Reviews

Current state of immunotherapy

for glioblastoma
Michael Lim1*, Yuanxuan Xia1, Chetan Bettegowda1 and Michael Weller2
Abstract | Glioma is the most common primary cancer of the central nervous system, and around
50% of patients present with the most aggressive form of the disease, glioblastoma. Conventional
therapies, including surgery, radiotherapy, and pharmacotherapy (typically chemotherapy with
temozolomide), have not resulted in major improvements in the survival outcomes of patients
with glioblastoma. Reasons for this lack of progress include invasive tumour growth in an essential
organ, which limits the utility of local therapy, as well as the protection of tumour cells by the
blood–brain barrier, their intrinsic resistance to the induction of cell death, and lack of
dependence on single, targetable oncogenic pathways, all of which impose challenges for
systemic therapy. Furthermore, the unique immune environment of the central nervous system
needs to be considered when pursuing immune-​based therapeutic approaches for glioblastoma.
Nevertheless, a range of different immunotherapies are currently being actively investigated in
patients with this disease, spurred on by advances in immuno-​oncology for other tumour types.
Herein, we examine the current state of immunotherapy for gliomas, notably glioblastoma, the
implications for combining the current standard-​of-care treatment modalities with
immunotherapies, potential biomarkers of response, and future directions for glioblastoma
immuno-​oncology.

Gliomas are intrinsic brain tumours thought to origi­ correlate with the natural disease course. These tumours
nate from neuroglial progenitor cells on the basis of their are further classified based on the absence or presence of
localization, morphological similarities to nonmalignant mutations in IDH1 (which encodes isocitrate dehydro­
neuroglial cells, and generation of experimental gliomas genase [NADP] cytoplasmic) or IDH2 (which encodes
by targeted manipulation of certain brain cells1–3. This isocitrate dehydrogenase [NADP], mitochondrial) and
disease entity encompasses a diverse range of central the absence or presence of 1p and 19q (1p/19q) chro­
nervous system (CNS) cancers, including astrocytomas, mosomal co-​deletion4,6. This classification results in the
oligodendrogliomas, and ependymomas — historical categorization of adulthood gliomas into three main
designations that reflect the putative tissue of origin4. groups: IDH-​mutant, 1p/19q-​co-deleted tumours with
Some of these tumours typically occur in childhood, a predominantly oligodendroglial morphology and a
such as pilocytic astrocytoma or subgroups of ependy­ favourable prognosis; IDH-​mutant, non-1p/19q-​co-
momas, whereas others, including glioblastoma, have a deleted tumours, which usually have an astrocytoma
peak incidence in the seventh decade of life5. Selected morphology and intermediate survival outcomes; and
subtypes of gliomas can be cured using surgery alone, IDH-​wild-type tumours that are mostly glioblastomas
but the largest group, diffuse gliomas of adulthood, is (WHO grade IV), exhibit gain of chromosome 7 and
generally resistant to all current standard-​of-care thera­ loss of chromosome 10, and have an unfavourable prog­
1
Department of peutic interventions (surgery, radiotherapy, and systemic nosis3. In fact, most glioblastomas are IDH wild type,
Neurosurgery, Johns Hopkins chemotherapy); glioblastoma, the most aggressive vari­ including the histologically defined subtypes giant-​cell
University School of Medicine, ant, is invariably lethal. The overall annual incidence of glioblastoma, gliosarcoma, and epithelioid glioblastoma.
Baltimore, MD, USA.
gliomas in the USA is ~6 cases per 100,000 individuals, By contrast, IDH-​mutant glioblastomas are now con­
2
Department of Neurology, with glioblastoma accounting for ~50% of cases, and the sidered a distinct disease entity related to IDH-​mutant
University Hospital and
University of Zurich, Zurich,
disease has a male predominance5. WHO II and III gliomas, and most of the following
Switzerland. According to the revised 2016 WHO classification of discussion relates mainly to classical IDH-​wild-type
*e-​mail: mlim3@jhmi.edu CNS tumours4, diffuse oligodendroglial and astrocytic glioblastoma.
https://doi.org/10.1038/ gliomas of adulthood are graded from grades II to IV In 2017, the European Association for Neuro-​
s41571-018-0003-5 on the basis of histological features that are known to Oncology published updated recommendations for

422 | JULY 2018 | volume 15 www.nature.com/nrclinonc

© 2018 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
 Reviews

Key points
alteration results in decreased expression of MGMT,
which is a DNA-​repair protein that limits the cyto­
• The current standard of care for patients with glioblastoma includes surgery, toxic activity of TMZ. Numerous failed efforts at fur­
temozolomide chemotherapy, radiotherapy, and corticosteroids, all of which have ther improving the clinical outcomes achieved with the
immunosuppressive effects; we must be cognizant of this complexity when TMZ/RT → TMZ regimen involved the addition of var­
developing immunotherapies.
ious antiangiogenic compounds, notably bevacizumab
• Evidence for immunostimulatory effects of these treatments in the clinic, including (an anti-​VEGF antibody), but also the αv integrin antag­
abscopal effects, induction of immunogenic cell death, and depletion of regulatory
onist cilengitide16–18. Tumour-​treating fields, also known
T cells by temozolomide, remains limited.
as alternating electric field therapy, is a novel approach
• Vaccination has been considered one of the most promising approaches to improving
to the treatment of glioblastoma and other cancers. The
the outcomes of patients with glioblastoma, although negative results from several
phase II and phase III trials challenge the current concept of vaccination as a single-​
addition of this modality to the TMZ/RT → TMZ regi­
modality immunotherapy. men (during the TMZ maintenance period) resulted in
• Oncolytic viruses might exert pro-​inflammatory responses that could potentially be
improved progression-​free survival (PFS) and overall
exploited in future combined modality immunotherapy studies, whereas the future of survival in patients with newly diagnosed glioblastoma
chimeric antigen receptor (CAR) T cell therapy for glioblastoma depends on the in an open-​label phase III trial19. However, whether and
identification of stably expressed and sufficiently tumour-​specific antigens. how this treatment should be integrated into the cur­
• Immune-​checkpoint inhibitors have promising therapeutic activity in preclinical rent standard of care remains controversial: this treat­
glioblastoma models, whereas the results emerging from clinical trials in patients with ment had no effect at recurrence20, no specific imaging
recurrent glioblastoma are disappointing; larger studies are underway in the frontline responses or pathological findings upon progression
treatment setting. and reoperation were reported, and thus the mode of
• Future immune-​based strategies are focused on combinations of different immune-​ action in vivo remains obscure. Moreover, the intense
checkpoint inhibitors with diverse treatment modalities that reverse local care provided by health care professionals for patients
immunosuppression in the microenvironment, converting a ‘cold’ tumour into a ‘hot’ enrolled into the experimental arm of this trial has led
tumour. to concerns that the prolonged survival was not entirely
attributable to the specific intervention.
All patients with glioblastoma eventually have disease
the diagnosis and treatment of adult diffuse gliomas, relapse. The management approaches used at tumour
including glioblastoma7. According to these guidelines, progression or recurrence are typically more individ­
the goal of surgery should be gross total resection, when ualized, accounting for patient-​specific and disease-​
feasible; whether macroscopically incomplete resections specific factors that include the radiological pattern of
confer a survival benefit relative to biopsy sampling relapse, time since diagnosis, previous treatment, and,
alone remains controversial8. Moreover, whether mac­ above all, general and neurological function. Favourable
roscopically resectable tumours share a less aggressive prognostic factors at glioblastoma progression include
biology than those for which gross total resection can­ younger age and MGMT-​promoter methylation as well
not be achieved, and hence have an intrinsically better as a long PFS duration with the first-​line treatment21,22.
prognosis, continues to be debated9. For several decades, The therapeutic options for patients with disease relapse
radiotherapy at a dose of up to 60 Gy, administered to include repeat surgery or radiotherapy, but mainly con­
a region consisting of the tumour plus a margin of non­ sist of pharmacological treatment with alkylating agent
malignant tissue in 30 fractions of 1.8–2.0 Gy, has been part chemotherapy or bevacizumab depending on local
of the standard of care for patients with glioblastoma. availability7. From the time of first progression or recur­
This approach resulted in a doubling of median survival rence, median overall survival durations in the range of
in early studies10,11. Patients with baseline characteris­ 6–9 months have been achieved in large series of patients
tics indicating a less favourable prognosis, specifically or in clinical trials23, but are shorter at the population
a Karnofsky performance status (KPS) of ≤60, or an level. Notably, a substantial proportion of patients (up to
age of >70 years, are now typically treated with hypo­ 50%) do not receive any second-​line anticancer ther­
fractionated radiotherapy with a similar biologically apy23,24. In fact, evidence that any therapeutic interven­
effective dose of 40 Gy administered in 15 fractions12. tion administered in the recurrent setting has a major
For newly diagnosed adult patients aged ≤70 years who effect on survival is lacking. For instance, in the EORTC
have a favourable general and neurological performance 26101 trial25, combining bevacizumab with lomus­
status, for example, defined as KPS of ≥70, the addition tine did not improve overall survival outcomes. More
of concomitant and maintenance temozolomide chemo­ importantly, the median PFS duration with lomustine
therapy to conventional radiotherapy (TMZ/RT → TMZ) alone — the current standard-​of-care systemic ther­
became the standard of care in 2005 after the publica­ apy for recurrent glioblastoma — was only 1.5 months,
tion of results demonstrating an improvement in 2-year but overall survival was 8.6 months, indicating that
overall survival from 10.4% with radiotherapy alone to the natural course of disease might be more relevant
26.5% with chemoradiotherapy13. In 2017, a clinical trial at recurrence than the current standard-​of-care treatment
of a similar treatment regimen essentially confirmed that options. Accordingly, novel approaches to the treatment of
elderly patients also derive benefit from chemotherapy recurrent glioblastoma are urgently needed. Most recur­
when added to hypofractionated radiotherapy14. TMZ rent tumours have previously been exposed to the geno­
provides greater benefit to patients with tumours that have toxic stress of irradiation and/or chemotherapy and are,
methylation of the 6-O-​methylguanine-DNA methyl­ therefore, predicted to have a higher mutational load
transferase (MGMT) gene14,15 because this epigenetic and to be more immunogenic than untreated tumours.

NaTUre RevIews | CLinicAL OncoLogy volume 15 | JULY 2018 | 423

© 2018 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
Reviews

Herein, we review the theoretical framework driving compared with other tumour types41, and these tumours
the development of immunotherapy for glioblastoma, profoundly affect the immune system both locally
describe the strengths and weaknesses of the immuno­ and systemically, although no evidence indicates an
therapy approaches that have been tested in the clinic increased incidence of infections typically associated
to date, and discuss future directions in this promising with immunosuppression in patients with glioblas­
area of therapy. toma42. The mechanisms underlying these immunolog­
ical effects are incompletely defined, but seem to involve
The CNS — an immune privileged system? both tumour-​intrinsic factors and host responses to
The concept that the CNS is immune privileged was tumour antigens originating from the CNS. For example,
based on initial experimental data reported >50 years in one study, investigators found that when melanoma
ago by the group of Peter Medawar, which showed cells expressing a model neoantigen were implanted
that foreign cells implanted into the brains of rodents in the brains of mice, neoantigen-​specific T cells were
became successfully engrafted, whereas the same actively deleted43. Furthermore, the T cells that escaped
cells were eradicated by the host immune system when deletion failed to produce pro-​inflammatory cytokines
they were placed in peripheral tissues26–28. Indeed, until or execute cytotoxic functions, even after encounter­
2015, the brain was thought to lack dedicated lymphatic ing the cognate neoantigen in peripheral lymphoid
channels, which was speculated to limit the presentation organs. By contrast, immune function was mostly pre­
of antigens originating in the brain to immune cells. In served in mice with comparably advanced flank or lung
addition, microglial cells were broadly considered to be tumours43. Brain-​tumour-related immune suppression
the major antigen-​presenting cells in the brain tumour was hypothesized to be mediated by TGFβ, which was
microenvironment, and these cells have been hypothe­ found to be secreted by microglia within the tumour
sized to skew T cells away from a cytotoxic phenotype29,30. microenvironment and was present at elevated levels
However, more recent data have refined our understand­ in the serum of mice bearing brain tumours compared
ing of immunological mechanisms that are active in the with mice with flank tumours or no brain tumours.
CNS. For example, we now know that the CNS is sub­ Pharmacological inhibition of TGFβ signalling partially
ject to active immunosurveillance and vigorous immune reversed the immune suppression but did not prolong
responses31 (Fig. 1). In retrospect, indications of immu­ the survival of mice with brain tumours43. In addition
nological activity in the CNS have long existed. Indeed, to tissue resident myeloid cells (microglia), evidence
follow-​up experiments by Medawar’s group revealed that suggests that migrating myeloid cells have an important
engraftment of foreign cells in the brains of rodents was role in glioblastoma-​associated immunosuppression.
prevented by vaccination of the animals against the same For example, Bloch et al.39 reported that macrophages
foreign cells before cell implantation26–28. Furthermore, isolated from the peripheral blood of patients with
in 2015, Louveau et al.32 defined a novel route of lym­ glioblastoma express increased levels of programmed
phatic egress from the brain along distinct channels that cell death 1 ligand 1 (PD-​L1), a ligand that activates
run parallel to dural venous sinuses. Thus, most antigen-​ the programmed cell death protein 1 (PD-1) immune-​
presenting cells exiting the brain are likely to travel to checkpoint receptor that restricts the activity of cytotoxic
the deep cervical lymph nodes, where they can prime T T cells. Correspondingly, these macrophages suppressed
and B lymphocytes32. These findings are corroborated the activation of co-​cultured T cells derived from the
by inflammatory conditions, such as multiple sclerosis same patients39. Furthermore, Fecci et al.44 found that
and cerebral brain abscesses, which demonstrate that glioblastomas induce broad sequestration of T cells out
immunogens present in the brain are capable of gener­ of the circulation into the bone marrow through down­
ating robust immune responses31,33. Taken together, these regulation of sphingosine 1-phosphate receptor 1 (S1P1)
findings support the notion that, while the brain is an in T cells.
immunologically distinct site, the immune microenvi­ Tumour-​intrinsic factors involved in glioblastoma-​
ronment offers adequate opportunities to implement associated immunosuppression include the induction of
immunotherapy for the treatment of brain tumours. signalling pathways that are known to suppress immune
responses. Wainwright et al.45 have demonstrated that
Unique mechanisms of immunosuppression glioblastoma cells express indoleamine 2,3-deoxygenase
Glioblastoma has been extensively studied as a par­ (IDO) enzymes, which catalyse the rate-​limiting step in
adigm for cancer-​associated immunosuppression34. the catabolism of tryptophan to kynurenine — a pathway
Glioblastoma rarely metastasizes to extracranial sites, that is involved in T cell immune tolerance and, there­
although circulating tumour cells have been detected in fore, immunosuppression. Moreover, Heimberger et al.46
patients with glioblastoma35,36. Many assume, therefore, characterized a role of signal transducer and activator
that glioma cells are either poorly suited to survive out­ of transcription 3 (STAT3) signalling in glioblastoma
side the brain or that, given the rapid progression of the cells in suppressing immune cell activity. STAT3 expres­
tumour, limited time is available for tumour cells to colo­ sion from human glioblastoma specimens seems to
nize extracranial sites37. Nonetheless, systemic immuno­ be driven by IL-10 in both tumour cells and immune
suppression has been demonstrated, as evidenced by cells. Indeed, soluble factors, notably TGFβ43,47, IL-10
impaired cellular immunity in patients with glioblas­ (Refs48,49), and prostaglandins50, were the first immuno­
toma and murine glioblastoma models38–40. Indeed, suppressive mediators identified in patients with
glioblastomas have a paucity of infiltrating T cells and glioblastoma, although subsequent studies have also
harbour a relatively low number of somatic mutations indicated that direct interactions between glioblastoma

424 | JULY 2018 | volume 15 www.nature.com/nrclinonc

© 2018 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
 Reviews

Microglial cell
TGFβ
IL-10
Treg cell
IL-10
TGFβ
IDO
Adenosine
LAG3
Exhausted CTL TIGIT
↑ PD-1 CTLA-4
↑ TIM3
↑ LAG3
Tumour cell
↑ CTLA-4
↓ MHC
↑ PD-L1
↑ STAT3
Tumour ↑ HLA-E
vascularization ↑ HLA-G
↑ FASL
TAM ↑ IDO
Blood CSF1R DC
• Treatment (TMZ)- ↑ PD-L1
related lymphopenia FLT3
• PD-L1 on circulating TGFβ
macrophages IDO
CXCR4

Bone marrow Lymphatic

• Treatment (TMZ)- system PD-L1
related lymphopenia PD-1
• Sequestration of
T cells
DC CTL

Spleen
Destruction of tumour-
antigen-specific T cells Deep cervical lymph nodes
Migration of APCs from the brain
via lymphatics, presentation of
tumour antigens, and priming of
T cells and B cells

Fig. 1 | Local and systemic immunosuppression in glioblastoma. The glioblastoma microenvironment is a highly
immunosuppressive milieu of tumour cells and immune cells. Tumour cells express increased levels of immunosuppressive
factors (such as programmed cell death 1 ligand 1 (PD-​L1) and indolamine 2,3-dioxygenase (IDO)) while limiting self-​
presentation of antigens through decreased MHC expression. Microglial cells secrete TGFβ and IL-10, which downregulate
the local myeloid and lymphoid immune cells and promote systemic immunosuppression. Myeloid cells, including tumour-​
associated macrophages (TAMs), have both immunosuppressive and tumour-​promoting effects through modified
expression of various intracellular and extracellular mediators. The lymphoid compartment also contributes to the
immunomodulating environment, with regulatory T (Treg) cells, in particular, mediating immunosuppressive effects through
upregulation of various soluble factors, immune-​checkpoint molecules, and metabolic pathways. This plethora of factors
contributes to the exhausted phenotype of cytotoxic T lymphocytes (CTLs), which express increased levels of exhaustion
markers such as programmed cell death protein 1 (PD-1). Among this milieu, dendritic cells (DCs) can traffic via the tumour
draining lymph nodes of the brain to the deep cervical lymph nodes and can present antigen to promote an adaptive
antitumour immune response, although this process might be abrogated in the context of the systemic
immunosuppression that is intrinsically associated with glioblastoma and can also be potentiated by the current standard-​
of-care treatments for this disease. For example, systemic temozolomide (TMZ) chemotherapy induces a lymphopenia
that is exacerbated by bone marrow sequestration of T cells. Furthermore, T cells specific to intracranial tumour antigens
are destroyed in the spleen, while circulating macrophages express increased levels of inhibitory immune-​checkpoint
ligands. APC, antigen-​presenting cell; CSF1R, colony-​stimulating factor 1 receptor; CTLA-4, cytotoxic T lymphocyte
antigen 4; CXCR4, CXC-​chemokine receptor 4; FASL, FAS ligand; LAG3, lymphocyte activation gene 3 protein; STAT3,
signal transducer and activator of transcription 3; TIGIT, T cell immunoreceptor with Ig and ITIM domains; TIM3, T cell
immunoglobulin mucin receptor 3.

cells and immune cells are involved in immune-​escape member 6 (TNFRSF6, commonly known as FAS)–FAS
mechanisms. Such interactions result in suppression ligand (FASL) interactions53, or triggering of inhibitory
of natural killer (NK) cell activity, which is mediated T cell checkpoints by PD-1 ligands54.
by atypical HLA molecules (including HLA-​E51 and Overall, glioblastoma appears to be a highly immuno­
HLA-​G52), direct induction of apoptosis in immune suppressive tumour. The exact mechanism of immune
cells via tumour necrosis factor receptor superfamily escape is unknown, although myeloid cells are probably

NaTUre RevIews | CLinicAL OncoLogy volume 15 | JULY 2018 | 425

© 2018 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
Reviews

key mediators of glioblastoma-​associated immuno­ The major disadvantage, however, is that EGFRvIII
suppression. Notably, resident microglia and macro­ is heterogeneously expressed on glioblastoma cells
phages outnumber infiltrating T cells in these tumours55, in vivo60, thus creating the potential for outgrowth of
and this paucity of T cells in the tumour microenviron­ tumour cells that lack this antigen. Furthermore, expres­
ment is in striking contrast to findings in other tumour sion of EGFRvIII is unstable throughout the course of
types, such as melanoma or lung cancer56. The results disease61,62, opening further avenues for immune escape
of some studies have suggested that the glioblastoma-​ of tumour cells that downregulate this protein. Three
associated myeloid cells are immunosuppressive, with uncontrolled phase II studies of rindopepimut vaccina­
an M2-like phenotype57, whereas other animal studies tion in favourably selected patients, that is, those with
have suggested that the ingression of myeloid cells, such gross total resection and no evidence of progression
as dendritic cells (DCs), from the periphery is required after completion of concomitant chemoradiotherapy,
to elicit an immune response58. Whether or not these have provided evidence of improved median survival
tumours are intrinsically non-​immunogenic, or whether of 24 months compared with historical controls63–65.
T cells are actively excluded, remains to be defined. These findings led to the initiation of ACT IV (Ref.66), an
Regardless, the myeloid compartment will probably be international phase III trial, in a large cohort of patients
a key target for future glioblastoma immunotherapy. with newly diagnosed EGFRvIII-​positive glioblastoma
Strategies focused on exploiting myeloid cells to foster (Table 1) . Of 745 patients enrolled, 405 fulfilled the
antitumour immune responses include reducing or neu­ predefined criteria for minimal residual disease (MRD),
tralizing the biological activity of immunosuppressive defined as the presence of <2 cm2 of contrast-​enhancing
molecules (such as TGFβ) and administering adjuvants tumour tissue after surgery and chemoradiotherapy66.
such as granulocyte–macrophage colony-​stimulating All patients were randomly assigned (1:1) to receive
factor (GM-​CSF) or Toll-​like receptor (TLR) agonists. either rindopepimut or keyhole limpet haemocyanin
Notably, the current focus of efforts to facilitate general (KLH; used in rindopepimut as the carrier protein for
antitumour immune responses in patients with glioblas­ the 13-amino acid EGFRvIII peptide epitope), together
toma is centred in the field of immune-​checkpoint inhi­ with maintenance TMZ; however, the primary end point
bition, but immune-​checkpoint inhibitors might need was prolonged overall survival in the MRD subgroup,
to be combined with therapies targeted at the myeloid specifically66. The trial was terminated early, after a pre-​
compartment59. planned interim analysis revealed the futility of treat­
Of note, the profound immunosuppressive properties ment. At the final analysis, no significant difference
of glioblastomas have also influenced the design of most in overall survival between the treatment arms was
immunotherapy trials to focus on patients with only a detected in the MRD population (median 20.1 months
limited tumour burden — that is, after gross total resec­ in the rindopepimut group versus 20.0 months in the
tion — and to minimize or omit the use of standard cor­ control group; HR 1.01, 95% CI 0.79–1.30; P = 0.93)66.
ticosteroid therapy and chemoradiotherapy, which can Interestingly, the hazard ratios seemed to be more
have immunosuppressive effects. However, these strate­ favourable for those patients with “significant residual
gies might limit the availability of tumour antigens that is disease”(REF.66) (>2 cm2; HR 0.79, 95% CI 0.61–1.02;
necessary for an effective anticancer immune response. P = 0.066), although this difference was statistically sig­
nificant only among patients treated in the USA (HR
Current state of glioblastoma vaccines 0.70, 95% CI 0.51–0.96; P = 0.027), raising the possibility
The field of innovative immunotherapeutic approaches that this result occurred by chance. Nevertheless, this
to treat glioblastoma is rapidly expanding. On the basis trend for potential benefit in patients with bulky residual
of the apparent failure of glioblastoma to metastasize out­ disease rather than only MRD supports the possibility
side the CNS, efforts to induce active immune surveil­ that a certain level of tumour tissue needs to be present to
lance against glioma cells in the brain by strengthening drive immune responses. Other important observations
the adaptive arm of the immune system, predominantly from that trial included the failure of a strong humoral
by vaccination (Fig. 2), have been pursued as a promis­ immune response to translate into clinical benefit and
ing path forward. Despite the relative paucity of data on the spontaneous loss of antigen expression, even in the
active immunotherapies in humans with glioblastoma, control arm, as observed in patients undergoing repeat
three vaccination approaches have reached phase III surgery66. Thus, EGFRvIII might not be a good target
clinical development, and numerous others are at earlier for immunotherapy owing to a lack of stable expression
stages of clinical testing. Rindopepimut (also known as even in the absence of EGFRvIII-​targeted treatment.
CDX-110 or PEPvIII) is a peptide vaccine that mimics While ACT IV was being conducted, a non-​
and thus targets EGFR variant III (EGFRvIII), which comparative phase II trial, known as ReACT67, was
is a constitutively active mutant form of EGFR that is performed to explore the combination of rindopepi­
expressed exclusively on glioblastoma cells in 25–30% of mut with bevacizumab in a smaller cohort of patients
patients with this disease60. This agent enables a simple with recurrent EGFRvIII-​positive glioblastoma (n = 72).
vaccination approach based on just one immunogenic The results of this trial suggested favourable outcomes
peptide rather than on the generation of a vaccine from relative to those obtained with bevacizumab plus KLH
patient-​derived immune and tumour cells. The advan­ only67 (Table 1). This evidence of the potential activity
tage of targeting the EGFRvIII neoantigen relates to the of rindopepimut combined with bevacizumab, taken
exclusivity of neoantigen expression on tumour cells, together with the negative outcome of ACT IV, lends
which limits the risk of ‘on-​target, off-​tumour’ toxicities. support for further trials of combined immunotherapy

426 | JULY 2018 | volume 15 www.nature.com/nrclinonc

© 2018 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
 Reviews

Immune-checkpoint
CTLA-4 blockade
CD80 or
CD86 CD28 Anti-CTLA-4
antibody
TCR CTL
DC +

Anti-PD-1
MHC antibody
class II CD3

PD-1 Anti-PD-L1
Antigen antibody CAR T cell therapy
presentation
PD-L1
Cytotoxicity MHC Glioblastoma-
Peptide vaccine class I associated
or tumour lysate antigen Anti-IL-13Rα2
CAR T cell
Vaccination
therapy

Glioblastoma Viral IL-13Rα2

cell replication EGFRvIII

Anti-EGFRvIII
CAR T cell

Viral infection
of tumour cell
• Cell lysis and release of
oncovirus
• Antigens that can
promote an antitumour Oncolytic viral therapy
immune response

Fig. 2 | Current immunotherapy modalities for the treatment of glioblastoma. Glioblastoma vaccine therapy relies on
dendritic cell (DC)-mediated presentation of glioblastoma-​associated peptides, antigens, or epitopes derived from tumour
lysates to T cells of the adaptive immune system through MHC class II–T cell receptor (TCR) (signal 1) and CD80 and/or
CD86–CD28 (signal 2) interactions. The cytotoxic T lymphocytes (CTLs) that are subsequently activated interrogate and
destroy tumour cells containing glioblastoma-​associated antigens presented on MHC class I molecules. However, tumour
cells often evade destruction by CTLs through upregulation of immune-​checkpoint ligands, such as programmed cell death
1 ligand 1 (PD-​L1) that can bind complementary receptors on the CTLs, such as programmed cell death protein 1 (PD-1) to
cause suppression of lymphocyte activation. Immune-​checkpoint blockade with monoclonal antibodies effectively
prevents this interaction. Similarly, antibody-​mediated blockade of cytotoxic T lymphocyte protein 4 (CTLA-4), an inhibitory
immune-​checkpoint molecule that binds CD80 and CD86 and prevents their interaction with CD28, can promote T cell
priming by DCs. Glioblastoma-​associated antigens, including IL-13 receptor subunit-​α2 (IL-13Rα2) and EGFR variant III
(EGFRvIII), are also presented on tumour cell surfaces independent of MHC class I, and these tumour-​associated antigens
are being exploited as specific targets of genetically modified chimeric antigen receptor (CAR) T cell therapies. Genetic
engineering is also used in oncolytic viral therapy to create viruses that selectively infect or replicate in tumour cells. The
resulting tumour cell lysis not only kills the infected tumour cells directly but can also activate immunogenic tumour cell
death pathways that can stimulate antigen presentation and an adaptive antitumour immune response.

and anti-​angiogenic therapy, specifically with inhibitors An IDH1-R132H-​mutated peptide vaccine is being uti­
of the VEGF pathway68,69. lized in both trials, although in NOA-16, the peptides
IDH1 peptide vaccines have also been investigated are being administered by subcutaneous injections con­
in clinical trials, further emphasizing the 2016 WHO currently with topical imiquimod (a TLR7 agonist that
classification of IDH-​mutant gliomas as a major subtype can activate myeloid cells71), whereas the vaccine is being
of this disease4. Preclinically, Schumacher et al.70 demon­ delivered in combination with the adjuvants GM-​CSF
strated CD4+ T helper 1 (TH1) cell-​mediated immune and Montanide ISA-51 in the RESIST trial. These trials
responses in a transgenic mouse model with humanized might provide a new avenue for the treatment of patients
MHC class I and MHC class II molecules loaded with with IDH-​mutant gliomas.
15-mer mutant peptides derived from IDH1-R132H. Expression of telomerase reverse transcriptase
Subsequently, two ongoing phase I clinical trials have (TERT) is increased in many cancers, including glio­
been initiated: NOA-16 and RESIST (NCT02454643 and mas; indeed, TERT promoter mutations are the genetic
NCT02193347, respectively; Supplementary Table 1). alterations most frequently detected in glioblastomas72,73.

NaTUre RevIews | CLinicAL OncoLogy volume 15 | JULY 2018 | 427

© 2018 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
Reviews

Table 1 | Completed clinical trials of therapeutic vaccines for glioblastoma

Trial name (ClinicalTrials. Experimental Control or comparator n Primary end point Results for primary
gov identifier) treatment treatment or outcomes outcome
Phase III trials
ACT IV (NCT01480479) Rindopepimut (EGFRvIII-​ KLH plus TMZ 745 OS Median OS: 20.1 versus
specific peptide and KLH 20.0 months (HR 1.01, 95% CI
conjugate) plus GM-​CSF 0.79–1.30; P = 0.93)66
and TMZ
Phase II trials
NA (NCT00323115) Autologous DC vaccine None 11 Tumour-​specific Increased mean frequency of
plus radiotherapy and CTL responses tumour-​specific CD4+ T cells
TMZ after vaccination (P = 0.004)177
NA (NCT01006044) Autologous DC vaccine Historical controls (SOC) 26 PFS PFS: 12.7 months (95% CI
plus SOC 7–16); comparison with
historical controls not
reported178
NA (NCT00576537) Autologous DC vaccine None 50 Safety and toxicity No grade 3–4 AEs were
of s.c. injection observed179,180
ACTIVATe (NCT00643097) Rindopepimut (EGFRvIII-​ Historical controls 40 PFS and • After vaccination, 0/5
specific peptide and KLH hypersensitivity to patients in standard TMZ
conjugate) plus GM-​CSF peptide on the basis group but 7/8 patients
and standard or dose-​ of a positive skin in dose-​intensified TMZ
intensified TMZ test of ≥ 5 mm in group had hypersensitivity
diameter (P = 0.005)
• Median PFS: 15.2 versus
6.3 months (HR 0.35, 95% CI
0.14–0.87; P = 0.024)65
HeatShock (NCT00905060) HSPPC-96 (peptide-​ None 46 Safety profile and • 34/46 patients had grade 1–2
complex vaccine made survival durations AEs; no grade 3–4 AEs were
using autologous tumour observed
lysates) plus TMZ • Median PFS 18.0 months
(95% CI 12.4–21.8)
• Median OS 23.8 months
(95% CI 19.8–30.2)181
NA (NCT00846456) DC vaccine made using None 20 Safety Grade 1–3 AEs
tumour stem cell mRNA (fatigue, anorexia, and
headache) were similar
to those of conventional
radiochemotherapy182
NA (NCT01920191) IMA-950 (peptide vaccine None 19 Tolerability and Vaccine was well tolerated
comprising multiple safety with local inflammation at
GAAs) and poly(I:C) plus injection site; a few patients
TMZ and radiotherapy developed cerebral oedema
that was easily managed using
steroids183
GBM-​Vax (NCT01213407) Trivax (DC-​based vaccine) SOC 87 PFS Not posted
plus SOC
NA (NCT01290692) TVI-​Brain-1 (tumour-​ None 86 PFS Not posted
based vaccine)
NA (NCT01081223)a TVI-​Brain-1 plus GM-​CSF None 14 Toxicity and PFS • Only minimal grade 1–2
inflammatory AEs
• Median OS 7.7 months (no
PFS reported)184
NA (NCT01280552) ICT-107 (autologous Unpulsed autologous 124 OS Median OS 18.3 versus
DC vaccine pulsed DC vaccine 16.7 monthsb (P > 0.05)77
with synthetic peptides
mimicking GAAs)
ReACT (NCT01498328) Rindopepimut and KLH plus bevacizumab 127 PFS • 6-Month PFS: 27% versus
GM-​CSF plus 11% (P = 0.048)
bevacizumab • ORR: 24% versus 17%
• Median OS: 12.0 versus
8.8 months (HR 0.47, 95% CI
0.25–0.91; P = 0.0208)67
ACT III (NCT00458601) Rindopepimut and None 82 PFS Median PFS from study entry:
GM-​CSF and then TMZ 9.2 months (95% CI 7.4–11.3)63

428 | JULY 2018 | volume 15 www.nature.com/nrclinonc

© 2018 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
 Reviews

Table 1 (cont.) | Completed clinical trials of therapeutic vaccines for glioblastoma

Trial name (ClinicalTrials. Experimental Control or comparator n Primary end point Results for primary
gov identifier) treatment treatment or outcomes outcome
Phase II trials (cont.)
NA (NCT00293423)a HSPPC-96 plus TMZ None 41 Safety, toxicity, • Minimal toxicity (mostly
and PFS injection site reactions);
1 patient had grade 3 fatigue
• Median PFS: 19.1 weeks
(95% CI 14.1–24.1)185
NA (NCT00014573) Combination None NR NR Not posted
chemotherapy (goal
(carmustine, cisplatin, of 30)
cyclophosphamide, and
paclitaxel), cytokine
therapy (G-​CSF, GM-​CSF,
and IL-2), autologous
tumour cell vaccine, and
autologous lymphocyte
infusion
NA (NCT00004024) Irradiated autologous None ~60 Unclear (possibly Preliminary results show that
tumour cell vaccine immunogenicity 17/19 patients had DTH; OS
and GM-​CSF, then and survival) was directly correlated with
muromonab-​ DTH response (P = 0.02)186
CD3-activated T cell
infusion and IL-2
Phase I trials
NA (NCT01222221) IMA-950 and GM-​CSF None 45 Causes of AEs • 4 grade 3–4 AEs were
and number of attributed to IMA-950;
patients with T cell the majority of toxicities
responses to GAAs were grade 1 injection site
on the basis of HLA reactions
multimer analysis • 36/40 (90%) immune-​
evaluable patients
responded to GAAs75
NA (NCT00576641) Autologous DC vaccine None 22 Safety and toxicity No grade 3–4 AEs were
(pulsed with tumour lysate) observed76
ZAP IT (NCT00626015) PEP-3 and TMZ (plus None 16 Suppression of One infusion of daclizumab
normal saline) or Treg cells on the basis significantly reduced
daclizumab or basiliximab of percent change circulating CD4+FOXP3+
(anti-​IL-2R antibodies) in CD4+FOXP3+ Treg cell abundance (P = 0.0464)
Treg cell frequency between values at baseline
and after the start of the
vaccine series187
NA (NCT00890032) DC vaccine loaded with None ~50 Feasibility and • Success rate for obtaining
BTSC mRNA safety RNA for DC pulsing
was > 80%
• No DLTs observed188
NA (NCT01250470) Survivin peptide vaccine None 9 Toxicity No grade 3–4 AEs related to
and Montanide ISA-51 the vaccine were observed;
(adjuvant) plus GM-​CSF the majority of AEs were of
grade 1 (REF.189)
NA (NCT01171469) DC vaccine made using None 8 MTD No DLTs observed between
BTSCs plus imiquimod 3 dosing levels of 5, 10, and
(Toll-​like receptor agonist) 15 million DCs190
NA (NCT01522820) CDX-1401 (fusion protein None 18 Safety Not posted
vaccine comprising
an antibody to the DC
receptor LY75 conjugated
to an NY-​ESO-1
peptide) ± sirolimus
(mTOR inhibitor)
NA (NCT00612001) Synthetic GAA peptide-​ None 8 DLT and MTD No grade 3–4 vaccine-​
pulsed DC vaccine related AEs were observed;
grade 1–2 AEs were mostly
flu-​like injection site reactions,
lymphadenopathy, and
rashes191

NaTUre RevIews | CLinicAL OncoLogy volume 15 | JULY 2018 | 429

© 2018 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
Reviews

Table 1 (cont.) | Completed clinical trials of therapeutic vaccines for glioblastoma

Trial name (ClinicalTrials. Experimental Control or comparator n Primary end point Results for primary
gov identifier) treatment treatment or outcomes outcome
Phase I trials (cont.)
NA (NCT00068510) Autologous tumour-​ None 28 DLT and MTD No grade 3–4 vaccine-​related
lysate-pulsed DC vaccine AEs were observed; grade
1–2 AEs were mostly flu-​like
injection site reactions,
lymphadenopathy, and
rashes191
NA (NCT00069940) Telomerase 540–548 None NR NR Not posted
peptide vaccine plus (goal
GM-​CSF of 35)
NA (NCT01621542) WT2725 (WT1-derived None 64 DLT and MTD No DLTs were observed; AEs
oligopeptide vaccine) were mostly grade 1 injection
site reactions192
All data were obtained from www.clinicaltrials.gov using the search terms “glioblastoma” and “vaccine” and the ‘Completed’ filter. ‘Not posted’ refers to studies
with results that have not been posted to www.clinicaltrials.gov or that were not found after a PubMed and Google Scholar search for the ClinicalTrials.gov
identifier, last name of the primary and/or lead investigator, and key words. Toxicity grades are based on the USA National Cancer Institute Common Terminology
Criteria for Adverse Events. AE, adverse event; BTSC, brain tumour stem cell; CTL , cytotoxic T lymphocyte; DC, dendritic cell; DLT, dose-​limiting toxicity ; DTH,
delayed type hypersensitivity ; EGFRvIII, EGFR variant III; FOXP3, forkhead box protein P3; G-​CSF, granulocyte colony-​stimulating factor ; GM-​CSF, granulocyte–
macrophage colony-​stimulating factor ; GAA , glioma-​associated antigen; HSPPC-96, heat shock protein peptide complex 96; IL-2R , IL-2 receptor ; KLH, keyhole
limpet haemocyanin; LY75, lymphocyte antigen 75; MTD, maximum tolerated dose; NA , not applicable; NR , not reported; OS, overall survival; PFS, progression-​free
survival; poly(I:C), polyriboinosinic–polyribocytidylic acid; s.c., subcutaneous; SOC, standard of care (surgery , chemotherapy , and radiotherapy); TMZ,
temozolomide; Treg cell, regulatory T cell; WT1, Wilms tumour protein. aPhase I/II clinical trials are tabulated only as phase II. bResults found in ‘Study Results’ tab of
www.clinicaltrials.gov webpage.

Capitalizing on these findings, vaccine therapy predi­ post-​surgery apheresis to chemoradiotherapy comple­
cated on the transfection of DCs to overexpress TERT tion and was intended to be administered together with
and thus present TERT antigens has been tested in maintenance TMZ. This strategy was chosen because
patients with pancreatic adenocarcinoma, with encour­ the generation of the vaccine required time, although
aging results reported in one patient74. However, no one might assume that an earlier initiation of vaccina­
clinical trials have yet been conducted to investigate tion might have benefits associated with the immunos­
TERT-​peptide-based vaccine therapy in patients with timulatory effects of radiotherapy. The phase I study
glioblastoma. of ICT-107 enrolled 17 patients with newly diagnosed
Vaccine strategies using more than one peptide have glioblastoma and 3 patients with recurrent glioblas­
also been developed. In comparison with rindopepi­ toma. Safety was confirmed, with a median follow-​up
mut, such approaches are more complex and clinically duration of 40.1 months, while the median PFS dura­
less advanced. IMA-950 is a multipeptide vaccine that tion was 16.9 months and median overall survival
is based on the administration of 11 tumour-​associated duration was 38.4 months in the patients with newly
peptides (nine HLA-​A*02 peptides, an elongated HLA diagnosed disease76. The results of a randomized phase II
class I peptide, and one HLA class II peptide) and trial suggest that this agent has some therapeutic activ­
the synthetic hepatitis B virus marker peptide IMA-​ ity, at least in HLA-​A2-positive patients77. In this trial77,
HBV-001. Results of a phase I trial confirmed the 124 patients were randomly assigned (2:1) to receive
tolerability of this vaccine, although two instances of ICT-107 or control treatment with autologous DCs that
dose-​limiting toxicity potentially related to the vaccine had not been exposed to the glioblastoma-​associated
(fatigue and anaphylaxis) were reported; the survival antigens; 77 patients were HLA-​A2-positive. Overall
data were not remarkable75. survival was not significantly improved (Table 1), but
ICT-107 is another multipeptide vaccine, which outcomes tended to favour the experimental therapy
was specifically designed for the treatment of glio­ in HLA-​A2-positive patients, potentially because four
blastoma76. This vaccine consists of patient-​derived of six peptides were predicted to be presented in an
DCs incubated ex vivo with six peptides from proteins HLA-​A2-dependent manner. However, the trial was
selected based on their over-​representation in the gene-​ underpowered to derive meaningful conclusions from
expression profiles of glioblastoma cells compared with further subgroup analyses. No safety concerns were
nonmalignant tissues: melanoma-​associated antigen 1 reported. A phase III trial of this vaccine, referred to as
(MAGEA1), HER2, interferon-​inducible protein AIM2, STING (NCT02546102), was opened for patient accrual
i-​dopachrome tautomerase (DCT), melanocyte protein in 2016 but was suspended owing to a lack of funding
(PMEL), and IL-13 receptor subunit-​α2 (IL-13Rα2). on 21 June 2017. The ICT-107 programme seems not
Preclinical proof of concept for this approach was dif­ to have been pursued further.
ficult to obtain because similar peptides might not DCVax-​L is a third therapeutic vaccine that has
have the same immunogenicity between humans reached the phase III stage of clinical testing in patients
and mice and because mouse glioma models might with glioblastoma (NCT00045968; Supplementary
not over­express the same genes as human tumours. Table 1). With the DCVax-​L approach, whole tumour
The vaccine was generated within a time frame from lysate from a patient’s resected glioblastoma is used to

430 | JULY 2018 | volume 15 www.nature.com/nrclinonc

© 2018 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
 Reviews

pulse analogous DCs and serves as a source for the entire Nevertheless, the strong induction of a humoral
spectrum of tumour antigens78. DCVax-​L probably has response observed in patients involved in the ACT IV
the longest history of any glioblastoma vaccine, with the trial66, although of no apparent clinical benefit, illustrates
approach confirmed preclinically in a rat model in which that inducing immune responses through vaccination is,
vaccination induced a substantial increase in tumour in principle, feasible. Furthermore, data demonstrating
T cell infiltration79. On the other hand, this vaccine is the induction of antitumour T cell responses have also
also the most logistically challenging to generate because been generated75. Thus, the experiences that have been
it requires the collection of tumour samples from each gathered with glioblastoma vaccination therapy to date
individual patient, which are then processed and used indicate that many vaccines are biologically active, but
to stimulate autologous DCs80. This strategy has the the degree of immune stimulation is generally insuffi­
theoretical advantage of being personalized but carries cient to translate into clinical benefit; combinatorial
the limitation of using autoantigens to which tolerance approaches might provide superior results.
has already been established. The phase I study of this
agent included 12 patients with newly diagnosed or Oncolytic viral therapies
recurrent glioblastoma, and the results suggested that a Initially, virus-​based anticancer therapies were consid­
low tumour burden and low levels of TGFβ2 expression ered a treatment strategy separate from immuno­
will help select patients who are more likely to benefit therapy and were predicated on selective viral replication
from this treatment. The pivotal DCVax-​L trial was initi­ in and subsequent destruction of cancer cells. However,
ated in December 2006 (NCT00045968; Supplementary the antitumour immune responses induced as a result
Table 1), but patient enrolment has been put on hold of oncolytic viral infections have blurred this distinc­
for unidentified reasons and outcomes data have not yet tion81 (Fig. 2). Viruses can activate the immune system
been made available. through pathogen-​associated molecular patterns and
Currently, a plethora of clinical trials are ongo­ pattern recognition receptors. Furthermore, viruses
ing in order to evaluate different antigens for vac­ often activate macrophages through receptors, such as
cine therapy against glioblastoma. At the time of TLRs82. As a secondary effect, activated myeloid cells can
this Review, 20 phase I, 16 phase II, and 2 phase III improve the infiltration of T cells into tumours to pro­
trials are ongoing (Supplementary Table 1). The two mote an inflamed microenvironment. As a result, viral
phase III trials, involving DCVax-​L (as discussed therapies are a very interesting approach to overcoming
above; NCT00045968) or an individual proteomics-​ the immunosuppression of glioblastoma. While initial
based approach (NCT01759810), reflect the ongoing strategies used replication-​incompetent viruses to avoid
paradigm shift in medicine towards a personalized complications of encephalitis83, contemporary oncolytic
medicine approach. This trend is also seen in the viral treatment approaches are increasingly utilizing
ongoing phase II studies that involve vaccines gener­ replication-​competent viruses, such as retroviruses,
ated using autologous tumour lysates (NCT01204684, adenoviruses, herpes simplex viruses (HSVs), polio­
NCT01635283, NCT02772094, NCT03018288, viruses, and measles viruses84,85 (Table 2; Supplementary
NCT02709616, and NCT02808364). The wide spec­ Table 2).
trum of phase II studies also illustrates the importance In May 2016, the recombinant oncolytic poliovirus
of the myeloid compartment in future immunother­ PVSRIPO received breakthrough therapy designation
apy approaches to the treatment of glioblastoma. from the FDA on the basis of the findings of an ongoing
For example, multiple trials are including the use of phase I study in patients with recurrent glioblastoma
myeloid-​activating adjuvants, such as polyriboinosinic– (NCT01491893; Supplementary Table 2). PVSRIPO
polyribocytidylic acid (poly(I:C); NCT02358187, is a genetically engineered form of the oral poliovirus
NCT00766753, NCT01204684, and NCT02078648). Sabin type 1, in which the internal ribosome entry site
Finally, the ongoing phase I trials reflect similar trends, is replaced with that of human rhinovirus type 2 in
with the use of myeloid-​modulating adjuvants and order to eliminate neurovirulence. PVSRIPO infects
personalized approaches to tumour antigens. and replicates within cells that express the poliovirus
In general, trials of DC-​based vaccine approaches receptor, which is an onco-​fetal cell adhesion molecule
are time and resource intensive, particularly if autol­ that is often expressed in glioblastoma, thus exploiting
ogous tumour tissue is required for generation of the natural affinity of the virus for CNS cells. The virus
the vaccine, as is this case for DCVax-​L. Even with is administered by convection-​enhanced delivery via a
a defined, ‘off-​t he-shelf ’ peptide antigen cocktail, catheter inserted directly into the tumour. Preliminary
such as the one used in ICT-107, patients still need data presented at the 2015 ASCO annual meeting indi­
to undergo apheresis after surgery in order to harvest cated that in 24 patients, 42% of whom had received
DCs to be expanded ex vivo and need to wait for weeks prior bevacizumab, the 2-year overall survival was
until the vaccine has been generated. Furthermore, 24%, and the median overall survival across all doses
for vaccine therapies to be implemented for broader of PVSRIPO was 12.5 months; updated results reported
use, such as a part of standard-​of-care treatment, in 2016 revealed that 3 patients (13%) remained alive at
challenges include logistical and regulatory require­ 3 years86. No full safety or preliminary efficacy data are
ments involved with transporting biological materi­ currently available in the public domain.
als to and from a site of vaccine generation, meeting Data from an ongoing phase I clinical trial (Supple­
Good Manufacturing Practice standards, and vaccine mentary Table 2) are also available for vocimagene amiretro­
quality control. repvec (Toca 511), which is a non-​lytic, replicating retrovirus

NaTUre RevIews | CLinicAL OncoLogy volume 15 | JULY 2018 | 431

© 2018 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
Reviews

Table 2 | Completed clinical trials of oncolytic viral therapy for glioblastoma

Trial name (ClinicalTrials. Active treatment Control or n Primary end Results for primary outcomes
gov identifier) comparator points or
treatment outcomes
Phase II trials
ParvOryx01 (NCT01301430) H-1 parvovirus (H-1PV) None 18 Safety and DLT No DLTs; 1 patient had a SUSAR that
was not considered a DLT193
BrTK02 (NCT00589875) AdV-​tk (adenoviral vector Historical 52 Safety and OS • 1 patient had transient grade 4
expressing HSV-​TK) plus controls with hemiparesis
valacyclovir (antiviral drug) SOC • Median OS: 17.1 versus
and SOC 13.5 months (HR 0.72, 95% CI
0.52–0.99; P = 0.0417)94
HGG-01 (NCT00870181) Intra-​arterial cerebral infusion Surgery and 47 6-month PFS • 6-month PFS: 54.5% versus 13.6%
of AdV-​tk and i.v. GCV after systemic (P = 0.001)
surgery chemotherapy • Median PFS: 29.6 versus 8.4 weeks96
Phase I trials
D24GBM (NCT01956734) DNX-2401 (Delta-24-RGD None 31 Number of No AEs related to DNX-2401 (REF.194)
adenovirus) and TMZ patients with AEs
NA (NCT00028158) G207 (modified oncolytic None 21 Safety No AEs were directly related to G207;
strain of HSV-1) by 5/21 patients had post-​inoculation
intratumoural inoculation stereotactic biopsies or extended
hospitalizations195
NA (NCT00528684) REOLYSIN (wild-​type reovirus None 18 MTD, DLT, • No DLTs were observed and no
formulation) by intratumoural and 6-month MTD was reached; no severe AEs
infusion response rate were seen
• 1 patient was free of progression at
6 months196
NAa Surgery and ONYX-015 None 24 Safety and DLT No serious AEs were related to
(adenovirus mutant) ONYX-015 treatment; no DLTs were
observed (up to a maximum dose of
1010 PFU)91
NA (NCT00390299) MV-​CEA (measles virus None 40 Toxicity and MTD Study suspended for unknown
expressing carcinoembryonic reason
antigen) before or after
surgery
NA (NCT02031965) HSV-1716 and surgery None 2 MTD Study terminated for unknown
reason
NA (NCT00751270) AdV-​tk plus valacyclovir None 15 Safety No AEs confirmed to be related to
viral vector or vaccine; some grade
3–4 AEs were probably related to
treatment but were transient95
NA (NCT00002824) H5.010RSVTK (recombinant None 18 MTD Not posted
adenovirus vector with HSV-​
TK) with surgery and GCV
NA (NCT00157703) G207 (modified oncolytic None 9 Toxicity 3/9 patients had serious AEs
strain of HSV-1) and (seizures) that were possibly related
radiotherapy to G207 administration197
NA (NCT01582516) Delta-24-RGD administered None 20 Toxicity Not posted
by CED into tumour
Most data were obtained from findings from www.clinicaltrials.gov using the search terms “glioblastoma”, “oncolytic”, and “viral” and the ‘Completed’ filter ; some
trials were identified through peer-​to-peer discussions. ‘Not posted’ refers to studies with results that have not been posted to www.clinicaltrials.gov or were not
found after a PubMed and Google Scholar search for the ClinicalTrials.gov identifier, last name of the primary and/or lead investigator, and key words. Toxicity
grades are based on the USA National Cancer Institute Common Terminology Criteria for Adverse Events. AE, adverse event; CED, convection-​enhanced delivery ;
DLT, dose-​limiting toxicity ; GCV, ganciclovir ; HSV, herpes simplex virus; HSV-​TK , HSV thymidine kinase; i.v., intravenous; MTD, maximum tolerated dose; NA , not
applicable; OS, overall survival; PFS, progression-​free survival; PFU, plaque-​forming unit; SOC, standard of care (surgery , chemotherapy , and radiotherapy); SUSAR ,
suspected unexpected serious adverse reaction; TMZ, temozolomide. aIdentified through peer-​to-peer discussion and does not have ClinicalTrials.gov identifier.

derived from the Moloney murine leukaemia virus that 45 patients undergoing surgery for recurrent or progres­
has been engineered to encode a modified yeast cyto­ sive high-​grade glioma, followed 6 weeks later by intra­
sine deaminase and preferentially infects tumour cells87. venous administration of Toca FC, an extended-​release
Although the virus infects both normal and tumour cells, formula of the prodrug 5-fluorocytosine. In virus-​infected
the tumour cells lack typical viral defence mechanisms that cells, this prodrug is converted to the antimetabolite
prevent viral DNA integration into their genome88. The 5-fluorouracil by the exogenous cytosine deami­
virus was injected into tissues lining the resection cavity in nase, which is not otherwise expressed in human cells,

432 | JULY 2018 | volume 15 www.nature.com/nrclinonc

© 2018 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
 Reviews

thereby providing some degree of selectivity for tumour Oncolytic viral therapies based on the measles
cells. The median overall survival duration of patients virus are another approach to the treatment of glio­
treated with vocimagene amiretrorepvec and Toca FC blastoma and are supported by promising preclinical
was 13.6 months, which was interpreted as being supe­ data. Specifically, a measles virus engineered to produce
rior to historical controls88. The proposed mechanism carcinoembryonic antigen (MV-​CEA), which serves
of improved efficacy is thought to be not only a direct as a serum marker for in vivo expression of the viral
tumoricidal effect but also a virally induced general­ genome, caused the regression of flank tumours and
ized antitumour immune response. Later-​phase studies markedly improved the survival of immunocompetent
of vocimagene amiretrorepvec are currently underway, glioblastoma-​bearing mice97. These findings led to a
including a randomized phase II trial in patients with phase I clinical trial to investigate MV-​CEA in patients
recurrent high-​grade glioma. with recurrent glioblastoma (NCT00390299; Table 2),
Adenoviruses have also been extensively studied in the although the trial has been suspended for unidentified
context of anticancer therapy (Table 2; Supplementary reasons. No other trials of a measles virus have been
Table 2) because they are common respiratory viruses initiated in patients with glioblastoma.
with relatively easy mechanisms of in vitro manipu­ HSVs are also well understood, and the use of HSV
lation and genetic engineering89; however, challenges strains genetically engineered to target cancer cells is
have included the identification of target receptors for another promising investigational approach to glio­
the virus. Adenovirus 5-Delta 24RGD (DNX-2401) is blastoma therapy98. Generations of HSV constructs
one oncolytic, conditionally replicative adenovirus that have been developed by selectively attenuating genes
achieves tumour cell target­ing through a 24-base dele­ to ensure that the virus predominantly targets repli­
tion of the transforming protein E1A and insertion of an cating cells in the CNS99. These manipulations include
Arg–Gly–Asp (RGD) motif onto a viral capsid protein thymidine kinase deletion, γ134.5 dual knockout, viral
for improved targeting towards αv integrins90. DNX-2401 ribonucleotide reductase disruption, and lacZ gene
has been investigated in a phase I trial in combina­ insertions into the viral ribonucleotide reductase
tion with TMZ and is currently under investigation gene promoter; extensive preclinical work has demon­
in early phase clinical studies in combination with the strated both the anti-​glioma effects and low neuro­
anti-​PD-1 antibody pembrolizumab (NCT02798406) or toxicity of such viruses99. Trials of several modified
IFNγ (NCT02197169; Supplementary Table 2). A phase HSV constructs, including G207 (NCT00028158,
I trial of a different conditionally replicating adenovirus, NCT00157703, and NCT02457845), HSV-1716
named ONYX-015, has been performed after resection (NCT02031965), G47Δ, and M032 (NCT02062827),
of recurrent glioma91 (Table 2). Tumour cell targeting have been conducted or are ongoing in patients with
was achieved by modifying the adenovirus protein E1B, glioblastoma (Table 2; Supplementary Table 2). The
which normally blocks p53-induced host cell apopto­ future direction of oncolytic viral therapies seems to be
sis in order to sustain viral replication: ONYX-015 was focused on combinations with immunotherapy strat­
designed with an attenuated E1B protein that cannot egies, such as those mentioned for DNX-2401, in the
bind to p53, rendering the virus capable of replicating in hope of exploiting the potentially durable anticancer
p53-deficient tumour cells only92. Notably, a maximum immune responses initiated by the viral infection to
tolerated dose was not defined; ONYX-015 was well elicit prolonged clinical responses.
tolerated at doses of up to 1010 plaque-​forming units91,
indicative of the excellent safety of many oncolytic viral Immune-​checkpoint inhibitors
therapies. The median overall survival duration, how­ Antibodies that reduce the activity of endogenous, neg­
ever, was only 6.2 months. This lack of efficacy might ative regulatory pathways limiting T cell activation are
be linked to efforts to optimize the safety of this virus: arguably the most important advance in cancer therapy
ONYX-015 might have been safe because it was highly made in the past decade, with major improvements in
attenuated, but for the same reason, it might have been the outcomes of patients with some difficult-​to-treat
less effective93. cancers, such as advanced-​stage melanoma or non-​
Adenoviruses have also been modified to serve as small-cell lung cancer100. To date, the most notable
tumoricidal gene delivery vectors, most notably agla­ examples have been antibodies that block the inhibitory
timagene besadenovec (AdV-​t k). This replication-​ immune-​checkpoint proteins cytotoxic T lymphocyte
incompetent adenovirus was transfected to express the antigen 4 (CTLA-4) and PD-1, which are expressed on
HSV thymidine kinase (HSV-​TK) gene, which converts T cells, or PD-​L1, which is expressed on certain subsets
the prodrug ganciclovir (GCV) into a toxic nucleotide of immune cells and is aberrantly expressed on tumour
analogue that can kill replicating tumour cells94. This cells of various histologies. Indeed, PD-​L1 has emerged
approach, termed gene-​mediated cytotoxic immuno­ as a biomarker of sensitivity to immune-​checkpoint
therapy, was found to be safe in the phase I clinical trial inhibition with anti-​PD-1 antibodies in the context of
BrTK01 (Ref.95; Table 2). Subsequently, two phase II tri­ many solid tumours101–103. This ligand is also expressed
als, BrTK02 (Ref.94) and HGG-01 (REF.96), have been con­ in a subset of glioblastomas ranging from ~2% to ~88%,
ducted using intratumoural AdV-​tk administration and although the extent of this expression remains the sub­
valacyclovir or intra-​arterial AdV-​tk administration ject of debate104,105. Moreover, higher PD-​L1 expression
and GCV, respectively. Together, the results of these two in glioblastoma has been correlated with poorer patient
trials demonstrated favourable PFS and overall survival prognoses in some studies105, which might be related
outcomes associated with AdV-​tk-based therapy (Table 2). to increased suppression of antitumour immunity,

NaTUre RevIews | CLinicAL OncoLogy volume 15 | JULY 2018 | 433

© 2018 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
Reviews

although this relationship was not detected in other Case reports have suggested that anti-​PD-1 ther­
studies104. Despite these contrasting findings, immune-​ apy can be effective for patients with glioblastoma.
checkpoint inhibitors have garnered considerable inter­ First, nivolumab was reported to result in long-​term
est from the glioblastoma community, considering the disease control in an adult patient with recurrent glio­
profound immunosuppression that is characteristic of blastoma114. Second, a case report published in 2016
this disease. Preclinical studies have demonstrated the recounted the impressive and durable responses to
impressive activity of immune-​checkpoint inhibitors nivolumab in two siblings with germ-​line, biallelic DNA-​
as monotherapies or in combination with radio­therapy repair defects and recurrent paediatric glioblastoma115.
in mouse models of glioma106–108. Notably, however, Finally, another adult patient with widely disseminated
caution in interpreting these data is prudent because glioblastoma showed a response to pembrolizumab116. In
orthotopic implantation of glioma cell lines with a high the latter two case reports, the patients had tumours with
mutational load might not enable the accurate predic­ a high mutation burden, which is a known predictor of
tion of responses in patients with spontaneously aris­ response to immune-​checkpoint inhibitors. Notably,
ing glioblastomas109. Indeed, glioblastomas typically in 2017, pembrolizumab was approved for patients
have a relatively low mutational load and a paucity of with microsatellite instability-​high or mismatch repair-​
T cell infiltration compared with other tumour types41. deficient solid tumours, independent of histology117;
Furthermore, the use of orthotopically implanted however, only a small fraction of all patients with glio­
tumour models, typically involving the GL-261 and blastomas have mutations affecting the DNA mismatch
SMA-560 cell lines, might introduce variables such as repair machinery118.
violation of the blood–brain barrier.
The anti-​PD-1 antibody nivolumab is the immune-​ Chimeric antigen receptor T cell therapy
checkpoint inhibitor for which clinical development has In addition to vaccines, oncolytic viruses, and immune-​
advanced furthest in patients with glioblastoma. Safety checkpoint inhibitors, another interesting immuno­
studies in this setting have revealed only grade 1 or 2 therapy approach leverages genetically modified T cells
toxicities from nivolumab monotherapy, and the tox­ (Table 3; Supplementary Table 4). T cells can be engi­
icities of combined treatment with nivolumab and the neered to express chimeric antigen receptors (CARs),
anti-​CTLA-4 antibody ipilimumab were similar to those which consist of the antigen recognition domains of
observed in patients with other tumour types110. In the antibodies linked to T cell activation domains derived
ongoing phase III CheckMate 143 trial (NCT02017717; from the T cell receptor CD3 ζ-​chain (CD3ζ) and co-​
Supplementary Table 3), the efficacy of nivolumab is stimulatory receptors (such as CD28 and/or TNFRSF9
being compared with that of bevacizumab in patients (commonly known as 4-1BB))119. The antigen recogni­
with glioblastoma across different lines of treatment. tion domains impart CAR T cells with specificity for
Data from this study have not yet been published in a tumour-​associated antigens, and these agents have shown
peer-​reviewed journal, but preliminary data reported at promise in the treatment of glioblastoma. Furthermore,
the 2017 World Federation of Neuro-​Oncology Societies the appeal of this approach lies in the capacity of CAR
meeting revealed that the primary end point of the trial T cells to recognize antigens that are not presented in
was not met: the median overall survival of patients with the context of MHC molecules, as is typically required
recurrent disease was 9.8 months with nivolumab ver­ for adaptive immune responses. Additionally, the CAR
sus 10.0 months with bevacizumab7,111. In addition, this T cells can be engineered to have an activated pheno­
trial contained exploratory phase I cohorts that included type (for example, through co-​expression of OX40 or
combined nivolumab and ipilimumab treatment arms, 4-1BB)120,121. These multiple features can help CAR
in which high rates of serious adverse events were T cells overcome some of the immunosuppressive effects
observed (adverse events leading to discontinuation of a tumour microenvironment.
occurred in 50% of patients across two nivolumab and Brown et al.122 reported a case study in which a
ipilimumab groups112); thus, this combination strategy patient with recurrent glioblastoma underwent repeat
is not being pursued further in the phase III stage of this resection of three of five intracranial lesions followed by
trial. The incidence and types of adverse events associ­ an infusion of CAR T cells targeting IL-13Rα2, which
ated with nivolumab monotherapy seem to be similar is often overexpressed on glioblastoma cells. The find­
in patients with glioblastoma and in those with other ings of a previous phase I study of a prior version of the
tumour types7,111,112, although the combination of this CAR T cell targeting IL-13Rα2 supported the safety
agent with TMZ/RT → TMZ was associated with an of this approach, with a low risk of common compli­
acceptable safety profile in exploratory cohorts com­ cations of CAR T cell therapy, such as cytokine-​release
prising patients with newly diagnosed glioblastoma113. syndrome123,124. The patient reported in the follow-​up
In two additional ongoing studies, CheckMate 498 and study by Brown et al.122 had leptomeningeal disease and,
CheckMate 548 (NCT02617589 and NCT02667587, therefore, the CAR T cells were administered intracra­
respectively; Supplementary Table 3), investigators are nially via two routes: initial weekly infusions of CAR
exploring nivolumab as an alternative to TMZ (both T cells into the resection cavity for 6 weeks, followed by
in combination with radiotherapy) in patients with intrathecal delivery into the ventricular system after the
MGMT-​promoter-unmethylated tumours and as an appearance of new lesions for a further ten infusions.
addition to the standard TMZ/RT → TMZ regimen in A dramatic radiographical response was observed, with
patients with MGMT-​promoter-methylated tumours, shrinkage of all lesions by 77–100%122. Nevertheless,
respectively. 7.5 months after the initiation of adoptive cell therapy,

434 | JULY 2018 | volume 15 www.nature.com/nrclinonc

© 2018 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
 Reviews

Table 3 | Completed clinical trials of chimeric antigen receptor T cell and other adoptive cell therapies for glioblastoma
Trial name (ClinicalTrials. Active treatment Control n Primary end point Results of primary outcomes
gov identifier) treatment or outcome
Phase II trials
NA (NCT00331526) Patient-​derived aldesleukin None 83 Toxicity, PFS, • Well tolerated, with most AEs
(recombinant IL-2)-stimulated LAK and OS attributable to craniotomy; one
cells infused intracranially at time possible intralesional haemorrhage
of craniotomy or within 4 weeks occurred that did not require
after craniotomy intervention
• Median survival 20.5 months (from
diagnosis)198
NA (NCT00003185) Infusion of T cells harvested from None 40 Toxicity and TTP • The trial is combined phase I/II, but
tumour-​draining lymph nodes plus no phase II results are available to
one dose of cyclophosphamide our knowledge
• Phase I results: mostly grade 1–2
AEs, but 2/12 patients had increased
seizure frequency easily controlled
with antiepileptics
• Median TTP > 16 months199
Phase I trials
NA (NCT01082926) Intratumoural infusion of None 6 Safety and toxicity No serious therapy-​related AEsa
GRm13Z40-2 CD8+ T cells (REF.200)
(allogeneic CTL line with deleted
glucocorticoid receptor sites and
modified to express IL-13–zetakine
(comprising an extracellular IL-13
domain fused to an activatory
CD3ζ domain, and HyTK)) and
intratumoural aldesleukin
NA (NCT00730613) Intratumoural infusion of None 3 Safety and No serious therapy-​related AEsa
autologous CTL line expressing feasibility (REF.200)
IL-13–zetakine and HyTK
ERaDICATe (NCT00693095) CMV-​ALT ± cutaneous inoculation None 23 T cell response Substantial expansion of
of CMV-​DCs to CMV-​DC with polyfunctional CMV-​specific CD8+
CMV-​ALT T cells expressing IFNγ, TNF, and CCL3
(REF.201)
ALECSAT-​GBM Infusion of ALECSAT (autologous None 23 Tolerability and AEs Not posted
(NCT01588769) natural killer cells and CTLs on the basis of KPS
activated and expanded ex vivo) and QoL interview
Most data were obtained from findings from www.clinicaltrials.gov using the search terms “glioblastoma”, “oncolytic”, and “viral” and the ‘Completed’ filter ; some
trials were identified through peer-​to-peer discussions. ‘Not posted’ refers to studies with results that have not been posted to www.clinicaltrials.gov or that were
not found after a PubMed and Google Scholar search for the ClinicalTrials.gov identifier, last name of the primary and/or lead investigator, and key words. Toxicity
grades are based on the USA National Cancer Institute Common Terminology Criteria for Adverse Events. AE, adverse event; CCL3, CC-​chemokine ligand 3; CD3ζ,
T cell receptor CD3 ζ -chain; CMV-​ALT, cytomegalovirus-​specific autologous lymphocyte transfer ; CMV-​DCs, dendritic cells pulsed with cytomegalovirus 65 kDa
phosphoprotein mRNA ; CTL , cytotoxic T lymphocyte; HyTK , hygromycin phosphotransferase gene fused to herpes simplex virus type 1 thymidine kinase gene; KPS,
Karnofsky performance score; L AK , lymphokine-​activated killer ; NA , not applicable; OS, overall survival; PFS, progression-​free survival; QoL , quality of life; TNF,
tumour necrosis factor ; TTP, time to progression. aNo paper or abstract was found specifically for the clinical trial, but the citation describes relevant results.

the patient ultimately had disease recurrence122. While syndrome or neurotoxicity126, which are common and
these results are encouraging, the recurrence indicates life-​threatening adverse effects of anti-​CD19 CAR T cell
that perhaps the tumour underwent immunoediting therapy in patients with B cell malignancies124. The over­
and ultimately selected for cells that were negative for all survival of the patients did not seem to be affected
IL-13Rα2; indeed, preliminary results from tissue ana­ by CAR T cell therapy, with only one patient having
lyses support this hypothesis122. Another notable feature disease stabilization (although lasting >18 months)126.
of this study is that the CAR T cells were administered Nevertheless, tumour infiltration of the CAR T cells
into the cerebrospinal fluid (CSF) rather than intrave­ was detected126. Furthermore, evidence of T cell activity
nously or directly into the tumour, suggesting that dif­ against tumour cells expressing EGFRvIII was observed
ferent ways to administer immunotherapies should be in tumour samples from seven patients who underwent
considered125. surgical resections after CAR T cell infusion, with sam­
In 2017, data from the first ten patients of a first-​in- ples from five of the seven having decreased tumoural
human clinical trial of CAR T cells directed at EGFRvIII EGFRvIII expression and concomitant upregulation
were published 126 (NCT02209376; Supplementary of immunosuppressive factors, including IDO1 and
Table 4). These ten patients with recurrent EGFRvIII-​ PD-​L1, and increased numbers of regulatory T (Treg)
positive glioblastoma received a single intravenous infu­ cells in the tumour microenvironment126. Together,
sion of autologous anti-​EGFRvIII CAR T cells126. Notably, the findings of these early clinical studies suggest that
treatment was safe, with no incidence of cytokine-​release glioblastomas can activate various adaptive responses to

NaTUre RevIews | CLinicAL OncoLogy volume 15 | JULY 2018 | 435

© 2018 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
Reviews

subvert anticancer immune responses and reinstate an patients receiving stereotactic body radiotherapy than
immunosuppressive milieu; these escape mechanisms in those treated with conventional hyperfractionated
will need to be overcome if we are to improve the effec­ radiotherapy133. The researchers hypothesized that the
tiveness of immunotherapy for this disease. greater extent of lymphopenia in the latter group results
Overall, evidence from clinical trials of CAR T cells from direct exposure of a larger number of circulating
suggests that the engineered cells can infiltrate glioblas­ immune cells to radiation owing to the less-​conformal
tomas, become activated, and, in one patient, eradicate delivery of radiotherapy133, supporting their previous
a considerable amount of malignant tissue122. However, findings in modelling studies relating to glioblastoma
in many solid tumour studies, CAR T cells alone had therapy134. These findings suggest that the use of shorter
insufficient antitumour activity127. While the exact courses of radiation should be considered, particularly
reasons for this observation are unknown, one could when immunotherapy is planned. TMZ can also cause
postulate that targeting one antigen in a highly hetero­ lymphopenia and results in changes to immune cell pop­
geneous tumour might not be sufficient to eradicate all ulations that can even persist in patients with recovery of
cancer cells. Indeed, to date, the best results with CAR normal blood counts. For example, TMZ can negatively
T cells have been achieved in patients with cancers that affect the number of memory T cells in a permanent
are highly clonal, such as leukaemias and lymphomas, fashion132. Finally, corticosteroids that are commonly
resulting in approvals in these settings. Nevertheless, used in the treatment of patients with glioblastoma can
in most patients, CAR T cells will probably need to be adversely affect the effectiveness of immunotherapies135.
administered in combination with other therapies, or In fact, a corticosteroid-​related gene-​expression signa­
CAR T cells targeting multiple different antigens will ture has been associated with unfavourable survival in
be required. Moreover, additional immunosuppressive patients with glioblastoma135,136. While many confound­
pressures within the tumour microenvironment are ing factors might have affected the results of these
likely to inhibit the anticancer activity of CAR T cells128. studies, valid concerns exist and the effects of cortico­
Thus, approaches to improving the efficacy of CAR steroids on immunotherapy should be studied further.
T cells might involve promoting antigen spreading, Taken together, deleterious effects of glioblastomas on
combination therapies (for example, with immune-​ the immune system might be compounded by conven­
checkpoint inhibitors129), or targeting of immuno­ tional therapies that are immunosuppressive and can
suppressive myeloid cells with agents such as IDO130 ultimately hinder an effective immunotherapy strategy.
or macrophage colony-​stimulating factor 1 receptor Therefore, considering strategies including stereotactic
(CSF1R)131 inhibitors. radiosurgery (SRS) or local chemotherapy and the judi­
cious use of corticosteroids might enable improved anti­
Immunotherapy and the current standard of care tumour immune responses to be achieved (as discussed
With immunotherapy now firmly established in the in a following section of this Review).
management of a variety of malignancies and having
demonstrated some promise in the treatment of glio­ Biomarkers for glioblastoma immunotherapy
blastoma, integrating immunotherapy into the current The lack of validated biomarkers is one of the current
standard-​of-care TMZ/RT → TMZ regimen is a crit­ challenges in treating patients with glioblastoma. MRI
ical next step to improving outcomes of patients with remains the gold standard method for determining
glioblastoma. The paradigm of administering systemic the disease burden, despite the well-​documented chal­
chemotherapy concurrently with radiotherapy demands lenges associated with accurately assessing disease status
particularly careful consideration, as this combination using MRI, including pseudoprogression and pseudo­
might have the unintended consequence of weak­ responses137,138. Unlike for some systemic cancers, repeat
ening the immune system by suppressing or perma­ tissue sampling, either through biopsy sampling or
nently ablating critical immune cell populations42,132. surgical resection, is fraught with a substantial risk of
Myelosuppression is commonly observed with chemo­ procedure-​related morbidities in patients with glioblas­
therapy in patients with various cancers, and the degree toma. These challenges are compounded in the setting
of immunosuppression is likely magnified in patients of immunotherapy, with reports of initial enlarge­
with brain tumours42,132: multiple aforementioned stud­ ment of tumour volume owing to immune infiltration
ies have found that glioblastoma can profoundly affect (pseudoprogression), true progression before a delayed
the peripheral immune system at baseline43,44. response, and contrast enhancement and associated
Indeed, the use of hyperfractionated radiation and oedema in patients with tumour types that are respon­
systemic chemotherapy with additional corticosteroids sive to immune-​checkpoint inhibition139,140 — including
to prevent or control toxicities seems to cripple the glioblastoma114,141.
immune system. Grossman et al.42 found that hyper­ Sporadic biomarkers have been reported to correlate
fractionated radiation correlated with a marked deple­ with response to immunotherapy in glioblastoma. Many
tion of CD4+ T cells in patients with glioblastoma, vaccine therapies using DCs have been and are currently
suggesting severe immunosuppression. These investi­ being investigated (Table 1; Supplementary Table 1), but
gators also found that the degree of radiation-​induced few biomarkers have been identified as prognostic indi­
immunosuppression was a negative prognostic indicator cators for immunotherapy. Everson et al.142 reported that
of survival outcomes. Subsequent studies by the same increased responsiveness of patient-​derived peripheral
group in patients with pancreatic cancer receiving only blood CD8+ T cells to immunostimulatory cytokines
radiation revealed that lymphopenia is less severe in after DC vaccination was associated with prolonged

436 | JULY 2018 | volume 15 www.nature.com/nrclinonc

© 2018 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
 Reviews

overall survival and identified patients who survived has the potential to help select appropriate patients for
for >2 years. IDO pathway flux, specifically the kynure­ immunotherapy and can be combined with conventional
nine:tryptophan ratio, has also been reported to be imaging studies to understand changes in disease burden
prognostic in patients with recurrent glioblastoma and disease characteristics after therapy has been initi­
undergoing surgery and HSPPC-96 vaccine treatment143. ated; incorporation of CSF-​based biomarkers should be
In the rapidly expanding field of immune-​checkpoint an important component of future clinical trials.
inhibitors, PD-​L 1 has emerged as a biomarker for
response to anti-​PD-1 or anti-​PD-L1 therapy in multiple The concept of cold versus hot tumours
solid tumour settings, as discussed previously102,103,144. On Impressive responses to immunotherapy have been
the other hand, biomarkers for a response to antagonists observed for a range of different tumour types, although
of the immune-​checkpoint protein CTLA-4, as well as certain tumours seem to be insensitive to the current
PD-1–PD-​L1 blockade, include a high mutational bur­ approaches; the concept of ‘cold’ versus ‘hot’ tumours has
den of tumours — although this association has been been used to explain this observation. With the report­
reported for non-​CNS tumours, such as melanoma145, ing of negative results from the CheckMate 143 trial of
case reports have demonstrated a similar trend in nivolumab after first relapse in patients with glioblas­
patients with glioblastoma, as mentioned previously115. toma7,111, many are starting to believe that glioblastoma
However, anti-​CTLA-4 treatment will probably require will fall into the category of cold tumours. The reasons
the development of biomarkers that are different from that tumours are unresponsive to immunotherapy are
those associated with PD-1–PD-​L1 blockade because likely multifactorial and include a highly immuno­
the effects of CTLA-4 inhibition are biologically distinct suppressive tumour milieu, defects in tumour antigen pres­
and abrogate global immunosuppression via effects on entation, and features of the physical microenvironment
CD4+ effector T cells and Treg cell activity101. In general, such as hypoxia and necrosis.
the mutational burden of each individual glioblastoma Multiple findings support the hypothesis that glio­
might influence the patient’s response to immuno­ blastoma is a cold tumour. Glioblastomas are known
therapy. This point is highlighted by genomic characteri­ to have relatively few tumour-​infiltrating lymphocytes
zations of breast and ovarian cancers that have shown (TILs) compared with other tumour types, suggesting
how deletions affecting the genes encoding members that they are quiescent tumours in terms of immune
of the apolipoprotein B mRNA editing enzyme cata­ reactivity157. Moreover, the TILs that are present have
lytic polypeptide (APOBEC) family of cytosine deam­ been shown to highly express markers of exhaustion,
inases lead to hypermutated tumours that can prompt such as the inhibitory immune-​checkpoint proteins
stronger immune system activation146. No studies of T cell immunoglobulin mucin receptor 3 (TIM3;
APOBEC mutations in glioblastoma have been reported, also known as hepatitis A virus cellular receptor 2)
although one could assume that higher mutational loads and lymphocyte activation gene 3 protein (LAG3)158.
will translate into a greater number of tumour neoanti­ Interestingly, instead of TILs, high numbers of
gens that are amenable to immune responses and thus myeloid cells, such as microglia and macrophages,
facilitate immunotherapy147. are present in the glioblastoma microenvironment
Given the surgical challenges associated with lesions and probably have predominantly immunosuppres­
located within the brain, liquid biopsy148 might be an sive activities41. Furthermore, defects in the antigen-​
attractive approach to analysing the genetic make-​up of presentation machinery can make the tumour cold
brain tumours. However, tumour-​derived DNA and cir­ with respect to T cell-​dependent immune responses.
culating tumour cells are rarely detectable in the blood For example, β2-microglobulin (β2M) is an essential
of patients with CNS neoplasms149,150, prompting sev­ subunit of the MHC class I antigen-​presenting com­
eral groups to explore the potential of CSF as a reservoir plex, and studies have demonstrated that antitumour
for tumour-​associated biomarkers149,151–155. Burgeoning immune responses against tumour cells containing
data suggest that tumour-​derived DNA in the CSF β2M mutations are attenuated159. Notably, mutations
can be evaluated qualitatively and quantitatively and in β 2M have been identified as an acquired resist­
that the levels of this DNA are correlated with disease ance mechanism in patients with melanoma treated
burden154,155. In a proof-​of-concept study, Wang et al.151 with pembrolizumab or ipilimumab159,160. Aberrations
demonstrated that whole-​exome sequencing can be per­ in the IFNγ–JAK–STAT pathway, which promotes
formed directly on DNA in CSF samples and can be used antigen presentation and drives the expression of
to explore the genetic landscape of brain tumours with­ immune-​c heckpoint molecules including PD-​L 1,
out the need for invasive procedures. This approach seem to confer both intrinsic and acquired resistance
could potentially be a powerful asset in determining to immune-​checkpoint inhibitors159,161. These mecha­
appropriate treatments. For instance, metabolomic ana­ nisms have been mainly shown in non-​CNS tumours,
lyses of these samples can identify the presence of IDH1 although downregulation of MHC expression, defects
mutations156. Metabolic biomarkers might therefore in β2M, and mutations in JAK and STAT proteins are
have utility as surrogate indicators of responses to ther­ not uncommon in glioblastomas and suggest similar
apy and of disease progression in IDH-​mutant tumours. defects in antigen presentation162,163.
One application of liquid biopsy to immunotherapy Finally, physical aspects of the glioblastoma micro­
involves the ability to detect EGFRvIII mutations that environment are thought to play important parts in
could inform decisions relating to the use of vaccines or attenuating antitumour immune responses. For example,
CAR T cell therapies152. Therefore, analysing CSF DNA one of the hallmark features of glioblastoma is necrosis,

NaTUre RevIews | CLinicAL OncoLogy volume 15 | JULY 2018 | 437

© 2018 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
Reviews

which seems to have a key role in immunosuppres­ immunotherapy with SRS108,166,171,173, as well as local
sion. This necrosis often stems from the disordered chemotherapy, oncolytic viral therapy, and laser abla­
tumour vasculature of glioblastoma, leading to areas of tion (NCT02311582, NCT01811992, NCT01205334,
hypoxia. Eli et al.164 showed that tumour necrosis results NCT02197169, NCT02798406, and NCT02576665);
in increased extracellular concentrations of potassium, however, many other modalities are being actively
which can inactivate tumour-​infiltrating T cells and investigated, such as nanoparticle therapies and various
might, therefore, limit the activity of immunotherapy in therapeutic devices (NCT03020017, NCT00734682,
patients with glioblastoma. NCT02340156, and NCT02766699).
The use of SRS has gained popularity across multiple
Novel and combinatorial approaches tumour types, owing to the potential to promote antigen
The initial results with immunotherapy in patients with release, limit the systemic immunosuppression associ­
glioblastoma have been disappointing; however, com­ ated with radiotherapy, and thereby aid in the initiation
bination therapies are being actively investigated in the of an antitumour immune response174. SRS has been
hope of transforming glioblastomas into hot tumours. shown to synergize with anti-​PD-1 antibody therapy
Given the rapidly expanding list of immunological tar­ in orthotopic mouse glioblastoma models, resulting in
gets implicated in the disease, as well as the large number increased numbers of activated CD8+ T cells express­
of immunotherapeutic agents at various stages of devel­ ing IFNγ and decreased numbers of tumour-​infiltrating
opment, the number of possible therapeutic combina­ Treg cells compared with those observed with either treat­
tions is prohibitively large for random testing. Instead, ment alone108. A durable survival benefit was achieved
rationally designed approaches are critical to the devel­ in mice with the combination approach, and this ben­
opment of effective treatment strategies. For example, efit was abrogated when CD8+ T cells were depleted108.
Koyama et al.165 found that resistance to anti-​PD-1 anti­ Moreover, evidence of immunological memory was
bodies in a mouse model of lung cancer was correlated provided by a lack of tumour cell engraftment upon
with the upregulation of TIM3. In addition, Kim et al.166 rechallenge of mice previously treated with SRS and
detected TIM3 expression in seven of eight human anti-​PD-1 therapy108. This paradigm has subsequently
glioblastoma specimens and showed that the number been expanded in preclinical models using focused radi­
of exhausted tumour-​infiltrating T cells (positive for ation in combination with targeting of other immune-​
both PD-1 and TIM3) increased over time in a mouse checkpoint molecules, such as CTLA-4, TIM3, GITR,
model of this disease; durable tumour control could be and 4-1BB108,166,171,173. Currently, ongoing clinical stud­
achieved when mice were treated with both anti-​PD-1 ies are evaluating the use of SRS with anti-​PD-1 anti­
and anti-​TIM3 antibodies. bodies in newly diagnosed and recurrent glioblastoma
Activation of multiple arms of the immune system (NCT02648633 and NCT02866747; Supplementary
is another potential approach to overcome immuno­ Table 3), with case reports relating to other tumour types
suppression. Lymphocytes are critical in mounting an supporting the viability of this strategy175.
antitumour immune response, although macrophages, Other groups are examining the role of ablative
NK cells, and suppressor cells (such as myeloid-​derived therapies, such as laser ablation, in combination with
suppressor cells (MDSCs) and Treg cells) could also be immune-​checkpoint inhibitors (Supplementary Table 3).
targeted in order to enhance antitumour immunity. The With regard to laser ablation, catheters are stereotac­
immune microenvironment of glioblastoma is charac­ tically implanted into glioblastomas and the tumours
terized by an abundance of M2-polarized myeloid cells are heated in a conformal manner to a temperature
and Treg cells and a paucity of NK cells57,167–169. Strategies that causes tumour cell death while minimizing injury
for activating the antitumour functions of macrophages, to nonmalignant tissue. In an ongoing phase I/II trial
DCs, and microglial cells by polarizing them to the (NCT02311582; Supplementary Table 3), the safety
M1-like phenotype and inhibiting MDSCs include anti­ and efficacy of MRI-​guided laser ablation combined
bodies to CSF1R and IDO inhibitors, both of which are with pembrolizumab are being evaluated in patients with
being investigated in glioblastoma models45,170 and in recurrent glioblastoma.
patients with this disease (NCT02052648). IDO inhib­ The approach to chemotherapy might need to be
itors also have effects on Treg cell accumulation45,170, and re-​examined in the context of immunotherapy. The
antibodies that target glucocorticoid-​induced TNFR-​ immuno­suppression associated with TMZ is empha­
related protein (GITR; also known as TNFRSF18) pro­ sized by the fact that clinicians are required to antici­
vide another avenue for targeting this cell type, with pate an increased susceptibility to infections176. Several
supporting evidence coming from mouse glioblastoma studies have evaluated the effects of chemotherapy on
models171. Finally, targeting NK cells with agents includ­ the antitumour immune response in glioblastomas.
ing the IL-15 superagonist ALT-803 has also shown For example, preclinical studies in a mouse glioblas­
promise in preclinical glioblastoma models172. toma model have revealed that the use of high doses of
Preclinical data have demonstrated that other local­ chemotherapy (either TMZ or carmustine) blunts the
ized treatment modalities can synergize with immuno­ antitumour immune responses invoked by anti-​PD-1
therapy108,166,171,173. The use of local therapies to increase antibodies132. Furthermore, the immunosuppressive
the availability of tumour antigens and immunotherapy effects of chemotherapy seem to be long-​lasting because
to drive an antitumour immune response provides the mice that survived after treatment with a combina­
rationale for this combination approach. Currently, tion of systemic chemotherapy and immunotherapy
this paradigm is mostly predicated on combining or systemic chemotherapy alone failed to generate an

438 | JULY 2018 | volume 15 www.nature.com/nrclinonc

© 2018 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
 Reviews

antitumour immune response upon tumour rechal­ anticancer immune response is not hindered during
lenge132. Interestingly, the animals treated with systemic maintenance chemotherapy.
chemotherapy could not be rescued with anti-​PD-1 In addition, when interpreting these interesting and
therapy at the time of tumour rechallenge132. Hence, the important immunological findings in mice, we must be
current strategy for immunotherapy studies in patients cognizant of the limitations of such preclinical models;
with glioblastoma, in which the experimental agents while the GL261 model is commonly used, the findings
are used after or even together with TMZ chemotherapy, might not necessarily translate to human glioblastoma.
might be suboptimal. Notably, when the mice were Hence, higher-​fidelity models of the disease are needed,
treated with local chemotherapy, synergistic activity including genetically engineered mouse models and
with immunotherapy was observed, resulting in a dura­ humanized mice, and confirmatory data from human
ble immune response132. Moreover, transfer of CD8+ tumour specimens should be sought in order to inform
T cells from these mice partially rescued the response the design of clinical studies.
of mice previously treated with systemic chemotherapy
to tumour rechallenge 132. The mechanism under­ Conclusions
lying these observations was hypothesized to involve Immunotherapy is clearly a revolution in cancer care.
the increased release of tumour antigens after local Dramatic responses have been observed across vari­
chemotherapy without systemic immunosuppression, ous tumour types with immunotherapies, particularly
resulting in augmented antigen presentation. immune-​checkpoint inhibitors and CAR T cells. Clearly,
Thus, sweeping generalizations about the immune however, not all tumours are susceptible to current
effects of systemic chemotherapy must be avoided. The immuno­therapies, and even among those patients who do
important clinical nuance must be recognized, con­ have a response, the effects are not always durable. Where
sidering that patients are treated with intense doses of glioblastoma falls in the spectrum of responsiveness will
systemic chemotherapy and then treated with a main­ soon become clear because several studies with large
tenance regimen of chemotherapy on 5 days out of a patient cohorts are set to mature. Early data suggest that
28-day cycle. Indeed, the effects of monthly TMZ on the most glioblastomas are cold tumours. Hence, ongoing
immune system might not be the same as those observed studies are using combination approaches with the aim
with the initial continuous 6-week course of treatment. of making these cold tumours hot and thus augmenting
Heimberger, Sampson, and colleagues46,65 have studied current immunotherapy strategies. The data will guide
the effects of monthly TMZ on the antitumour immune the way in which immunotherapy is implemented as part
response in patients receiving the EGFRvIII vaccine rin­ of the standard of care for patients with glioblastoma.
dopepimut and found that TMZ preferentially ablated
Treg cells over effector cells, thereby suggesting that the Published online 11 April 2018

1. Lathia, J. D., Mack, S. C., Mulkearns-​Hubert, E. E., 12. Roa, W. et al. Abbreviated course of radiation therapy literature-​based meta-​analysis from 91 trials.
Valentim, C. L. L. & Rich, J. N. Cancer stem cells in in older patients with glioblastoma multiforme: a Neuro. Oncol. 16, 696–706 (2014).
glioblastoma. Genes Dev. 29, 1203–1217 prospective randomized clinical trial. J. Clin. Oncol. 23. Weller, M., Cloughesy, T., Perry, J. R. & Wick, W.
(2015). 22, 1583–1588 (2004). Standards of care for treatment of recurrent
2. Hambardzumyan, D., Amankulor, N. M., Helmy, K. Y., 13. Stupp, R. et al. Radiotherapy plus concomitant and glioblastoma — are we there yet? Neuro. Oncol. 15,
Becher, O. J. & Holland, E. C. Modeling adult gliomas adjuvant temozolomide for glioblastoma. N. Engl. 4–27 (2013).
using RCAS/t-​va technology. Transl Oncol. 2, IN6 J. Med. 352, 987–996 (2005). 24. Gramatzki, D. et al. Glioblastoma in the Canton of
(2009). 14. Perry, J. R. et al. Short-​course radiation plus Zurich, Switzerland revisited: 2005 to 2009. Cancer
3. Weller, M. et al. Glioma. Nat. Rev. Dis. Primers 1, temozolomide in elderly patients with glioblastoma. 122, 2206–2215 (2016).
15017 (2015). N. Engl. J. Med. 376, 1027–1037 (2017). 25. Wick, W. et al. Lomustine and bevacizumab in
4. Louis, D. N. et al. The 2016 World Health 15. Hegi, M. E. et al. MGMT gene silencing and benefit progressive glioblastoma. N. Engl. J. Med. 377,
Organization Classification of Tumors of the Central from temozolomide in glioblastoma. N. Engl. J. Med. 1954–1963 (2017).
Nervous System: a summary. Acta Neuropathol. 131, 352, 997–1003 (2005). 26. Billingham, R. E., Brent, L. & Medawar, P. B. Actively
803–820 (2016). 16. Chinot, O. L. et al. Bevacizumab plus radiotherapy-​ acquired tolerance of foreign cells. Nature 172,
5. Ostrom, Q. T. et al. CBTRUS statistical report: primary temozolomide for newly diagnosed glioblastoma. 603–606 (1953).
brain and central nervous system tumors diagnosed in N. Engl. J. Med. 370, 709–722 (2014). 27. Billingham, R. E., Brent, L., Medawar, P. B. &
the United States in 2008–2012. Neuro. Oncol. 17 17. Gilbert, M. R., Sulman, E. P. & Mehta, M. P. Sparrow, E. M. Quantitative studies on tissue
(Suppl. 4), iv1–iv62 (2015). Bevacizumab for newly diagnosed glioblastoma. transplantation immunity. I. The survival times of skin
6. Reifenberger, G., Wirsching, H.-G., Knobbe-​Thomsen, N. Engl. J. Med. 370, 2048–2049 (2014). homografts exchanged between members of different
C. B. & Weller, M. Advances in the molecular genetics 18. Stupp, R. et al. Cilengitide combined with standard inbred strains of mice. Proc. R. Soc. B Biol. Sci. 143,
of gliomas — implications for classification and treatment for patients with newly diagnosed 43–58 (1954).
therapy. Nat. Rev. Clin. Oncol. 14, 434 (2017). glioblastoma with methylated MGMT promoter 28. Medawar, P. B. Immunity to homologous grafted skin;
7. Weller, M. et al. European Association for Neuro-​ (CENTRIC EORTC 26071–22072 study): a the fate of skin homografts transplanted to the brain,
Oncology (EANO) guideline on the diagnosis and multicentre, randomised, open-​label, phase 3 trial. to subcutaneous tissue, and to the anterior chamber
treatment of adult astrocytic and oligodendroglial Lancet. Oncol. 15, 1100–1108 (2014). of the eye. Br. J. Exp. Pathol. 29, 58–69 (1948).
gliomas. Lancet Oncol. 18, e315–e329 (2017). 19. Stupp, R. et al. Effect of tumor-​treating fields plus 29. Woodroofe, M. N., Bellamy, A. S., Feldmann, M.,
8. Kreth, F.-W. et al. Gross total but not incomplete maintenance temozolomide versus maintenance Davison, A. N. & Cuzner, M. L. Immunocytochemical
resection of glioblastoma prolongs survival in the era temozolomide alone on survival in patients with characterisation of the immune reaction in the central
of radiochemotherapy. Ann. Oncol. 24, 3117–3123 glioblastoma: a randomized clinical trial. JAMA 318, nervous system in multiple sclerosis. Possible role
(2013). 2306–2316 (2017). for microglia in lesion growth. J. Neurol. Sci. 74,
9. Beiko, J. et al. IDH1 mutant malignant astrocytomas 20. Stupp, R. et al. NovoTTF-100 A versus physician’s 135–152 (1986).
are more amenable to surgical resection and have a choice chemotherapy in recurrent glioblastoma: a 30. Schiffer, D., Mellai, M., Bovio, E. & Annovazzi, L.
survival benefit associated with maximal surgical randomised phase III trial of a novel treatment The neuropathological basis to the functional role of
resection. Neuro. Oncol. 16, 81–91 (2014). modality. Eur. J. Cancer 48, 2192–2202 microglia/macrophages in gliomas. Neurol. Sci. 38,
10. Walker, M. D. et al. Evaluation of BCNU and/or (2012). 1571–1577 (2017).
radiotherapy in the treatment of anaplastic gliomas: a 21. Weller, M. et al. MGMT promoter methylation is a 31. Waksman, B. H. & Adams, R. D. Allergic neuritis: an
cooperative clinical trial. J. Neurosurg. 49, 333–343 strong prognostic biomarker for benefit from dose-​ experimental disease of rabbits induced by the
(1978). intensified temozolomide rechallenge in progressive injection of peripheral nervous tissue and adjuvants.
11. Walker, M. D. et al. Randomized comparisons of glioblastoma: the DIRECTOR trial. Clin. Cancer Res. J. Exp. Med. 102, 213–236 (1955).
radiotherapy and nitrosoureas for the treatment of 21, 2057–2064 (2015). 32. Louveau, A. et al. Structural and functional features of
malignant glioma after surgery. N. Engl. J. Med. 303, 22. Han, K. et al. Progression-​free survival as a surrogate central nervous system lymphatic vessels. Nature
1323–1329 (1980). endpoint for overall survival in glioblastoma: a 523, 337–341 (2015).

NaTUre RevIews | CLinicAL OncoLogy volume 15 | JULY 2018 | 439

© 2018 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
Reviews

33. Canessa, A., Del Bono, V., Miletich, F. & Pistoia, V. lymphocytes in human glioma. Cancer Immun. Arch. newly diagnosed glioblastoma (GBM) patients
Serum cytokines in toxoplasmosis: increased levels of 7, 12 (2007). [abstract]. J. Clin. Oncol. 32 (Suppl), 2005 (2014).
interferon-​gamma in immunocompetent patients with 57. Komohara, Y., Ohnishi, K., Kuratsu, J. & Takeya, M. 78. Polyzoidis, S. & Ashkan, K. DCVax®-L — developed by
lymphadenopathy but not in AIDS patients with Possible involvement of the M2 anti-​inflammatory Northwest Biotherapeutics. Hum. Vaccin. Immunother.
encephalitis. J. Infect. Dis. 165, 1168–1170 macrophage phenotype in growth of human gliomas. 10, 3139–3145 (2014).
(1992). J. Pathol. 216, 15–24 (2008). 79. Liau, L. M. et al. Treatment of intracranial gliomas
34. Nduom, E. K., Weller, M. & Heimberger, A. B. 58. Greter, M. et al. Dendritic cells permit immune with bone marrow-​derived dendritic cells pulsed with
Immunosuppressive mechanisms in glioblastoma. invasion of the CNS in an animal model of multiple tumor antigens. J. Neurosurg. 90, 1115–1124
Neuro. Oncol. 17 (Suppl. 7), vii9–vii14 (2015). sclerosis. Nat. Med. 11, 328 (2005). (1999).
35. Schweitzer, T., Vince, G. H., Herbold, C., Roosen, K. & 59. Preusser, M., Lim, M., Hafler, D. A., Reardon, D. A. & 80. Liau, L. M. et al. Dendritic cell vaccination in
Tonn, J.-C. Extraneural metastases of primary brain Sampson, J. H. Prospects of immune checkpoint glioblastoma patients induces systemic and
tumors. J. Neurooncol. 53, 107–114 (2001). modulators in the treatment of glioblastoma. Nat. Rev. intracranial T-​cell responses modulated by the local
36. Westphal, M. & Lamszus, K. Circulating biomarkers for Neurol. 11, 504–514 (2015). central nervous system tumor microenvironment.
gliomas. Nat. Rev. Neurol. 11, 556 (2015). 60. Weller, M. et al. Assessment and prognostic Clin. Cancer Res. 11, 5515–5525 (2005).
37. Müller, C. et al. Hematogenous dissemination of significance of the epidermal growth factor receptor 81. Lichty, B. D., Breitbach, C. J., Stojdl, D. F. & Bell, J. C.
glioblastoma multiforme. Sci. Transl. Med. 6, vIII mutation in glioblastoma patients treated with Going viral with cancer immunotherapy. Nat. Rev.
247ra101 (2014). concurrent and adjuvant temozolomide Cancer 14, 559 (2014).
38. Roszman, T., Elliott, L. & Brooks, W. Modulation of radiochemotherapy. Int. J. Cancer 134, 2437–2447 82. Akira, S., Takeda, K. & Kaisho, T. Toll-​like receptors:
T-​cell function by gliomas. Immunol. Today 12, (2014). critical proteins linking innate and acquired immunity.
370–374 (1991). 61. van den Bent, M. J. et al. Changes in the EGFR Nat. Immunol. 2, 675 (2001).
39. Bloch, O. et al. Gliomas promote immunosuppression amplification and EGFRvIII expression between paired 83. Martuza, R. L., Malick, A., Markert, J. M.,
through induction of B7-H1 expression in tumor-​ primary and recurrent glioblastomas. Neuro. Oncol. Ruffner, K. L. & Coen, D. M. Experimental therapy of
associated macrophages. Clin. Cancer Res. 19, 17, 935–941 (2015). human glioma by means of a genetically engineered
3165–3175 (2013). 62. Felsberg, J. et al. Epidermal growth factor receptor virus mutant. Science 252, 854–856 (1991).
40. Chae, M. et al. Increasing glioma-​associated variant III (EGFRvIII) positivity in EGFR-​amplified 84. Lawler, S. E., Speranza, M.-C., Cho, C.-F. &
monocytes leads to increased intratumoral and glioblastomas: prognostic role and comparison Chiocca, E. A. Oncolytic viruses in cancer treatment: a
systemic myeloid-​derived suppressor cells in a murine between primary and recurrent tumors. Clin. Cancer review. JAMA Oncol. 3, 841–849 (2017).
model. Neuro. Oncol. 17, 978–991 (2015). Res. 23, 6846–6855 (2017). 85. Foreman, P. M., Friedman, G. K., Cassady, K. A. &
41. Li, B. et al. Comprehensive analyses of tumor 63. Schuster, J. et al. A phase II, multicenter trial of Markert, J. M. Oncolytic virotherapy for the treatment
immunity: implications for cancer immunotherapy. rindopepimut (CDX-110) in newly diagnosed of malignant glioma. Neurotherapeutics 14, 333–344
Genome Biol. 17, 174 (2016). glioblastoma: the ACT III study. Neuro. Oncol. 17, (2017).
42. Grossman, S. A. et al. Immunosuppression in patients 854–861 (2015). 86. Desjardins, A. et al. Patient survival on the dose
with high-​grade gliomas treated with radiation and 64. Sampson, J. H. et al. Immunologic escape after escalation phase of the Oncolytic Polio/Rhinovirus
temozolomide. Clin. Cancer Res. 17, 5473–5480 prolonged progression-​free survival with epidermal Recombinant (PVSRIPO) against WHO grade IV
(2011). growth factor receptor variant III peptide vaccination malignant glioma (MG) clinical trial compared to
43. Jackson, C. M. et al. Systemic tolerance mediated by in patients with newly diagnosed glioblastoma. historical controls [abstract]. J. Clin. Oncol. 34 (Suppl),
melanoma brain tumors is reversible by radiotherapy J. Clin. Oncol. 28, 4722–4729 (2010). 2061 (2016).
and vaccination. Clin. Cancer Res. 22, 1161–1172 65. Sampson, J. H. et al. Greater chemotherapy-​induced 87. Perez, O. D. et al. Design and selection of Toca 511 for
(2016). lymphopenia enhances tumor-​specific immune clinical use: modified retroviral replicating vector with
44. Chongsathidkiet, P. et al. Downregulation of responses that eliminate EGFRvIII-​expressing tumor improved stability and gene expression. Mol. Ther. 20,
sphingosine-1-phosphate receptor type 1 mediates cells in patients with glioblastoma. Neuro. Oncol. 13, 1689–1698 (2012).
T-​cell sequestration in bone marrow amidst 324–333 (2011). 88. Cloughesy, T. F. et al. Phase 1 trial of vocimagene
glioblastoma. J. Neurosurg. 126, 1442 (2017). 66. Weller, M. et al. Rindopepimut with temozolomide for amiretrorepvec and 5-fluorocytosine for recurrent
45. Wainwright, D. A. et al. Durable therapeutic efficacy patients with newly diagnosed, EGFRvIII-​expressing high-​grade glioma. Sci. Transl Med. 8, 341ra75
utilizing combinatorial blockade against IDO, CTLA-4, glioblastoma (ACT IV): results of a randomized, (2016).
and PD-​L1 in mice with brain tumors. Clin. Cancer Res. double-​blind, international phase 3 trial. Lancet 89. Sonabend, A. M., Ulasov, I. V., Han, Y. & Lesniak, M. S.
20, 5290–5301 (2014). Oncol. 18, 1373–1385 (2017). Oncolytic adenoviral therapy for glioblastoma
46. Heimberger, A. B. et al. Immunological responses in a 67. Reardon, D. A. et al. ReACT: Overall survival from multiforme. Neurosurg. Focus 20, E19 (2006).
patient with glioblastoma multiforme treated with a randomized phase II study of rindopepimut 90. Lamfers, M. L. M. et al. Potential of the conditionally
sequential courses of temozolomide and (CDX-110) plus bevacizumab in relapsed glioblastoma. replicative adenovirus Ad5-Δ24RGD in the treatment
immunotherapy: case study. Neuro. Oncol. 10, J. Clin. Oncol. 33, 2009 (2015). of malignant gliomas and its enhanced effect with
98–103 (2008). 68. Khan, K. A. & Kerbel, R. S. Improving immunotherapy radiotherapy. Cancer Res. 62, 5736–5742 (2002).
47. Bodmer, S. et al. Immunosuppression and outcomes with anti-​angiogenic treatments and vice 91. Chiocca, E. A. et al. A phase I open-​label, dose-​
transforming growth factor-​beta in glioblastoma. versa. Nat. Rev. Clin. Oncol. https://doi.org/10.1038/ escalation, multi-​institutional trial of injection with an
Preferential production of transforming growth nrclinonc.2018.9 (2018). E1B-​attenuated adenovirus, ONYX-015, into the
factor-​beta 2. J. Immunol. 143, 3222–3229 69. Fukumura, D., Kloepper, J., Amoozgar, Z., Duda, D. G. peritumoral region of recurrent malignant gliomas, in
(1989). & Jain, R. K. Enhancing cancer immunotherapy using the adjuvant setting. Mol. Ther. 10, 958–966 (2004).
48. Huettner, C., Czub, S., Kerkau, S., Roggendorf, W. & antiangiogenics: opportunities and challenges. 92. Bischoff, J. R. et al. An adenovirus mutant that
Tonn, J.-C. Interleukin 10 is expressed in human Nat. Rev. Clin. Oncol. https://doi.org/10.1038/ replicates selectively in p53-deficient human tumor
gliomas in vivo and increases glioma cell proliferation nrclinonc.2018.29 (2018). cells. Science 274, 373–376 (1996).
and motility in vitro. Anticancer Res. 17, 3217–3224 70. Schumacher, T. et al. A vaccine targeting mutant IDH1 93. Coffin, R. S. From virotherapy to oncolytic
(1997). induces antitumour immunity. Nature 512, 324 immunotherapy: where are we now? Curr. Opin. Virol.
49. Huettner, C., Paulus, W. & Roggendorf, W. Messenger (2014). 13, 93–100 (2015).
RNA expression of the immunosuppressive cytokine 71. Prins, R. M. et al. The TLR-7 agonist, imiquimod, 94. Wheeler, L. A. et al. Phase II multicenter study of
IL-10 in human gliomas. Am. J. Pathol. 146, 317 enhances dendritic cell survival and promotes tumor gene-​mediated cytotoxic immunotherapy as adjuvant
(1995). antigen-​specific T cell priming: relation to central to surgical resection for newly diagnosed malignant
50. Lauro, G. M., Di Lorenzo, N., Grossi, M., Maleci, A. & nervous system antitumor immunity. J. Immunol. 176, glioma. Neuro. Oncol. 18, 1137–1145 (2016).
Guidetti, B. Prostaglandin E 2 as an 157–164 (2006). 95. Chiocca, E. A. et al. Phase IB study of gene-​mediated
immunomodulating factor released in vitro by human 72. Tchirkov, A. et al. Clinical implications of quantitative cytotoxic immunotherapy adjuvant to up-​front surgery
glioma cells. Acta Neuropathol. 69, 278–282 (1986). real-​time RT–PCR analysis of hTERT gene expression and intensive timing radiation for malignant glioma.
51. Wischhusen, J., Friese, M. A., Mittelbronn, M., in human gliomas. Br. J. Cancer 88, 516 (2003). J. Clin. Oncol. 29, 3611–3619 (2011).
Meyermann, R. & Weller, M. HLA-​E protects glioma 73. Killela, P. J. et al. TERT promoter mutations occur 96. Ji, N. et al. Adenovirus-​mediated delivery of herpes
cells from NKG2D-​mediated immune responses frequently in gliomas and a subset of tumors derived simplex virus thymidine kinase administration
in vitro: implications for immune escape in vivo. from cells with low rates of self-​renewal. Proc. Natl improves outcome of recurrent high-​grade glioma.
J. Neuropathol. Exp. Neurol. 64, 523–528 (2005). Acad. Sci. USA 110, 6021–6026 (2013). Oncotarget 7, 4369–4378 (2016).
52. Wiendl, H. et al. A functional role of HLA-​G expression 74. Suso, E. M. I. et al. hTERT mRNA dendritic cell 97. Phuong, L. K. et al. Use of a vaccine strain of measles
in human gliomas: an alternative strategy of immune vaccination: complete response in a pancreatic cancer virus genetically engineered to produce
escape. J. Immunol. 168, 4772–4780 (2002). patient associated with response against several carcinoembryonic antigen as a novel therapeutic agent
53. Didenko, V. V., Ngo, H. N., Minchew, C. & Baskin, D. S. hTERT epitopes. Cancer Immunol. Immunother. 60, against glioblastoma multiforme. Cancer Res. 63,
Apoptosis of T lymphocytes invading glioblastomas 809–818 (2011). 2462–2469 (2003).
multiforme: a possible tumor defense mechanism. 75. Rampling, R. et al. A Cancer Research UK first time in 98. Russell, S. J., Peng, K.-W. & Bell, J. C. Oncolytic
J. Neurosurg. 96, 580–584 (2002). human phase I trial of IMA950 (novel multipeptide virotherapy. Nat. Biotechnol. 30, 658–670 (2012).
54. Parsa, A. T. et al. Loss of tumor suppressor PTEN therapeutic vaccine) in patients with newly diagnosed 99. Wollmann, G., Ozduman, K. & van den Pol, A. N.
function increases B7-H1 expression and glioblastoma. Clin. Cancer Res. 22, 4776–4785 Oncolytic virus therapy of glioblastoma
immunoresistance in glioma. Nat. Med. 13, 84 (2016). multiforme–concepts and candidates. Cancer J. 18,
(2007). 76. Phuphanich, S. et al. Phase I trial of a multi-​epitope- 69 (2012).
55. Parney, I. F., Waldron, J. S. & Parsa, A. T. Flow pulsed dendritic cell vaccine for patients with newly 100. Luke, J. J., Flaherty, K. T., Ribas, A. & Long, G. V.
cytometry and in vitro analysis of human diagnosed glioblastoma. Cancer Immunol. Targeted agents and immunotherapies: optimizing
glioma–associated macrophages. J. Neurosurg. 110, Immunother. 62, 125–135 (2013). outcomes in melanoma. Nat. Rev. Clin. Oncol. 14, 463
572–582 (2009). 77. Wen, P., Reardon, D. A., Phuphanich, S. & Aiken, R. A (2017).
56. Dunn, G. P., Dunn, I. F. & Curry, W. T. Focus on TILs: randomized, double-​blind, placebo-​controlled phase 2 101. Topalian, S. L., Taube, J. M., Anders, R. A. & Pardoll,
prognostic significance of tumor infiltrating trial of dendritic cell (DC) vaccination with ICT-107 in D. M. Mechanism-​driven biomarkers to guide immune

440 | JULY 2018 | volume 15 www.nature.com/nrclinonc

© 2018 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
 Reviews

checkpoint blockade in cancer therapy. Nat. Rev. 125. Brown, C. E. et al. Optimization of IL13Rα2-targeted 148. Cohen, J. D. et al. Detection and localization of
Cancer 16, 275 (2016). chimeric antigen receptor T cells for improved anti-​ surgically resectable cancers with a multi-​analyte
102. Topalian, S. L. et al. Safety, activity, and immune tumor efficacy against glioblastoma. Mol. Ther. 26, blood test. Science 359, 926–930 (2018).
correlates of anti-​PD-1 antibody in cancer. N. Engl. 31–44 (2018). 149. Bettegowda, C. et al. Detection of circulating tumor
J. Med. 366, 2443–2454 (2012). 126. O’Rourke, D. M. et al. A single dose of peripherally DNA in early- and late-​stage human malignancies.
103. Lipson, E. J. et al. Antagonists of PD-1 and PD-​L1 in infused EGFRvIII-​directed CAR T cells mediates Sci. Transl Med. 6, 224ra24 (2014).
cancer treatment. Semin. Oncol. 42, 587–600 (2015). antigen loss and induces adaptive resistance in 150. Figueroa, J. M. & Carter, B. S. Detection of
104. Berghoff, A. S. et al. Programmed death ligand 1 patients with recurrent glioblastoma. Sci. Transl Med. glioblastoma in biofluids. J. Neurosurg. https://doi.
expression and tumor-​infiltrating lymphocytes in 9, eaaa0984 (2017). org/10.3171/2017.3.JNS162280 (2017).
glioblastoma. Neuro. Oncol. 17, 1064–1075 (2015). 127. Dai, H., Wang, Y., Lu, X. & Han, W. Chimeric antigen 151. Wang, Y. et al. Detection of tumor-​derived DNA in
105. Nduom, E. K. et al. PD-​L1 expression and prognostic receptors modified T-​cells for cancer therapy. J. Natl. cerebrospinal fluid of patients with primary tumors of
impact in glioblastoma. Neuro. Oncol. 18, 195–205 Cancer Inst. 108, djv439 (2016). the brain and spinal cord. Proc. Natl Acad. Sci. USA
(2016). 128. Fesnak, A. D., June, C. H. & Levine, B. L. Engineered 112, 9704–9709 (2015).
106. Fecci, P. E. et al. Systemic CTLA-4 blockade T cells: the promise and challenges of cancer 152. Figueroa, J. M. et al. Detection of wtEGFR
ameliorates glioma-​induced changes to the CD4+T cell immunotherapy. Nat. Rev. Cancer 16, 566 (2016). amplification and EGFRvIII mutation in CSF-​derived
compartment without affecting regulatory T-​cell 129. Morales-​Kastresana, A., Labiano, S., Quetglas, J. I. & extracellular vesicles of glioblastoma patients. Neuro.
function. Clin. Cancer Res. 13, 2158–2167 (2007). Melero, I. Better performance of CARs deprived of the Oncol. https://doi.org/10.1093/neuonc/nox085
107. Reardon, D. A. et al. Glioblastoma eradication PD-1 brake. Clin. Cancer Res. 19, 5546–5548 (2017).
following immune checkpoint blockade in an (2013). 153. Huang, T. Y. et al. Detection of Histone H3 mutations in
orthotopic, immunocompetent model. Cancer 130. Ninomiya, S. et al. Tumor indoleamine 2, cerebrospinal fluid-​derived tumor DNA from children
Immunol. Res. 4, 124–135 (2016). 3-dioxygenase (IDO) inhibits CD19-CAR T cells and is with diffuse midline glioma. Acta Neuropathol.
108. Zeng, J. et al. Anti-​PD-1 blockade and stereotactic downregulated by lymphodepleting drugs. Blood 125, Commun. 5, 28 (2017).
radiation produce long-​term survival in mice with 3905–3916 (2015). 154. Pentsova, E. I. et al. Evaluating cancer of the central
intracranial gliomas. Int. J. Radiat. Oncol. Biol. Phys. 131. Pyonteck, S. M. et al. CSF-1R inhibition alters nervous system through next-​generation sequencing
86, 343–349 (2013). macrophage polarization and blocks glioma of cerebrospinal fluid. J. Clin. Oncol. 34, 2404–2415
109. Weller, M. et al. Vaccine-​based immunotherapeutic progression. Nat. Med. 19, 1264 (2013). (2016).
approaches to gliomas and beyond. Nat. Rev. Neurol. 132. Mathios, D. et al. Anti-​PD-1 antitumor immunity is 155. De Mattos-​Arruda, L. et al. Cerebrospinal fluid-​
13, 363–374 (2017). enhanced by local and abrogated by systemic derived circulating tumour DNA better represents the
110. Sampson, J. H. et al. Preliminary safety and activity of chemotherapy in GBM. Sci. Transl Med. 8, 370ra180 genomic alterations of brain tumours than plasma.
nivolumab and its combination with ipilimumab in (2016). Nat. Commun. 6, 8839 (2015).
recurrent glioblastoma (GBM): CHECKMATE-143 133. Wild, A. T. et al. Lymphocyte-​sparing effect of 156. Locasale, J. W. et al. Metabolomics of human
[abstract]. J. Clin. Oncol. 33 (Suppl.), 3010 (2015). stereotactic body radiation therapy in patients with cerebrospinal fluid identifies signatures of malignant
111. Reardon, D. A. et al. Randomized phase 3 study unresectable pancreatic cancer. Int. J. Radiat. Oncol. glioma. Mol. Cell. Proteom. 11, M111.014688
evaluating the efficacy and safety of nivolumab versus Biol. Phys. 94, 571–579 (2016). (2012).
bevacizumab in patients with recurrent glioblastoma: 134. Yovino, S., Kleinberg, L., Grossman, S. A., 157. Hao, C. et al. Cytokine and cytokine receptor mRNA
CheckMate 143 [abstract]. Neuro. Oncol. 19 Narayanan, M. & Ford, E. The etiology of treatment-​ expression in human glioblastomas: evidence of Th1,
(Suppl. 3), OS10.3 (2017). related lymphopenia in patients with malignant Th2 and Th3 cytokine dysregulation. Acta
112. Omuro, A. et al. Nivolumab with or without gliomas: modeling radiation dose to circulating Neuropathol. 103, 171–178 (2002).
ipilimumab in patients with recurrent glioblastoma: lymphocytes explains clinical observations and suggests 158. Wherry, E. J. T cell exhaustion. Nat. Immunol. 12, 492
results from exploratory phase 1 cohorts of methods of modifying the impact of radiation on (2011).
CheckMate 143. Neuro Oncol. https://doi. immune cells. Cancer Invest. 31, 140–144 (2013). 159. Zaretsky, J. M. et al. Mutations associated with
org/10.1093/neuonc/nox208 (2017). 135. Horvat, T. Z. et al. Immune-​related adverse events, acquired resistance to PD-1 blockade in melanoma.
113. Lim, M. et al. Nivolumab (nivo) in combination with need for systemic immunosuppression, and effects on N. Engl. J. Med. 375, 819–829 (2016).
radiotherapy (RT) ± temozolomide (TMZ): updated survival and time to treatment failure in patients with 160. Sade-​Feldman, M. et al. Resistance to checkpoint
safety results from CheckMate 143 in pts with melanoma treated with ipilimumab at Memorial blockade therapy through inactivation of antigen
methylated or unmethylated newly diagnosed Sloan Kettering Cancer Center. J. Clin. Oncol. 33, presentation. Nat. Commun. 8, 1136 (2017).
glioblastoma (GBM) [abstract]. Ann. Oncol. 28 3193–3198 (2015). 161. Gao, J. et al. Loss of IFN-​γ pathway genes in tumor
(Suppl. 5), 3250 (2017). 136. Pitter, K. L. et al. Corticosteroids compromise survival cells as a mechanism of resistance to anti-​CTLA-4
114. Roth, P., Valavanis, A. & Weller, M. Long-​term control in glioblastoma. Brain 139, 1458–1471 (2016). therapy. Cell 167, 397–404 (2016).
and partial remission after initial pseudoprogression 137. Hygino da Cruz, L. C., Rodriguez, I., Domingues, R. C., 162. Yeung, J. T. et al. LOH in the HLA class I region at
of glioblastoma by anti-​PD-1 treatment with Gasparetto, E. L. & Sorensen, A. G. 6p21 is associated with shorter survival in newly
nivolumab. Neuro. Oncol. 19, 454–456 (2017). Pseudoprogression and pseudoresponse: imaging diagnosed adult glioblastoma. Clin. Cancer Res. 19,
115. Bouffet, E. et al. Immune checkpoint inhibition for challenges in the assessment of posttreatment glioma. 1816–1826 (2013).
hypermutant glioblastoma multiforme resulting from AJNR Am. J. Neuroradiol. 32, 1978–1985 (2011). 163. Ferguson, S. D., Srinivasan, V. M. & Heimberger, A. B.
germline biallelic mismatch repair deficiency. 138. Ryken, T. C. et al. The role of imaging in the The role of STAT3 in tumor-​mediated immune
J. Clin. Oncol. 34, 2206–2211 (2016). management of progressive glioblastoma: a suppression. J. Neurooncol. 123, 385–394 (2015).
116. Johanns, T. M. et al. Immunogenomics of hypermutated systematic review and evidence-​based clinical practice 164. Eil, R. et al. Ionic immune suppression within the
glioblastoma: a patient with germline POLE deficiency guideline. J. Neurooncol. 118, 435–460 (2014). tumour microenvironment limits T cell effector
treated with checkpoint blockade immunotherapy. 139. Chiou, V. L. & Burotto, M. Pseudoprogression and function. Nature 537, 539–543 (2016).
Cancer Discov. 6, 1230–1236 (2016). immune-​related response in solid tumors. J. Clin. 165. Koyama, S. et al. Adaptive resistance to therapeutic
117. [No authors listed.] FDA grants accelerated approval Oncol. 33, 3541–3543 (2015). PD-1 blockade is associated with upregulation of
to pembrolizumab for first tissue/site agnostic 140. Hodi, F. S. et al. Evaluation of immune-​related alternative immune checkpoints. Nat. Commun. 7,
indication. U.S. Food & Drug Administration https:// response criteria and RECISTv 1.1 in patients with 10501 (2016).
www.fda.gov/Drugs/InformationOnDrugs/ advanced melanoma treated with pembrolizumab. 166. Kim, J. E. et al. Combination therapy with anti-​PD-1,
ApprovedDrugs/ucm560040.htm (2017). J. Clin. Oncol. 34, 1510–1517 (2016). anti-​TIM-3, and focal radiation results in regression of
118. Maxwell, J. A. et al. Mismatch repair deficiency does 141. Okada, H. et al. Immunotherapy Response Assessment murine gliomas. Clin. Cancer Res. 23, 124–136
not mediate clinical resistance to temozolomide in in Neuro-​Oncology (iRANO): a report of the RANO (2017).
malignant glioma. Clin. Cancer Res. 14, 4859–4868 Working Group. Lancet Oncol. 16, 534–542 167. Wu, A. et al. Glioma cancer stem cells induce
(2008). (2015). immunosuppressive macrophages/microglia.
119. Jena, B., Dotti, G. & Cooper, L. J. N. Redirecting T-​cell 142. Everson, R. G. et al. Cytokine responsiveness of Neuro. Oncol. 12, 1113–1125 (2010).
specificity by introducing a tumor-​specific chimeric CD8 + T cells is a reproducible biomarker for the 168. Heimberger, A. B. et al. Incidence and prognostic
antigen receptor. Blood 116, 1035–1044 (2010). clinical efficacy of dendritic cell vaccination in impact of FoxP3+ regulatory T cells in human gliomas.
120. Morgan, R. A. et al. Recognition of glioma stem cells glioblastoma patients. J. Immunother. Cancer 2, 10 Clin. Cancer Res. 14, 5166–5172 (2008).
by genetically modified T cells targeting EGFRvIII and (2014). 169. Stevens, A., Klöter, I. & Roggendorf, W. Inflammatory
development of adoptive cell therapy for glioma. 143. Zhai, L. et al. The kynurenine to tryptophan ratio as a infiltrates and natural killer cell presence in human
Human Gene Therapy 23, 1043–1053 (2012). prognostic tool for glioblastoma patients enrolling in brain tumors. Cancer 61, 738–743 (1988).
121. Finney, H. M. Akbar, A. N. & Lawson, A. D. G. immunotherapy. J. Clin. Neurosci. 22, 1964–1968 170. Quail, D. F. et al. The tumor microenvironment
Activation of resting human primary T cells with (2015). underlies acquired resistance to CSF-1R inhibition in
chimeric receptors: costimulation from CD28, 144. Patel, S. P. & Kurzrock, R. PD-​L1 expression as a gliomas. Science 352, aad3018 (2016).
inducible costimulator, CD134, and CD137 in series predictive biomarker in cancer immunotherapy. 171. Patel, M. A. et al. Agonist anti-​GITR monoclonal
with signals from the TCRζ chain. J. Immunol. 172, Mol. Cancer Ther. 14, 847–856 (2015). antibody and stereotactic radiation induce
104–113 (2004). 145. Snyder, A. et al. Genetic basis for clinical response to immune-​mediated survival advantage in murine
122. Brown, C. E. et al. Regression of glioblastoma after CTLA-4 blockade in melanoma. N. Engl. J. Med. 371, intracranial glioma. J. Immunother. Cancer 4, 28
chimeric antigen receptor T-​cell therapy. N. Engl. 2189–2199 (2014). (2016).
J. Med. 375, 2561–2569 (2016). 146. Cescon, D. W., Haibe-​Kains, B. & Mak, T. W. 172. Mathios, D. et al. Therapeutic administration of IL-15
123. Brown, C. E. et al. Bioactivity and safety of IL13Rα2- APOBEC3B expression in breast cancer reflects superagonist complex ALT-803 leads to long-​term
redirected chimeric antigen receptor CD8 + T cells in cellular proliferation, while a deletion polymorphism is survival and durable antitumor immune response in
patients with recurrent glioblastoma. Clin. Cancer Res. associated with immune activation. Proc. Natl Acad. a murine glioblastoma model. Int. J. Cancer 138,
21, 4062–4072 (2015). Sci. USA 112, 2841–2846 (2015). 187–194 (2016).
124. Neelapu, S. S. et al. Chimeric antigen receptor T-​cell 147. Wu, A. & Lim, M. Issues to consider in designing 173. Belcaid, Z. et al. Focal radiation therapy combined
therapy — assessment and management of toxicities. immunotherapy clinical trials for glioblastoma with 4-1BB activation and CTLA-4 blockade yields
Nat. Rev. Clin. Oncol. 15, 47 (2018). management. J. Cancer Ther. 7, 573 (2016). long-​term survival and a protective antigen-​specific

NaTUre RevIews | CLinicAL OncoLogy volume 15 | JULY 2018 | 441

© 2018 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
Reviews

memory response in a murine glioma model. PLOS 185. Bloch, O. et al. Heat-​shock protein peptide 197. Markert, J. M. et al. A phase 1 trial of oncolytic
ONE 9, e101764 (2014). complex-96 vaccination for recurrent glioblastoma: HSV-1, G207, given in combination with radiation
174. Sharabi, A. B. et al. Stereotactic radiation therapy a phase II, single-​arm trial. Neuro. Oncol. 16, for recurrent GBM demonstrates safety and
augments antigen-​specific PD-1-mediated antitumor 274–279 (2014). radiographic responses. Mol. Ther. 22, 1048–1055
immune responses via cross-​presentation of tumor 186. Sloan, A. E. et al. Adoptive immunotherapy in (2014).
antigen. Cancer Immunol. Res. 3, 345–355 (2015). patients with recurrent malignant glioma: preliminary 198. Dillman, R. O. et al. Intralesional lymphokine-​activated
175. Postow, M. A. et al. Immunologic correlates of the results of using autologous whole-​tumor vaccine killer cells as adjuvant therapy for primary
abscopal effect in a patient with melanoma. N. Engl. plus granulocyte-​macrophage colony–stimulating glioblastoma. J. Immunother. 32, 914–919 (2009).
J. Med. 366, 925–931 (2012). factor and adoptive transfer of anti-​CD3–activated 199. Plautz, G. E. et al. T cell adoptive immunotherapy of
176. Chen, J. Y., Hovey, E., Rosenthal, M., Livingstone, A. & lymphocytes. Neurosurg. Focus 9, e9 (2000). newly diagnosed gliomas. Clin. Cancer Res. 6, 2209
Simes, J. Neuro-​oncology practices in Australia: a 187. Sampson, J. H. et al. A pilot study of IL-2Rα (2000).
Cooperative Group for Neuro-​Oncology patterns of care blockade during lymphopenia depletes regulatory 200. Thaci, B. et al. Significance of interleukin-13 receptor
study. Asia. Pac. J. Clin. Oncol. 10, 162–167 (2014). T-​cells and correlates with enhanced immunity in alpha 2-targeted glioblastoma therapy. Neuro. Oncol.
177. Fadul, C. E. et al. Immune response in patients with patients with glioblastoma. PLoS ONE 7, e31046 16, 1304–1312 (2014).
newly diagnosed glioblastoma multiforme treated with (2012). 201. Reap, E. et al. Dendritic cells enhance polyfunctionality
intranodal autologous tumor lysate-​dendritic cell 188. Vlahovic, G. et al. Feasibility and safety study of GBM of adoptively transferred T cells which target
vaccination after radiation chemotherapy. stem cell tumor amplified RNA immunotherapy in cytomegalovirus in glioblastoma. Cancer Res. 78,
J. Immunother. 34, 382–389 (2011). recurrent glioblastoma. Neuro. Oncol. 15, iii68–iii74 256–264 (2017).
178. Inogés, S. et al. A phase II trial of autologous dendritic (2013).
cell vaccination and radiochemotherapy following 189. Fenstermaker, R. A. et al. Clinical study of a survivin Acknowledgements
fluorescence-​guided surgery in newly diagnosed long peptide vaccine (SurVaxM) in patients with The authors thank A. Wu of the Johns Hopkins University
glioblastoma patients. J. Transl Med. 15, 104 recurrent malignant glioma. Cancer Immunol. School of Medicine for her help in formatting the manuscript.
(2017). Immunother. 65, 1339–1352 (2016).
179. Wheeler, C. J. et al. Vaccination elicits correlated 190. Olin, M. R. et al. Vaccination with dendritic cells loaded Author contributions
immune and clinical responses in glioblastoma with allogeneic brain tumor cells for recurrent All authors contributed to researching data for the article,
multiforme patients. Cancer Res. 68, 5955–5964 malignant brain tumors induces a CD4+IL17+response. discussions of content, and writing, reviewing, and editing the
(2008). J. Immunother. Cancer 2, 4 (2014). manuscript.
180. Jouanneau, E. et al. Intrinsically de-​sialylated 191. Prins, R. M. et al. Comparison of glioma-​associated
CD103+CD8 T cells mediate beneficial anti-​glioma antigen peptide-​loaded versus autologous tumor Competing interests
immune responses. Cancer Immunol. Immunother. 63, lysate-​loaded dendritic cell vaccination in malignant M.L. has received research funding from Accuray, Agenus,
911–924 (2014). glioma patients. J. Immunother. 36, 152–157 Altor, Arbor, BMS, Celldex, and Immunocellular, and has been
181. Bloch, O. et al. Autologous heat shock protein peptide (2013). a consultant for Agenus, Baxter, BMS, Boston Biomedical,
vaccination for newly diagnosed glioblastoma: impact 192. Fu, S. et al. Initial phase 1 study of WT2725 dosing Oncorus, Regeneron, SQZ Biotechnologies, Stryker, and
of peripheral PD-​L1 expression on response to emulsion in patients with advanced malignancies. Tocagen. M.W. has received research grants from Acceleron,
therapy. Clin. Cancer Res. 23, 3575–3584 (2017). J. Clin. Oncol. 35, 2066 (2017). Actelion, Bayer, Merck (EMD), MSD, Novocure, OGD2,
182. Vik-​Mo, E. O. et al. Therapeutic vaccination against 193. Geletneky, K. et al. Oncolytic H-1 parvovirus shows PIQUR, and Roche, and has received honoraria for lectures,
autologous cancer stem cells with mRNA-​transfected safety and signs of immunogenic activity in a first advisory board participation, or consulting from AbbVie, BMS,
dendritic cells in patients with glioblastoma. phase I/IIa glioblastoma trial. Mol. Ther. 25, Celldex, Merck (EMD), MSD, Novocure, Pfizer, Roche, Teva,
Cancer Immunol. Immunother. 62, 1499–1509 2620–2634 (2017). and Tocagen. Y.X. and C.B. declare no competing interests.
(2013). 194. Alonso, M. M. et al. Oncolytic virus DNX-2401 with
183. Dutoit, V. et al. IMA950 multipeptide vaccine a short course of temozolomide for glioblastoma at Publisher’s note
adjuvanted with poly-​ICLC in combination with first recurrence: clinical data and prognostic Springer Nature remains neutral with regard to jurisdictional
standard therapy in newly diagnosed HLA-​A2 biomarkers [abstract]. Cancer Res. 77 (Suppl.), CT027 claims in published maps and institutional affiliations.
glioblastoma patients [abstract]. Ann. Oncol. 28 (2017).
(Suppl. 11), 11PD (2017). 195. Markert, J. M. et al. Conditionally replicating herpes Supplementary information
184. Salacz, M. E., Camarata, P. J., Ots, M., Mcintire, J. & simplex virus mutant, G207 for the treatment of Supplementary information is available for this paper at
Lovick, D. TVI-​Brain-1 — a phase I study to test the malignant glioma: results of a phase I trial. Gene Ther. https://doi.org/10.1038/s41571-018-0003-5.
safety of a combination of autologous cancer cell 7, 867–874 (2000).
vaccination, adoptive transfer of cancer antigen-​ 196. Kicielinski, K. P. et al. Phase 1 clinical trial of
specific effector T cells and low-​dose interleukin 2 intratumoral reovirus infusion for the treatment of Related links
during treatment of patients with recurrent grade III/IV recurrent malignant gliomas in adults. Mol. Ther. 22, ClinicalTrials.gov database: https://clinicaltrials.gov/
glioma. Neuro. Oncol. 14, vi43–vi49 (2012). 1056–1062 (2014).

442 | JULY 2018 | volume 15 www.nature.com/nrclinonc

© 2018 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.