J Neurosurg (Suppl) 105:194–201, 2006

A simple dose gradient measurement tool to complement the

conformity index

IAN PADDICK, M.SC., AND BODO LIPPITZ, M.D.

Gamma Knife Centre, Cromwell Hospital and the London Gamma Knife Centre/St. Bartholomew’s
and the Royal London Hospital, London, United Kingdom; and Department of Neurosurgery,
Karolinska Hospital, Stockholm, Sweden

A dose gradient index (GI) is proposed that can be used to compare treatment plans of equal conformity. The steep
dose gradient outside the radiosurgical target is one of the factors that makes radiosurgery possible. It therefore makes
sense to measure this variable and to use it to compare rival plans, explore optimal prescription isodoses, or compare
treatment modalities.
The GI is defined as the ratio of the volume of half the prescription isodose to the volume of the prescription
isodose. For a plan normalized to the 50% isodose line, it is the ratio of the 25% isodose volume to that of the 50%
isodose volume.
The GI will differentiate between plans of similar conformity, but with different dose gradients, for example, where
isocenters have been inappropriately centered on the edge of the target volume.
In a retrospective series of 50 dose plans for the treatment of vestibular schwannoma, the optimal prescription
isodose was assessed. A mean value of 40% (median 38%, range 30–61%) was calculated, not 50% as might be
anticipated. The GI can show which of these prescription isodoses will give the steepest dose falloff outside the target.
When planning a multiisocenter treatment, there may be a temptation to place some isocenters on the edge of the
target. This has the apparent advantage of producing a plan of good conformity and a predictable prescription isodose;
however, it risks creating a plan that has a low dose gradient outside the target. The quality of this dose gradient is
quantified by the GI.

KEY WORDS r Gamma Knife surgery r dose planning r radiosurgery r

conformity r selectivity r dose gradient

quest for a simple and universal scoring system mal prescription isodose, so that the steepest possible dose

T
HE
that can be used to objectively measure the quality
of radiosurgical treatment has been important but,
as yet, not entirely successful. Many authors have tried to
quantify the quality of a treatment plan with conformity
indices.17,24,30 This parameter is an objective measure of
how well the distribution of radiation conforms to the
size and shape of a target, but the dose falloff outside the
target is of equal, if not greater, importance as a mea-
sure of treatment plan quality, particularly with regard to
complication prediction. The lower isodoses outside the
prescription isodose volume cover significant amounts of
normal tissue and are responsible for most normal tissue
complications (Fig. 1).5,11 The GI is a powerful tool that
can be used to objectively measure this dose falloff outside
the target and can also be used to demonstrate the opti-
Abbreviations used in this paper: GI = gradient index; GKS =
Gamma Knife surgery; PIV = prescription isodose volume.
falloff for any given isocenter configuration is achieved.
Recent increases in the sophistication of radiosurgery
delivery techniques have enabled plans of excellent con-
formity to be routinely achievable, even for the most com-
plex of target volumes. It is no longer adequate to produce
a plan with good conformity of the prescription isodose
to the target outline. An equally important consideration
is to ensure optimal dose falloff outside the target, espe-
cially in proximity to critical structures. This is achieved
by ensuring that all the primary beams are incident on the
target, and not on normal tissue, as should be the case for
all forms of external-beam radiotherapy. This rule is vio-
lated when isocenters are placed on the edge of the target
(Fig. 2).
To produce a conformal dose plan, multiple isocenters
must be placed in the delineated target. If too many of
these isocenters overlap, then an undesirable “hot spot” is
created. In this area the dose is significantly higher than
anywhere else in the treatment plan. In Leksell Gamma-
Plan (Elekta Instruments AB, Stockholm, Sweden),

194 J. Neurosurg. / Volume 105 / December, 2006

Simple dose gradient tool and the conformity index

FIG. 1. Neuroimage of a treatment plan for a cavernous sinus

meningioma, revealing the 75%, 50%, 25%, 15%, and 10% iso-
dose lines. There is excellent conformity of the 50% isodose line
and the periphery of the target; however, the 25% isodose volume
covers a significant volume of brainstem and temporal lobe.

treatment plan isodoses are normalized to the maximum

dose (hot spot) in the treatment plan. Control of this hot
spot is critical in producing a treatment plan to a particular
isodose. The requirement to use a particular prescription
isodose volume (usually 50%), often forces compromise
on the position of an isocenter, necessitating its place-
ment toward the periphery of the target, away from the
hot spot (Fig. 3). This can result in placement of isocen-
ters on the edge of, or even just outside, the target. For
such isocenters, the majority of the dose is deposited in
the normal tissue outside of the target volume, and this
will clearly work against the production of an optimal
radiosurgical treatment plan. Although the plan may re-
main highly conformal, lower isodoses will enclose sig-
nificantly larger volumes compared with an optimal plan.
Hence, depending on the method of dose plan construc-
tion, equally conformal plans can be produced that have
remarkably different dose falloff characteristics (Fig. 2).
The GI is designed to differentiate between these plans of
the same conformity but with different dose gradients.
Clinical Material and Methods
The treatment plans of two groups of patients were
analyzed. Group 1 consisted of 58 consecutively pro-
duced dose plans that were retrospectively examined, with
the GI being calculated. The patient characteristics are
J. Neurosurg. / Volume 105 / December, 2006
FIG. 2. Neuroimages of three plans for treatment for the
same vestibular schwannoma. All three plans have equal con-
formity indices as described by Paddick17 (PCI), but the vol-
umes enclosing half the prescription dose are markedly at vari-
ance. a: Plan 1 consists of 5 ⫻ 14–mm isocenters prescribed
to the 46% isodose, PCI = 0.87, PIV46% = 3.7 cm3 , PIV23% =
12.2 cm3 , GI = 12.2/3.7 = 3.30. b: Plan 2 consists of 16 ⫻ 8–mm
isocenters prescribed to the 50% isodose, PCI = 0.87, PIV50% =
3.6 cm3 , PIV25% = 10.6 cm3 , GI = 10.6/3.6 = 2.94. c: Plan 3
consists of 2 ⫻ 14–mm collimators, 2 ⫻ 8–mm collimators, and
a single 4-m collimator, PCI = 0.87, PIV42% = 3.7 cm3 , GI =
9.8/3.7 = 2.65. The prescription isodose of 42% has the greatest
deviation from the gold standard of 50%, but nevertheless has the
sharpest dose falloff outside the target. This contradicts the gen-
erally held belief that many small isocenters and a 50% isodose
line produce a sharper dose falloff.
195
 I. Paddick and B. Lippitz

TABLE 2
Treatment plan parameters for Group 2
Factor Value

no. of dose plans 50

target vol (cm3 )
mean 2.9
median 1.7
range 0.20–12.9
prescription isodose (%)
mean 49.4
median 48.5
range 40–75
no. of isocenters
mean 5.9
median 6
range 2–13
Shaw conformity index
mean 1.14
median 1.09
range 1.00–1.60
Paddick conformity index
mean 0.81
median 0.84
range 0.62–0.93
FIG. 3. Neuroimage of dose planning for a vestibular schwan-
noma. The 45% and 95% isodose lines are shown. The addition
of the final 8-mm isocenter (large circle) causes contraction of
the isodoses, because of its proximity to the hot spot. The dose from LGP and analyzed using Excel (Microsoft, Inc.,
planner has three choices: 1) Move the 8-mm isocenter further out Redmond, WA).
into the brainstem. 2) Accept a renormalized plan and use a lower
prescription isodose. 3) Replace the 8-mm isocenter with 4-mm
isocenters to move the isodose line closer to the border (no need Results
to renormalize the plan).
The proposed GI is the ratio of the volume of half the
prescription isodose to the volume of the prescription iso-
summarized in Table 1. Plans with a target volume of less dose. The index can be used for any prescription isodose.
than 0.2 cm3 were excluded because of the compromises For a plan normalized to the 50% isodose line, it is the ratio
made with conformity and the unavoidable spillage of of the 25% isodose volume (PIV25% ) to the 50% isodose
dose outside the target inherent in such treatments. Mul- volume (PIV50% ), whereas for a plan normalized to the
tiple metastases treatments were also excluded because 60% isodose line, it is PIV60% /PIV30% . This simple ratio
of the complex interaction between different treatment has the advantage of allowing unbiased comparison be-
sites. tween alternative plans normalized to different isodoses.
Group 2 consisted of dose plans for 50 arbitrarily This is difficult with previously proposed GIs.24,30
selected patients with vestibular schwannoma and were
examined for optimal GIs by varying the prescription iso- Comparison of Treatment Plans
dose. Vestibular schwannoma treatments were chosen be- In a study of 58 consecutive dose plans for patients
cause of their significant dependence on plan conformity treated at a single institution over a 1-year period, the mean
and selectivity. Treatment plan parameters are summa- GI was 2.83 (median 2.81, range 2.4–3.3). Ten treatment
rized in Table 2. plans had a ratio greater than 3.00. These were examined
Measurements for both groups were made on LGP and found to have isocenters inappropriately placed on the
(version 5.32; Elekta Instruments AB) configured for the edge of, or outside, the target. The plan with the lowest
Gamma Knife unit (model B/C; Elekta Instruments AB). GI (2.4) was one in which a single 14-mm isocenter was
Numerical dose–volume histogram data were exported placed.

TABLE 1 Optimal Isodoses

Tumor type and treatment plan parameters for Group 1∗ What prescription isodoses are currently used by
Gamma Knife surgeons, radiation oncologists, and
Tumor Type No. of Dose Plans physicists and why? In an arbitrary selection of 10
vestibular schwannoma 30 papers4,6,9,11–13,20,21,28,31 published since 2000 in which the
meningioma 11 treatment of vestibular schwannoma by GKS was de-
arteriovenous malformation 7 scribed, prescription isodoses varied between 30% and
pituitary adenoma 5 94%,4,20 although some centers (in three of 10 papers) re-
single metastasis 4
trigeminal schwannoma 1
ported strict adherence to the 50% isodose line,12,13,21 and
one group reported strict adherence to the 65% isodose
∗
Prescription isodose: mean 50%; median 50%; range 40 to 68%. line.28

196 J. Neurosurg. / Volume 105 / December, 2006

Simple dose gradient tool and the conformity index

In the treatment of cerebral metastases, prescription iso-

dose variation tends to be wider because single isocen-
ter treatments are more common. A selection of 10
papers2,3,8,14,16,18,24,26,27,29 published since 2000 showed
prescription isodoses ranging from 25 to 91%,18 and no au-
thor reported strict adherence to any particular value iso-
dose. When treating with a single isocenter, adjustment
of the volume of the prescription isodose to fit around
the target is only possible by changing the value of the
prescription isodose line. This makes it difficult to treat
lesions of different diameter with the same value isodose.
An example of this imposed deviation from the ideal 50%
is for an 8-mm diameter target, where single isocenter
coverage is produced either with the 4-mm collimator at
the 29% isodose line or with the 8-mm collimator at the
83% isodose line. By comparing a lower isodose that rep- FIG. 4. Graph showing the variation of the proposed GI against
resents a consistent fraction of the prescription isodose the prescription isodose line for individual collimators.
against the prescription isodose volume, the GI caters for
such a wide variation in prescription isodose and is im-
mune to any variation in the prescription isodose between Multiple Isocenter Plans
different treatment plans. The GI was retrospectively plotted against varying pre-
scription isodoses for 50 multiisocenter plans for treat-
ment of vestibular schwannoma (Fig. 6), and the lowest
Single Isocenters GI value with its corresponding prescription isodose was
noted. The optimal prescription isodose averaged 39.8%
Gradient Indices were plotted against the prescription (median 38%, range 30–61%). Forty seven of 50 plans
isodose for the four collimators (Table 3, Fig. 4). The GI’s had a superior dose falloff at prescription isodoses of less
were calculated for the 50% isodose line, and optimal than the 50% isodose, none at 50%, and three greater
prescription isodoses were found that gave the lowest GI than 50%. This demonstrates that in the majority of cases,
(Table 3). These varied dramatically between collimators. a prescription isodose below 50% gives a superior dose
Although the largest collimator has the lower ratio, it gradient outside the target. The original prescription iso-
does not necessarily mean that the linear dose gradient doses used for treatment averaged 49.4%, (median 49.5%,
is greater for the 18-mm than for the 4-mm collimator. range 40–75%). Forty-eight of 50 plans would have ben-
It does imply that the volume included within half efited from renormalization to a lower prescription iso-
the prescription isodose is smaller relative to the volume dose, and two would have benefited from a higher iso-
of the prescription isodose. The larger isodose volume dose prescription. Optimal GI’s were 7.1% lower than
of the 18-mm collimator creates the lowest ratio those actually achieved in the original dose plans, 9.4%
because the falloff of volume is related to the cube of lower than those achieved if the 50% prescription iso-
the linear dimensions. However, when the cubed root of dose had been strictly used for every plan, and 47.1%
isodose volumes are compared, the optimal prescription lower than those achieved if the 65% isodose had been
isodose values remain the same. used for every plan. In this study, plans were not adjusted
Data from Fig. 4 can be presented in a clinically useful to maintain conformity. Although this means that many
format by plotting the GI against the diameter of the cor- plans with optimal GI’s had poor conformity (as the result
responding prescription isodose (Fig. 5). These data show of over or undercoverage), these data do show that for a
which collimator size gives the optimal dose falloff out-
side the target for any particular target/prescription iso-
dose diameter (Table 4). This knowledge can be useful
when deciding which collimator to select when using a
single isocenter plan. These data are correct for measure-
ments in the x and y direction, but assume a good fit of
the prescription isodose to the target in the z direction.

TABLE 3
Summary of data contained in Fig. 4
Factor Value

collimator size (mm) 4 8 14 18

GI for 50% prescription isodose 2.71 2.71 2.52 2.51
optimal prescription isodose (%) 38 52 66 68
GI for optimal prescription isodose 2.63 2.60 2.37 2.34
percent difference between GI FIG. 5. Graph showing the variation of the GI with prescrip-
for 50% & optimal GI (%) 3.0 4.1 6.0 6.8 tion isodose diameter in the x and y planes for the four single
collimators.

J. Neurosurg. / Volume 105 / December, 2006 197


FIG. 6. Graph showing the variation of mean GI with prescrip-
tion isodose for 50 vestibular schwannoma treatment plans. The
flat bottom of the curve shows 1.5% variation in the GI between
the 33% and 43% isodose lines.
series of 50 different isocenter configurations dose falloff
can generally be increased by lowering the prescription
isodose.
The optimal prescription isodose and optimal GI val-
ues were investigated for correlation with the number of
isocenters and target volume by using regression anal-
ysis. There was a strong correlation found between the
optimal GI and the number of isocenters (p ⬍ 0.001). Sur-
prisingly, a greater number of isocenters created a poorer
dose falloff outside the target, as measured by the GI
(Fig. 7).
Discussion
The conformity index, in its various manifestations over
the years, represents an attempt to measure objectively
how well the distribution of radiation follows the shape of
the radiosurgical target. It is generally accepted that con-
formity of a radiosurgical plan is important for successful
treatment.
Shaw, et al.,24 were the first to propose such an index
for use in radiosurgery. They defined the conformity index
(CI) as the prescription isodose volume (PIV) divided by
the target volume (TV):
PIV
CI = . (1)
TV
This index is still in use at many Gamma Knife centers,
but the new index proposed by Paddick17 is now used by an
TABLE 4
Collimator sizes for optimal dose falloff for targets up to
30 mm in diameter
Collimator Diameter Optimal Isodose Diameter
(mm) (mm)
4 0–9
8 10–14
14 15–19
18 20–30
198
I. Paddick and B. Lippitz
FIG. 7. Graph showing the optimal GI plotted against the num-
ber of isocenters for a series of 50 vestibular schwannoma treat-
ment plans.
increasing number of centers because it does not produce
false perfect scores. This is described as
TVPIV2
PCI = , (2)
TV ⫻ PIV
where TVPIV is the volume of the target covered by the
prescription isodose.
Wagner, et al.,30 were the first to propose a confor-
mity index containing a component that measured the
dose falloff outside the prescription isodose volume. The
Conformity Gradient Index (CGI) is defined as
CGI = (CGIc ⫹ CGIg)/2
CGIc = (TV/PIV) ⫻ 100% (3)
CGIg = 100 – {100[(Reff,50%Rx –Reff,Rx )–3 mm]}
where Reff,50%Rx = effective radius of isodose that is 50%
of PI and Reff,Rx = effective radius of prescription isodose.
The Conformity Gradient Index relies on an effective
radius of the target, which is cumbersome to calculate. A
3-mm fall off is allowed from the PIV to half the PIV.
Any more than 3 mm results in a numerical penalty. In
addition, the gradient element of the index is combined
with Shaw’s TV/PIV conformity index, which is known
to produce false perfect scores.17
Nakamura, et al.,15 looked at the relationship between
conformity and clinical complication rates for a series
of 1338 lesions treated at their institution between 1993
and 1998. Unexpectedly, they found a direct correlation
between treatment conformity and toxicity. This was ex-
plained by the fact that the treatment plans for very small
lesions (⬍1 cm3 ) were of poor conformity, but no com-
plications occurred. However, when excluding the lowest
prescription volume quartile (to exclude small lesions with
poor conformity), their results did not change. This is not
entirely surprising. The larger the lesion, the easier it is to
produce a conformal dose plan; however, radiation toxic-
ity is more likely to occur because of the greater volume
irradiated.
J. Neurosurg. / Volume 105 / December, 2006
Simple dose gradient tool and the conformity index
Liščák, et al.,11 ignored the conformity index and looked
at the ratio of lower isodose volumes to the prescription
isodose. They looked for a correlation between these pa-
rameters and complications following GKS in a series of
121 patients treated for vestibular schwannoma. A posi-
tive correlation was found between vertigo, dizziness, and
seventh cranial nerve complications and the ratio of the
volume of the prescription isodose minus 20% and the vol-
ume of the prescription isodose. This correlation clearly
demonstrates the importance of minimizing lower isodose
volumes in radiosurgery. The parameters used by Liščák
and colleagues had the disadvantage of being dependent
on the prescription isodose. For a plan normalized to the
40% isodose, this ratio would be PIV20% /PIV40% (ratio
of the volume of half the prescription dose to the PIV),
whereas for a plan normalized to the 60% isodose line
it would be PIV40% /PIV60% (ratio of the volume of two
thirds of the prescription dose to the PIV). The significant
advantage of the GI proposed in this paper is that its value
is independent of the prescription isodose.
Flickinger, et al.,5 demonstrated a correlation between
complications following GKS for arteriovenous malfor-
mations and the 12-Gy isodose volume. The 12-Gy dose is
approximately half the prescription dose for this group of
85 patients in whom complications developed. The 12-Gy
isodose volume, which includes healthy brain surround-
ing the target, is now generally regarded as a complication
predictor for a large number of conditions,10 which again
demonstrates how important the lower isodoses are in the
prediction of complications.
It may surprise some readers that the authors are ad-
vocating prescription isodoses lower than 50%, when this
level is already considerably lower than that used in linear
accelerator–based radiosurgery.
Traditionally, the prescription dose for GKS has cor-
responded to the 50% isodose line and this is still the
most common practice today, with a few exceptions.28
The original basis for adopting this practice was built on
the premise that when viewing the sigmoid curve of the
dose profile of a single isocenter, the steepest dose gra-
dient is around the half maximum: the 50% isodose line.
This figure of 50% is, however, only an approximation.
Using data from Leksell GammaPlan, the actual positions
of maximum dose gradient for the 4-, 8-, 14-, and 18-
mm collimators are in the region of the 63, 68, 70, and
78% isodoses, respectively. The use of multiple isocenters
complicates the matter further (Fig. 8).
What potential differences in treatment effects might
we expect from dropping our nominal prescription isodose
from around 50% to 40%?
Complications
By lowering the prescription isodose, the maximum
dose in the center of the target will accordingly rise. As
already demonstrated in this paper, lowering the prescrip-
tion isodose to 40% will, in the majority of cases, de-
crease the dose to normal tissues immediately outside
the target volume, lowering the complication risk as re-
ported by Liščák, et al.11 A controversial area is the case
in which functional tissue/critical structures are believed
to be present inside the target volume, for example in the
case of vestibular schwannomas or cavernous sinus menin-
giomas. Plowman19 suggested that in the case of treatment
J. Neurosurg. / Volume 105 / December, 2006
FIG. 8. A simplified dose profile along a line through a multi-
isocenter dose plan. When planning using multiple isocenters, the
maximum dose gradient is still at approximately the 70% isodose
line of each individual isocenter. As multiple isocenter configu-
rations produce maxima that are much higher than the individual
isocenter maxima, the renormalization of the plan means that the
area of steepest dose gradient is found at much lower isodose
volumes. In this example it is found at 40% of the peak dose of
the plan.
for vestibular schwannoma, high maximum doses inside
the target may be undesirable because the hot spot may be
incident on a length of the cochlear nerve running through
a portion of tumor inside the internal auditory meatus.
Following reports of excellent hearing preservation af-
ter linear accelerator–based radiotherapy and surgery, in
which higher prescription isodoses were used1,7 (typically
80–90%), Van Eck, et al.,28 reported on a series of 95 pa-
tients treated with GKS with 13 Gy to the 65% isodose
line. The 1-year follow-up results suggested that the in-
creased hearing preservation (86%) in this group might
be due to this technique. Their group of patients was one
of the first to benefit from the increased conformity made
possible by the recently introduced automatic position-
ing system, and the comparisons made were with the
Gamma Knife unit model U/B–treated group, in which
conformity was generally inferior. This difference, along
with the relatively short follow-up time, may account for
the good results recorded. Microanatomical data do not
suggest that an increase in maximum dose should be of
concern to the surgeon. In a series of 609 patients, Sam-
path, et al.,22 recorded the incidence of facial and cochlear
nerves encompassed by vestibular schwannomas in which
the tumor was less than 2.5 cm in diameter. There was
a 3.4% incidence of the facial nerve passing through the
tumor, and a single case of involvement with the cochlear
nerve (0.16%). This demonstrates that in many patients,
vestibular schwannomas do not encompass any critical
structure.
In the case of cavernous sinus meningiomas when the
tumor has encompassed, for example, the third or sixth
nerves, care must be taken to ensure that the hot spot is
away from the critical structure, as should be the case
when planning to any isodose.
In other lesions treated with GKS, in which there is
no critical structure inside the target volume, there is no
indication that the maximum dose within the target has
any influence on the therapeutic outcome, although overall
there is no study demonstrating a relationship between the
maximum dose in GKS and complication rates. Indeed,
the wide variations in prescription isodose level used when
treating multiple metastases, in which peripheral isodoses
199

of 40 to 90% are not unusual have failed to yield any
correlation between maximum dose and outcomes, with
the possible exception local control.25
It is important to note that a dose plan may have opti-
mal conformity at one isodose but optimal GI at another
isodose. Therefore, it is not sufficient to merely construct
a conformal dose plan, find the optimal GI, and select its
corresponding prescription isodose. The GI will, however,
let the dose planner know whether the prescription isodose
selected is close to producing the maximum dose gradient
for a particular isodose configuration. Through an iterative
process, a dose plan with optimal conformity and GI can
then be achieved. For example, a multiisocenter plan is
produced, with the 50% isodose line conformally cover-
ing the target. After dose matrix histogram is exported and
analyzed, the optimal isodose for a minimum GI is found
to be 40%. The plan is then adjusted by moving isocen-
ters further into the target, increasing the hot spot, which
renormalizes the plan until the 40% isodose conformally
covers the target. The optimal GI is then rechecked and
if necessary the plan is adjusted again. From a practical
point of view, time will be saved if the 40% isodose line
is initially selected when planning commences. If a dose
planner starts planning with a particular isodose showing,
it is far more likely that they will finish a plan with that
isodose covering the target volume. In practice, the first
author (I.P.) now plans initially to the 40 to 45% prescrip-
tion isodose line. This usually gives a near optimal GI,
and replanning is rarely needed. In the series of vestibular
schwannomas, 45 of 50 plans have a GI within 4% of their
lowest value if prescribed to the 40% isodose line, whereas
the average deviation from the optimal GI for all 50 plans
would only be 1.5%, compared with 10.4% for the 50%
isodose, when only 21 of 50 plans would be within 4%
of the lowest GI. Figure 6 shows a 1.5% variation from
the optimal GI over a 10% range of prescription isodose
level. This illustrates that a near-optimal isodose will give
a GI of negligible difference to the optimal GI.
As different dose planners plan with different tech-
niques, it may be that the mean optimal prescription iso-
dose slightly varies from one planner to another. This can
be easily verified by retrospective analysis of plans created
at each institution.
Conclusions
The GI is simple in construction, yet provides the quan-
titative means with which to compare two competing dose
plans. A more favorable GI reflects a steeper dose gradi-
ent and, therefore, a lower applied radiation dose to the
healthy brain and ultimately a lower complication rate.
This is particularly valuable in critical anatomical loca-
tions or larger volumes where the GI helps to select the
dose plan with the lowest “penumbra dose.” This is partic-
ularly important for multiisocenter treatments, which are
used in the majority of GKS dose plans. From our retro-
spective review of 58 clinical treatments, a GI of less than
3.0 generally reflects a reasonably selected prescription
isodose level combined with a well-placed configuration
of appropriately sized isocenters.
Not only is the GI useful for comparing two potential
treatment plans for the same patient, but it can also be
used to compare different methods of radiosurgery, for
200
I. Paddick and B. Lippitz
example, different models of the Gamma Knife unit, lin-
ear accelerator, CyberKnife, tomotherapy, and protons. In
addition, the index can be quoted with clinical results,
giving other users an indication of the dose gradients
present in particular series of patients. Furthermore, it
does not require the outlining of the target, the require-
ment of which has led to a slow uptake of the conformity
index.
Like all dose plan scoring tools, this index can never be
a substitute for clinical expertise in evaluating a treatment
plan.
There are good grounds for believing that the GI may
be a useful predictor of adverse outcomes in the treatment
of targets in proximity to critical structures, in the same
way that the 12-Gy isodose volume is used to predict radi-
ation necrosis complications in arteriovenous malforma-
tions and Liščák’s ratio is used to predict complications in
the treatment of vestibular schwannomas. Assuming that
the peak dose inside the target volume (and therefore the
value of the prescription isodose) is not detrimental, then
prescribing to isodoses around 40% will usually improve
the dose falloff around the target volume for multiisocen-
ter plans. The isodose that gives the steepest dose gradient
will depend on the isocenter configuration of the individ-
ual plan; however, the GI can be used to indicate whether
a proposed isodose is near optimal, and if not, what the
optimal prescription isodose is.
Acknowledgments
The authors would like to express their thanks to Professor Chris-
ter Lindquist, for his many helpful discussions, and Mr. Phil Black-
burn and Mrs. Diane Paddick for their assistance in preparing this
manuscript.
References
1. Andrews D, Suarez O, Goldman H, Downes M, Bednarz G,
Corn B, et al: Stereotactic radiosurgery and fractionated
stereotactic radiotherapy for the treatment of acoustic schwan-
nomas: comparative observations of 125 patients treated at
one institution. Int J Radiat Oncol Biol Phys 50:1265–1278,
2001
2. Bhatnagar AK, Flickinger JC, Kondziolka D, Lunsford LD:
Stereotactic Radiosurgery for four or more intracranial metas-
tases. Int J Radiat Oncol Biol Phys 64:898–903, 2006
3. Chen JC, Petrovich Z, O’Day S, Morton D, Essner R, Gian-
notta SL, et al: Stereotactic radiosurgery in the treatment of
metastatic disease in the brain. Neurosurgery 47:268–279,
2000
4. Chung WY, Liu KD, Shiau CY, Wu H, Wang L, Guo W,
et al: Gamma Knife surgery for vestibular schwannoma: 10-
year experience of 195 Cases. J Neurosurg (Suppl) 102:87–
96, 2005
5. Flickinger JC, Kondziolka D, Lunsford LD, Kassam A,
Phuong LK, Liščák R, et al: Development of a model to predict
permanent symptomatic postradiosurgery injury for arteriove-
nous malformation patients. Int J Radiat Oncol Biol Phys
46:1143–1148, 2000
6. Flickinger JC, Kondziolka D, Niranjan A, Lunsford LD: Re-
sults of acoustic neuroma radiosurgery: an analysis of 5 years’
experience using current methods. J Neurosurg 94:1–6, 2001
7. Fuss M, Debus J, Lohr F, Huber P, Rhein B, Engenhart-
Cabillic R, et al: Conventionally fractionated stereotactic ra-
diotherapy (FSRT) for acoustic neuromas. Int J Radiat Oncol
Biol Phys 48:1381–1387, 2000
J. Neurosurg. / Volume 105 / December, 2006
Simple dose gradient tool and the conformity index
8. Hasegawa T, Kondziolka D, Flickinger JC, Germanwala A,
Lunsford LD: Brain metastases treated with radiosurgery
alone: an alternative to whole brain radiotherapy? Neuro-
surgery 52:1318–1326, 2003
9. Hasegawa T, Kida Y, Kobayashi T, Yoshimoto M, Mori Y,
Yoshida J: Long-term outcome in patients with vestibular
schwannoma treated using Gamma Knife surgery: 10-year
follow up. J Neurosurg 102:10–16, 2005
10. Korytko T, Radivoyevitch T, Colussi B, Wessels V, Pillai R,
Maciunas R, et al: 12 Gy Gamma Knife radiosurgical volume
is a predictor for radiation necrosis in non-AVM intracranial
tumors. Int J Radiat Oncol Biol Phys 64:419–424, 2006
11. Liščák R, Novotný J, Urgošik D, Vladyka V, Šimonová G:
Statistical analysis of risk factors after Gamma Knife radio-
surgery of acoustic neurinomas. Radiosurgery 3:205–213,
2000
12. Lunsford LD, Niranjan A, Flickinger JC, Maitz A, Kondzi-
olka D: Radiosurgery of vestibular schwannoma: summary of
experience in 829 cases. J Neurosurg (Suppl) 102:195–199,
2005
13. Massager N, Nissim O, Delbrouck C, Devriendt D, David P,
Desmedt F: Role of intracanalicular volumetric and dosi-
metric parameters on hearing preservation after vestibular
schwannoma radiosurgery. Int J Radiat Oncol Biol Phys 64:
1331–1340, 2006
14. Mindermann T: Tumor recurrence and survival following
Gamma Knife surgery for brain metastases. J Neurosurg
(Suppl) 102:287–288, 2005
15. Nakamura JL, Verhey LJ, Smith V, Petti P, Lambourn K, Lar-
son D, et al: Dose conformity of Gamma Knife radiosurgery
and risk factors for complications. Int J Radiat Oncol Biol
Phys 51:1313–1319, 2001
16. Nam TK, Lee JI, Jung YJ, Im YS, An HY, Nam DH, et al:
Gamma Knife surgery for brain metastases in patients harbor-
ing four or more lesions: survival and prognostic factors. J
Neurosurg (Suppl) 102:147–150, 2005
17. Paddick I: A simple scoring ratio to index the conformity of
radiosurgical treatment plans. J. Neurosurg (Suppl 3) 93:
219–222, 2000
18. Pan HC, Sheehan J, Stroila M, Steiner M, Steiner L: Gamma
Knife surgery for brain metastases from lung cancer. J Neu-
rosurg (Suppl) 102:128–133, 2005
19. Plowman PN: Post-radiation sensorineuronal hearing loss. Int
J Radiat Oncol Biol Phys 52:589–591, 2002
20. Prasad D, Steiner M, Steiner L: Gamma Knife surgery for
vestibular schwannoma. J Neurosurg (Suppl 3) 93:32–36,
2000
21. Rowe JG, Radatz MWR, Walton L, Soanes T, Rodgers
J, Kemeny AA: Clinical experience with Gamma Knife
radiosurgery in the management of vestibular schwannoma
J. Neurosurg. / Volume 105 / December, 2006
secondary to type 2 neurofibromatosis. J Neurol Neurosurg
Psychiatry 74:1288–1293, 2003
22. Sampath P, Rini D, Long D: Microanatomical variations in
the cerebellopontine angle associated with vestibular schwan-
nomas (acoustic neuromas): a retrospective study of 1006
consecutive cases. J Neurosurg 92:70–78, 2000
23. Serizawa T, Iuchi T, Ono J, Saeki N, Osato K, Odaki M, et al:
Gamma Knife treatment for multiple brain tumors compared
with whole-brain radiation therapy. J Neurosurg (Suppl 3)
93:32–36, 2000
24. Shaw E, Kline R, Gillin M, Souhami L, Hirschfeld A, Dinapoli
R, et al: Radiation Therapy Oncology Group: radiosurgery
quality assurance guidelines. Int J Radiat Oncol Biol Phys
27:1231–1239, 1993
25. Shaw E, Scott C, Souhami L, Dinapoli R, Kline R, Loeffler
J, Farnan N: Single dose radiosurgical treatment of recurrent
previously irradiated primary brain tumors and brain metas-
tases: final report of RTOG protocol 90–05. Int J Radiat
Oncol Biol Phys 47:291–298, 2000
26. Shaw E, Scott C, Souhami L, Dinapoli R, Kline R, Loeffler J, et
al: Single dose radiosurgical treatment of recurrent previously
irradiated primary brain tumors and brain metastases: final
report of the RTOG protocol 90-05. Int J Radiat Oncol Biol
Phys 47:291–298, 2000
27. Sneed PK, Suh JH, Goetsch SJ, Sanghavi SN, Chappell R,
Buatti JM, et al: A multi-institutional review of radiosurgery
alone vs. radiosurgery with whole brain radiotherapy as the
initial management of brain metastases. Int J Radiat Oncol
Biol Phys 53:519–526, 2002
28. Van Eck A, Horstmann G: Increased functional hearing after
Gamma Knife surgery for vestibular schwannoma. J Neuro-
surg (Suppl) 102:204–206, 2005
29. Varlotto JM, Flickinger JC, Niranjan A, Bhatnagar AK,
Kondziolka D, Lunsford LD: Analysis of tumor control and
toxicity in patients who have survived at least one year after
radiosurgery for brain metastases. Int J Radiat Oncol Biol
Phys 57:452–464, 2003
30. Wagner TH, Bova FJ, Friedman WA, Buatti JM, Bouchet
LG, Meeks SL: A simple and reliable index for scoring rival
stereotactic radiosurgery plans. Int J Radiat Oncol Biol Phys
57:1141–1149, 2003
31. Wowra B, Muacevic A, Jess-Hempen A, Hempel JM, Müller-
Schunk S, Tonn JC: Outpatient Gamma Knife surgery for
vestibular schwannoma: definition of the therapeutic pro-
file based on a 10-year experience. J Neurosurg (Suppl)
102:114–118, 2005
Address reprint requests to: Ian Paddick, M.Sc., Gamma Knife
Centre, Cromwell Hospital, Cromwell Road, London, SW5 OTU,
United Kingdom. email: ian@physicsconsulting.co.uk.
201