J Neurooncol
DOI 10.1007/s11060-015-1782-5
CLINICAL STUDY
Evaluation of pre-radiotherapy apparent diffusion coefficient
(ADC): patterns of recurrence and survival outcomes analysis
in patients treated for glioblastoma multiforme
Andrew Elson1 • Eric Paulson1 • Joseph Bovi1 • Malika Siker1 • Chris Schultz1
Peter S. Laviolette2
Received: 9 July 2014 / Accepted: 5 April 2015
Ó Springer Science+Business Media New York 2015
Abstract Purpose: To investigate the association of pre-
radiotherapy apparent diffusion coefficient (ADC) abnor-
malities with patterns of recurrence and outcomes in pa-
tients with glioblastoma multiforme (GBM). Materials and
Methods: Fifty-two patients with recurrent GBM were
retrospectively evaluated. Diffusion MRI images were ac-
quired for all patients postoperatively prior to radiotherapy.
ADC images were evaluated for geographic regions of
diffusion restriction (hypointensity) within the FLAIR
volume. If identified, the ADC map and the T1?C MRI at
the time of recurrence were registered to the original plan
to determine the pattern of recurrence and the coverage of
the ADC abnormality by the 60 Gy isodose line (IDL).
Progression-free and overall survival was determined for
patients with and without an ADC hypointensity. Results:
An ADC hypointensity was identified in 32 (62 %) of
cases. The recurrence pattern in these cases was central in
27/32 (84 %), marginal in 4/32 (13 %) and distant in 1/32
(3 %). The recurrence overlapped with the ADC hy-
pointensity in 28 (88 %) patients. The ADC hypointensity
was covered by 95 % of the 60 Gy IDL in all cases. Ka-
plan–Meier analysis revealed inferior progression free
survival and overall survival in patients with an ADC hy-
pointensity compared to those without, despite similarities
between the groups in terms of age, RT dose, performance
& Andrew Elson
aelson1672@gmail.com
1
Department of Radiation Oncology, Medical College of
Wisconsin, 9200 W. Wisconsin Avenue, Froedtert Hospital
East Clinics 3rd Floor, Milwaukee, WI 53226, USA
2 characterize the diffusion environment on a voxelwise
Department of Radiology Imaging Research, Froedtert &
Medical College of Wisconsin, 9200 W Wisconsin Avenue,
Milwaukee, WI 53226, USA
•
status, and extent of resection. Conclusions: The presence
of an ADC hypointensity on pre-radiotherapy diffusion-
weighted imaging is associated with the location of tumor
recurrence as demonstrated by frequent overlap in this
series, and is associated with a trend toward inferior out-
comes. This abnormality may reflect a high risk region of
hypercellularity and warrants consideration with respect to
radiotherapy planning.
Keywords Glioblastoma multiforme
Radiation
therapy
Apparent diffusion coefficient
Diffusion
weighted imaging
Introduction
Glioblastoma multiforme (GBM) is the most malignant pri-
mary brain tumor and carries a poor prognosis, with a minority
of patients surviving beyond 2 years [1]. Postoperative temo-
zolomide based chemoradiation continues to represent the
standard of care [2, 3]. Radiotherapy targeting is commonly
based on the MR contrast enhancing and T2/FLAIR hyperin-
tense region, as pathologic correlation has confirmed solid
tumor components as well as infiltrative tumor cells within
these regions [4]. Shortcomings of the use of contrast en-
hancement and T2/FLAIR alone for planning purposes include
the possibility that solid tumor may be present without contrast
enhancement, as well as the confounding effect of edema.
MRI diffusion-weighted imaging (DWI) is an evolving
tool with many applications in neuro-oncology. DWI
probes the Brownian motion of water molecules and can
generate apparent diffusion coefficient (ADC) maps which
basis [5, 6]. Regions of diffusion restriction may be found
in various pathophysiologic processes including acute
123

stroke, brain abscess, hypoxic-ischemic encephalopathy,
demyelinating diseases, leukodystrophies, vasculitis, and
primary brain neoplasms. Diffusion restriction resulting
from tumor hypercellularity as a result of sequestration of
fluid into the intracellular compartment and decreased ex-
tracellular space is of particular interest in neuro-oncology
[7].
Previous studies utilizing biopsy and resection samples
have established cellularity, nuclear–cytoplasmic ratio, and
tumor grade as inversely proportional to ADC [8–10].
DWI has shown utility in predicting treatment response
in patients with GBM. Li et al. demonstrated that an in-
crease in normalized ADC postradiotherapy was consistent
with treatment response [11]. Lutz et al. reported that pa-
tients with tumor progression had significantly lower ADC
values postradiotherapy than patients with stable disease
[12]. A quantitative evaluation of ADC changes known as
the functional diffusion map (fDM) has shown predictive
value for treatment response in patients with gliomas [13].
ADC changes have also been shown to have prognostic
implications in patients treated with bevacizumab, in which
significant alterations in contrast enhancement can occur
[14–18].
Although most studies evaluating ADC values do so
within regions of abnormal T2/FLAIR or contrast en-
hancement, Gupta et al. reviewed 27 patients with GBM
exhibiting an isolated ADC hypointensity without corre-
sponding contrast enhancement. In 85 % of these cases,
contrast enhancement eventually developed at the site of
ADC hypointensity [19].
At our Institution, patients with GBM postoperatively
undergo CT/MRI simulation for radiotherapy targeting
purposes. DWI sequences and ADC maps are routinely
acquired and generated during these MRI simulation exams
in addition to standard sequences. The purpose of our study
was to determine the frequency with which regions of
diffusion restriction occur on the pre-radiotherapy simula-
tion DWI/ADC images, whether such abnormalities are
included within the high dose radiotherapy volume, and
their association with the spatial location of recurrence. We
hypothesized that the presence of an ADC hypointensity
represents an area of high cellularity and therefore a high
risk region for recurrence.
Methods and materials
Patients
Fifty-two adult patients (32M/20F median age 61 years,
range 32–85 years) with a pathologically confirmed diag-
nosis of GBM who completed postoperative external beam
radiation therapy to 60 Gy in 30 fractions with
123
J Neurooncol
temozolomide were included for this retrospective analysis.
Patients were required to have undergone postoperative,
pre-radiotherapy DWI during the time of the MR simula-
tion. Patients included for analysis had known tumor re-
currence such that patterns of failure could be analyzed.
Radiotherapy target volumes were based on RTOG
guidelines without references to the diffusion images.
Target delineation was performed by contouring the T2/
FLAIR edema field with a 2 cm clinical target volume
(CTV) and 3–5 mm planning target volume (PTV) ex-
pansion as the 46 Gy volume. The 60 Gy volume consisted
of a 2 cm CTV and 3–5 mm PTV expansion on the re-
section cavity and residual contrast enhancement. Treat-
ment delivery consisted of 3D conformal radiotherapy with
95 % prescription dose PTV coverage. All patients re-
ceived daily temozolomide during radiation. All aspects of
this study were approved by the institutional IRB.
MRI DWI acquisition and ADC map generation
All patients underwent MRI simulation prior to radio-
therapy for treatment planning purposes. Patients were set
up and imaged in the treatment position on a Siemens
Verio 3T MRI (Siemens Healthcare, Erlangen, Germany)
using flexible phased-array radiofrequency coils [20]. In
addition to conventional T2/FLAIR and pre-/post-contrast
T1 images, diffusion-weighted images were acquired and
corresponding ADC maps were generated using software
integrated within the MR unit.
Follow-up and determination of recurrence
Patients were routinely continued on temozolomide for 12
monthly cycles unless this was precluded by intolerance or
recurrence. MR imaging was performed 1 month post-ra-
diation and every 2 months thereafter. All cases of poten-
tial recurrence were reviewed at a multidisciplinary tumor
board. Recurrence was defined by radiographic evidence
accompanied by either a change in management or surgical
re-resection with histologic confirmation. There was no
uniform institutional protocol for management at the time
of recurrence however this most commonly consisted of
administration of bevacizumab. In order to address the
potential impact of salvage therapy on overall survival after
disease recurrence, the medical records of all the patients in
this cohort were evaluated to determine the salvage
therapies administered relative to ADC status.
ADC and patterns of recurrence analysis
Analyzed radiotherapy plans were originally generated
using the Xio treatment planning system (Elekta AB,
Stockholm, Sweden) and were restored with the original
J Neurooncol
structure sets and dose distributions. The pre-radiotherapy
ADC maps were registered to the planning CT, matching
the registration of T1?C and T2/FLAIR MR images
originally used for planning. The post-radiotherapy T1?C
MRI revealing tumor recurrence was registered to the
original planning CT such that the pattern of recurrence
could be determined. MR image registration to the origi-
nal CT Simulation dataset was performed with Focal
Fusion (Elekta, AB, Stockholm, Sweden) which allows for
software based rigid auto-registration followed by manual
adjustment. All pre-radiotherapy DWI/ADC study sets
were evaluated for the presence of a discrete zone of
diffusion restriction outside of the resection cavity but
within the FLAIR volume. The identification of an ADC
abnormality required the presence of a geographic zone of
DWI hyperintensity corresponding to an area of ADC
hypointensity to verify the presence of diffusion restriction
without the confounding effect of underlying T2 signal
known as ‘‘T2-shine through [21, 22].’’ The ADC hy-
pointensity was not excluded if it exhibited overlap with
contrast enhancement or was located adjacent to the re-
section cavity. Thirty-two cases were found to have such
an abnormality, and for these cases the original radio-
therapy plans were analyzed to determine the geographic
relationship of the ADC abnormality to the 60 Gy isodose
line (IDL) as well as to the subsequent T1 enhancing re-
currence. Contours of the ADC hypointensity were
manually generated along with contours of the contrast
enhancing component of the recurrence. Overlap of the
ADC abnormality with the recurrent tumor was deter-
mined by any overlap of the respective contours, and the
percentage of the ADC volume occupied by recurrence
was calculated. The pattern of recurrence was defined with
respect to the high dose (60 Gy) IDL according to Bran-
des et al. in which recurrences are defined as central,
marginal, or distant when [80, 20–80, or \20 % respec-
tively of the recurrence volume is contained within 95 %
of the 60 Gy IDL, respectively (Fig. 1). This categoriza-
tion was determined by analyzing the dose–volume his-
togram (DVH) of the recurrence volume. Likewise, the
dosimetric coverage of the ADC abnormality was deter-
mined from the DVH of the ADC volume.
Evaluation of MGMT methylation status
MGMT methylation status was assessed in 32 of 52 (62 %)
patients using methylation-specific polymerase chain re-
action. The remainder of the patients had unknown MGMT
status. The determination of whether or not to perform
MGMT methylation testing was made clinically by the
treating neuro-oncologist in a multidisciplinary setting.
Statistical analysis
Progression free survival (PFS) and overall survival (OS)
for this review were defined as time elapsed from the
completion of radiotherapy to documentation of progres-
sion or death, and patients were censored at the date of last
follow up. PFS and OS were analyzed using the Kaplan–
Meier method and hazard ratios were calculated using the
log-rank test. A p value \0.05 was considered statistically
significant. Multivariate analysis was performed using a
Cox proportional hazards model. Statistics were performed
using SAS statistics software version 9.3 (The SAS Insti-
tute, Cary, NC).
Results
ADC analysis and patterns of recurrence
Thirty-two of 52 patients (62 %) were found to have a
discrete region of diffusion restriction within the FLAIR
volume and outside of the resection cavity as defined by
DWI hyperintensity and ADC hypointensity. For these
patients the original radiotherapy plan was restored and the
pattern of recurrence was determined. The pattern of re-
currence was central in 27/32 (84 %), marginal in 4/32
(13 %) and distant in 1/32 case (3 %). The mean recur-
rence volume covered by 95 % of the 60 Gy IDL was
99.6 % in the central group, 64.4 % in the marginal group,
and for the single distant recurrence 14.8 % of the volume
was covered. Coverage of the ADC hypointensity volume
was evaluated relative to the 60 Gy IDL as well as overlap
with the recurrence volume as defined by the contrast en-
hanced T1 MRI. The mean volume of ADC hypointensity
covered by 95 % of the 60 Gy IDL was 99.6 %. The ADC
hypointensity volume overlapped with the recurrence vol-
ume in 28/32 (88 %) cases including 25/27 (93 %) central
recurrences, 3/4 (75 %) marginal recurrences, and 0/1
(0 %) distant recurrences. The mean ADC hypointensity
overlap ratio was 0.6. Pattern of recurrence data is sum-
marized in Table 1.
Patient characteristics and outcome
Baseline patient characteristics are presented in Table 2.
Patients with and without an ADC hypointensity did not
differ with respect to age, performance status, radiotherapy
dose, or extent of resection. Patients with an ADC hy-
pointensity more frequently exhibited MGMT non-methy-
lation. The median follow up time was 29 months (range
1.4–35.4 months).
123

Fig. 1 Representative screenshots of image analysis and contours.
Panels show the following elements of analysis: a MR diffusion
weighted image with a hyperintensity (adjacent to the anterior horn of
the right lateral ventricle) and b the corresponding hypointensity on
the ADC map generated from the diffusion image. c DWI image with
ADC hypointensity contoured with an additional superimposed
contour representing the contrast enhancing recurrence volume
depicted in the T1?C MR (panel d). All images were registered
For the entire group the median progression free sur-
vival and overall survival from the completion of radio-
therapy were 4.9 (95 % CI 3.17–6.2) and 12.6 months
(95 % CI 10.2–18.9), respectively.
123
J Neurooncol
and the contours were transposed to original CT dataset (e). Panel
f depicts analysis of the dose plan (60 Gy inner IDL, 57 Gy outer
IDL) relative to the ADC hypointensity contour and the recurrence
volume contour. Examples of central (g), marginal (h), and distant
(i) recurrence pattern based on the recurrence volume relative to the
IDL. Isodose lines shown represent 100 % (inner) and 95 % (outer)
of the 60 Gy IDL
When analyzed by ADC status, the median PFS (with
95 % CI) for patients with an ADC hypointensity versus
without was 3.2 (2.0–4.9) versus 8.0 (6.0–11.6) months
(HR 1.96, 95 % CI 1.14–3.39, p = 0.013).
J Neurooncol

Table 1 Dosimetric parameters by pattern of recurrence

Central n = 27 (84 %) Marginal n = 4 (13 %) Distant n = 1 (3 %) Total (n = 32)

Mean ADC volume (cm3) 4.6 2.8 9.9 4.5

Recurrence volume (cm3) 49.8 75.0 4.3 48.1
ADC overlap (%) 64 50 0 60
Recurrence covered by 95 % IDL(%) 99.6 64.4 14.8 92.7
ADC covered by 95 % IDL (%) 99.6 98.6 100 99.6
ADC overlaps recurrence 25/27 (93 %) 3/4 (75 %) 0/1 (0 %) 28/32 (88 %)
ADC ADC hypointensity

Table 2 Baseline patient

ADC hypointensity present ADC hypointensity absent Total p
characteristics
N 32 (62 %) 20 (38 %) 52 (100 %)
Male/female 19/13 13/7 32/20
Median age 61 61 61 0.87a
Median KPS 90 90 90 0.69a
KPS C80 29 (91 %) 18 (90 %) 47 (90 %)
KPS \80 3 (9 %) 2 (10 %) 5 (10 %)
Mean dose (Gy) 60 60 60 0.24a
Resection status
GTR 6 (19 %) 5 (25 %) 11 (21 %) 0.9b
STR 21 (66 %) 14 (70 %) 35 (67 %) 0.31b
Bx 5 (16 %) 1 (5 %) 6 (11 %) 0.22b
MGMT status
Methylated 3 (9 %) 8 (40 %) 11 (21 %) 0.23b
Unmethylated 17 (53 %) 4 (20 %) 21 (40 %) 0.007b
Unknown 12 (38 %) 8 (20 %) 20 (38 %) 0.5b
a
t test
b
Binomial test

The median OS (with 95 % CI) for patients with an
ADC hypointensity versus without was 11.3 (7.9–14.8)
months versus 18.9 (10.2–24.2) months (HR 1.77, 95 % CI
0.98–3.19, p = 0.059).
Kaplan–Meier survival curves for PFS and OS for the
entire cohort as well as by ADC abnormality are depicted
in Fig. 2.
Univariate analysis (Kaplan–Meier) was performed for
the additional prognostic factors of age, extent of resection,
and performance status to determine the effect on survival.
Univariate analysis revealed that age greater than or equal
to 65 years was significantly associated with reduced PFS
and OS.
Cox proportional hazards analysis was performed for
multivariate analysis of the presence of ADC hypointensity
and age with respect to PFS and OS. This analysis revealed
that age was a significant predictor of adverse OS, and the
absence of an ADC hypointensity was a significant
predictor of favorable PFS (Table 3). Kaplan–Meier PFS
and OS curves are depicted in Fig. 2.
Salvage therapies administered at the time of disease
recurrence included re-resection, bevacizumab, pulsed low
dose rate re-irradiation (PLDR), and isotretinoin, or various
combinations thereof. Both the group with an identifiable
ADC abnormality and the group without underwent re-
resection or bevacizumab therapy at a similar frequency
(25 and 85 %, respectively). The salvage therapies ad-
ministered by ADC abnormality status are depicted in
Table 4.
Discussion
In this study patients were found to have a predominantly
central failure pattern with the majority (84 %) of recur-
rences having been covered[80 % by the 60 Gy IDL. This

123

Fig. 2 Kaplan–Meier survival curves for a entire group PFS, b entire group OS, c OS by ADC abnormality (present or absent), d PFS by ADC
abnormality (present or absent)
is similar to previous studies analyzing recurrence patterns
in patients treated with temozolomide based chemora-
diation which report central recurrences in 67–90 % of
cases [23–28]. Brandes et al. reported an association be-
tween pattern of recurrence, MGMT status, and time to
progression, with central or marginal recurrence pattern in
only 58 % of patients with MGMT Methylation versus
85 % without methylation [23]. In addition, distant recur-
rence occurred at a median of 15 versus 9 month for central
recurrence. This pattern was also noted by Minniti et al. in
which MGMT methylated patients exhibited central re-
currence in 64 % of cases versus 91 % of non-methylated
patients [27]. A potential rationale for this observation is
that MGMT methylated tumors are more susceptible to
local eradication, shifting the patterns of failure distantly.
Given that distant recurrences have been found to occur
later than the more common central recurrences, this may
imply that control of disease near the origin may improve
PFS.
123
J Neurooncol
Efforts to improve local control through radiotherapy
dose escalation have not found a benefit in the pattern of
recurrence or overall patient outcomes in the setting single
fraction SRS [29–31], fractionated SRS [32], or fraction-
ated dose escalation up to 90 Gy [33]. A randomized trial
of SRS boost followed by chemoradiation with BCNU to
60 Gy versus chemoradiation alone revealed no survival
benefit to SRS dose escalation [34]. A 2005 ASTRO con-
sensus guideline recommended against the use of single
fraction or hypofractionated radiosurgery boost [35].
Nevertheless, there is continued interest in dose esca-
lation for local control particularly in the era of temo-
zolomide chemotherapy where improvements in outcomes
have been achieved. Incorporating advanced imaging
techniques into radiotherapy planning is under investiga-
tion, as the majority of dose escalation protocols have re-
lied on the contrast enhancement, as do the current
guidelines for standard fractionated radiotherapy. The po-
tential advantage of advanced imaging includes the
J Neurooncol

Table 3 Univariate and

Univariate analysis
multivariate analysis for PFS
and OS Prognostic factor N PFS median (months) p OS median (months) p
Age
C65 16 3.5 0.048 8.9 0.01
\65 36 5.0 14.8
Resection
\GTR 41 5.0 0.67 13.1 0.61
GTR 11 3.7 8.7
KPS
\80 5 0.9 0.92 NAa 0.77
C80 47 4.9 12.6
ADC abnormality
Present 32 3.2 0.013 11.3 0.059
Absent 20 8.0 18.9
Multivariate analysis
Prognostic factor PFS (HR) (95 % CI) p OS (HR) (95 %CI) p
Age [65 1.8 (0.99–3.35) 0.056 2.4 (1.2–4.9) 0.014
ADC abnormality absent 0.49 (0.28–0.88) 0.017 0.56 (0.29–1.04) 0.071
a
Median OS not reached

Table 4 Therapies
ADC abnormality Present % Absent % Total %
administered at the time of
recurrence by ADC abnormality No salvage therapy 3 9.4 2 10.0 5 9.6
Bevacizumab only 21 65.6 11 55.0 32 61.5
Bevacizumab, PLDR 0 0.0 1 5.0 1 1.9
Bevacizumab, isotretinoin, PLDR 0 0.0 1 5.0 1 1.9
Re-resection 1 3.1 1 5.0 2 3.8
Re-resection, bevacizumab 3 9.4 2 10.0 5 9.6
Re-resection, PLDR 1 3.1 0 0.0 1 1.9
Re-resection, PLDR, bevacizumab 2 6.3 1 5.0 3 5.8
Re-resection, PLDR, bevacizumab, isotretinoin 1 3.1 1 5.0 2 3.8
Re-resection as any component of salvage 8 25.0 5 25.0 13 25.0
Bevacizumab as any component of salvage 27 84.4 17 85.0 44 84.6
PLDR pulsed low dose rate re-irradiation

possibility of more precise targeting of the high risk areas
as well as normal tissue avoidance.
Park et al. reviewed 23 patients treated for GBM with
post-operative radiotherapy who had undergone MR
spectroscopic imaging (MRSI) prior to radiotherapy.
Spectroscopic volumes containing a choline to NAA ratio
[2 were generated as a marker of high risk tumor region,
and registration of these volumes to the radiotherapy plans
revealed that in several cases that inclusion of the spec-
troscopic data could have provided improved coverage of
the recurrent volume and spared normal brain tissue [36].
Similarly, a prospective dose escalation trial for patients
with GBM has evaluated 11C-Methionine PET (MET PET)
to better define high risk regions of potential recurrence. In
this protocol patients were treated with IMRT based dose
escalation up to 81 Gy, and review of MET PET imaging
revealed that this modality identifies abnormal tissue not
delineated by standard MR imaging which was associated
with an increased risk of recurrence particularly with
suboptimal coverage by the high dose volume [37, 38].
Interest in treatment volume reduction has been driven
by the goal of achieving normal tissue sparing with
equivalent tumor control. Studies of reduced margin ra-
diotherapy to 60 Gy [25, 39] or in conjunction with dose
escalation [40] have shown similar patterns of recurrence
to more traditional target delineation however with im-
proved normal tissue sparing. Imaging modalities which
identify high risk regions for tumor recurrence may further

123

assist in efforts to target the most critical regions from the
disease control standpoint while allowing for reduced
margins for normal tissue sparing.
The present study to our knowledge is the first to
correlate patterns of recurrence with postoperative, pre-
radiotherapy DWI abnormalities. MR diffusion imaging is
commonly acquired in CNS diseases and has the advan-
tage of being a fast, widely available, non-invasive
imaging procedure without the requirement of contrast
administration. The hypothesis that regions of ADC hy-
pointensity reflects a high risk region of local hypercel-
lularity-mediated diffusion restriction was supported by
our findings that these abnormalities overlap with the
eventual recurrence in 88 % of cases and the proportion
of overlap is 60 %, suggestive of geographical co-local-
ization. Our analysis indicates that the ADC hypointensity
itself volumetrically is much smaller than the recurrence
volume. In addition, patients with a region of diffusion
restriction exhibited a significantly reduced median pro-
gression free survival of 3.2 versus 8.0 months for those
without such an abnormality, and a trend toward reduced
overall survival of 11.3 versus 18.9 months. This suggests
that this abnormality represents an adverse prognostic
feature even when age, performance status, and extent of
resection are taken into account. The median PFS of
3.2 months in patients with an ADC hypointensity is
congruent with the findings of Gupta et al. in which re-
gional diffusion restriction in patients treated for GBM
was found to precede eventual contrast enhancing recur-
rence at a median of 3.0 months [19]. One possible ex-
planation for these findings is that tumor hypercellularity
may be identified prior to the alterations required to allow
tumor associated disruption of the blood brain barrier and
contrast extravasation, and may manifest on the ADC
map as a smaller region than the eventual recurrence
volume.
Dosimetric analysis revealed that the pre-radiotherapy
ADC abnormality is well covered by the 60 Gy IDL in the
majority of cases. It is therefore unlikely that MR diffusion
imaging would alter standard RTOG based planning pro-
cedures. However, for the purposes of localized conformal
dose escalation as well as volume reduction for normal
tissue sparing, the DWI/ADC maps may provide useful
information regarding the identification of high risk areas.
Methylation of the O6–methylguanine–DNA methyl-
transferase (MGMT) is a known predictive and prognostic
feature of GBM which confers a significant survival ad-
vantage particularly with the use of temozolomide [41]. In
this study, only 61 % of patients had known MGMT
methylation status. The ADC hypointensity group exhib-
ited more frequent MGMT non-methylated status however
due to the incomplete MGMT data any conclusions re-
garding this are extremely limited. The incomplete MGMT
123
J Neurooncol
data in this cohort represents one of the limitations of this
study.
An unanticipated finding of this study was that the 11
patients having undergone a GTR defined as no residual
contrast enhancement on immediate post-operative imag-
ing appeared to have inferior overall and progression-free
survival although this was not statistically significant. To
further investigate this however, patient characteristics
were analyzed to identify any potential underlying adverse
risk factors. Patient age and KPS did not differ between the
groups having achieved a GTR versus the overall cohort.
Methylation status revealed that 2 of the 11 were MGMT
methylated, 2 were unmethylated, and 7 had unknown
status. Therefore given the importance of this marker on
prognosis, MGMT methylation status could have played a
role in the trend toward poorer survival in the GTR cohort,
particularly if a large proportion of the seven patients
having undergone GTR with unknown methylation status
were in fact unmethylated. To assess the matter further,
pre-operative and immediate post-operative contrast MRI
images were volumetrically analyzed to determine the
extent of resection as a percentage of post-operative resi-
dual contrast enhancement versus pre-operative T1?C tu-
mor volume. Using a threshold of [95 % resection versus
\95 % resection in a univariate analysis, the patient sub-
group achieving [95 % resection had a median OS of 13.1
versus 12.5 month for \95 %, which did not reach statis-
tical significance. Therefore when analyzing patients who
underwent essentially a near GTR, the result was more
consistent with the expected effect of extent of resection on
outcome.
Additional limitations of this study include the retro-
spective nature of the investigation as well as patient se-
lection criteria, which may inhibit the comparison of
outcome data from this study with respect to other series.
Progression-free survival and overall survival reported here
are comparable to other series when taking into account
that outcomes reported are from the time of radiotherapy
completion. Interpretation of overall survival data must be
done cautiously given that there was not a uniform pro-
cedure for management at the time of progression, however
analysis of salvage therapies administered at the time of
recurrence indicated that a similar proportion of patients in
both the group with an ADC abnormality and the group
without underwent re-resection or bevacizumab therapy as
a component of salvage therapy, thus lessening the impact
of differential salvage therapy as a confounding factor.
Although diffusion restriction on DWI has been corre-
lated with increased cellularity in the setting of primary
brain tumors, in any given patient the underlying cause of a
region of diffusion restriction cannot be known with cer-
tainty radiographically. Diffusion restriction may reflect
various entities including local infarct, operative
J Neurooncol
mechanical effects, necrosis, or a combination thereof.
Despite the range of possible mechanisms, the results of
this study support the hypothesis that these regions repre-
sent high-risk areas for recurrence and therefore likely in-
clude components of hypercellularity. In addition, for the
purposes of radiotherapy targeting or dose escalation, it is
unlikely that any region of diffusion restriction reflects
normal functioning brain tissue. Given the retrospective
nature of this study in a limited number of patients, further
evaluation would be warranted to determine the extent to
which diffusion restriction may assist treatment planning
by identifying high risk regions in the post-operative, pre-
radiotherapy setting.
Despite the limitations of this study, the results indicate
a rationale for the inclusion of MR DWI as a component of
radiotherapy treatment planning for GBM. As treatments
for this disease site evolve, more emphasis may be placed
on advanced imaging techniques for target delineation
beyond the standard T1?C and T2/FLAIR imaging, as well
as efforts at normal tissue sparing through appropriately
designed margin reductions. Further studies are warranted
to determine the role of MR diffusion weighted imaging in
radiotherapy planning and as a potential tool for dose
escalation and margin reduction. In particular, the ongoing
development and implementation of advanced imaging
techniques such as DWI, dynamic susceptibility based
imaging, dynamic contrast enhanced imaging, and
metabolic imaging may eventually provide composite data
reflecting multiple high risk tumor features such as hy-
percellularity, neoangiogenesis, hypoxia, and hyperme-
tabolism which could provide a more comprehensive
evaluation of tumor biology.
Conflict of interest The authors declare that they have no conflict
of interest. The authors have no financial disclosures with regard to
this work.
References
1. DeAngelis LM (2001) Brain tumors. N Engl J Med
344(2):114–123
2. Stupp R, Mason WP, van den Bent MJ et al (2005) Radiotherapy
plus concomitant and adjuvant temozolomide for glioblastoma.
N Engl J Med 352(10):987–996
3. Stupp R, Hegi ME, Mason WP et al (2009) Effects of radio-
therapy with concomitant and adjuvant temozolomide versus
radiotherapy alone on survival in glioblastoma in a randomised
phase III study: 5-year analysis of the EORTC-NCIC trial. Lancet
Oncol 10(5):459–466
4. Kelly PJ, Daumas-Duport C, Kispert DB, Kall BA, Scheithauer
BW, Illig JJ (1987) Imaging-based stereotaxic serial biopsies in
untreated intracranial glial neoplasms. J Neurosurg 66(6):865–
874
5. Mori S, Barker PB (1999) Diffusion magnetic resonance imaging:
its principle and applications. Anat Rec 257(3):102–109
6. Hagmann P, Jonasson L, Maeder P, Thiran JP, Wedeen VJ, Meuli
R (2006) Understanding diffusion MR imaging techniques: from
scalar diffusion-weighted imaging to diffusion tensor imaging
and beyond. Radiographics 26(Suppl 1):S205–S223
7. Karaarslan E, Arslan A (2008) Diffusion weighted MR imaging
in non-infarct lesions of the brain. Eur J Radiol 65(3):402–416
8. Ellingson BM, Malkin MG, Rand SD et al (2010) Validation of
functional diffusion maps (fDMs) as a biomarker for human
glioma cellularity. J Magn Reson Imaging 31(3):538–548
9. Sugahara T, Korogi Y, Kochi M et al (1999) Usefulness of dif-
fusion-weighted MRI with echo-planar technique in the evalua-
tion of cellularity in gliomas. J Magn Reson Imaging 9(1):53–60
10. Guo AC, Cummings TJ, Dash RC, Provenzale JM (2002) Lym-
phomas and high-grade astrocytomas: comparison of water dif-
fusibility and histologic characteristics. Radiology 224(1):177–
183
11. Li Y, Lupo JM, Polley MY et al (2011) Serial analysis of imaging
parameters in patients with newly diagnosed glioblastoma mul-
tiforme. Neuro Oncol 13(5):546–557
12. Lutz K, Wiestler B, Graf M (2013) Infiltrative patterns of
glioblastoma: identification of tumor progress using apparent
diffusion coefficient histograms. J Magn Reson Imaging
39(5):1096–1103
13. Moffat BA, Chenevert TL, Lawrence TS et al (2005) Functional
diffusion map: a noninvasive MRI biomarker for early stratifi-
cation of clinical brain tumor response. Proc Natl Acad Sci USA
102(15):5524–5529
14. Ellingson BM, Cloughesy TF, Lai A et al (2011) Graded func-
tional diffusion map-defined characteristics of apparent diffusion
coefficients predict overall survival in recurrent glioblastoma
treated with bevacizumab. Neuro Oncol 13(10):1151–1161
15. Pope WB, Lai A, Mehta R et al (2011) Apparent diffusion co-
efficient histogram analysis stratifies progression-free survival in
newly diagnosed bevacizumab-treated glioblastoma. Am J Neu-
roradiol 32(5):882–889
16. Pope WB, Qiao XJ, Kim HJ et al (2012) Apparent diffusion
coefficient histogram analysis stratifies progression-free and
overall survival in patients with recurrent GBM treated with
bevacizumab: a multi-center study. J Neuro Oncol 108(3):491–
498
17. Paldino MJ, Desjardins A, Friedman HS, Vredenburgh JJ, Bar-
boriak DP (2012) A change in the apparent diffusion coefficient
after treatment with bevacizumab is associated with decreased
survival in patients with recurrent glioblastoma multiforme. Br J
Radiol 85(1012):382–389
18. Ellingson BM, Cloughesy TF, Lai A, Nghiemphu PL, Pope WB
(2012) Nonlinear registration of diffusion-weighted images im-
proves clinical sensitivity of functional diffusion maps in recur-
rent glioblastoma treated with bevacizumab. Magn Reson Med
67(1):237–245
19. Gupta A, Young RJ, Karimi S et al (2011) Isolated diffusion
restriction precedes the development of enhancing tumor in a
subset of patients with glioblastoma. Am J Neuroradiol
32(7):1301–1306
20. Paulson ES, Erickson B, Schultz C, Li XA (2014) Comprehensive
MRI simulation methodology using a dedicated MRI scanner in
radiation oncology for external beam radiation treatment plan-
ning. Med Phys 42:28–39
21. Provenzale JM, Engelter ST, Petrella JR, Smith JS, MacFall JR
(1999) Use of MR exponential diffusion-weighted images to
eradicate T2 ‘‘shine-through’’ effect. Am J Roentgenol
172(2):537–539
22. Schaefer PW, Grant PE, Gonzalez RG (2000) Diffusion-weighted
MR imaging of the brain. Radiology 217(2):331–345
23. Brandes AA, Tosoni A, Franceschi E et al (2009) Recurrence
pattern after temozolomide concomitant with and adjuvant to
123

radiotherapy in newly diagnosed patients with glioblastoma:
correlation with MGMT promoter methylation status. J Clin
Oncol 27(8):1275–1279
24. Dobelbower MC, Burnett Iii OL, Nordal RA et al (2011) Patterns
of failure for glioblastoma multiforme following concurrent ra-
diation and temozolomide. J Med Imaging Radiat Oncol
55(1):77–81
25. McDonald MW, Shu HK, Curran WJ Jr, Crocker IR (2011)
Pattern of failure after limited margin radiotherapy and temo-
zolomide for glioblastoma. Int J Radiat Oncol Biol Phys
79(1):130–136
26. Milano MT, Okunieff P, Donatello RS et al (2010) Patterns and
timing of recurrence after temozolomide-based chemoradiation
for glioblastoma. Int J Radiat Oncol Biol Phys 78(4):1147–1155
27. Minniti G, Amelio D, Amichetti M et al (2010) Patterns of failure
and comparison of different target volume delineations in patients
with glioblastoma treated with conformal radiotherapy plus
concomitant and adjuvant temozolomide. Radiother Oncol
97(3):377–381
28. Ogura K, Mizowaki T, Arakawa Y (2013) Initial and cumulative
recurrence patterns of glioblastoma after temozolomide-based
chemoradiotherapy and salvage treatment: a retrospective cohort
study in a single institution. Radiat Oncol 8:97
29. Buatti JM, Friedman WA, Bova FJ, Mendenhall WM (1995)
Linac radiosurgery for high-grade gliomas: the university of
florida experience. Int J Radiat Oncol Biol Phys 32(1):205–210
30. Gannett D, Stea B, Lulu B, Adair T, Verdi C, Hamilton A (1995)
Stereotactic radiosurgery as an adjunct to surgery and external
beam radiotherapy in the treatment of patients with malignant
gliomas. Int J Radiat Oncol Biol Phys 33(2):461–468
31. Shrieve DC, Alexander E 3rd, Black PM et al (1999) Treatment
of patients with primary glioblastoma multiforme with standard
postoperative radiotherapy and radiosurgical boost: prognostic
factors and long-term outcome. J Neurosurg 90(1):72–77
32. Cardinale R, Won M, Choucair A et al (2006) A phase II trial of
accelerated radiotherapy using weekly stereotactic conformal
boost for supratentorial glioblastoma multiforme: RTOG 0023.
Int J Radiat Oncol Biol Phys 65(5):1422–1428
123
J Neurooncol
33. Chan JL, Lee SW, Fraass BA et al (2002) Survival and failure
patterns of high-grade gliomas after three-dimensional conformal
radiotherapy. J Clin Oncol 20(6):1635–1642
34. Souhami L, Seiferheld W, Brachman D et al (2004) Randomized
comparison of stereotactic radiosurgery followed by conventional
radiotherapy with carmustine to conventional radiotherapy with
carmustine for patients with glioblastoma multiforme: report of
radiation therapy oncology group 93-05 protocol. Int J Radiat
Oncol Biol Phys 60(3):853–860
35. Tsao MN, Mehta MP, Whelan TJ et al (2005) The american
society for therapeutic radiology and oncology (ASTRO) evi-
dence-based review of the role of radiosurgery for malignant
glioma. Int J Radiat Oncol Biol Phys 63(1):47–55
36. Park I, Tamai G, Lee MC et al (2007) Patterns of recurrence
analysis in newly diagnosed glioblastoma multiforme after three-
dimensional conformal radiation therapy with respect to pre-ra-
diation therapy magnetic resonance spectroscopic findings. Int J
Radiat Oncol Biol Phys 69(2):381–389
37. Tsien CI, Brown D, Normolle D et al (2012) Concurrent temo-
zolomide and dose-escalated intensity modulated radiation ther-
apy in newly diagnosed glioblastoma. Clin Cancer Res
18(1):273–279
38. Lee IH, Piert M, Gomez-Hassan D et al (2009) Association of
11C-methionine PET uptake with site of failure after concurrent
temozolomide and radiation for primary glioblastoma multi-
forme. Int J Radiat Oncol Biol Phys 73(2):479–485
39. Chang EL, Akyurek S, Avalos T et al (2007) Evaluation of
peritumoral edema in the delineation of radiotherapy clinical
target volumes for glioblastoma. Int J Radiat Oncol Biol Phys
68(1):144–150
40. Lee SW, Fraass BA, Marsh LH et al (1999) Patterns of failure
following high-dose 3-D conformal radiotherapy for high-grade
astrocytomas: a quantitative dosimetric study. Int J Radiat Oncol
Biol Phys 43(1):79–88
41. Hegi ME, Diserens AC, Gorlia T et al (2005) MGMT gene si-
lencing and benefit from temozolomide in glioblastoma. N Engl J
Med 352(10):997–1003