresultfromtheabsence ofafunctionalimmuneresponsetoapoorlyimmunogenictumor.Themagnitudeofresistanceisinfluencedbydifferencesin:(1)non-synonymousmutationalloadandneo- antigenexpression,(2)thepresenceofintratumoraldendriticcellscapableofantigentraffickingandpresentation,(3)thegenerationorrecruitmentoftumor-specificTcells and(4)immuneinhibitionbyinhibitoryimmunecellpopulationsintheTME.Continuoustherapeuticpressuremayresultinthedevelopmentofsecondary(acquired) resistance.Mechanismsinclude(5)upregulationofotherco-inhibitorymoleculesand(6)lossoftumor(neo)antigenexpression.

(5)

cellsintheTME[57].Inlinewiththishypothesisisdatafrom

a

PD- 1- and

a

CTLA-4-treated patients tumors showing increased presenceof memoryTcell-and (activated) DCgene signatures in ICB responders, in contrast to MDSC, Treg and monocyte signaturesinthenon-respondingpatients[58].Findingswaysto shift this balance preferably from both sides will be key in improvingICBresponsiveness.

3.2.SecondaryICBresistance

Overtime,relapsewilloccurinamajorityofpatientsinitially responsive toICB therapy. A possible phenomenon underlying secondaryresistancearenewmutationsacquiredbytumorcells that have expanded under continuous therapeutic pressure (immune editing) and have eventually grown out (immune evasion)(Fig.2).

Tumorintrinsicmutationsthathaveevolvedoverthecourseof ICB-treatmentcanhavehighlyvariableconsequences totumor- immuneinteractions.Ithasbeenknownforseveralyearsthatloss ofantigendisplaybytumorcellsduetomutationsintheantigen- processingmachinery(e.g.TAP) orproteins involvedinantigen presentation(

b

2-microglobulin,HLA)cancauselackofrecogni- tion by CD8+ T-cells following immunotherapy [59]. Recently, similar mutations were detected in patients who relapsed following

a

PD-1ICB[60].Anotherpathwaythatcanbesilenced by mutations following ICB is the interferon-gamma receptor (IFNGR)pathway,consistingoftheIFNGR,JAK1/JAK2andSTAT1 which promotestranscription ofinterferon-induced genes[61].

ThecytokineIFN-

g

isknowntohavedichotomousimmunological properties by inducing apoptosis of tumor cells, blood vessel

disruptionandupregulationofMHC-expressionontheonehand, butexpressionofIDO,PD-L1andotherco-inhibitorymarkerson theotherhand[61–64].Theseco-inhibitorymoleculesincluding LAG-3andTIM-3synergizewithCTLA-4andPD-1inpromotingT- cellexhaustion[65,66]andareknowntobeupregulatedfollowing initiationofICBtherapy[67].InactivatingmutationsintheIFNGR- pathwayhavebeendocumentedinpatientsandarehypothesized tooccurinsettingsofcheckpointblockadewhichleavestumors cellsexposedonlytotheanti-tumorpropertiesofIFN-

g

,causing selective pressure [60,63]. Paradoxically, chronic exposure to interferons including IFN-

g

was found to also induce immune resistanceduetoPD-L1-dependentand-independentmechanisms [64]. This may occur in settings of chronic (ICB-induced) inﬂammationwherethepro-tumorfunctionsofIFNsprevailover theanti-tumorones,leadingtoimmuneresistance.

Besides speciﬁc mutations in immune-related pathways, tumors may lose neo-antigens and thereby escape immune control. In two melanoma patients, immunogenic neo-antigens were lost duringtumor progression indicating immune-editing [68].ImmuneeditingwasalsoreportedinNSCLCpatientswhose lesion(s) initially responded to PD-1-inhibition but later pro- gressed. Therelapsedtumorsweredevoidof severalmutations encoding forneo-antigensthatwerepresentprior totreatment [69].

4.Therapeuticinterventionsaimedat(re-)sensitizingtumorsto ICB

Increasing the response rates to ICB will require rational combinations of conventional anti-cancer therapies and other

Fig.3.Therapeuticoptionstotargetimmuneresistance;sensitizingtumorstocheckpointblockadetherapycanbeachievedbyincreasingimmunerecognition,targetingthe TMEto removeimmunesuppression,combiningantibodiestoco-inhibitory/stimulatorymoleculesonTcells,increasingeffector Tcellsandbyincreasingtumor immunogenicity.

(6)

immunotherapiesona perpatientbasis tooptimallyprimethe tumorforICBstohaveeffect.Asmanyofthesetherapiesactby alleviating both primary and secondary forms of immune resistancetheyshallbeaddressedperindividualclassoftherapy (Fig.3).

4.1.ModulatingtheT-cell:novelimmunecheckpointsinvolvingco- inhibition/co-stimulation

FollowingthediscoveryofPD-1andCTLA-4,numerousother co-inhibitorymoleculesontheT-cellsurfacehavebeencharacter- izedand shown tocontribute toT-cell exhaustion [5].It could therefore be beneﬁcial or even necessary to target multiple inhibitory molecules at the same time to attempt reversal of exhaustion [70]. It should be noted, however, that T-cell dysfunctionin canceris a multifactorial process depending on manyfactorsbesidesco-inhibitoryreceptorsignaling[5].More- over,co-expressionofmultipleinhibitorymoleculesbesidesPD-1, includingLAG-3andTIM-3indicatesastateof‘hyperexhaustion’

thatisnotrecoverablebyICB-treatment[71].Upregulationofco- inhibitorymoleculeshasbeenshowntooccurinmiceandhumans followingPD-1-inhibition(TIM-3,LAG-3")[67]andincaseofanti- CTLA-4-treatment(VISTA,PD-L1")[72].Theseﬁndingsprovidea clinicalincentivetocombinedifferentICB-therapiestopotentially sensitize tumors previously thought to be ICB-resistant (e.g.

prostatecancer).

Paradoxically,dysfunctionalTcells intheTMEareknownto expressco-stimulatoryreceptorssimultaneouslywithco-inhibi- torymoleculessuchas4-1BB(CD137),ICOSandOX40,suggestinga possible balance that can be therapeutically exploited [72,73].

Preliminary data frompre-clinical mouse models indeed show beneﬁt of combining agonistic antibodies to co-stimulatory molecules with antagonistic antibodies targeting co-inhibitory molecules[73,74].Itmaythereforebebeneﬁcialto‘pushthepedal’ bytargetingco-stimulatorymoleculesonthehand,and‘release thebrakes’usingco-inhibitorycheckpointblockingantibodieson theotherhandtofullyexploitT-celleffectorfunction.

4.2.Chemotherapy,radiotherapyandoncolyticviruses:aimingtore- establishanti-tumorimmunity

Manyconventionalanti-cancertherapiessuchaschemo-and radiotherapy, including oncolytic viral therapy was previously thoughttoprincipallyactbyarrestingtumorcellproliferationand causingcelldeath.However,novelinsightshaveledtoachangein paradigmwheremanyofthese‘traditional’anti-cancertreatment strategiesare now appreciated tofunction at least partiallyby modulatingtheimmunesystem[75,76].Asmentioned before,a majorcontributortoprimaryICB-resistanceislackoffunctional DCsinthetumorcapableofprimingTcellsinlymphoidorgans.

Bothradiotherapyandcertainclassesofchemotherapy,butalso severaloncolyticvirusesarecapableofcausingimmunogeniccell death(ICD)whichincreasesantigenavailabilitytodendriticcells intheTME[76].Besidesreleasingantigens,tumorscellsrelease damageassociatedmolecularpatterns(DAMPs)thatarecapableof attractingand stimulatinginnateimmune cellstosubsequently phagocytosecellulardebrisandpresentantigentotumor-speciﬁc T cells [75–78]. A thorough appraisal of the various immune modulatingfunctionsofthedifferentclassesofchemotherapy,and to a lesser extend radiotherapies, is beyond the scope of this review.However,itisimportanttonotethatevendrugswithinthe sameclassofchemotherapiese.g.oxaliplatinandcisplatin,may havedifferenteffectsontheimmunesystem,beitICDorenhanced expression of co-stimulatory markers on APCs, respectively [77,79].

Chemo-andradiotherapyhavealsobeenshowntoupregulate type I interferons in the tumor microenvironment, thereby attractingTcells byincreasedchemokine productionin caseof anthracyclines [80], or by activating dendritic cells critical for adaptiveimmuneinduction[81].TherapyelicitedtypeIinterfer- onscanalsoimproveresponsesin thesetting ofsecondary ICB resistancewhereMHC-moleculesonthe tumorcellsurface are downregulated,butcanbepotentlyre-expressedwhenexposedto typeIinterferon[82].Reinstatingimmunityfollowingprimaryor secondaryimmuneresistancebyconventionaltherapieshasbeen shownto(re-)sensitizetumorstoICBtherapy[83,84].Inastudyby Twyman-SaintVictoretal.,melanomapatientsreceivedradiation on one index lesion followed by systemic CTLA-4-blocking antibodies.Besides a fewresponses includingone patientwith abscopalresponses(regressionofunirradiateddistanttumors),the majorityofpatientsprogressed[10].Theywentonfurthertoshow that upregulation of PD-L1 on the tumor following radio- immunotherapy signiﬁcantly abrogated effective immune responses, which could be reversed by administering PD-1- inhibitingantibodies.Similarphenomenaalsooccurinthesetting ofoncolyticviraltherapywherevirustreatmentisabletoinﬂame immunologically silent tumors and upregulate immune check- pointsthatcouldbetargetedbyICB[85,86].Ithastobenotedthat severalstudieshavealsoreportednegativeeffectsofradiotherapy on anti-tumor immunity including the increase of immune suppressingcellsintheTME(Tregs,MDSCsandTAMs)[75].Also inpatientsreceiving radiotherapy,immunemonitoringofblood showedincreasedmyeloidcellanddecreasedlymphoidcellcounts and immune reactivity following radiotherapy in contrast to standardchemotherapy[87,88].Someofthesediscrepanciesmay becausedbyopposingbiological pathwaysunderlyingdifferent radiation regimens aswas recently reportedbyDemaria et al., showingthatmultiplelow-doseirradiationcyclessynergizedwith

a

CTLA-4 antibodies in contrast to one single higher dose of radiotherapy in pre-clinical tumor models. Lower doses of radiation induced local type I IFN-productionand concomitant recruitment of DCs, whereas high dose irradiation activated a cytosolic DNA-degradationpathway,preventing immuneinduc- tion[89].Novelmechanismsunderlyingthesedivergenteffectsof radiotherapy will have to be addressed and may involve modiﬁcationofthetreatmentscheduleanddose(fractionatedor highdose)andtherequirementforfuturecombinationstrategies (e.g.TMEtargeteddepletion,ICB).

4.3.Cytoreductionbysurgery:an(neo-)adjuvantroleforICBin treatinglocallyadvanceddisease?

Theadditionofimmunotherapytoconventionalcytoreductive surgerymayimprovepatientsurvivalbyextendingrecurrencefree survivalfollowing(incomplete)tumorresection.Fromanimmu- nological perspective, the major advantage of surgery is the reductionoftumor-andassociatedantigenload.Chronicantigen exposure is known to be a main contributor to exhaustion of effectorTcellsandoccursalreadyearlyintumorigenesis[90].The persistence of T cell exhaustion could eventually lead to the irreversibility to reinvigorate T cell function with ICB therapy [71,90].Moreover,increasedtumorsizecorrelateswithextended immune suppression [91], suggesting that manually reducing tumorsizecouldalleviateimmuneinhibitionandT-cellexhaustion.WhetherICBshouldbeadministeredinanadjuvantorneo- adjuvantsettinghasbeenrecentlyinvestigatedinmurinebreast cancermodels.In thesemodelsLiu etal.showedsuperiorityof neo-adjuvant anti-PD-1therapyover adjuvanttreatmentin the contextofsurgery[92].Micetreatedwithneo-adjuvantICBhad signiﬁcantly longer recurrence free survival due to higher frequenciesofcirculatingtumorspeciﬁcmemoryTcellscapable

(7)

of surveying the body for micro-metastasis [92]. The reported immune response kinetics resemble what is observed in the setting of acute infection, where a decrease in antigen load followingclearanceofthepathogensupportsinductionofaproper memory T-cell pool [93]. Furthermore, recent insights into biomarkers associated with response to

a

PD-1 therapy have implicatedelevatedCD8+PD1+T-cellproliferationinasettingof lowtumorloadtobepredictiveofresponse[94].Itispossiblethat inthefuture,surgerymayfulﬁlapivotalroleinestablishingsucha settinginthecaseofextensivetumorburden.However,itshould benotedthatsurgerymayalsoinducetheinﬂuxofimmunosup- pressive cells abrogating T-cell function as part of a systemic

‘wound healing response’ [95] (De Goeje, Aerts, unpublished results).

4.4.Immunotherapy:passiveandactiveimmunizationapproachesto inducenovelimmuneresponses

PrimaryimmuneresistancetoICBcanresultfromtheinability orlackofendogenousDCscapableofpriminganti-tumorTcells (non-inflamedtumor)orthepresenceoftumorinfiltratingTcells thatareeither irreversiblyexhaustedornot specificfortumor- antigens[21,71].Inthesecases,novelimmuneresponsesneedto beinducedthat intimecanbefurtherenhanced bycheckpoint blockade.

Tumorvaccinesenableinductionofnovelimmuneresponses orreinstatepre-existingimmuneresponsestowardsaspeciﬁcor wide array of tumor antigens formulated in the vaccine [1].

Althoughcancervaccinesoffersignificantadvantagesincluding highspecificity,afavorablesafetyprofile,of-the-shelfapplicabil- ity and the premiseof life-long anti-tumor immunity, clinical efficacyisoftenlimitedinovertcancer[2].Severalstudieshave highlightedthe importance andpower of neo-antigen specific immune responses in establishing tumor control [30,96].

Exploitingnoveltoolsfromtheﬁeldofcancerimmunogenomics enablesthecharacterizationofimmunogenicneo-antigensthat canbesubsequentlyproducedand incorporatedintopersonal- izedvaccines [97,98].Severaltrialsare underwayinvestigating thesafetyandclinicalefﬁcacyofthesepersonalizedvaccines[97].

Besidespeptidevaccines,itispossibletocircumventendoge- nousantigenpresentationandexpose invitroculturedautolo- gousdendriticcellstotumorantigensandstimuli[99].Thisform of immunotherapy called DC-therapy was found to be safe, capableofinducinganti-tumorimmuneresponsesandeffective inasubgroupofadvancedcancerpatients[2,100].Additionally, DC-immunotherapy was shown to induce epitope spreading, elicitingnovelT-cellresponsesspeciﬁctoantigensnotformulat- edinthevaccine,andcapableofinducingbothCD8+andCD4+T- cellresponsesinvivo[101].Bothformsofactiveimmunization werefoundtosynergizewithcheckpointblockadetherapyinpre- clinicaltumormodels,possiblybyelicitinganewpoolofTcells thatis susceptible to re-invigoration in a (PD-L1 high)tumor [102,103]. In case of tumors lacking a functional antigen- presentationpathway(mutationsinTAP,lowMHC-I;secondary immuneresistance),itmaybepossibleinthefuturetovaccinate with TEIPPs(Tcell epitopes associated with impaired peptide processing),astheseantigensareselectivelypresentedinsettings ofabnormalantigenprocessingsuchascancer[104].

Instead of actively inducing endogenous anti-tumor T-cell responses using (DC-)vaccines, one can directly infuse large numbersoftumorantigen-specificT-cellsderivedfromresected tumor tissue (TIL-therapy) or from PBMCs following genetic modificationTCR-engineeredorchimericantigenreceptor(CAR)T- celltherapy[105].Theseforms oftherapyarecurrentlyrevolu- tionizingthefieldofhemato-oncologywiththeimplementationof

CD19-speciﬁcTcells,andhaveyieldedanecdotalresultsinsolid tumors[106].However,asthemajorityofcancerpatientsarenot eligible for TIL-therapy, and safe and effective targets for engineeredTcellsarestilllackingaswellasthechallengesinT- cell penetration and persistence for most solid tumors, T-cell therapystillhasalongroadahead.

4.5.Targetingkeyplayersofthetumormicroenvironment–making anexampleofTAMs

We recently identified TAMs to be critically involved in determiningtheexhaustionstatusofvaccine-inducedT-cells,as tumor infiltrating T cells expressed lower levels of the co- inhibitorymoleculesPD-1,LAG-3andTIM-3followingM-CSFRi- mediated TAM-depletion [127]. As this PD-1 low/intermediate expressingphenotypeisparticularlysensitivetore-invigorationby PD-1-blocking antibodies[71],M-CSFR-inhibition enhanced the efficacyofICBinmouseofmodelsofpancreaticcancer[107].

BesidesdepletingTAMs(e.g.bytargetingtheM-CSF-receptoror homingreceptorssuchasCCR2),skewingofTAMstoamorepro- inﬂammatory‘M1’phenotypemay beeven moreefﬁcacious in inducingtumorregression. Skewingof TAMsbyCD40-agonistic antibodies was shown to result in loss of desmoplasia and inductionof tumorregression incombinationwithgemcitabine in pancreatic cancer patients and pre-clinical models of PDAC [108,109].Similarobservationsweremadefollowingpharmaco- logicalinhibitionofPI3K

g

inmultipletumormodels,wherePI3K

g

was identiﬁed as a key molecular switch governing the M2 macrophagephenotype[110,111].SkewingofTAMscouldtherefore ameliorate primary immune resistance caused by mechanical obstructionofT-cellinﬁltrationbythecollagen-richstroma[112].

In supportofthis arethemarkedlyincreasedT-cellnumbersin tumorstreatedwithPI3Ky-inhibitionorCD40-agonisticantibodies [111,113]. Importantly, resistance to ICB in pre-clinical models could be overcome by combination with both TAM-skewing compounds,highlightingtheroleof myeloidcellsinperturbing anti-tumor immunity and ICB-efﬁcacy [114,115]. As PD-1 is thought to act primarily on T-cells at the effector site, it is tempting to speculate whether skewing of TAMs to a M1- phenotypecouldprovideB7-costimulatorymoleculescapableof binding CD28 onT-cells in thetumor. As PD-1-blockadecould enable proper signaling through the CD28-B7-axis, this could provide anotherexplanationfor theobserved synergybetween thesedifferentformsofimmunotherapy.

The composition of the TME varies extensively between different tumor types, requiring tailored approaches to target speciﬁcimmunepopulations[116].BesidesTAMs,othermyeloid cells suchas neutrophils, MDSCs and tolerogenic DCs but also regulatoryTcellscanposesigniﬁcantobstaclestothegenerationof effective anti-tumor immunity. In line with TAM-targeting therapies, strategiesaimedat depletingMDSCs(e.g.anti-CXCR2 or CCR2 antibodies, multikinase inhibitors e.g. cabozantinib) [117–119] or Tregs (Fc-optimized aCD25-antibodies) [120] all synergizewithICB-therapies.

5.Apersonalizedmedicineapproachtooptimallystratifyand treatcancerpatientswithICB

Atpresent,theidentiﬁcationofpredictivefactorsdetermining theresponsetoICBtreatmenthasremaineddifﬁcult.Extensively reviewedbiomarkerssuchasPD-L1ontumor-andmyeloidcells havefailedtodeliverrobustresultsacrossmultiplecancers[121].

Similar to PD-L1, tumor mutational load has been found to contributetoICB-response but itsdiscriminative valueremains insufﬁcient[41].Amoreholisticandcompletecharacterizationof

(8)

thetumoranditsTMEwilllikelyimprovetheaccuracyofcurrent predictivemarkers[58].Thismayincludeassessingthepresenceof aCD8+T-cellinﬁltrateincombinationwiththePD-L1statusofa tumortofurtherdelineatewhetheratumormightbesensitiveto ICBor that other therapiesare required toprime the immune systemﬁrst[122].

Assessing primary immune resistance can be achieved by employing novel tools in immunogenomics including next- generation sequencing on baseline tumor samples [98]. Using genome-wide approaches or eventually specified sets of genes correspondingtospecificresistancemodules,itwillbepossibleto determineboththetumorantigen-andimmunologicallandscape oftumors[36].Recentlydiscoveredmultipleximmunohistochem- istrytools will offer localization of certain cell types on often alreadyavailableparaffinembeddedtissuetofurtheraidpatient stratification [123]. Elegantly, optimized pipelines designed to predictneo-epitopesusingtheaforementionedtechniquesoffer theopportunityforpersonalizedimmunotherapyusingvaccines andTCR-modified/CAR-T-cellapproaches[97].

Incontrasttoprimarytumortissuewhichisreadilyavailable upon disease diagnosis,samples acquiredduring and afterICB treatmentareoftendifficulttoobtain,therebylimitingmonitoring oftreatmentovertime.Asseveralgroupshavedemonstratedthe predictivevalueoftumortissueearlyduringcourseoftreatment [124,125] it will be challenging to find more non-invasive biomarkersthatcanguideimmunotherapy.Attemptshavebeen made to define such markers in peripheral blood of patients yieldingpromising resultsbycharacterizing proliferatingPD-1+

CD8+T-cellsfollowing

a

PD-1treatment[94,126].Extendingthe scopetoothercirculatingimmunecellssuchasmyeloidcellscould furtherimprovethesensitivityoftheseanalysis.

6.Conclusion

Arecentappreciationoftheroleourimmunesystemplaysin tumors has led tothe widespread implementation of immune modulating drugs such as ICBs for the treatment of advanced cancer; with unprecedented clinical success. However, as the majority of patients fails to demonstrate durable responses, rational combinations of conventional- and novel anti-cancer therapieswillneedtobeemployedonanindividualizedbasisto ensurethebestpossibleresponses.

Funding

Thisresearchdidnotreceiveanyspeciﬁcgrantfromfunding agenciesinthepublic,commercial,ornotfor-proﬁtsectors.

Conﬂictofinterest

FlorisDammeijer:None.

SaiPingLau:None.

CasperH.J.vanEijck:None.

SjoerdH.vanderBurg:Norelationshiptodiscloseinrelationto the submitted work. Relevant activities outside the submitted work: Scientiﬁc advisory board, DC prime, the Netherlands;

strategicadvisory board,ISA pharmaceuticals, theNetherlands;

Advisor,NKI,TheNetherlands.Heisalsonamedasinventoronthe patentfortheuseofsyntheticlongpeptidesascancervaccine.

JoachimG.J.V.Aerts:Norelationshiptodiscloseinrelationto the submitted work. Relevant ﬁnancial activities outside the submittedwork:StockorOtherOwnership:AmpheraConsulting orAdvisoryRole:Bristol-MyersSquibb,MSDOncology,Boehringer

Ingelheim,Eli-Lilly,RocheSpeakersBureau:AstraZeneca.Research Funding:Genentech(Inst),BoehirngerIngelheim(inst).Patents, Royalties,OtherIntellectualProperty:Patent:Tumorcelllysatefor dendritic cell loading (Inst), SNP analyses for immunotherapy (Inst).

References

[1]S.H.vanderBurg,R.Arens,F.Ossendorp,T.vanHall,C.J.Melief,Vaccinesfor establishedcancer:overcomingthechallengesposedbyimmuneevasion, Nat.Rev.Cancer16(2016)219–233.

[2]F.Dammeijer,L.A.Lievense,G.D.Veerman,H.C.Hoogsteden,J.P.Hegmans,L.

R.Arends,etal.,Efﬁcacyoftumorvaccinesandcellularimmunotherapiesin non-small-celllungcancer:asystematicreviewandmeta-analysis,J.Clin.

Oncol.34(2016)3204–3212.

[3]M.A.Postow,M.K.Callahan,J.D.Wolchok,Immunecheckpointblockadein cancertherapy,J.Clin.Oncol.33(2015)1974–1982.

[4]S.L.Topalian, C.G.Drake, D.M.Pardoll, Immune checkpointblockade:a commondenominatorapproachtocancertherapy,CancerCell27(2015) 450–461.

[5]E.J.Wherry,M.Kurachi,MolecularandcellularinsightsintoTcellexhaustion, Nat.Rev.Immunol.15(2015)486–499.

[6]T.L.Walunas,D.J.Lenschow,C.Y.Bakker,P.S.Linsley,G.J.Freeman,J.M.Green, etal.,CTLA-4canfunctionasanegativeregulatorofTcellactivation, Immunity1(1994)405–413.

[7]M.F.Krummel,J.P.Allison,CD28andCTLA-4haveopposingeffectsonthe responseofTcellstostimulation,J.Exp.Med.182(1995)459–465.

[8]K.Wing,Y.Onishi,P.Prieto-Martin,T.Yamaguchi,M.Miyara,Z.Fehervari, etal.,CTLA-4controloverFoxp3+regulatoryTcellfunction,Science322 (2008)271–275.

[9]T.R.Simpson,F.Li,W.Montalvo-Ortiz,M.A.Sepulveda,K.Bergerhoff,F.Arce, etal.,Fc-dependentdepletionoftumor-infiltratingregulatoryTcellsco- definestheefficacyofanti-CTLA-4therapyagainstmelanoma,J.Exp.Med.

210(2013)1695–1710.

[10]C. Twyman-SaintVictor,A.J. Rech, A. Maity,R. Rengan,K.E.Pauken, E.

Stelekati,etal.,Radiationanddualcheckpointblockadeactivatenon- redundantimmunemechanismsincancer,Nature520(2015)373–377.

[11]D.M. Pardoll, The blockade of immune checkpoints in cancer immunotherapy,Nat.Rev.Cancer12(2012)252–264.

[12]W.Zou,J.D.Wolchok,L.Chen,PD-L1(B7-H1)andPD-1pathwayblockadefor cancertherapy:mechanisms,responsebiomarkers,andcombinations,Sci.

Transl.Med.8(2016)328rv4.

[13]J.Lau,J.Cheung,A.Navarro,S.Lianoglou,B.Haley,K.Totpal,etal.,Tumour andhostcellPD-L1isrequiredtomediatesuppressionofanti-tumour immunityinmice,Nat.Commun.8(2017)14572.

[14]E.Hui,J.Cheung,J.Zhu,X.Su,M.J.Taylor,H.A.Wallweber,etal.,Tcell costimulatoryreceptorCD28isaprimarytargetforPD-1-mediated inhibition,Science355(2017)1428–1433.

[15]A.O.Kamphorst,A.Wieland,T.Nasti,S.Yang,R.Zhang,D.L.Barber,etal., RescueofexhaustedCD8TcellsbyPD-1-targetedtherapiesisCD28- dependent,Science355(2017)1423–1427.

[16]F.S.Hodi,S.J.O’Day,D.F.McDermott,R.W.Weber,J.A.Sosman,J.B.Haanen, etal.,Improvedsurvivalwithipilimumabinpatientswithmetastatic melanoma,N.Engl.J.Med.363(2010)711–723.

[17]J.D.Wolchok,B.Neyns,G.Linette,S.Negrier,J.Lutzky,L.Thomas,etal., Ipilimumabmonotherapyinpatientswithpretreatedadvancedmelanoma:a randomised,double-blind,multicentre,phase2,dose-rangingstudy,Lancet Oncol.11(2010)155–164.

[18]C.Robert,J. Schachter,G.V. Long,A.Arance, J.J.Grob,L. Mortier,etal., Pembrolizumabversusipilimumabinadvancedmelanoma,N.Engl.J.Med.

372(2015)2521–2532.

[19]C. Robert,G.V. Long, B. Brady, C. Dutriaux,M.Maio, L. Mortier,et al., NivolumabinpreviouslyuntreatedmelanomawithoutBRAFmutation,N.

Engl.J.Med.372(2015)320–330.

[20]L.A.Lievense,D.H.Sterman,R.Cornelissen,J.G.Aerts,Checkpointblockadein lungcancerandmesothelioma,Am.J.Respir.Crit.CareMed.(2017)epub aheadofprint.

[21]P.Sharma,S.Hu-Lieskovan,J.A.Wargo,A.Ribas,Primary,adaptive,and acquiredresistancetocancerimmunotherapy,Cell168(2017)707–723.

[22]J.Larkin,V.Chiarion-Sileni,R.Gonzalez,J.J.Grob,C.L.Cowey,C.D.Lao,etal., Combinednivolumabandipilimumabormonotherapyinuntreated melanoma,N.Engl.J.Med.373(2015)23–34.

[23]J.Larkin,R.Gonzalez,P.Rutkowski,J.Grob,C.L.Cowey,C.D.Lao,etal.,Overall survival(OS)resultsfromaphaseIIItrialofnivolumab(NIVO)combined withipilimumab(IPI)intreatment-naïvepatientswithadvancedmelanoma (CheckMate067),AACR(2017)AbstractnrCT075.

[24]J.Brahmer,K.L.Reckamp,P.Baas,L.Crino,W.E.Eberhardt,E.Poddubskaya, etal.,Nivolumabversusdocetaxelinadvancedsquamous-cellnon-small-cell lungcancer,N.Engl.J.Med.373(2015)123–135.

[25]H.Borghaei,L.Paz-Ares,L.Horn,D.R.Spigel,M.Steins,N.E.Ready,etal., Nivolumabversusdocetaxelinadvancednonsquamousnon-small-celllung cancer,N.Engl.J.Med.373(2015)1627–1639.