The role of immune checkpoint inhibition in the treatment of ovarian cancer

14 Pages·2016·0.76 MB·English

by  Stéphanie L. Gaillard

Gaillardetal.GynecologicOncologyResearchandPractice (2016) 3:11
DOI10.1186/s40661-016-0033-6
REVIEW Open Access
The role of immune checkpoint inhibition
in the treatment of ovarian cancer
Stéphanie L. Gaillard1*, Angeles A. Secord2 and Bradley Monk3
Abstract
Theintroductionofimmunecheckpointinhibitorshasrevolutionizedtreatmentofmultiplecancersandhasbolstered
interest in this treatment approach. So far, emerging clinical data show limited clinical efficacy of these agents in
ovarian cancer with objective response rates of 10–15% with some durable responses. In this review, we present
emerging clinical data of completed trials of immune checkpoint inhibitors and review ongoing studies. In
addition we examine the current knowledge of the tumor microenvironment of ovarian cancers with a focus
on the significance of PD-L1 expression and tumor-infiltrating lymphocytes on predicting response to immune
checkpoint blockade. We evaluate approaches to improve treatment outcomes through the use of predictive
biomarkers and patient selection. Finally, we review management considerations including immune related
adverse events and response criteria.
Keywords: Ovarian cancer, Fallopian tube cancer, Primary peritoneal cancer, Immunotherapy, Immune
checkpoint inhibitors, PD-1, PD-L1, CTLA-4
Background andthepotentialroleforimmunotherapyinthetreatment
Roleofimmunecheckpointsanddevelopmentofimmune ofthisdisease.
checkpointinhibitors Immunotherapy refers to treatment designed to en-
Ovarian cancer is the most lethal of the gynecologic ma- hance an individual’s own immune function to eradicate
lignancies. Over 22,000 new cases of ovarian cancer are malignant cells. While there have been various ap-
diagnosed each year in the United States resulting in proaches, from cancer vaccines to adoptive immune cell
greater than 14,000 deaths per year [1]. The five year therapies, immune checkpoint inhibitors have caused a
survival rate is less than 25% for women diagnosed with paradigm shift in cancer treatment. These therapies are
advanced stage disease (stage III or IV) despite aggres- now FDA-approved for a variety of cancers including
sive treatment with surgery and adjuvant chemotherapy. melanoma, non-small cell lung cancer (NSCLC), renal
Although >80% of patients will have a response to initial cell carcinomas (RCC), bladder cancer, and classical
therapy, epithelial ovarian cancer ultimately recurs in Hodgkin lymphoma. The enthusiasm for this approach
the majority of patients. Recurrence is associated with a stems from evidence of complete and long-lasting tumor
poor prognosis because of the eventual development of regression in malignancies that are often refractory to
chemotherapy-resistant disease. Thus there is a great chemotherapy.
need, and opportunity, to improve ovarian cancer out- T-cell mediated cancer cell death requires the produc-
comes by understanding the immune milieu of ovarian tion of effector T-cells (T ) through the coordinated
eff
cancers and harnessing the power of immunotherapy. initiation of a multi-step process involving antigen pres-
This review will focus on the current understanding entation, priming and activation, T-cell trafficking and
of the immune microenvironment of ovarian cancers infiltration into the tumor, recognition of cancer cells,
and cancer cell elimination [2]. ThisT-cell mediated im-
mune response is regulated by a number of stimulatory
and inhibitory signals. Inhibitory signals serve to prevent
*Correspondence:[email protected]
1DepartmentofMedicine,DivisionofMedicalOncology,DukeCancer pathologic over-activation of the immune system, as an
Institute,200TrentDrive,Durham,NC27710,USA uncontrolled inflammatory response could result in the
Fulllistofauthorinformationisavailableattheendofthearticle
©TheAuthor(s).2016OpenAccessThisarticleisdistributedunderthetermsoftheCreativeCommonsAttribution4.0
InternationalLicense(http://creativecommons.org/licenses/by/4.0/),whichpermitsunrestricteduse,distribution,and
reproductioninanymedium,providedyougiveappropriatecredittotheoriginalauthor(s)andthesource,providealinkto
theCreativeCommonslicense,andindicateifchangesweremade.TheCreativeCommonsPublicDomainDedicationwaiver
(http://creativecommons.org/publicdomain/zero/1.0/)appliestothedatamadeavailableinthisarticle,unlessotherwisestated.
Gaillardetal.GynecologicOncologyResearchandPractice (2016) 3:11 Page2of14
development of autoimmune or inflammatory disorders. regulates T-cell priming and activation, activities that
However, inhibition of the T response against cancer occur in the early phases of the immune response. Inhib-
eff
cells contributes to immune evasion. These inhibitory itionofCTLA-4duringtheT-cellpriming/activationstep
signals may come from extrinsic sources, such as regula- leads to dysregulated expansion of auto-reactive T cells,
tory T-cells (T ) and inhibitory cytokines, or intrinsic including tumor-specific T-cells. Anti-CTLA-4 inhibitors
regs
sources, such as immune checkpoint proteins expressed havebeenassociatedwithsignificantimmune-relatedtox-
on the surface of T . It is the balance of these signals icities which are likely a result of the indiscriminate and
eff
that determines the success or failure of the immune unselectedactivationofauto-reactiveT-cells.
systemtoeliminate cancer cells. PD-1 is a cell surface receptor that is upregulated dur-
T play a critical role in the extrinsic suppression of ing normal T-cell activation and modulates the activity
regs
anti-tumor immunity. When T are the dominant T- of antigen-experienced effector T-cells. Interaction of
regs
cell population in the tumor microenvironment, they PD-1 with either of its two known ligands, PD-L1 and
inhibit tumor-antigen specific immunity and promote PD-L2, results in inhibition of T-cell signaling and cyto-
tumor growth. Depletion of these T can restore anti- kine production as well as decreased effector T-cell
regs
tumor immune activity. Similarly, other suppressive numbers due to limited T-cell proliferation and in-
immune cells [e.g. myeloid derived suppressor cells creased susceptibility to apoptosis. Of the two ligands,
(MDSCs), M2 macrophages] influence the balance of PD-L1 appears to be the more relevant in the tumor
regulatorysignals. microenvironment and is expressed on a wide range of
Immune checkpoint receptors, such as cytotoxic T tumor cells. Tumor-infiltrating lymphocytes can induce
lymphocyte-associated protein 4 (CTLA-4) and pro- PD-L1 expression on tumor cells leading to reduced
grammed cell death protein 1 (PD-1), have emerged as anti-tumor immunity [3]. PD-L1 expression may also be
critical intrinsic modulatory mechanisms impairing nat- regulated through gene amplification or via oncogenic
ural anti-neoplastic immunity (Fig. 1). These receptors signaling pathways [4–6]. Antibodies directed at either
are negative regulators which attenuate normal T-cell PD-1 or PD-L1 result in abrogation of the negative
activation to prevent pathologic over-activation. Interfer- signal, thusrestoringT-cellfunction.
ing with immune checkpoint signaling has been shown An important distinction between CTLA-4 and PD-1/
to enhance anti-tumor immune responses through the L1 inhibitors is their location of action [7]. Because
recovery of T-cell function. The CTLA-4 and PD-1 im- CTLA-4 regulates T-cell priming and activation, anti-
mune checkpoint proteins function at different points in CTLA-4 antibodies lead to activation of T-cells in
the process, which may explain their differential activ- lymphoidperipheraltissues.Anti-PD-1/L1 effectsappear
ities and toxicities. The CTLA-4 immune checkpoint to be limited to the tumor microenvironment without
Fig.1Costimulatory and coinhibitory pathways regulate the T-cell response to antigen. APC: antigen-presenting cell, CTLA-4: cytotoxic T
lymphocyte-associated protein 4, MHC: major histocompatibility complex, PD-1: programmed cell death protein 1; PD-L1: PD-1 ligand, TCR:
T-cell receptor
Gaillardetal.GynecologicOncologyResearchandPractice (2016) 3:11 Page3of14
evidence of recirculation.Thusa numberofpharmacody- on PD-L1 expression. Thus, other suppressive signals
namicmarkersforanti-CTLA-4activityhavebeenidenti- are likely present and inhibition of other checkpoint re-
fied in the peripheral blood, whereas no biomarkers of ceptors may be beneficial. Although this stratification
PD-1 activity have been isolated from peripheral blood system is based on studies in melanoma and presents
thusfar. several caveats (ref. [9]), it provides a framework for
understanding the tumor microenvironment and ra-
Evidenceforusingcheckpointinhibitorsinovariancancer tionale for the benefit of immune checkpoint inhibitors
Two central tenets have emerged to predict effective in ovarian cancer.
treatment with immune checkpoint inhibitors: 1) acces-
sibility of the tumor by effector immune cells and 2) Prognosticsignificanceofovariancancertumor
dominance of the immune checkpoint pathways as the microenvironment
mechanism suppressing anti-tumor immunity. The first Evidence of the importance of the local tumor immune
is frequently defined by the presence of tumor- microenvironment in ovarian cancer emerged in 2003
infiltrating lymphocytes or the ratio of effector immune when Zhang et al. showed that infiltration of treatment
cells [i.e. T , dendritic cells (DCs), M1 macrophages] to naïve tumors with T-cells was associated with a signifi-
eff
immune suppressive immune cells [i.e. T , myeloid cantly improved median progression free (22.4 vs
regs
derived suppressor cells (MDSCs), M2 macrophages]. 5.8 months, p<0.001) and overall survival (50.3 vs
The second principle is less well defined as no accurate 18.0 months, p<0.001) compared to tumors with no T-
biomarker has been identified, although multiple ap- cells present [10]. However, we have since learned that
proaches are being evaluated. Expression of PD-L1 on not only is presence of T cells important, but that the
tumor cells has been suggested as a predictive biomarker type of T-cell present influences outcomes. The propor-
to identify cancers that may be more responsive to PD-1/ tionofT inthetumor negatively impactsclinical out-
regs
PD-L1inhibitors[8].Basedonworkinitiallyperformedin comes and was a predictor of increased risk of death in
melanomas, tumors have been classified into 4 groups a multi-variate analysis [11, 12]. Multiple studies have
based on the presence of tumor-infiltrating lymphocytes since confirmed that the ratio of immune suppressive to
(TILs) and PD-L1 expression (Table 1) [3, 9]. Type I tu- effector immune infiltrates within ovarian tumors is as-
morsexhibit apatternofadaptive immuneresistanceand sociated with clinical outcome [13–17]. Immune re-
may be most likely to respond to immune checkpoint in- sponses to ovarian cancer appear to vary by histologic
hibitors. Conversely,Type II tumors show no discernable subtype with high-grade serous cancers most likely asso-
immune reaction and single agent checkpoint blockade is ciated with a prognostically favorable tumor-infiltrating
unlikely to be successful. Alternative approaches that in- lymphocyte response [18, 19]. Classification of different
clude methods to recruit effector immune populations to histologic subtypes of ovarian cancers based on TIL and
the tumor (e.g. vaccines), possibly in combination with PD-L1 revealed that type I patterns were more common
immune checkpoint inhibitors, are predicted to be in high-grade serous cancers while type IV patterns pre-
necessary. Type III tumors exhibit intrinsic expression of dominatedinotherhistologicsubtypes(Table2)[18].
PD-L1, possibly through oncogenic stimulation, with no
immune reactivity. This highlights that tumor expression ExpressionofPD-L1onovariancancercells
ofPD-L1alonecannotbeusedasanindicatorofpotential Hamanishi and colleagues first reported that high ex-
benefit of PD-1/L1 inhibition as without effector immune pression of PD-L1 onovariancancer cellswasassociated
cellsinthetumor,singleagentimmunecheckpointinhib- with poorer outcomes [20]. The 5-year survival rate for
itionisunlikelytobebeneficial.SimilartotypeIItumors, patients with high- versus low-expressing PD-L1 tumors
approaches to stimulate immune trafficking to the tumor was 52.6±7.7% versus 80.2±8.9%, p=0.016, respect-
willbenecessary.FinallyTypeIVtumorsdisplayapattern ively. PD-L2 expression was also associated with poorer
of tolerance to immune infiltration that is not dependent
Table2Classificationofovariancancersbytypeofimmune
microenvironment(basedonWebbetal.[18])
Table1Classificationoftumorsbasedonpresenceoftumor
%Totalforhistologicsubtype
infiltratinglymphocytes(TIL)andPD-L1expression(basedon
Histologicsubtype N TypeI TypeII TypeIII TypeIV
Tengetal.[9])
High-gradeserous 112 57.4 5.1 0 37.4
TypeI:Adaptiveimmuneresistance TypeII:Immunologicalignorance
TIL+ TIL- Low-gradeserous 11 0 9.1 0 90.9
PD-L1+ PD-L1-
Mucinous 30 26.7 16.7 0 56.7
TypeIII:Intrinsicinduction TypeIV:Tolerance
Endometrioid 125 22.4 14.4 1.6 61.6
TIL- TIL+
PD-L1+ PD-L1- Clearcell 129 16.2 30.2 0 53.5
Gaillardetal.GynecologicOncologyResearchandPractice (2016) 3:11 Page4of14
outcomes but was not statistically significant. High ex- expression may be a marker of a tumor poised to respond
pression of PD-L1 onovariancancer cellswasassociated toimmunestimulatoryeffectsofchemotherapyorperhaps
with reduced infiltration of cytotoxicT lymphocytes into because PD-L1 may suppress the activity of immune-
tumors suggesting that PD-L1 expression promotes an suppressiveimmune cells (i.e. T ), PD-L1 expression
regs
immunosuppressive microenvironment by inhibiting T- on immune cells could tip the balance towards a
cell infiltration [20]. Both PD-L1 expression and TIL more favorable immune microenvironment [24]. Thus
were independent prognostic factors, though PD-L1 evaluating PD-L1 expression on tumor cells in isola-
expression was inversely correlated with survival. In pre- tion is not sufficient to predict immune response and
clinical models, PD-L1 expression can be induced by efficacy of immune checkpoint blockade in ovarian
interferon-gamma (often produced by TILs) and admin- cancer.
istration of chemotherapy, suggesting a balance that
maintains an immune suppressive environment [21, Trialsofimmunecheckpointinhibitorsinovariancancer
22]. PD-1/L1 blockade causes regression of ovarian Several antibodies directed against PD-1, PD-L1, and
tumors in a syngeneic ovarian cancer mouse model CTLA-4 have been developed and are being tested clin-
further validating the importance of this regulatory ically inpatientswith ovarian cancer.Table 3reflectsthe
pathway [23]. latest data from studies that have reported outcomes.
PD-L1 expression is not limited to tumor cells and has Table4showsongoingovariancancertrialswithimmune
been reported on immune cells including antigen- checkpoint inhibitors as monotherapy or combined with
presentingcells,T-cells,andB-cells.Arecentstudyshowed otheragents.Theschemaforongoingorplannedphase3
thatPD-L1expressionispredominantlyexpressedbymac- studiesareshowninFig.2.
rophages in ovarian cancer rather than on the ovarian
cancer cells themselves; in this context, macrophage asso- Nivolumab
ciatedPD-L1expressionwasamarkeroffavorableprogno- Nivolumab is a fully humanized IgG4 monoclonal anti-
sis [18]. The differences between this study and the one body targeting the PD-1 receptor and is FDA approved
abovemaybeduetothedifferencesintheantibodiesused, for the treatment of melanoma, NSCLC, renal cell car-
butalsoreflectthedevelopingunderstandingofPD-L1ex- cinoma,andHodgkin’slymphoma.Astudyofnivolumab
pression and its prognostic role in ovarian cancer. PD-L1 in recurrent ovarian cancer was the first to be published
Table3Studiesofimmunecheckpointinhibitorsinovariancancerwithreportedresults
Immunotherapy Trialnumber Diseasestatus Phase N Results(N;duration) G3/4adverseevents Reference
agent(s)
Ipilimumab recurrentEOC, I 9 PR(1;35+mos.) diarrhea Hodietal.[50]
previouslytreated SD(3;1for6+mos.)
withGVAXvaccine
BMS-936559 NCT00729664 recurrentEOC I 17 6%PR(1;1.3+mos.) infusion-relatedreaction, Brahmeretal.[80]
(anti-PD-L1) 18%SD(3;6+mos.) adrenalinsufficiency
Nivolumab platinumresistant II 20 10%CR(2;11+mos.) lymphocytopenia, Hamanishietal.[25]
EOC 5%PR(1;11+mos.) hypoalbuminemia,
30%SD(6;1for11+mos.) elevatedALT,rash,fever,anemia
Pembrolizumab NCT02054806 recurrentEOC, Ib 26 4%CR(1;6+mos.) transaminitis Vargaetal.[26]
PD-L1positive 8%PR(2;6+mos.)
23%SD(8;2for6+mos.)
Ipilimumab NCT01611558 recurrentEOC II 40 10%BRR(4;NA) NA clinicaltrials.gov[27]
Avelumab NCT01772004 recurrentEOC Ib 124 10%PR(12;4for rash,edema,elevatedamylase/ Disisetal.[28]
6+mos.)44% lipase,arthritis,colitis,
SD(55;NA) hyperglycemia/DM
Durvalumab NCT02484404a recurrentEOC I/II 10 PR(1;11+mos.) Lymphopenia,anemia Leeetal.[29]
+Olaparib SD(7;4+mos.)
Durvalumab 4 PR(1;7mos.) Lymphopenia,anemia,nausea,
+Cediranib SD(2;1for6mos.) diarrhea,hypertension,PE,
pulmonaryhypertension,fatigue,
headache
Abbreviations:Nnumberofovariancancerpatientstreated,EOCepithelialovariancancer,CRcompleteresponse,PRpartialresponse,SDstabledisease,ALT
alanineaminotransferase,BRRbestresponserate(CR/PRstatusnotprovided),mos.months,NAnotavailable,DMdiabetesmellitus;PE,pulmonaryembolism
aAsofdatacut-offdate:May10,2016
Gaillardetal.GynecologicOncologyResearchandPractice (2016) 3:11 Page5of14
Table4Ongoingstudiesofimmunecheckpointinhibitorsinovariancancer
Phase Trialnumber Trial Diseasestatus Immunotherapy Concurrenttherapy
agent(s)
3 NCT02580058 AStudyOfAvelumabAloneOrInCombination recurrentplatinum Avelumab LiposomalDoxorubicin
WithPegylatedLiposomalDoxorubicinVersus resistant
PegylatedLiposomalDoxorubicinAloneIn
PatientsWithPlatinumResistant/Refractory
OvarianCancer(JAVELINOvarian200)
3 NCT02718417 AvelumabinPreviouslyUntreatedPatientsWith primary Avelumab CarboplatinPaclitaxel
EpithelialOvarianCancer(JAVELINOVARIAN100)
3 ENGOT-ov29- Arandomized,double-blinded,phaseIIIstudyof recurrentplatinum Atezolizumab Carboplatin-based
GCIG atezolizumabversusplaceboinpatientswithlate sensitive chemotherapy
relapseofepithelialovarian,fallopiantube,or Bevacizumab
peritonealcancertreatedbyplatinum-based
chemotherapyandbevacizumab
2 NCT02440425 DoseDensePaclitaxelWithPembrolizumab recurrentplatinum Pembrolizumab DoseDensePaclitaxel
(MK-3475)inPlatinumResistantOvarianCancer resistant
2 NCT02498600 NivolumabWithorWithoutIpilimumabinTreating recurrentplatinum Nivolumab+/-
PatientsWithPersistentorRecurrentEpithelial sensitive/resistant Ipilimumab
Ovarian,PrimaryPeritoneal,orFallopianTubeCancer
2 NCT02520154 PembrolizumabinCombinationWithChemotherapy primary Pembrolizumab CarboplatinPaclitaxel
inFrontlineOvarianCancer
2 NCT02659384 Anti-programmedCellDeath-1Ligand1(aPDL-1) recurrentplatinum Atezolizumab Bevacizumab
AntibodyAtezolizumab,Bevacizumaband resistant AcetylsalicylicAcid
AcetylsalicylicAcidinRecurrentPlatinumResistant
OvarianCancer
2 NCT02674061 EfficacyandSafetyStudyofPembrolizumab recurrentplatinum Pembrolizumab
(MK-3475)inWomenWithAdvancedRecurrent sensitive/resistant
OvarianCancer(MK-3475-100/KEYNOTE-100)
2 NCT02764333 TPIV200/huFR-1(AMulti-EpitopeAnti-Folate recurrentplatinum Durvalumab TPIV200/huFR-1(anti-
ReceptorVaccine)PlusAnti-PD-L1MEDI4736 resistant folatereceptorvaccine)
(Durvalumab)inPatientsWithPlatinum
ResistantOvarianCancer
2 NCT02766582 PhaseII:Pembrolizumab/Carboplatin/Taxolin suboptimally Pembrolizumab CarboplatinPaclitaxel
EpithelialOvaryCancer cytoreducedprimary
1/2 NCT02431559 APhase1/2StudyofMotolimod(VTX-2337)and recurrentplatinum Durvalumab MotolimodPegylated
MEDI4736inSubjectsWithRecurrent,Platinum- resistant LiposomalDoxorubicin
ResistantOvarianCancerforWhomPegylated
LiposomalDoxorubicin(PLD)isIndicated
1/2 NCT02484404 Phase1and2StudyofMEDI4736inCombination recurrentplatinum Durvalumab OlapariborCediranib
WithOlapariborCediranibforAdvancedSolid sensitive/resistant
TumorsandRecurrentOvarianCancer
1/2 NCT02485990 StudyofTremelimumabAloneorCombinedWith recurrentorpersistent Tremelimumab Olaparib
OlaparibforPatientsWithPersistentEOC(Epithelial
Ovarian,FallopianTubeorPrimaryPeritonealCarcinoma)
1/2 NCT02571725 PARP-inhibitionandCTLA-4BlockadeinBRCA-deficient BRCA-deficient Tremelimumab Olaparib
OvarianCancer recurrentplatinum
sensitive/resistant
1/2 NCT02657889 StudyofNiraparibinCombinationWithPembrolizumab recurrentplatinum Pembrolizumab Niraparib
(MK-3475)inPatientsWithTriple-negativeBreastCancer resistant
orOvarianCancer(KEYNOTE-162)
1/2 NCT02726997 MatchedPairedPharmacodynamicsandFeasibility primary Durvalumab CarboplatinPaclitaxel
StudyofDurvalumabinCombinationWith
ChemotherapyinFrontlineOvarianCancer
1 NCT02737787 AStudyofWT1VaccineandNivolumabForRecurrent ≥2ndremission Nivolumab WT1vaccine
OvarianCancer
0 NCT02728830 AStudyofPembrolizumabontheTumoral primary Pembrolizumab
ImmunoprofileofGynecologicCancers
Gaillardetal.GynecologicOncologyResearchandPractice (2016) 3:11 Page6of14
Fig.2Ongoing or planned phase 3 trials in ovarian cancer with immune checkpoint inhibitors. a NCT02718417: Javelin Ovarian 100.
b ENGOT-ov29-GCIG: ATALANTE. c NCT02580058: Javelin Ovarian 200. d NRG-GY009
for this patient population [25]. In this study 20 patients phase Ib study (KEYNOTE-028, NCT02054806) was
with platinum-resistant ovarian cancer were treated in 2 conducted of single-agent pembrolizumab in ovarian
cohorts either with 1 or 3 mg/kg nivolumab every cancer patients [26]. Eligibility requirements included
2weeksuntilprogressionor upto48 weeks.Best overall expression of PD-L1 in 1% of tumor nests or PD-L1 ex-
response was the primary endpoint. Grade 3 or 4 ad- pression in stroma. Pembrolizumab 10 mg/kg was given
verse events occurred in 8 patients (20%) and two expe- every 2 weeks until progression, intolerable adverse ef-
rienced severe adverse events (grade 3 disorientation, fects or for up to 2 years. Twenty-six patients were
gait disorder, fever in 1 patient and grade 3 fever, deep treated. Objective response rate was 11.5% with 1 CR, 2
venous thrombosis in the other). The best overall re- partial responses (PR), and 23% stable disease (SD).
sponse was 15%. Four patients experienced prolonged Durable responses were noted and the median time to
disease control (2 patients in each dose cohort) with 2 responsewas8weeks.
patients in the 3 mg/kg cohort experiencing a durable
complete response (CR). While response rates were Ipilimumab
similar to what has been seen with chemotherapy in Ipilimumab is a recombinant, IgG1 human monoclonal
platinum resistant disease, the durable responses are antibody targeting CTLA-4 that is FDA-approved for
atypical in this disease and a cause for enthusiasm par- the treatment of melanoma. In a phase II study of ipili-
ticularly in a very heavily pre-treated population. PD-L1 mumab monotherapy in recurrent platinum-sensitive
expression did not significantly correlate with objective ovarian cancer (NCT01611558), 40 patients were treated
response. Fourteen of 16 patients with PD-L1 high ex- with 10 mg/kg ipilimumab every 3 weeks x 4 doses (in-
pression did not show a response while 1 of 4 patients duction phase) followed by 10 mg/kg every 12 weeks
with lowexpression wasaresponder. until progression orunacceptabletoxicity [27].Ofthe 40
who started the study, 38 (95%) did not complete the in-
Pembrolizumab duction phase because of disease progression (14, 35%),
Pembrolizumab is an anti-PD-1 humanized IgG4 mono- drug toxicity (17, 42.5%), death (1, 2.5%), or other/unre-
clonal antibody FDA-approved for the treatment of ported (6, 15%). Twenty patients (50%) experienced
melanoma and NSCLC. A non-randomized, multicohort drug-related adverse events of grade 3 or higher. The
Gaillardetal.GynecologicOncologyResearchandPractice (2016) 3:11 Page7of14
objectiveresponserate(ORR)was10.3%[95%confidence Hodgkin’s lymphoma where >65% of patients had a re-
interval(CI)2.9to34.2%]byRECISTcriteria.Ofnote,the sponse to treatment and 17–21% achieved a complete
10 mg/kg dose is higher than the FDA approved dose for response [30, 31], but more consistent with response
the treatment of unresectable or metastatic melanoma rates in previously treated patients with melanoma
(3mg/kg) but is equivalent to the dose used for the adju- (28%), NSCLC (18%), and renal cell carcinoma (27%)
vanttreatmentofmelanoma. [32]. The phase Ib avelumab study suggests that ovarian
cancer patients who have been treated with fewer prior
Avelumab courses of chemotherapy may have a greater benefit
Avelumab is a fully humanized monoclonal anti-PD-L1 from these agents, thus chemotherapy may induceT-cell
IgG1 antibody that does not block PD-1 interaction with exhaustion or have other irreversible immunosuppres-
PD-L2.InaPhaseIb(NCT01772004,Javelinsolidtumor sive effects[33].
study) [28], 124 patients with refractory or recurrent Combining the nivolumab and avelumab studies cited
ovarian cancer (progression within 6 months, or after above, it is notable that 4 of the 5 patients who experi-
2nd/3rd line treatment) were treated with 10 mg/kg every enced durable responses had tumors with clear cell hist-
2 weeks until progression or unacceptable toxicity. The ology [25, 28]. Histology of responding patients on other
median duration of treatment was 12 weeks. Grade 3/4 studies has not been reported. These observations run
adverse events occurred in 6.4% of patients and 8.1% of counter to the prediction that ovarian cancers with clear
patients discontinued treatment secondary to an adverse cell histology would be less likely to respond to PD-1
event. Twelve patients experienced a partial response for inhibitorsbasedonlowPD-1expressionandlowTILin-
an ORR of 9.7%. Disease control rate (DCR, defined as filtration (Table 2). However, they are particularly intri-
ORR+SD) was 54%. ORR was 12.3% in PD-L1+ tumors guing given the characteristic chemorefractory nature of
and 5.9% in PD-L1- tumors (based on>=1% threshold). ovarian clear cell carcinomas (OCCC) [34, 35]. Because
Differences in median PFS and OS were not statistically OCCC are genomically remarkably similar to RCC, it
significantbasedonPD-L1expression.Therearecurrently has been postulated that therapies effective for RCC may
two Phase 3 trials of avelumab for ovarian cancer; one for be similarly effective for OCCC [36]. Nivolumab was re-
front-line therapy in combination with carboplatin and centlyapprovedforthetreatmentofadvancedRCCbased
paclitaxel (Javelinovarian100)and the otherforrecurrent onthe phaseIII CheckMate025 trial inwhichnivolumab
platinum-resistantdisease(Javelinovarian200)(Fig.2). showed an ORR of 25% and a 5 month OS benefit over
everolimus (ORR 5%) [37]. Whether immune checkpoint
Durvalumab inhibitors will result in such dramatic benefits in OCCC
Durvalumab is an Fc optimized IgG1 monoclonal anti- remainstobedeterminedinlargercohorts.
body directed against PD-L1,recently given breakthrough
therapy designation by the FDA for PD-L1 positive Opportunitiestoimproveoutcomeswithimmune
urothelial bladder cancer. In an ongoing phase I/II study checkpointinhibitors
of durvalumab (NCT02484404) in combination with ei- Identificationofpredictivebiomarkers
therthePARPinhibitor,olaparib,or theVEGFRinhibitor, A critical need in this field is the development of bio-
cediranib, there was 1 PR in 9 evaluable ovarian cancer markers that can predict response to therapy, provide
patientslasting>6monthswiththecombinationofdurva- early indication of efficacy, and warn of the development
lumabandolapariband1PRin5evaluableovariancancer of adverse effects. The most promising have focused on
patientstreatedwithdurvalumabandcediranib[29]. prediction of response to PD-1/L1 therapy. Indications
in melanoma trials that tumor PD-L1 expression, density
Otherimmunecheckpointinhibitors ofTILs,andproportionofTcellsexpressingPD-1orPD-
Atezolizumab is an Fc-engineered, humanized, non- L1 was associated with response led to the categorization
glycosylated IgG1 kappa monoclonal antibody targeting schema outline above in an attempt to identify subsets of
PD-L1thatisFDA-approvedforthetreatmentofbladder/ melanoma patients who would be most likely to respond
urothelial carcinomas. Tremelilumab is a fully humanized totreatment (Table 1)[3,9]. Further validationisstill ne-
antibody against CTLA-4. To date no studies have re- cessary to determine whether this categorization predicts
ported outcomes for patients with ovarian cancer treated better outcomes. Individually none of these factors is a
withatezolizumabortremelilumab. reliablepredictorofresponse.
Although cross trial comparisonsarenot feasiblegiven
the early stage ofdevelopment and different trial eligibil- PD-L1expressionandTILs
ity parameters, it is remarkable that all of the studies so Several studies of anti-PD-1/L1 therapeutic antibodies in
far have similar ORR (10–15%). This is markedly lower multiple tumor types, including melanoma and NSCLC,
than was seen in early trials of PD-1 inhibitors for have suggested that PD-L1 expression is associated with
Gaillardetal.GynecologicOncologyResearchandPractice (2016) 3:11 Page8of14
agreaterlikelihoodof benefit[8,32,38–40].Thesestud- has not yet been standardized. Additionally, there are
ies typically categorized tumors as PD-L1 positive if at challenges associated with the use of different anti-
least 5% of tumor cells showed cell-surface PD-L1 stain- bodies, fixation and staining techniques, and the subject-
ing. While initial studies suggested that PD-L1 negative ive interpretation of staining thresholds. Added to this
tumorsdidnotshowresponse[32,38],subsequentstudies complexity is defining the importance of the subset of
in multiple tumor types have shown objective responses cells (immune versus tumor) upon which PD-L1 is
in up to 20% of PD-L1 negative tumors [39, 41, 42]. In expressed. At this point, there is no indication that PD-
comparison, the phase 2 nivolumab study in ovarian L1 expression or TIL by IHC should be used as an abso-
cancer patients showed that only 2 of 16 patients with luteselectioncriterion fortherapy.
high PD-L1 expression showed a response, while 1 of 4
patients with low expression responded [25]. Similarly, Mutationalload
the avelumab study showed that even with a staining Genetic alterations within a tumor (including mutations,
cut-off level of ≥1% of tumor cells in ovarian cancer, 1 DNA rearrangements,deletions,andinsertions)havethe
of 17 patients with a PD-L1 negative tumor showed an potential to generate neo-antigens which are associated
objective response [28]. Thus, it is unclear whether PD- with clinical response to immune checkpoint therapies.
L1 can be used reliably as a predictive biomarker for Tumors with higher mutational loads, such as melan-
anti-PD-1/L1 directed therapy. By contrast, PD-L1 ex- oma, NSCLC, and bladder cancer, have shown the great-
pression status does not appear to influence response est response rates to anti-PD-1/PD-L1 therapy, while
to anti-CTLA-4 therapy. In a study of previously un- cancers with relatively low mutation rates (pancreatic
treated melanoma, median PFS (mPFS) in response to and prostate cancers) have shown low response to these
ipilimumab was unaffected by PD-L1 status (PD-L1 therapies [51]. In melanoma, mutational load was associ-
positive 3.9 months, 95% CI 2.8 to 4.2 months versus ated with the degree of clinical benefit to CTLA-4 block-
PD-L1 negative 2.8 months, 95% CI 2.8 to 3.1 months) ade and pembrolizumab [52–54]. However, there are
while response to nivolumab was influenced by PD-L1 patients withlow mutationalburdenwho have responded
status (14.0 months, 95% CI 9.1 to not reached versus and those with highmutational burden that have not and
5.3 months, 95% CI 2.8 to 7.1 in PD-L1 positive versus nospecificcut-offpointcouldbedeterminedunderwhich
PD-L1 negative tumors, respectively) [43]. patientswouldnotderivebenefit[54].
Other attempts at identifying predictive biomarkers Mismatched repair (MMR) deficiency, defined by de-
have focused on T-cell infiltration. In melanoma, T-cell fects in one or more of 6 genes involved in the DNA
density, particularly at the invasive tumor border, has mismatch repair complex, results in 10–100 fold in-
been associated with response to anti-PD-1 therapy, creases in tumor mutational burden compared to MMR-
however tumors with low Tcell density have also shown competent tumors. MSI tumors express high levels of
response [44]. However, a separate study evaluating fac- multiple immune checkpoint molecules including PD-1,
tors associated with response to anti-PD-1 therapy in PD-L1, CTLA-4 and lymphocyte activation gene 3
multiple solid tumors treated on a phase I clinical trial (LAG3) [55]. Le et al. showed that MSI-high status in
showed that the presence of TILs did not correlate with colorectal cancer and mismatch-repair deficient non-
clinical outcomes [8]. Response to anti-CTLA-4 therapy colorectal cancers was able to predict clinical response
has been associated with a more inflamed tumor micro- to pembrolizumab in a phase 2 trial [56, 57]. There was
environment and potential markers include increased an ORR of 48% across tumor histologies with 12 month
expression of the activation marker, inducible T-cell co- OS and PFS rates of 79 and 54%, respectively. Germline
stimulator (ICOS), on peripheral blood CD4+ cells and MMR gene inactivation only occurs in ~2% of ovarian
tumor-infiltrating lymphocytes, and an increase in Teff:- cancers,howeversomaticlossofexpressioncan occur in
Treg cell ratio in tumor tissues [45–50]. However no up to 29% of ovarian cancers [58]. Whether microsatel-
validated predictive biomarkers for CTLA-4 therapy are lite instability status (MSI) status may be a predictive
yetavailable. biomarker to identify genetic subsets of ovarian tumors
Several factors may account for the difficulty in using with an improved likelihood of response to immune
receptor expression and T-cell populations as predictive checkpoint inhibition remains tobedetermined.
biomarkers. PD-L1 expression and T-cell infiltration are In non-MMR deficient ovarian cancers, the predomin-
dynamic processes soevaluation of tissue archived at the ant genetic abnormality is copy number alteration and
time of surgery may not reflect the level of expression at mutation rate is generally low [59]. Despite relatively
the time of recurrence or planned treatment. Small low mutation burdens compared to other cancers, in-
tumor specimens may miss focal expression of PD-L1 or creased neo-antigen presentation may result from other
T-cells localized only at the leading edge of the tumor. genetic alterations. Patients with BRCA-associated tu-
Scoring of immunohistochemistry for PD-L1 and TILs mors may be more likely to have a higher burden of
Gaillardetal.GynecologicOncologyResearchandPractice (2016) 3:11 Page9of14
genetic alterations (copy number alterations, deletions, Similarly, combining immune checkpoint inhibitors
amplifications) given the role of BRCA in homologous with other anti-neoplastic treatments to enhance thera-
recombination DNA repair [60]. BRCA1-associated peutic outcomes is an active area of investigation (see
ovarian cancers have also been associated with increased Table 4 for trials for ovarian cancer). Chemotherapy,
intra-tumoral T-cell infiltration [61]. Thus it has been radiotherapy, tyrosine kinase inhibitors, and epigenetic
suggested that these patients may derive greater benefit modulators may be synergistic adjuncts to immunother-
from immune checkpointinhibitors. Contrarytothis hy- apy through their ability to increase tumor immunogen-
pothesis, no responses were seen in patients with a icity [71–74]. A particular area of interest for ovarian
BRCA mutation in the avelumab phase Ib study (ORR cancer is the combination of immune checkpoint inhibi-
16% in BRCA-wildtype tumors), DCR was 11.1% in tors with anti-angiogenic agents and/or PARP inhibitors.
BRCA-mutation carriers versus 48.0% in BRCA-wildtype Both anti-angiogenic agents and PARP inhibitors influ-
[28]. Thus at this time, thereis noreason to suggest that ence the ovarian cancer immune microenvironment in in
immunotherapy trials should be limited to this patient vivomodelsandcombinationtherapyissupportedbypre-
population,beyondthestated objectiveofthestudy. clinical studies [75–78]. Clinical trials of combination
therapyareongoing,sofarphaseIresultsofdurvalumab/
Functionalassays cediranibanddurvalumab/olaparib(NCT02484404)show
Other approaches to developing predictive biomarkers the combinations are feasible. It will be necessary to
include assessing functional capacity of the immune develop individualized approaches to determine which
cells within the tumor microenvironment. These ap- treatment strategy is most likely to be effective for in-
proaches include intracellular cytokine staining to dividual patients.
measure interferon-gamma signaling and T-cell poly-
functionality, measurement of local inhibitory cytokine Managementconsiderationswithimmunecheckpoint
production (IL-10, TGF-beta), measurement of T-cell inhibitors
activation or proliferation potential, and T-cell clonal- Immune-relatedtoxicities
ity/repertoire [62–69]. However, as of yet, none of Related to their mechanism of action of impairing T-cell
these approaches has been validated to be predictive of inhibition, immune checkpoint inhibitors can cause a
response to therapy. loss of self-tolerance and thus the development of
immune-related adverse effects (irAEs). While the fre-
Combinatorialtherapytoimprovetherapeuticoutcomes quency of irAEs with these agents is common (~60%
Improved therapeutic efficacy has been demonstrated with anti-CTLA-4 therapy and 40% with anti-PD-1/L1
with the combination of anti-PD-1 and anti-CTLA-4 therapy), in general serious toxicity [grade 3–5 using the
inhibitors. In previously untreated melanoma patients, Common Terminology Criteria for Adverse Events
median PFS was 11.5 months (95% CI 8.9–16.7 months) (CTCAE)] is more likely with anti-CTLA-4 therapy
with nivolumab and ipilimumab compared with (>40%)thanwithanti-PD-1/L1therapy(~5%)[32,79,80].
2.9 months (95% CI 2.8–3.4) with ipilimumab and While the immune side effects could involve any organ
6.9 months (95% CI 4.3 to 9.5) with nivolumab alone system, the most common irAEs with both anti-CTLA-4
[43]. Interestingly, in patients whose tumors did not ex- and anti-PD-1/L1 therapy are dermatologic (rash, pruri-
press PD-1, median PFS was improved in patients re- tis), gastrointestinal (diarrhea), rheumatologic (arthralgia,
ceiving both drugs compared to those who received arthritis, myalgia, myositis), endocrine disorders (thyroid-
nivolumab alone; there was no difference in median PFS itis, hypothyroidism, hypophysitis), and infusion-related
between these two treatment groups in patients with reactions. Serious gastrointestinal toxicities, such as
PD-1 expressing tumors. The frequency of treatment re- immune-mediated colitis and hepatitis, are more likely
lated toxicities was also increased with combination with anti-CTLA-4 therapy. Involvement of the following
therapy; 55% of patients experienced grade 3/4 events in organ systems is much less common, especially with
the nivolumab and ipilimumab group, 16.3% in the nivo- anti-PD-1/L1 pathways inhibitors, but have been ob-
lumab group, and 27.3% in the ipilimumab group. In served including: pulmonary (pneumonitis, sarcoidosis),
mouse models of ovarian cancer, 1/3 to 1/2 of TILs hematologic (hemolytic anemia, aplastic anemia, neutro-
coexpressed PD-1 and CTLA-4 [70]. This subpopulation penia), ocular (uveitis, conjunctivitis), cardiac (myocardi-
exhibited poor effector functions with diminished cap- tis, pericarditis), neurologic (myasthenia gravis, Guillan
acity to secrete effector cytokines and proliferate. Dual Barre, Bell’s palsy, posterior reversible leukoencephalopa-
blockade of PD-1 and CTLA-4 increased T-cell activity thy, and aseptic meningitis). While irAEs associated with
and tumor regression. This treatment strategy is cur- these therapies are generally reversible, without prompt
rently being evaluated for recurrent ovarian cancer in management they can evolve into life-threatening condi-
theNRGOncologystudyGY003(NCT02498600). tions.Earlyrecognitionofthedevelopmentoftheseevents
Gaillardetal.GynecologicOncologyResearchandPractice (2016) 3:11 Page10of14
is critical to prevention of the progression to severe ad- evaluated the effect of initiation of immunomodulatory
verseeffects.Vigilance,bothonthepartofthepatientand agents for ipilimumab irAEs in melanoma patients [84].
thetreatingphysician,isnecessarytoidentifysubtledevel- Of the 298 patients, 85% experienced an irAE of any
opingsignsofirAEsandspecialtyconsultationmaybene- grade, 35% required corticosteroid treatment, and 10%
cessary when irAEs are suspected but not definitive. required anti-TNFalpha therapy. Overall survival and
Immune-relatedAEscandevelopatanytimeduringtreat- time to treatment failure (median 5.7 months) was not
ment and even after discontinuation of therapy. However, affected by the occurrence of irAEs or the use of sys-
theonsetofirAEsfollowsacharacteristicpatternofdevel- temic corticosteroids. Continued anti-tumor activity has
opment. For the CTLA-4 inhibitor, ipilimumab, most been observed in patients treated with high-dose ste-
irAEs develop during the initial induction period (usually roids for irAEs [85]. However,data on the effects ofster-
4 doses given every 3 weeks). Dermatologic reactions oid use on immune checkpoint inhibitor efficacy are still
more commonly occur early during treatment, frequently limited. Thus, it is recommended to avoid prophylactic
during the first few weeks, while diarrhea and colitis steroids andlimittherapeuticsteroids asneeded.
develop later. Endocrine disorders may be late effects,
frequently developing between 7 and 20 weeks after Measurementofresponse:specificimmunerelatedcriteria
initiation of treatment and sometimes being identified One of the challenges faced in assessing the therapeutic
after discontinuation of treatment [81]. Combining value of these agents is determining the most appropri-
CTLA-4 inhibitor and PD-1 inhibitors, while signifi- atemeasurementofefficacy.Whileresponserates toim-
cantly increasing response also results in substantially mune checkpoint inhibitors as single agents is relatively
more irAEs. low, the impressive duration in responding patients
Spain et al. recently provided a substantive review of suggests that overall survival may be a better measure of
the management of irAEs [82]. Briefly, management is efficacy. In fact, there have been patients who achieve
dependent on the severity of the event, usually graded long-term survival benefit without evidence of clinical
using CTCAE. All references to grade in this article will response [53]. In addition, early studies noted that some
be using CTCAE version 4.0 [83]. Typically grade 1 patients had responses after initial apparent progression
symptoms can be monitored and may not require inter- ofdisease,while others showed amixed responseornew
ruption of therapy. For Grade 2 symptoms, the immune lesions despite an overall decrease in tumor burden.
checkpoint inhibitor therapy should be withheld until Since these response patterns were not adequately
symptoms improve and treatment with an immunomod- captured by RECIST1.1, new immune-related response
ulatory medication may be considered. Immune-related criteria (irRC) were developed to specifically accommo-
AEs with a higher risk of resulting in serious organ dys- date the response patterns seen after treatment with im-
function(such as colitis, hepatitis, pneumonitis, nephritis, mune checkpoint inhibitors [86]. Unlike in RECIST1.1,
neurologic symptoms) likely warrant initiation of cortico- new lesions do not automatically signal progression and
steroid treatment early (Grade 2) rather than waiting for apparent progressive disease must be confirmed 4 weeks
symptoms to worsen. When the severity of the irAE war- after initial assessment to qualify for true progression. In
rantsthereversalofinflammation(≥Grade3),corticoste- melanoma patients treated with either ipilimumab or
roids are the first immunomodulatory medication to be pembrolizumab,‘pseudoprogression’ occurred in~7% of
administered. Careful and timely monitoring of response patients and has been attributed to peritumoral lympho-
tosteroidtherapyisrequiredtoidentifysteroid-refractory cyte infiltration or delayed immune activity [87–89].
cases. In these situations, the use of more potent im- RECIST1.1 was noted to underestimate the benefit of
munomodulatory agents may be necessary such as the pembrolizumab in this population by up to 15% [88].
anti-TNF-alpha antibody infliximab, the anti-metabolite However, because this phenomenon occurs relatively in-
mycophenylate mofetil, anti-thymocyte globulin, and/or frequently, many studies continue to use RECIST1.1. In
calcineurin inhibitors (i.e. tacrolimus and cyclosporine). order to adequately assess efficacy across studies, it will
Consultation with, and if indicated hospitalization at, a be necessary to harmonize response assessments across
centerfamiliarwithsteroid-refractoryimmunecheckpoint studies and identify more refined radiographic or bio-
inhibitorirAEsisadvised. logic markers of early efficacy. In addition, the develop-
ment of other immune-specific clinical trial endpoints
Steroiduseduringtreatment may be necessary to account for prolonged duration of
Because of the effect of steroids on inhibiting T-cell responseafter initialprogression [90].
activation, patients receiving supraphysiologic doses of Rates of pseudoprogression in ovarian cancer have not
corticosteroids have generally been excluded from trials been reported, but trials to date suggest it occurs less
of immune checkpoint inhibitors. In the most compre- frequently than in melanoma [25, 26, 28]. Thus, in con-
hensive retrospective review on the subject, Horvat et al. trast to the management of melanoma, progression by