Pattern of Failure after Limited Margin Radiotherapy
and Temozolomide for Glioblastoma
Mark W. McDonald, MD, Hui-Kuo G. Shu, MD, PhD, Walter J. Curran, Jr, MD, FACR, Ian R. Crocker, MD, FACR
Department of Radiation Oncology, Emory University School of Medicine, Atlanta, GA
Based on our experience, limited margin radiation with temozo-
lomide does not appear to result in more marginal or distant tumor
recurrences. The volume of brain treated to 60 Gy was, on average,
70% smaller than what would have been treated using the 2.5 cm
margin recommended on RTOG clinical trials. Using a boost PTV
margin of ≤ 1 cm in 92% of our patients, 5% had a marginal failure
and 2% a distant failure. This is comparable to modern series, as
shown in Table 1.
Although the prognosis for patients with glioblastoma remains
poor, the addition of temozolomide to treatment has improved sur-
vival, with 10% of patients remaining alive at 5 years on the EORTC-
NCIC trial of concurrent and adjuvant temozolomide. Reducing the
volume of irradiated brain, especially the boost volume (which re-
ceives the highest dose), may reduce the development of radiation
necrosis and late neurotoxicity in longer term survivors, as the de-
gree of neurotoxicity appears related to the volume of brain irradi-
ated. Limited margin radiotherapy may also reduce the acute tox-
icity of radiation and improve tolerance of new chemotherapy and
molecular agents with overlapping acute toxicities.
Introduction Materials & Methods
Treatment records of the Department of Radiation Oncology were queried to
identify 62 adult patients age 18 years or greater consecutively treated with ra-
diation therapy for newly diagnosed, pathologically confirmed glioblastoma be-
tween 2006 and 2008.
All patients underwent preoperative brain MRI with IV gadolinium contrast.
Patients were evaluated by neurosurgery and underwent either maximal safe
tumor resection or stereotactic biopsy alone. A postoperative brain MRI was ob-
tained within 48 hours and usually on postoperative day one. All patients were
evaluated by neurooncology and received concurrent systemic therapy. Sixty pa-
tients (97%) received concurrent and planned adjuvant temozolomide, and two
patients (3%) were enrolled on a clinical trial investigating the use of concurrent
Radiation therapy was initiated 2–4 weeks after surgery and sooner in patients
who underwent biopsy alone. The initial clinical target volume (CTV) included
postoperative MRI T2 signal abnormality with a median margin of 0.7 cm (range
0.5–1.3 cm). The boost CTV included residual T1 enhancing tumor and resection
cavity with a median margin of 0.5 cm (range: 0–1 cm). CTVs were modified to
respect anatomical barriers to tumor spread. The CTV did not extend into the
ventricles, bone, outside of brain tissue, or into the brainstem or the contralateral
brain hemisphere except along the corpus callosum or cerebral peduncle where
Glioblastoma is the most common malignant primary brain tumor
in adults. The current standard of care is maximal safe neurosurgi-
cal resection followed by focal radiation treatment with concurrent
and adjuvant temozolomide, an oral alkylating agent. Despite ag-
gressive treatment, most patients develop local tumor recurrence or
progression in less than a year.
With the advent of CT imaging to better localize the tumor extent,
the standard radiation treatment changed from whole brain radia-
tion to focal radiation. Contemporary radiation techniques use the
postoperative MRI to define the residual tumor extent and typically
deliver an initial dose to the T2 signal abnormality (a mixture of ede-
ma and infiltrative tumor cells) followed by a “boost” to the residual
T1 enhancing tumor and the tumor resection cavity. A safety mar-
gin is applied to account for uncertainties in defining the true tumor
volume and variability inherent in daily radiation setup. The size of
this safety margin had traditionally been 2–3 cm based on early ana-
tomic and clinical research in the era of nascent CT imaging.
In 2006, based on improvements in neuroimaging, image registra-
tion, and daily treatment localization (kV-kV matching), we adopted
a policy of using reduced margins in radiation planning for patients
with glioblastoma. By reducing the volume of brain receiving high
doses of radiation, acute and late neurotoxicity of treatment may be
reduced, but may also result in more marginal and/or distant tumor
recurrences, which have represented approximately 10% of recur-
rences in modern series. This review examined the pattern of tumor
failure in our patients treated with limited margin radiotherapy and
temozolomide to ascertain the pattern of tumor failure.
An anatomically modified boost CTV margin of 0.5 cm with 0.3 cm
for PTV expansion does not appear to increase the risk of marginal
and/or distant tumor failures when compared to other contemporary
published series. The RTOG-defined boost PTV margin of 2.5 cm
may not be required, as our experience suggests treatment margins
can be safely reduced. Whether a reduction in treatment volume
decreases toxicity is yet to be defined. Applicability of these results
is limited to patients treated with similar quality image registration,
immobilization, and localization.
pathways to contralateral spread exist. Planning target volumes (PTVs) were cre-
ated by adding a uniform 0.3 or 0.5 cm margin to CTVs. The total margin for the
boost PTV was ≤ 1 cm in 92% of patients.
Prescribed doses were 46 or 54 Gy (initial PTV) and 60 Gy (boost PTV). Pa-
tients underwent daily kilovoltage image localization (kV-kV matching) prior to
A hypothetical boost PTV was created for each patient using the 2.5 cm mar-
gin recommended by the Radiation Therapy Oncology Group (RTOG) in their
cooperative group clinical trials. The volume of the treated boost PTV was com-
pared to the hypothetical RTOG boost PTV to assess the percent volume reduc-
tion achieved by using more limited margins.
Patients underwent repeat brain MRI approximately 4 weeks after comple-
tion of radiation therapy and were scheduled for multi-disciplinary follow-up
and brain MRI every 3 months, and sooner if clinical symptoms suggested tumor
progression. Radiographic criteria of tumor progression were development of
new areas of T1 enhancement, enlargement of existing T1 enhancement by more
than 25%, or recurrence of previously resolved areas of T1 enhancement.
Assessment of Pattern of Failure
The MRI at the time of tumor recurrence or progression was coregistered to
the radiation treatment planning data including the treatment volumes and the
radiation dose distribution. New T1 enhancement comprised the volume of re-
current tumor, which was categorized relative to the percent falling within the 60
Gy isodose line (IDL) as infield (>80%), marginal (20–80%), or distant (<20%).
The median follow-up time for all patients was 1 year, and 15
months for those remaining alive. At the time of analysis, 43 of the 62
patients had developed radiographic tumor progression. The medi-
an time to tumor progression was 6 months. Six patients had not yet
progressed after a median follow-up time of 17 months (range 5–28
months). Eight patients died of other medical causes (eg. pulmonary
embolis, aspiration pneumonia, etc.) and had recent MRIs showing
no evidence of tumor progression. Three patients died of presumed
tumor progression but without radiographic imaging. Two patients
have been lost to follow-up.
Forty-one of the 43 patients with radiographic tumor progression
had imaging available for pattern of failure analysis. As shown in
Table 1, 38 (93%) recurrences were infield, 2 (5%) were marginal, and
1 (2%) was a distant recurrence relative to the 60 Gy IDL. The one
distant failure was also distant relative to the hypothetical RTOG
The median volume of the treated boost PTV was 140 cm3 (range
30–394 cm3), 70% less than the 477 cm3 median hypothetical boost
PTV using RTOG margins (range 155–887 cm3) (P < 0.001).
Table 1: Pattern of Failure, with Comparison to Published Data
Institution Year pts Imaging Dose
MDACC 2007 48
U. Mich 1999 36
# Boost Boost
70-80 Gy 1.5 cm
*median boost margin, †boost margin to block edge, ‡MRI in 69%, MDACC = M.D. Anderson Cancer Center, U.Mich = University of Michigan
Pattern of Failure
2.5 cm margin
2.5 cm margin
2 cm margin
Example Patient Case
A heterogeneously enhancing lesion in the superior
medial right frontoparietal lobe was visible on contrast
enhanced T1 MRI (upper left and above).
T2 hyperintensity surrounded the lesion (lower left).
A subtotal resection was performed. The resection
cavity and residual T1 enhancing tumor is encompassed by
the inner red line. The boost PTV (outer red line) included
a 0.8 cm margin, and was treated to 60 Gy. The yellow line
shows the size of a traditional 2.5 cm margin.
The postoperative T2 abnormality (lower left) was
treated with a 0.8 cm margin (outer blue line), to 54 Gy.
The green line shows the size of a traditional 2 cm margin.
A central recurrence developed 5 months after
treatment. 99% of the volume of recurrent tumor was
encompassed within the original boost PTV (red line). The
yellow line shows the size of a traditional 2.5 cm margin.
At lower left, the volume of recurrent tumor is shown
in relation to the original resection cavity and residual T1
enhancement (inner red line) and boost PTV (outer red