Bevacizumab plus hypofractionated radiotherapy versus radiotherapy alone in elderly patients with glioblastoma: The randomized, open-label, phase II ARTE trial

Abstract

Background:
The addition of bevacizumab to temozolomide-based chemoradiotherapy (TMZ/RT→TMZ) did not prolong overall survival (OS) in patients with newly diagnosed glioblastoma in phase III trials. Elderly and frail patients are underrepresented in clinical trials, but early reports suggested preferential benefit in this population.

Patients and methods:
ARTE was a 2:1 randomized, multi-center, open-label, non-comparative phase II trial of hypofractionated RT (40 Gy in 15 fractions) with bevacizumab (10 mg/kg x 14 days) (arm A, N = 50) or without bevacizumab (arm B, N = 25) in patients with newly diagnosed glioblastoma aged ≥65 years. The primary objective was to obtain evidence for prolongation of median OS by the addition of bevacizumab to RT. Response was assessed by RANO criteria. Quality of life (QoL) was monitored by the EORTC QLQ-C30/BN20 modules. Exploratory studies included molecular subtyping by 450k whole methylome and gene expression analyses.

Results:
Median PFS was longer in arm A than in arm B (7.6 and 4.8 months, P = 0.003), but OS was similar (12.1 and 12.2 months, P = 0.77). Clinical deterioration was delayed and more patients came off steroids in arm A. Prolonged PFS in arm A was confined to tumors with the receptor tyrosine kinase (RTK) I methylation subtype (HR 0.25, P = 0.014) and proneural gene expression (HR 0.29, P = 0.025). In a Cox model of OS controlling for established prognostic factors, associations with more favorable outcome were identified for age <70 years (HR 0.52, P = 0.018) and Karnofsky performance score (KPS) 90-100% (HR 0.51, P = 0.026). Including molecular subtypes into that model identified an association of the RTK II gene methylation subtype with inferior OS (HR 1.73, P = 0.076).

Conclusion:
Efficacy outcomes and exploratory analyses of ARTE do not support the hypothesis that the addition of bevacizumab to RT generally prolongs survival in elderly glioblastoma patients. Molecular biomarkers may identify patients with preferential benefit from bevacizumab.

Clinical trial registration number:
NCT01443676.

Full text

ORIGINAL ARTICLE
Bevacizumab plus hypofractionated radiotherapy
versus radiotherapy alone in elderly patients with
glioblastoma: the randomized, open-label, phase II
ARTE trial
H.-G. Wirsching
1,2†
, G. Tabatabai
1,2
, U. Roelcke
3
, A. F. Hottinger
4,5
,F.Jo¨rger
6
, A. Schmid
7
, L. Plasswilm
8
,
D. Schrimpf
9,10
, C. Mancao
11
, D. Capper
9,10
, K. Conen
12
, T. Hundsberger
13
, F. Caparrotti
14
,
R. von Moos
15
, C. Riklin
16
, J. Felsberg
17
, P. Roth
1,2
, D. T. W. Jones
10,18
, S. Pfister
10,18
, E. J. Rushing
1,19
,
L. Abrey
20
, G. Reifenberger
17,21
, L. Held
22
, A. von Deimling
9,10
, A. Ochsenbein
7
& M. Weller
1,2
*
1
Brain Tumor Center Zurich;
2
Department of Neurology, University Hospital and University of Zurich, Zurich;
3
Department of Neurology, Brain Tumor Center Aarau,
Cantonal Hospital Aarau, Aarau; Departments of
4
Clinical Neurosciences;
5
Medical Oncology, University Hospital Lausanne, Lausanne;
6
Clinical Trials Center,
University Hospital and University of Zurich, Zurich;
7
Department of Medical Oncology, University Hospital Bern, Bern;
8
Department of Radiation Oncology,
Cantonal Hospital St. Gallen, St. Gallen, Switzerland;
9
Department of Neuropathology, Ruprecht-Karls-University Heidelberg;
10
German Cancer Research Center
(DKFZ), Heidelberg, Germany;
11
Genentech, Oncology Biomarker Development, Basel;
12
Department of Medical Oncology, University Hospital Basel, Basel;
13
Department of Neurology, Cantonal Hospital St. Gallen, St. Gallen;
14
Department of Radiation Oncology, University Hospital Geneva, Geneva;
15
Department of
Medical Oncology, Cantonal Hospital Chur, Chur;
16
Department of Medical Oncology, Cantonal Hospital Lucerne, Lucerne Switzerland;
17
Department of
Neuropathology, Heinrich Heine University Du¨ sseldorf, Du¨sseldorf;
18
Department of Pediatric Hematology and Oncology, Ruprecht-Karls-University Heidelberg,
Heidelberg, Germany;
19
Department of Neuropathology, University Hospital Zurich, Zurich;
20
F. Hoffmann-La Roche, Pharma Division, Product Development
Oncology, Basel, Switzerland;
21
German Cancer Research Center, Essen/Du¨ sseldorf, Germany;
22
Biostatistics Department, University of Zurich, Zurich, Switzerland
*Correspondence to: Prof. Michael Weller, Department of Neurology, University Hospital and University of Zurich, Frauenklinikstrasse 26, CH-8091 Zu¨rich, Switzerland.
Tel: þ41-44-255-5500; Fax: þ41-44-255-4507; E-mail: michael.weller@usz.ch
†
Present address: Human Biology Division, Fred Hutchinson Cancer Research Center, 1100 Fairview Avenue North, Mail Stop C3-168, Seattle, WA 98109, USA.
Background: The addition of bevacizumab to temozolomide-based chemoradiotherapy (TMZ/RT !TMZ) did not prolong
overall survival (OS) in patients with newly diagnosed glioblastoma in phase III trials. Elderly and frail patients are underrepresented
in clinical trials, but early reports suggested preferential benefit in this population.
Patients and methods: ARTE was a 2 : 1 randomized, multi-center, open-label, non-comparative phase II trial of
hypofractionated RT (40 Gy in 15 fractions) with bevacizumab (10 mg/kg14 days) (arm A, N¼50) or without bevacizumab (arm
B, N¼25) in patients with newly diagnosed glioblastoma aged 65 years. The primary objective was to obtain evidence for
prolongation of median OS by the addition of bevacizumab to RT. Response was assessed by RANO criteria. Quality of life (QoL)
was monitored by the EORTC QLQ-C30/BN20 modules. Exploratory studies included molecular subtyping by 450k whole
methylome and gene expression analyses.
Results: Median PFS was longer in arm A than in arm B (7.6 and 4.8 months, P¼0.003), but OS was similar (12.1 and
12.2 months, P¼0.77). Clinical deterioration was delayed and more patients came off steroids in arm A. Prolonged PFS in arm A
was confined to tumors with the receptor tyrosine kinase (RTK) I methylation subtype (HR 0.25, P¼0.014) and proneural gene
expression (HR 0.29, P¼0.025). In a Cox model of OS controlling for established prognostic factors, associations with more
favorable outcome were identified for age <70 years (HR 0.52, P¼0.018) and Karnofsky performance score 90%–100% (HR 0.51,
P¼0.026). Including molecular subtypes into that model identified an association of the RTK II gene methylation subtype with
inferior OS (HR 1.73, P¼0.076).
Conclusion: Efficacy outcomes and exploratory analyses of ARTE do not support the hypothesis that the addition of
bevacizumab to RT generally prolongs survival in elderly glioblastoma patients. Molecular biomarkers may identify patients with
preferential benefit from bevacizumab.
V
CThe Author(s) 2018. Published by Oxford University Press on behalf of the European Society for Medical Oncology.
All rights reserved. For permissions, please email: journals.permissions@oup.com.
Annals of Oncology 29: 1423–1430, 2018
doi:10.1093/annonc/mdy120
Published online 10 April 2018

Clinical trial registration number: NCT01443676.
Key words:glioblastoma, elderly, bevacizumab, radiotherapy, molecular subtype
Introduction
Almost 50% of all patients with glioblastoma are older than
65 years, <20% of these survive the first year [1]. Many elderly
patients with glioblastoma are deemed too frail to tolerate com-
bined post-operative temozolomide chemoradiotherapy (TMZ/
RT !TMZ) [2,3], rendering monotherapy regimens of hypo-
fractionated RT [4] or TMZ alone preferred treatment options
for some of these patients [5–7].
The vascular endothelial growth factor (VEGF)-targeting anti-
body bevacizumab (Avastin
TM
) was considered an optimal can-
didate drug to improve overall survival (OS) of elderly
glioblastoma patients based on results of early, uncontrolled clin-
ical reports that suggested preferential benefit from bevacizumab
in frail and elderly patients (supplementary Note S1, available at
Annals of Oncology online). Yet, bevacizumab failed to prolong
OS in two phase III trials and there was only a trend for more
benefit from bevacizumab in patients with unfavorable prognos-
tic factors, although frail elderly patients were underrepresented
in these trials since being eligible for TMZ/RT !TMZ was an in-
clusion criterion in both trials [8,9].
Patients and methods
Study design
ARTE (NCT01443676) was designed as a two-arm, randomized, non-
comparative, open-label phase II trial. The primary objective was to de-
termine the efficacy and tolerability of the combination of bevacizumab
and radiotherapy in elderly patients with newly diagnosed glioblastoma.
The primary end point was median OS. Major inclusion criteria were: age
65 years or older, newly diagnosed supratentorial glioblastoma, eligible
for first infusion of bevacizumab 28 and 49 days after surgery for glio-
blastoma, Karnofsky performance score (KPS) of 60 or more, stable or
decreasing corticosteroid dose within 5 days before enrolment, availabil-
ity of paraffin-embedded tissue for central pathology review and deter-
mination of O
6
-methylguanine DNA methyltransferase (MGMT)
promoter methylation status, and adequate hematological, renal and
liver function. An amendment (November 2013) requested the absence
of MGMT promoter methylation when it became clear that MGMT pro-
moter methylation predicted larger benefit from temozolomide alone
than from RT alone in patients with glioblastoma aged 65 years [5,6].
All patients gave written informed consent, and the study was approved
by the local ethical committees and competent authorities (KEK-ZH No.
2011-0135).
Central pathology review and molecular analyses
Histological diagnoses were reviewed centrally (ER) based on the 2007
World Health Organization (WHO) classification [10]. Molecular assess-
ments included IDH mutation, MGMT promoter methylation, genome-
wide CpG methylation and nCounter gene expression arrays (supplementary
Note S2, available at Annals of Oncology online).
Study treatment, assessments and end points
Patients were allocated to treatment arms using a web-based randomiza-
tion system without stratification in a 2 : 1 distribution. RT was adminis-
tered to the gross tumor volume plus a 2 cm margin over 3 weeks, in 15
fractions of 2.66 Gy, to a total 40.0 Gy. Bevacizumab was administered
intravenously at 10 mg/kg bodyweight every 2 weeks. Dose delays were
foreseen for adverse reactions associated with bevacizumab that were
considered likely to evolve into serious or life-threatening events.
Bevacizumab was stopped for disease progression or unacceptable tox-
icity, including life-threatening events or subjective patient-related fac-
tors. Details on clinical assessments and end points are provided in
supplementary Note S3, available at Annals of Oncology online.
Statistical analyses
The initial sample size was 40 patients for arm A and 20 patients for arm
B. Sample size calculations were based on the experimental arm only for
the null hypothesis that, based on NOA-08 [6], median OS was
10 months, for 80% power to detect an improvement of 2 months using
a two-sided alpha level of 10%. Per amendment, the study design was
modified to aim for enrollment of 60 patients with tumors without
MGMT promoter methylation. Details on the applied statistical tests are
summarized in supplementary Note S4, available at Annals of Oncology
online.
Results
Patient characteristics
At 9 sites in Switzerland, 75 patients aged 65 years or older were
randomized from March 2013 to August 2015 (supplementary
Figure S1, available at Annals of Oncology online). Arms A and B
were overall well balanced (Table 1). More patients in arm A had
cognitive impairment (42% versus 15%, P¼0.032). The MGMT
promoter methylation status was available for 72 patients (96%),
55 tumors (76%) had an unmethylated MGMT promoter.
Genome-wide methylation arrays revealed 16 tumors (28%) with
receptor tyrosine kinase (RTK) I, 23 tumors (40%) with RTK II
and 16 tumors (28%) with mesenchymal (MES) methylation pat-
terns. Gene expression subtyping revealed 15 proneural (PN)
(29%), 19 classical (CL) (37%) and 17 MES (33%) glioblastomas.
At the time of databank closure (19 August 2016), 74 deaths were
documented, of which 70 were attributed to tumor progression,
1 was of unknown cause and 3 from pulmonary embolism, myo-
cardial infarction and gastrointestinal perforation, respectively.
One patient was still on study treatment, no patient was lost to
follow up for OS.
Safety and tolerability
Adverse events are summarized in supplementary Table S1, avail-
able at Annals of Oncology online. Grade 3–5 events were overall
infrequent. More adverse events occurred in arm A than in arm
B, but there was also a longer treatment period in arm A. The rate
of thromboembolic events was similar in both groups (19%
Original article Annals of Oncology
1424 | Wirsching et al. Volume 29 | Issue 6 | 2018

versus 20%), but more severe and life-threatening thrombo-
embolic events occurred in arm A (16% versus 8%), including
one death each from pulmonary embolism and myocardial in-
farction. Bevacizumab was discontinued for reasons other than
progression in four patients, including death in one patient, pul-
monary embolism in two patients and withdrawal of consent in
one patient.
Efficacy estimates by treatment arm
The median time from surgery to randomization was 25 days
(range 16–41) in arm A and 24 days (range 16–37) in arm B
(P¼0.60). Forty-seven of 50 patients in arm A and 23 of 25
patients in arm B completed the intended dose of 40 Gy of hypo-
fractionated RT. The median number of bevacizumab adminis-
trations in arm A was 12 (range 0–86). Median PFS was
7.6 months in arm A and 4.8 months in arm B (Figure 1A; supple-
mentary Table S2, available at Annals of Oncology online,
P¼0.003) and PFS at 6 months (PFS6) was 78% and 28% in
arms A and B (supplementary Table S2, available at Annals of
Oncology online, P<0.001). Median PFS was also more favorable
in arm A when analyses were restricted to the per protocol popu-
lation, i.e. to patients with tumors with an unmethylated MGMT
promoter and confirmation of the histopathological diagnosis by
central neuropathology review (supplementary Figure S2, avail-
able at Annals of Oncology online). Median PFS in the per proto-
col population was 7.6 months in arm A and 4.7 months in arm B
(supplementary Table S3, available at Annals of Oncology online,
P¼0.071). Twenty-four of 48 patients in arm A (50%), but 18 of
25 patients (72%) in arm B received any salvage therapy at pro-
gression (P¼0.071). Median OS in the intent to treat population
was 12.1 months in arm A and 12.2 months in arm B (Figure 1B;
supplementary Table S2, available at Annals of Oncology online,
P¼0.77), and survival rates at 12 months (OS12) were 54% and
56% in arms A and B (supplementary Table S2, available at
Annals of Oncology online, P¼0.85). Similarly, no apparent
benefit from bevacizumab with respect to OS emerged in the per
protocol population (supplementary Table S3, available at
Annals of Oncology online).
Central radiology review and response assessment
Serial MRI of 66 patients was available for post hoc central review.
At study entry, all but two patients, both from arm A, had
contrast-enhancing lesions, which qualified for measurable dis-
ease. Response measures were balanced between treatment arms
(P¼0.15; supplementary Table S4, available at Annals of
Oncology online). The locally determined time point of progres-
sion was centrally confirmed in 54 patients. It was antedated in
arm A by one scan in eight patients, two scans in two patients and
four scans in one patient, all of which qualified for pseudores-
ponse by RANO criteria because of progression on T2-FLAIR
sequences. Among patients from arm B, progression was ante-
dated one scan in one patient. Central review confirmed longer
PFS in arm A than in arm B (median PFS 7.0 versus 4.7 months,
P¼0.029).
Table 1. Patient characteristics
Arm A N550 Arm B N525 P
Age at diagnosis (years)
Median 70 70
Range 65–87 65–79 0.93
Gender, N(%)
Male 32 (64) 16 (64)
Female 18 (36) 9 (36) 1.00
KPS at study entry, N(%)
90–100 25 (50) 16 (64)
70–80 21 (42) 6 (24)
60 4 (8) 3 (12) 0.30
MMSE at study entry (points)
27 26 (52) 17 (68)
<27 19 (38) 3 (12) 0.032
No data 5 (10) 5 (20)
Steroids at study entry, N(%)
Yes 22 (44) 12 (44)
No 27 (54) 13 (56) 0.80
No data 1 (2) 0 (0)
CE-T1 (cm
2
)
Median 5.9 3.8
Range 0.0–32.2 0.5–17.8 0.12
Tumor burden on T2 (cm
2
)
Median 15.3 12.1
Range 2.3–43.2 1.8–39.0 0.13
Central histopathology, N(%)
Glioblastoma 46 (92) 23 (96)
Anaplastic astrocytoma 1 (2) 0 (0)
Anaplastic ependymoma 0 (0) 1 (4) 1.00
No data 3 (6) 1 (4)
IDH,N(%)
Mutated 1 (2) 0 (0)
Wildtype 39 (78) 19 (76) 1.00
No data 10 (20) 6 (24)
MGMT promoter, N(%)
Methylated 10 (20) 6 (24)
Unmethylated 37 (74) 18 (72) 0.72
No data 3 (6) 1 (4)
Gene methylation, N(%)
a
RTK I 11 (22) 5 (20) 0.67
RTK II 15 (30) 8 (32) 0.73
MES 11 (22) 5 (20) 0.88
MYCN 1 (2) 0 (0.0) n.a.
IDH mut 1 (2) 0 (0.0) n.a.
No data 11 (22) 7 (28) n.a.
Gene expression, N(%)
b
PN 9 (18) 6 (24) 0.39
CL 11 (22) 8 (32) 0.20
MES 15 (30) 2 (8) 0.033
No data 15 (30) 9 (36) n.a.
a
Jones et al. [11].
b
Verhaak et al. [12].
P-values <0.05 may be considered statistically significant.
CE-T1, contrast enhancement on T1-weighted images; CI, confidence
interval; CL, classical; IDH, isocitrate dehydrogenase; KPS, Karnofsky per-
formance score; MMSE, mini-mental status examination; MES, mesenchy-
mal; MGMT, O
6
-methylguanine DNA methyltransferase; MYCN, N-myc
proto-oncogene; PN, proneural; RTK, receptor tyrosine kinase.
Annals of Oncology Original article
Volume 29 | Issue 6 | 2018 doi:10.1093/annonc/mdy120 | 1425

Quality of life, cognitive functioning and steroid
use
Adherence of the study population to quality of life measure-
ments at each scheduled study visit was in the range of 65%–96%
in arm A and 44%–80% in arm B. Median deterioration-free
survival from baseline was 5.7 months in arm A and 2.8 months
in arm B (Figure 1C; supplementary Table S3, available at Annals
of Oncology online, P<0.001). Before progression, no differences
Figure 1. Clinical outcome by treatment arm. Progression-free survival (A) and overall survival (B) were analyzed in the intention-to-treat
population in patients treated with bevacizumab and radiotherapy (arm A) versus radiotherapy alone (arm B) utilizing the log-rank test. (C)
Deterioration-free survival in the intention-to-treat population in arm A versus arm B. (D) Individual functional and symptom scores from
EORTC QLQ-C30/BN20 modules before tumor progression were analyzed in a generalized linear mixed model that controlled for time treat-
ment interactions. Percent effects and 95%confidence intervals are depicted. (E) Cognitive functioning before tumor progression was
assessed by mini-mental status examination (MMSE). Boxes define the median and interquartile range, whiskers define the range. Differences
in treatment arms were compared by Student’s t-test for each timepoint using the Holm–Sidak method to correct for multiple comparisons.
Original article Annals of Oncology
1426 | Wirsching et al. Volume 29 | Issue 6 | 2018

were detected for individual scales in a generalized linear mixed
model, except for less favorable values in arm A for global health
(P¼0.048) and pain (P¼0.027) (Figure 1D).
Cognitive functioning was additionally assessed by serial
MMSE. It remained stable before progression in both groups and
no difference between arms was detected at individual visits
(Figure 1E), or in a linear regression model (P¼0.43 for differen-
ces in slopes). Six individuals with a decline in MMSE of 5 points
or more before progression were identified, all of which were
treated in arm A. Two of these six patients qualified for pseudor-
esponse by RANO.
Among 34 patients who were on steroids at study entry, 21 of
22 patients (95%) in arm A and 8 of 12 patients (66%) in arm B
came off steroids before progression (P¼0.024). The median
time on steroids from study entry was 1.6 months in arm A and
2.6 months in arm B (P¼0.38) (censoring patients at
progression).
Exploratory analyses of treatment effects
Treatment effects on PFS were noted throughout most clinically
defined subgroups and were pronounced in patients with poor
general condition at baseline (KPS 80%, HR 0.16, 95% CI
0.06–0.45, P<0.001) and in patients with cognitive impairment
at baseline (MMSE <27 points, HR 0.08, 95% CI 0.02–0.41,
P¼0.002). In contrast, no improved PFS was noted in patients
with better cognitive functioning (MMSE 27, HR 0.85, 95% CI
0.42–1.72, P¼0.66; supplementary Figure S3, available at Annals
of Oncology online).
Exploratory analyses of treatment effects with respect to OS
identified no subgroup with benefit from bevacizumab, albeit a
tendency to prolonged OS was noted in patients with poor gen-
eral condition (KPS 60%–80%, HR 0.55, 95% CI 0.24–1.24,
P¼0.15) and patients who did not receive any second line ther-
apy (HR 0.44, 95% CI 0.17–1.15, P¼0.092, supplementary
Figure S3, available at Annals of Oncology online).
Utilizing median PFS or OS as a cut-off, receiver operating
characteristics analyses failed to identify prognostic cut-offs for
baseline planimetric estimates of contrast enhancement on T1-
weighted images, or for tumor sizes on T2-weighted images,
including separate analysis of the entire study cohort, arm A or
arm B (data not shown).
Clinical characteristics and outcome by gene
methylation and gene expression
Unsupervised clustering of gene methylation similarities reflected
the methylation classifier-based assignment of glioblastoma sub-
types (Figure 2A). Groups defined by methylation subtypes did
not differ with respect to established prognostic factors, including
age (P¼0.44), KPS (P¼0.80), steroid intake at study entry
(P¼0.41) or MGMT promoter methylation (P¼0.91) and no
differences in PFS or OS were detected (supplementary Figure S4,
available at Annals of Oncology online).
Longer PFS in arm A than in arm B was detected solely in the
subgroup of RTK I glioblastomas (HR 0.25, 95% CI 0.07–0.82,
P¼0.014; Figures 2B, S3 and S4) and there was a trend towards a
higher rate of radiographic complete responses of RTK I com-
pared with RTK II or MES glioblastomas (supplementary Table
S4, available at Annals of Oncology online). No gene methylation
subtype with prolonged OS in arm A versus arm B was identified
(Figure 2C; supplementary Figures S3 and S4, available at Annals
of Oncology online).
There was an association of methylation subtypes with gene ex-
pression subtypes (P¼0.011), including RTK I methylation in
57% of PN, RTK II methylation in 73% of CL and MES methyla-
tion in 53% of glioblastomas with MES gene expression (Figure
2A). No differences in PFS and OS were detected between gene
expression subtypes (Figure S5). PFS was specifically longer
among bevacizumab-treated patients with IDH wildtype PN glio-
blastomas (HR 0.29, 95% CI 0.08–0.98, P¼0.025) (Figure 2D;
supplementary Figures S3 and S5, available at Annals of Oncology
online). No gene expression subtype with differences in OS be-
tween treatment arms was identified (Figure 2E; supplementary
Figures S3 and S5, available at Annals of Oncology online).
Outcome analyses of the MES gene expression subtype were pre-
cluded by imbalance between treatment arms (Table 1;supple-
mentary Figure S5, available at Annals of Oncology online).
Multivariate modeling of outcome
Cox proportional hazards modeling for PFS revealed treatment
with bevacizumab and higher KPS as relevant prognostic factors,
whereas no role for age, steroid intake at study entry or MGMT
promoter methylation emerged (Table 2). Applying the same
model to OS revealed age and KPS as relevant prognostic factors
(Table 2). Univariate analyses are summarized in supplementary
Table S5, available at Annals of Oncology online. Inclusion of gene
methylation and gene expression subtypes into this model as add-
itional single variables identified an association of the RTK II sub-
type with inferior PFS and OS (supplementary Table S6, available
at Annals of Oncology online).
We also considered the possibility that second line treatment
masked effects from study treatment on OS. However, in our
model the inclusion of any salvage therapy versus best supportive
care at recurrence did not render treatment arm a prognostic fac-
tor (HR 1.01, 95% CI 0.55–1.87, P¼0.96), and neither were sur-
gery at recurrence (HR for arm A versus arm B 0.89, 95% CI
0.34–2.30, P¼0.80), or cross-over to receive bevacizumab at re-
currence in arm B (HR for arm A versus arm B 0.63, 95% CI
0.29–1.32, P¼0.22) prognostic. In arm B, 13 patients (52%)
received bevacizumab at first or second recurrence, including two
patients (8%) who received bevacizumab at first and second re-
currence (supplementary Figure S1, available at Annals of
Oncology online).
Discussion
The ARTE trial was designed to provide evidence for a benefit
from first-line bevacizumab antiangiogenic therapy when added
to RT in elderly glioblastoma patients considered non-eligible for
combined modality treatment, based on standards of care when
the trial was initiated [13]. The amendment of the trial to include
only patients with an unmethylated MGMT promoter was moti-
vated ethically when it became clear that temozolomide and not
RT was the preferred first-line monotherapy for elderly glioblast-
oma patients with a methylated MGMT promoter considered
Annals of Oncology Original article
Volume 29 | Issue 6 | 2018 doi:10.1093/annonc/mdy120 | 1427

Figure 2. Molecular glioblastoma subtypes. (A) Unsupervised hierarchical clustering by gene methylation patterns. Columns represent
patients and rows CpG probes. Pearson’s correlation was carried out as distance measure and average linkage utilizing the 5000 probes with
highest standard deviations. Progression-free survival (B, D) and overall survival (C, E) by treatment arm in patients with RTK I gene methyla-
tion (B, C), and in patients with PN gene expression (D, E) glioblastoma subtypes. One patient with IDH mutated glioblastoma was excluded
from outcome analysis in (C) and (E).
Original article Annals of Oncology
1428 | Wirsching et al. Volume 29 | Issue 6 | 2018

non-eligible for combined modality treatment [5,6]. Given that
benefit from RT is not associated with absence or presence of
MGMT promoter methylation in this patient population, we
focused our analyses on the intention-to-treat population.
We found indeed a small increase in OS of 2 months over the
historical benchmark of the NOA-08 trial, but a similar increase
was also seen in the patients treated with RT alone, suggesting an
effect of patient selection or improved standards of care with
time [14], although cross-over to receive bevacizumab in arm B
contributing to OS cannot be ruled out. Overall the outcome
data reported in ARTE compare well with contemporary patient
populations (supplementary Table S7, available at Annals of
Oncology online). That the addition of bevacizumab increased
PFS, but apparently not OS is consistent with the phase III trials
concluded while ARTE was ongoing [8,9]. The benefit in PFS
with bevacizumab according to central review was consistent
with the benefit according to investigator assessment, albeit the
timepoint of progression was antedated in 22% of patients in arm
A due to more rigorous application of imaging criteria.
Early uncontrolled reports of bevacizumab therapy in glio-
blastoma patients suggested that benefit may be pronounced in
elderly and frail patients [15–17]. Accordingly, an OS benefit
from bevacizumab in patients who did not qualify for second line
therapy was reported from secondary analyses of the AVAGlio
trial [18]. We also noted a tendency towards prolonged OS in
patients with lower KPS and in patients who did not receive fur-
ther treatment at tumor progression, albeit these subgroup analy-
ses suffer from the lack of statistical power.
In line with reports from the AVAGlio trial [8,19], patient-
reported quality of life was maintained under bevacizumab at least
until progression and more patients came off steroids with bevaci-
zumab. In contrast, an association of bevacizumab treatment with
inferior QoL and worse cognitive functioning was reported from
the RTOG-0825 trial [9], fostering speculation on putative neuro-
toxicity of bevacizumab. In the ARTE trial, bevacizumab was over-
all well tolerated, but indeed more neuropsychiatric adverse events
and cognitive decline before progression were noted in arm A.
Although neurotoxicity from bevacizumab cannot be ruled out,
other causes potentially contributing to cognitive impairment in
individual patients in arm A included the prolonged treatment
period, imbalances in cognitive functioning at baseline and unrec-
ognized tumor progression, as determined by the post hoc radi-
ology review. The largest fraction of adverse events in both
treatment arms were classifiedas other neurologic events, compris-
ing mostly sensorimotor symptoms likely related to tumor burden.
Secondary analyses of outcome measures also included stratifi-
cation by molecular markers. Genome-wide analysis of gene
methylation identified an association of the RTK I methylation
subtype and the proneural gene expression subtype with PFS
benefit from bevacizumab. Although sample size was small, these
results confirm the previously reported link of preferential benefit
from bevacizumab in patients with tumors with a proneural gene
expression pattern [20]. This may provide an important path for-
ward for patient selection in future clinical trials exploring
VEGF-targeted therapies.
In conclusion, ARTE did not confirm the hypothesis that the
combination of bevacizumab with hypofractionated RT prolongs
OS in elderly glioblastoma patients, but exploratory analyses
yielded novel molecular genetics and imaging biomarkers that
may help to enrich patient populations for future trials on novel
anti-VEGF compounds.
Acknowledgements
Nadine Bosbach and Dr Gabriela Senti, Clinical Trial Center
Zurich, for administrative support.
Funding
The study was supported by a grant from Roche Pharma (Basel,
Switzerland). No grant numbers apply.
Disclosure
AFH, HGW, LP, MW, PR, RvM and TH received speaker hono-
raria for advisory boards from Roche. MW received financial
support for research to his institution by Roche. CM and LA are
employees and stock owners of Roche / Genentech. The remain-
der authors declare no potential conflict of interest.
References
1. Ostrom QT, Gittleman H, Fulop J et al. CBTRUS statistical report: pri-
mary brain and central nervous system tumors diagnosed in the United
States in 2008–2012. Neuro Oncol 2015; 17(Suppl 4): iv1–iv62.
Table 2. Multivariate analyses of associations with PFS and OS
PFS OS
Hazard ratio and 95%CI PHazard ratio and 95%CI P
Treatment arm:arm A versus arm B 0.36 (0.20–0.65) 0.001 1.09 (0.63–1.89) 0.75
Age at study entry:65–69 versus 70þyears 1.13 (0.66–1.92) 0.651 0.52 (0.30–0.89) 0.018
KPS:90%–100% versus 60%–80% 0.50 (0.28–0.89) 0.018 0.51 (0.28–0.92) 0.026
Steroids at study entry:no versus yes 1.10 (0.66–1.84) 0.704 0.87 (0.52–1.44) 0.59
MGMT promoter: methylated versus unmethylated
a
0.69 (0.37–1.30) 0.250 0.79 (0.42–1.49) 0.46
a
Note that enrollment was restricted per amendment 1 to unmethylated patients (see supplementary Figure S2, available at Annals of Oncology online).
Annals of Oncology Original article
Volume 29 | Issue 6 | 2018 doi:10.1093/annonc/mdy120 | 1429

2. Stupp R, Mason WP, van den Bent MJ et al. Radiotherapy plus concomi-
tant and adjuvant temozolomide for glioblastoma. N Engl J Med 2005;
352(10): 987–996.
3. Perry JR, Laperriere N, O’Callaghan CJ et al. Short-course radiation plus
temozolomide in elderly patients with glioblastoma. N Engl J Med 2017;
376(11): 1027–1037.
4. Roa W, Brasher PM, Bauman G et al. Abbreviated course of radiation
therapy in older patients with glioblastoma multiforme: a prospective
randomized clinical trial. JCO 2004; 22(9): 1583–1588.
5. Malmstrom A, Gronberg BH, Marosi C et al. Temozolomide versus
standard 6-week radiotherapy versus hypofractionated radiotherapy in
patients older than 60 years with glioblastoma: the Nordic randomised,
phase 3 trial. Lancet Oncol 2012; 13(9): 916–926.
6. Wick W, Platten M, Meisner C et al. Temozolomide chemotherapy alone
versus radiotherapy alone for malignant astrocytoma in the elderly: the
NOA-08 randomised, phase 3 trial. Lancet Oncol 2012; 13(7): 707–715.
7. Weller M, van den Bent M, Tonn JC et al. European Association for Neuro-
Oncology (EANO) guideline on the diagnosis and treatment of adult astro-
cytic and oligodendroglial gliomas. Lancet Oncol 2017; 18(6): e315–e329.
8. Chinot OL, Wick W, Mason W et al. Bevacizumab plus radiotherapy-
temozolomide for newly diagnosed glioblastoma. N Engl J Med 2014;
370(8): 709–722.
9. Gilbert MR, Dignam JJ, Armstrong TS et al. A randomized trial of bevacizumab
for newly diagnosed glioblastoma. N Engl J Med 2014; 370(8): 699–708.
10. Louis DN, Ohgaki H, Wiestler OD et al. The 2007 WHO classification of
tumours of the central nervous system. Acta Neuropathol 2007; 114(2):
97–109.
11. Jones DT, Capper D, Sill M et al. Next-generation neuropathology:
improving diagnostic accuracy for brain tumors using DNA methylation
array-based molecular profiling. Neuro Oncol 2014; 16: iii4A.
12. Verhaak RG, Hoadley KA, Purdom E et al. Integrated genomic analysis
identifies clinically relevant subtypes of glioblastoma characterized by
abnormalities in PDGFRA, IDH1, EGFR, and NF1. Cancer Cell 2010; 17:
98–110.
13. Weller M, van den Bent M, Hopkins K et al. EANO guideline for the
diagnosis and treatment of anaplastic gliomas and glioblastoma. Lancet
Oncol 2014; 15(9): e395–e403.
14. Gramatzki D, Dehler S, Rushing EJ et al. Glioblastoma in the Canton of
Zurich, Switzerland revisited: 2005 to 2009. Cancer 2016; 122(14):
2206–2215.
15. Nghiemphu PL, Liu W, Lee Y et al. Bevacizumab and chemotherapy for
recurrent glioblastoma: a single-institution experience. Neurology 2009;
72(14): 1217–1222.
16. Lai A, Tran A, Nghiemphu PL et al. Phase II study of bevacizumab plus
temozolomide during and after radiation therapy for patients with newly
diagnosed glioblastoma multiforme. JCO 2011; 29(2): 142–148.
17. Kreisl TN, Kim L, Moore K et al. Phase II trial of single-agent bevacizu-
mab followed by bevacizumab plus irinotecan at tumor progression in
recurrent glioblastoma. JCO 2009; 27(5): 740–745.
18. Chinot OL, Nishikawa R, Mason W et al. Upfront bevacizumab may ex-
tend survival for glioblastoma patients who do not receive second-line
therapy: an exploratory analysis of AVAglio. Neuro Oncol 2016; 18(9):
1313–1318.
19. Taphoorn MJ, Henriksson R, Bottomley A et al. Health-related quality of
life in a randomized phase III study of bevacizumab, temozolomide, and
radiotherapy in newly diagnosed glioblastoma. JCO 2015; 33(19):
2166–2175.
20. Sandmann T, Bourgon R, Garcia J et al. Patients with proneural glio-
blastoma may derive overall survival benefit from the addition of bevaci-
zumab to first-line radiotherapy and temozolomide: retrospective
analysis of the AVAglio trial. JCO 2015; 33(25): 2735–2744.
Original article Annals of Oncology
1430 | Wirsching et al. Volume 29 | Issue 6 | 2018