See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/12235221

Development of Radiation Therapy Optimization

Article in Acta oncologica (Stockholm, Sweden) · February 2000

DOI: 10.1080/028418600750013267 · Source: PubMed

CITATIONS READS
102 1,519

1 author:

Anders Brahme
Karolinska Institutet
365 PUBLICATIONS 11,426 CITATIONS

SEE PROFILE

All content following this page was uploaded by Anders Brahme on 27 March 2015.

The user has requested enhancement of the downloaded file.

 REVIEWS AND CONTROVERSIES

Development of Radiation Therapy Optimization

Anders Brahme

From the Department of Medical Radiation Physics, Karolinska Institutet and Stockholm University,
Stockholm, Sweden

Correspondence to: Anders Brahme, Department of Medical Radiation Physics, Karolinska Institutet and
Stockholm University, Box 260, S-171 76 Stockholm, Sweden

Acta Oncologica Vol. 39, No. 5, pp. 579–595, 2000

Acta Oncol Downloaded from informahealthcare.com by 177.155.212.60 on 05/20/14

The principal radiobiological problems in the treatment of advanced tumors and the solution of many of them by radiobiologically
optimized intensity-modulated radiation therapy are presented. Considerable improvements of the treatment outcome using radiobiolog-
ically optimized intensity-modulated treatments are achieved by: (a) increasing the tumor dose and dose per fraction; (b) keeping constant
or even reducing slightly the dose and dose per fraction to organs at risk; (c) reducing the overall treatment time and the number of
treatment fractions. The merits of the new radiation modalities and advanced intensity-modulated treatment techniques are compared in
terms of equipment costs per patient cured. It is predicted that the new development of radiobiologically optimized intensity-modulated
radiation therapy will rapidly become an important clinical tool, increasing the efficiency of the collaboration between radiation
physicists, radiation biologists and radiation oncologists. Not only does it allow the optimal treatment of every patient, but it also
promotes an efficient feedback of treatment outcome and complication data to improve the accuracy of known dose response relations
to further augment future treatment results. Equipment costs may go up during a transition period until efficient interfaces between new
diagnostic equipment, treatment-planning systems and intensity-modulated treatment units are fully developed. From then onwards the
cost of high quality biologically optimized intensity-modulated treatments will decrease and so will the treatment time and personnel
For personal use only.

requirements, at the same time as the treatment quality is greatly improved particularly for more advanced tumors.
Recei6ed 11 No6ember 1999
Accepted 6 April 2000

During the first century of radiation therapy many dra- review was given by Milford Schulz in his 1974 Janeway
matic advances have taken place. The improvements of Lecture: ‘The supervoltage story’ (1). To illustrate the
treatment outcome have mainly been linked to develop- broad range of devices that have been used throughout
ments in the areas of molecular and clinical oncology, the years some of the main radiation sources are given in
radiation biology and radiation physics. The present dis- Table 1.
cussion will concentrate mainly on the radiation aspects It is surprising that even though such a multitude of
even though in recent years an important potential for radiation devices have been used there are very few ran-
improvements is emerging in the area of molecular oncol- domized clinical trials that have been performed in order
ogy. The radiobiological developments have been numer- to quantify the possible clinical merits of each new treat-
ous in areas like dose fractionation, understanding of the ment modality. This is generally explained by the rather
role of hypoxic tumor cells and the relative biological modest and obvious clinical improvements expected so a
efficiency of different radiation modalities at low and large-scale randomized study would not really be feasible.
high doses. The increased understanding of the mecha- For example the improvements with regard to skin spar-
nisms behind cell cycle progression and check point con- ing going from low or medium energy x-rays to mega-
trol, cell survival curves and dose response relations for voltage photons like 60Co and above are generally
different tumors and normal tissues, as well as different accepted by most people. Often the publications of Allt
repair mechanisms of damaged DNA has been impor- (2) and Bush (3) are cited with regard to the improve-
tant. In the area of radiation physics the development ments obtained going from 60Co to betatron
includes accurate dosimetry, treatment planning and di- bremsstrahlung. However, it is not clear which property
agnostic techniques in addition to the development of a of the betatron beam that is responsible for the improve-
large number of new radiation modalities. Almost all ment. Allt (2) suggested the increased field size used with
radiation sources that have been developed by accelerator the betatron but that is not linked to the new higher
physicists have been tested with regard to their possible photon energy unless it was feasible just because the
advantages in the treatment of cancer. A nice historical integral dose was lower with the high-energy beam.

© Taylor & Francis 2000. ISSN 0284-186X Acta Oncologica

 580 A. Brahme Acta Oncologica 39 (2000)

More recently some clear-cut evidence for the advantage time, the dose should be increased by 0.35 Gy to compen-
of multileaf collimation compared to more conventional sate for accelerated tumor cell proliferation. If we go
beam blocking was demonstrated in the American neutron through all the troubles to make the clinical data sets
therapy trial where a considerable reduction of the compli- compatible well defined dose response relations exist for
cations was seen with the isocentric double focused multi- most tumors and normal tissues and can be determined
leaf collimator in Seattle as compared to the more and used as an invaluable historical reference data set for
conventional collimation techniques used at the other US treatment prediction and optimization. The aim of the
centers (4). Such a large influence could be expected with present review is to discuss the clinical possibilities and the
neutrons, because normal tissue damage is a major prob- cost effectiveness of the different new intensity-modulated
lem since the repair potential of neutron induced damage treatment techniques that are rapidly being developed and
to normal tissues is very low. In photon beams the gain introduced in the clinic today.
would probably be smaller and more difficult to detect in
a clinical trial. THE RADIOBIOLOGICAL BASIS FOR THE
Even though one preferably would like to test most new DEVELOPMENT OF RADIATION THERAPY
treatment modalities and irradiation techniques by ran- In order to accurately quantify the relative merits of new
domized clinical trials, small improvements in irradiation treatment modalities and treatment techniques clinically
techniques may be more accurately quantified by well
Acta Oncol Downloaded from informahealthcare.com by 177.155.212.60 on 05/20/14

relevant radiobiological response models are needed. For

designed radiobiological models. Holthusen (5) was one of that purpose it is important to adopt biologically sound
the first to discuss the role of dose response relations for cell survival models for tumors and normal tissues to
tumors and normal tissues. A few decades ago some clinically observed dose response data so that the model
researchers did not even believe in the existence of dose describes the response of standard radiation therapy. As
response relations and there are still those who question mentioned in the introduction and discussed in great detail
their relevance (6, 7). However, today when we can ac- in a number of publications (9 – 11) a large amount of
count for most factors that influence the radiation re- clinical dose response data is being accumulated. Several
sponse, like actuarial survival, total dose, dose per introductions to the modeling of tumors and normal tis-
For personal use only.

fraction, dose rate and total treatment duration accurate sues have recently been published, so the reader is referred
dose response relations can be established. For example, in to them for more details about the underlying radiobiol-
the study by A, gren-Cronqvist et al. (8) the tumor control ogy and the fundamental properties of the biological mod-
of five different larynx data sets was studied as a function els (12 – 19). The present focus is therefore on the major
of tumor stage. After bringing all the data sets together on challenges of curative therapy of advanced disease.
the same response format the standard deviation of the A typical dose response data set (19) is illustrated in Fig.
50% control dose, D50, was as small as 3% around 60 Gy 1, where the dose response curves for small and advanced
in 2 Gy fractions in 6 weeks. For each day beyond this head and neck tumors and the associated fatal complica-

Table 1
Radiation modalities for radiation therapy

Radiation modalities used for radiation therapy Approximate year of Approximate cost range
introduction (M$)

X-ray tube and high voltage source combinations 1896 0.2–0.5

Radionuclides for Brachy and external beam therapy 1900 0.5–1
Radium–60Co–137Cs–127I–252Cf–103Pl
Bremsstrahlung and electrons 1930 1–5
Cascade transformers
Van de Graaff generators
Betatrons
Synchrotrons
Traveling and standing wave linear accelerators
Circular microtrons
Racetrack accelerators
p mesons, neutrons, protons and light ions 1935 5–40
D-T generators
Cyclotrons
Proton linear accelerators
Synchrotrons
Heavy ions from various ion sources 1975 40–150
Radiofrequency quadruple, linear accelerator and synchrotron
combinations
 Acta Oncologica 39 (2000) De6elopment of radiation therapy optimization 581
Acta Oncol Downloaded from informahealthcare.com by 177.155.212.60 on 05/20/14
For personal use only.

Fig. 1. The problem of the diminishing therapeutic window in curative radiation therapy of small glottic and advanced head and neck
tumors. The large horizontal arrows illustrate the shift of the tumor control (PB) and normal tissue complication curves (PI) as the tumor
stage is increased from small glottic tumors to larger more advanced tumors (19). The insert illustrates one slice of the approximately
uniform dose delivery used to obtain the clinically observed data points.

tions are plotted. The mean dose to the tumor, or more shifted to the right to higher doses (left most dotted curve).
precisely the internal target volume (20), is used as inde- However, as the larger tumor is irradiated to higher doses
pendent dose variable. This data set, in a nutshell, illus- the fatal normal tissue reactions increase substantially and
trates the main problem of radiation therapy of advanced the complication curve (PI, the probability for fatal injury)
tumors. For the small glottic tumors (solid curves) the moves instead to the left or lower doses. Thus, for ad-
therapeutic window between tumor control (PB and left vanced disease the therapeutic window is substantially
most sigmoid curve) and fatal normal tissue damage (PI decreased making it very difficult to cure all patients. This
and right most sigmoid) is very large. For this reason the is the key problem of curative radiation therapy and we
probability to achieve complication free cure (almost 85% have to solve it in order to cure patients who present with
at the optimal dose level) is very high. However, for the advanced disease.
advanced tumors (dotted curves) the therapeutic window Figure 1 also illustrates a few other interesting phenom-
has shrunk and the optimal complication free cure is now ena that are common in many clinical materials. Not only
about 35%. Figure 1 illustrates the case of more or less are the dose response curves for the larger tumors shifted
standard uniform beam radiation treatments where the to larger doses but their dose response slopes, as expressed
increased tumor burden for the larger tumors requires by their normalized dose response gradient, g, are de-
substantially larger radiation fields and tumor doses to be creased. Conversely, the slope is increased for the normal
eradicated, in accordance with most existing radiobiologi- tissues making it even more difficult to achieve a high
cal dose response models. Therefore, the tumor control probability of complication free cure. This further reduc-
curve (PB, the probability for beneficial treatment) is tion of the therapeutic window for advanced tumors is
 582 A. Brahme Acta Oncologica 39 (2000)

owing to an increased tumor heterogeneity and increased also depend on that most genetic alterations in the normal
mass of normal tissues being irradiated resulting in more tissues are preserved in the tumor, such as for example the
shallow and steeper slopes, respectively (16, 19, 20). ATM gene, which makes both the tumor and the normal
Furthermore, for the advanced tumor stages the clini- tissues more sensitive than usual (19, 21). To be able to use
cally observed complication free cure (solid triangles on clinical dose response data for treatment optimization it is
the middle dashed bell-shaped curve) is lower than most therefore important to record not only tumor cure and
traditional response models predict (upper dashed bell- normal tissue damage but also to register the degree of
shaped curve). This is so since the traditional models correlation between these two major endpoints as it
assume that cure and fatal complications are statistically strongly influences the location of the optimal dose level as
independent processes resulting in a complication free cure seen by the dashed bell-shaped curves in Fig. 1, cf. A, gren
or positive treatment outcome given by the product of the et al. (19).
probability for cure and no fatal injury or P + =PB(1− In order to illustrate how the merits of the new biologi-
PI). However, in the clinical material almost all patients cally based treatment optimization techniques using inten-
(80%) who suffer fatal complications are also cured so sity-modulated beams can handle these problems, we will
the clinically observed P + is much closer to the totally now compare them with standard uniform external beam
correlated response with P + =PB −PI (lowest dashed treatment configurations on an advanced head and neck
bell-shaped curve). The main reason why the tumor con- tumor similar to that of the dotted curves in Fig. 1. The
Acta Oncol Downloaded from informahealthcare.com by 177.155.212.60 on 05/20/14

trol and severe normal tissue complications often are intensity-modulated dose distributions have been calcu-
strongly correlated in clinical trials is that both are caused lated with one of the most advanced and flexible optimiza-
by high uniform dose levels both in the tumor and the tion algorithms available today (22 – 25).
normal tissues, for example when parallel opposed beam The resultant dose response relations are presented in
techniques are being used. However, the correlation may Fig. 2 and they illustrate the considerable therapeutic
For personal use only.

Fig. 2. Illustration of the possible improvement of the therapeutic window and the clinical outcome by using biologically optimized
intensity-modulated treatments. By three-dimensional intensity-modulation the dose to the tumor is increased at the same time as it is
reduced to the normal tissues. It should be pointed out that this figure is based on a more uniform subset of the advanced head and neck
tumors in Fig. 1 with shorter treatment time and, in addition, the dose per fraction was increased instead of the number of fractions.
Therefore, the observed steepnesses of the response curves and the cure (dotted lines) are higher than in Fig. 1 (cf. 14, 26). The
considerable increase of the therapeutic window and complication free cure by using intensity-modulated dose delivery is clearly seen
(solid curves).
 Acta Oncologica 39 (2000) De6elopment of radiation therapy optimization 583

advantages achieved by using intensity-modulated radia- therapy technique developed by Wright et al. (27) and
tion beams. The inserted isodose diagrams compare in the Takahashi (28). Simultaneously, the first optimization al-
same slice of a head and neck target classical and intensity- gorithms with essentially uniform or wedged beams shaped
modulated beams and the associated dose response curves according to the projection of the target volume were
for the two treatments (dotted and solid curves, respec- developed (29, 30). This is also the method used by most
tively). This figure allows a comparison of the two treat- workers today saying they are doing conformal radiation
ment techniques since the horizontal dose axis expresses therapy. However, from Fig. 3 it is clear that considerable
the mean dose to the tumor or more exactly the internal improvements in treatment outcome are not assured until
target volume. Therefore, the dose-response curves for the we enter the area of full intensity-modulation and physi-
tumor will practically be the same in the two cases (left cally and preferably biologically optimized treatments. It is
most solid sigmoidal curve). However, the normal tissue well known today that the considerable dose optimization
complication curves are different since in the intensity- effort that took place from the mid-sixties throughout the
modulated plan the dose to the tumor is significantly seventies was not so successful due to the low number of
increased at the same time as the dose to the critical free variables used, cf. Brahme (31). We need thousands of
normal tissues is reduced somewhat. Since all curves are free variables whereas in the sixties a few dozen of fields
related to the mean tumor dose and it has increased over with or without wedges were used. However, when radio-
the normal tissue dose by almost 15 Gy compared to the biological dose optimization is used it is even possible to
Acta Oncol Downloaded from informahealthcare.com by 177.155.212.60 on 05/20/14

plan with essentially uniform dose, the solid complication reduce the number of free variables using few field multi-
curve for the intensity-modulated plan has thus moved segmented treatments, cf. Svensson et al. (32).
about 15 Gy to the right from its location with essentially As important as the dose level delivered to the target
uniform dose delivery (right most sigmoidal curve). tissues is the fractionation schedule employed. Consider-
As a consequence, the therapeutic window has opened able advances have taken place during the last few
up substantially and the bell-shaped curve for complica- decades. First, it was understood that large doses per
tion-free cure (P + ) has increased from almost 50% to fraction are not tolerated by the late responding normal
85%. This significant increase in treatment outcome is tissues that normally set a limit to what can be achieved by
For personal use only.

achieved because biologically optimized intensity-modu- radiation therapy. It is well proven today that this phe-
lated dose delivery results in an increased dose to the nomenon is linked to the extended shoulder region of the
tumor at the same time as the normal tissues are spared to cell survival curve (33) which often is due to a larger
a greater extent. By comparison of Figs. 1 and 2 it is seen compartment of resting G0 cells in such organs. Late
that with intensity-modulation the treatment outcome for responding organs therefore have a longer recovery phase
advanced tumors can be brought up to the level of small after irradiation allowing them to repair a larger fraction
tumors and uniform dose delivery and even somewhat of the sub-lethal injury induced by each treatment. The
further, provided the disease is well localized. late response of such organs may also be due to multi-step
This also implies that the classic problem of double processes where the radiation effect on the microscopic
trouble in dose escalation is converted to a double or even capillary vasculature in addition to direct damage to tissue
triple advantage. Not only is the relative dose to the specific cells may contribute.
normal tissues reduced, but the dose per fraction and the The shape of the shoulder of the cell survival curve is
dose rate are also reduced and thus the risk for severe late often described by the a/b ratio of the linear quadratic cell
complications is substantially reduced. As mentioned in survival model:
the figure caption the response curves in Fig. 2 were 
− aD 1 +
D 
derived with a varying dose per fraction rather than with s=e a/b [1]
varying the number of 2 Gy fractions as in Fig. 1. This
data set is therefore well suited for use with non-uniform An a/b value of 3 Gy is common for late responding
dose delivery where the dose per fraction will vary with the tissues which means that at 3 Gy the linear cell kill is equal
intensity-modulation (26). The comparison in Fig. 2 thus to the quadratic repairable kill of the D 2 term. Tumors
illustrates the clinical advantages that can be gained by and acutely responding normal tissues have an a/b of
biologically optimized intensity-modulated dose delivery. about 10 Gy and large doses per fraction are better
A schematic view of the different steps in the develop- tolerated.
ment of treatment techniques that took place during the It has been known for a very long time that the cell kill
first century of radiation therapy is illustrated in Fig. 3. is purely exponential (34) at very large doses and not
Obviously, all radiation therapy aims at being conformal: quadratic as equation 1 indicates. More recently it has
that is to wrap the ‘therapeutic’ isodose as close as possible been shown that the linear quadratic model is not so well
around the tumor, or more exactly, around the internal suited to describe the response at low doses either, even
target volume. In the present review the term conformal is though it works quite well in the standard clinical 2
used to describe this aim similar to the conformation Gy/fraction range. At low doses the cell kill seems to be
 584 A. Brahme Acta Oncologica 39 (2000)
Acta Oncol Downloaded from informahealthcare.com by 177.155.212.60 on 05/20/14
For personal use only.

Fig. 3. Schematic overview of the development of modern radiation therapy techniques. Approximate year of introduction and the
resultant complication free cure, P + , compared to that of standard radiation therapy are also indicated. It is seen that the commonly
used term conformal therapy does not necessarily imply a very large improvement in P + . Obviously the whole lower row of panels are
special cases of IMRT and CRT but in general such high degrees of conformity is rarely obtained just by absorbers or dynamic
collimation, respectively.

much steeper than previously realized. The increased sensi- A very interesting effect of clinical relevance can be seen in
tivity at low doses as shown in Fig. 4 is more pronounced Fig. 4 by studying the shallowest possible secant of the
when the shoulder of the cell survival curve is large (35) survival curve through the origin to find the smallest
such as for late responding tissues with a low a/b value. effective cell kill for a given dose to the tissue in question.
There are already indications that this low dose hypersen- For the lung epithelial cells in Fig. 4 this is the tangent
sitivity phenomenon may be clinically relevant since low close to the 2 Gy per fraction dose which means that this
and high doses per fraction to the skin may cause similar dose per fraction is the one which produces the least
skin reactions even if the total skin doses differ substan- possible cell kill per unit dose to the tissue. Thus, the low
tially (36). From the above points of view a better descrip- dose hypersensitivity not only makes the tissue quite sensi-
tion of the cell survival curve is obtained by an expression tive for low doses per fraction but it also forms a fraction-
of the type: ation window where a certain dose level per fraction
produces the least possible damage to a given tissue per
s=e − aD +bD 2e − bD [2]
unit dose when trying to irradiate an underlying tumor. It
where the a term describes the irreparable cell kill and the is very likely that this mechanism is responsible for the
b term the potentially lethal part which can be repaired if observation in most clinical trials that the best possible
the right environmental conditions for repair prevail. The therapeutic effect is obtained in the dose per fraction range
thick solid curve in Fig. 4 is from equation 2 and the dots around 2 Gy per fraction.
with error bars are experimental data for lung epithelial At higher doses per fraction the late responding tissues
cells (35). By comparing with the dashed linear quadratic introduce more complications (32) whereas at lower doses
curve due to equation 1 the improved fitting particularly at per fraction the low dose hypersensitivity may cause more
low but also at high doses outside the figure are obvious. complications as indicated in Fig. 4. Thus, when introduc-
 Acta Oncologica 39 (2000) De6elopment of radiation therapy optimization 585

ing optimized intensity-modulated radiation therapy to low 2 Gy per fraction. These problems of intensity-mod-
increase the dose in the tumor and often simultaneously ulated and multi-portal conformal therapy may therefore
reduce it to the normal tissues (cf. Fig. 2) we should partly be solved in the clinic by one and the same ap-
make sure that the dose per fraction to the normal tis- proach: a shortening of the overall treatment time by a
sues is largely left unchanged to ensure a substantially reduction of the number of fractions when using opti-
increased tumor cure and an unchanged level of damage. mized intensity-modulated beams.
Since the low dose hypersensitivity is most pronounced Even if the reduction is not more than a few fractions
in late responding tissues with broad shouldered survival it represents a new kind of advantage since the effect of
curves (a/b :3) one would expect that this phenomenon accelerated tumor repopulation that often sets in after
is not very pronounced for the tumors (a/b] 10) in fair the first 2 – 3 weeks of treatment will be less of a prob-
agreement with existing experimental data, cf. Lambin et lem when the total treatment time is reduced (37). In
al. (35). general, the delivered dose could be reduced by 0.4 –0.6
A further conclusion of high relevance to some forms Gy per day of shortening of the total duration of the
of modern conformal treatments may be drawn, since treatment compared to the reference situation of 30 frac-
the dose is often spread out from a large number of tions in 6 weeks. Therefore, if the total dose is left un-
beam portals covering a large part of the body volume. changed, an equivalent tumor boost of about 0.5 Gy
This will make the dose per fraction to a large percent- times the number of days of treatment time reduction is
Acta Oncol Downloaded from informahealthcare.com by 177.155.212.60 on 05/20/14

age of the patient’s normal tissues low and an increased obtained.

cell kill due to the low dose hypersensitivity may in- In conclusion, there are basically five kinds of im-
crease complications. It may therefore be desirable to provement of the treatment outcome possible when using
increase the dose per fraction to the normal tissues back biologically optimized intensity-modulation:
to the classically established level of around or just be-
– Increase of the tumor dose and dose rate
– Increase of the tumor dose per fraction
– Unchanged or slightly reduced dose per fraction to
For personal use only.

the normal tissues

– A reduced treatment time and an associated addi-
tional tumor boost
– Less work in the clinic as fewer treatment fractions
are used.
All of these advantages can simultaneously be effective
at a varying degree or all the advantages may be focused
on one or two of them making the remaining largely
unaffected leaving considerable room for advanced ra-
diobiological treatment optimization.
It is often claimed by physicians that radiation ther-
apy in the future will be replaced by molecular biology
approaches and gene therapy. Surely we will see consid-
erable improvements in these areas. However, they will
also augment the power of radiation therapy as we can
use the new genetic markers for radiation responsiveness
and radio-resistance. Furthermore, gene therapy will
probably fail on large bulky solid tumors and could gain
considerably by combination with radiation therapy.
We should therefore also expect considerable improve-
ments in radiation therapy when these new tools come
into clinical use as indicated by the last step of develop-
ment in Fig. 3. It is likely that patient individual pre-
dictive assay based on molecular or cellular methods
Fig. 4. Illustration of the normal tissue fractionation window when combined with biologically based therapy opti-
caused by low dose hypersensitivity. Both at lower doses and mization can improve the treatment outcome by 5 – 15%
higher doses per fraction the normal tissue damage is increased
over that at the standard 2 Gy/fraction level used in classical
compared to using the average value of the response
radiotherapy. For simplicity a fixed survival fraction at 2 Gy of over a population, at least for locally advanced tumors
0.3 is assumed for the tumor. (12, 13, 19).
 586 A. Brahme Acta Oncologica 39 (2000)
Acta Oncol Downloaded from informahealthcare.com by 177.155.212.60 on 05/20/14

Fig. 5. Illustration of the main types of beam shaping and dose delivery systems developed today. The greater flexibility in dose delivery
with the narrow pencil beam and fan beam approaches is obtained at the cost of an increased treatment time. This time loss is not
obtained if dynamic multileaf collimation is combined with scanned narrow photon beams that are directed only where the collimator is
opened.
For personal use only.

COMPARISON OF NEW RADIATION MODALITIES may not be as large as the reduction of the beam area since
AND IRRADIATION TECHNIQUES the narrow beam approaches may be used without beam
As seen from Table 1 a large number of radiation therapy flattening filter thus increasing the output by a factor of
modalities have been developed and used over the years. two. Still the treatment time with a fan beam is about
The majority of the dose delivery systems have been of the 5 – 10 times larger than for conventional treatments and
classical type using block collimation sometimes combined almost 100 times larger for the narrow pencil beam ap-
with wedge or compensator filters as illustrated by the proaches unless the beam current and speed of motion of
third panel in Fig. 5. However, in recent years most new the accelerator are substantially decreased. The treatment
radiation modalities are linked to more flexible beam time can also be increased by removing the filter when
shaping or intensity-modulation. The other three panels in using the dynamic leaf collimation approach. If scanned
the top row of Fig. 5 illustrate various approaches from beams are used, the treatment time can be brought back to
robot mounted linear accelerators producing fairly narrow standard uniform beam values or even faster when sub-
photon beams that can be moved around the patient, via stantial intensity-modulation of the beam is required. It is
intensity-modulated fan beams like in the tomotherapy therefore clear that dynamic multileaf collimation is here
and peacock devices to full dynamic multileaf collimation to stay as a practical and cost effective way of realizing
in the fourth panel of Fig. 5. Obviously, this last approach intensity-modulation.
has the greatest flexibility since by almost closing all the Today multileaf collimation combined with scanned
collimator leaves a narrow slit beam and the fan beam beams is available for clinical electron and photon therapy
geometry is obtained and by further closing all but one through the Racetrack accelerator (38), but it has also
leaf pair completely a pencil beam delivery is possible. The been developed for ion beam therapy (39) proton therapy
lower set of panels in Fig. 5 illustrates the few field (40, 41) and been proposed for neutron therapy (42). It is
treatment techniques employed with standard and dy- clear that such techniques have important advantages with
namic multileaf collimation and the arc techniques of regard to the speed and quality of the treatment and may
tomotherapy and robotic dose delivery. be increasingly implemented and used for radiation ther-
From the above description it is immediately clear that apy in the new millennium. The large number of new
the treatment time will be significantly increased as the treatment techniques that are being developed today cov-
collimator is closed down and only a narrow fan or pencil ers the entire spectrum of radiation modalities in Table 1
beam is delivered. The effective increase in treatment time and their radiobiological and dose distributional properties
 Acta Oncologica 39 (2000) De6elopment of radiation therapy optimization 587

are summarized in Fig. 6. It is no major surprise that the This normal tissue damage syndrome is present particu-
time and cost axia are running parallel in Table 1 since the larly in neutron beams due to their high LET, being
most costly techniques have been developed most recently. almost as high as for heavy ions as seen in Fig. 6. This also
After the extensive and very interesting test period of explains why combined neutron and photon treatments
heavy ion beams Berkeley was closed down in 1992, only have shown interesting results in clinical trials (45, 46). The
the advanced developmental projects at Chiba, Japan and microscopically more uniform low LET photon part helps
Darmstadt, Germany, are being used. It is clear from a remove microscopic regions with too low dose associated
cost/effectiveness point of view (in this context: equipment with high LET dose delivery. Owing to their extensive use,
cost per patient cured) that these projects still need to be neutron beams have shown more clear-cut evidence that
clinically proven. They have probably their most impor- they are advantageous for some largely hypoxic tumors,
tant role in the study and development of high LET such as salivary glands, prostate and sarcomas. In fact, the
radiobiology for a deeper understanding of the role of ions optimal use of a high LET beams of neutrons or heavy
in radiation therapy. However, since the optimistic days of ions should combine with low LET beams in the often
the late seventies (43) the role of high LET heavy ions has well-oxygenated microscopically invasive part of the tumor
not really met up to the expectations largely due to the as illustrated in Fig. 7. The high LET component should
increased normal tissue damage in microscopically invasive thus be focused only on the gross tumor. This radiobiolog-
parts of the tumor. Today this is at least partly understood ically optimized photon and neutron beam combination
Acta Oncol Downloaded from informahealthcare.com by 177.155.212.60 on 05/20/14

since high LET beams have a high RBE at low doses. was based on experimental cell survival data for the two
However, at curative doses the high microscopic hetero- beam components for well-oxygenated and hypoxic cells
geneity makes the low dose RBE misleading since tumor (47). Still the cost effectiveness of neutrons is a problem
clonogens will survive in microscopic cold spots (44). This owing to the high installation cost and the high expertise
microscopic heterogeneity is understandable as the high of the personal required for dosimetry and the handling of
ionization density of the heavy ions and the low dose normal tissue reactions. The interest in boron capture
delivered ( 20 Gy) by necessity will make some tumor therapy in recent years using epithermal neutrons from
clonogens escape lethal hits. Therefore, to be curative with nuclear reactors that otherwise had been dismantled may
For personal use only.

high LET beams considerably larger doses must be deliv- reduce the cost but will still suffer from the microscopic
ered than indicated by the low dose RBE, causing severe heterogeneity, specially in invasive regions with low boron
acute and late reactions in the normal tissues. This normal uptake and low neutron fluence.
tissue problem, augmented by the poor repair of high LET Today p mesons are not really realistic either, partly due
damage, sometimes overshadows the advantage of a high to the high accelerator costs involved. They do have some
LET in hypoxic tumors. interesting therapeutic properties though, since they like

Fig. 6. Comparison of the radiobiological (effective LET and OER) and the dose distributional properties (penumbra and tumor to
superficial tissue dose ratio for a parallel opposed beam pelvic irradiation) of different radiation beam modalities. The approximate cost
per typical installation is also indicated (cf. 46).
 588 A. Brahme Acta Oncologica 39 (2000)

ulated dose delivery with electrons and photons has re-

duced the number of tumors where protons are required.
Because the cost of a proton installation is about 10 times
higher than for advanced photon and electron therapy
units with intensity-modulation, mainly very difficult tu-
mor sites are indicated for proton therapy. Among tumors
that are of interest for proton and light ion application are
those located close to serially organized tissues where a
small local overdose can cause fatal complication such as
most tumors close to the spinal cord. The steep dose
gradients outside the target volume make protons and
light ions an interesting alternative in such cases (cf. Fig.
6).
To quantify the merits of high-energy proton therapy as
compared to radiobiologically optimized intensity-modu-
lated photon and electron beams some comparisons have
been made as illustrated in Figs. 8 and 9. For an advanced
Acta Oncol Downloaded from informahealthcare.com by 177.155.212.60 on 05/20/14

cervix cancer with extensive nodal involvement all the

Fig. 7. The radiobiologically optimized combination of high LET three radiation modalities were compared using radiobio-
neutrons and low LET photons on a hypoxic tumor. Neutrons logically optimized intensity-modulation. In Fig. 8 the
dominate over the hypoxic region whereas photons dominate in
angular variation of the maximum achievable level of
the well-oxygenated microscopically invasive region. The compli-
cation free cure with only photons (upper solid curve) is inferior complication free cure (P + ) is plotted as a function of the
to that with neutron addition owing to the high survival of the angle of incidence for a single intensity-modulated beam
hypoxic cell fraction in a pure photon beam. portal. Obviously, this target would never be treated with
a single portal of any of the tested modalities since the
the light ions He, Li and Be combine low and high LET inevitable dose reduction from most directions can be
For personal use only.

dose delivery in one single modality. Of the more costly improved considerably from other directions (cf. Fig. 9).
radiation modalities light ions and protons are most inter- Since every proton portal generally requires multiple range
esting, since they combine advanced precise dose delivery shifted proton beams we have compared them with elec-
with a not too high LET and efficient repair in normal trons and photons alone or in combination with multiple
tissues. However, the rapid development of intensity-mod- beam energies through one and the same beam portal.

Fig. 8. Comparison of the single beam portal complication free cure of some of the main low LET radiation modalities in Fig. 6 as a
function of the angle of incidence while treating an advanced cervix tumor. The optimal angle of each treatment modality is indicated
by a star.
 Acta Oncologica 39 (2000) De6elopment of radiation therapy optimization 589
Acta Oncol Downloaded from informahealthcare.com by 177.155.212.60 on 05/20/14
For personal use only.

Fig. 9. Same as Fig. 8 but now considering two beams and plotting the resultant complication free cure, (P + ), using color coding. It is
striking that all the radiation modalities reach approximately the same maximum complication free cure but for greatly differing angles
of incidence.

Owing to their very different depth dose characteristics optimal direction is now around 45° close to the maximum
(cf. Fig. 6) the beam direction, where photons and electrons with only electrons. If we now combine 10 pre-selected
alone have their most interesting use (highest P + value), electron beam energies, the P + is increased by 21% over the
differs considerably as seen in the lower part of Fig. 8. For single 50 MeV electron energy and 12% over photons alone
50 MeV electrons the therapeutic range is not enough for to 74% indicating the high value of simultaneous intensity-
lateral irradiation but for some anterior and all posterior and energy-modulation with electrons (see also Fig. 10).
directions intensity-modulated electrons do much better If we similarly use 10 pre-selected proton energies uni-
than photons owing to the finite range and low exit dose of formly covering the maximum depth range of the tumor, a
the electrons. The high complication free cure with lateral further increase by 12% is seen to a total of 86%. If higher
50 MV photon beams is owing to the high photon energy energy electrons were allowed such that their therapeutic
with a deep extended dose maximum and the possibility to range covered the maximum tumor depth better over 70%
give a high dose to the gross tumor without damaging would be achieved with 100 MeV alone (cf. Fig. 9) and at
rectum. To see the improvement in dose delivery through least 80 – 82% would be reached combining 10 different
a single portal the proceeding curve is for 50 MeV and 50 lower electron energies and probably a few percent more if
MV combined from the same direction by biologically a single high-energy photon beam was replacing one of the
optimizing the total dose distribution delivered by simulta- electron energies (not shown). In conclusion, with a single
neous intensity-modulating of both the electron and photon beam portal 10 intensity-modulated proton beams probably
components (cf. 48, 49). The resultant improvement is give some 3 – 5% higher P + than 10 electron and photon
considerable from some directions specially anterially but beams combined through one portal.
the best P + value is only increased by 3% to 65% from 62% It should be pointed out that all of this proton advantage
with photons alone and 53% with electrons alone. The would be lost if not biologically optimized three-dimen-
 590 A. Brahme Acta Oncologica 39 (2000)

sional Bragg peak spot scanning would be used for the this reason unfortunate that the majority of all uniform beam
protons and it is thus a fundamental prerequisite for future photon therapy is performed with parallel opposed or
proton installations, cf. Brahme (46). Unfortunately, this is four-field box techniques where the entrance and exit portals
not yet done systematically at any proton therapy center to coincide and give a high dose to the normal tissues.
date and significantly lower values would be obtained by Perpendicular beams or most preferably 100° – 120° beams
most standard fixed range modulation techniques presently and symmetric three-field techniques are generally more
in use. advantageous for photons (50). However, for electron and
If we now switch from one to arbitrary two beam portal proton beams the situation is different as they have almost
combinations the situation becomes much more complex as zero exit dose and therefore less interaction with the normal
we need to optimize some 650 dose plans covering all possible tissues. At an electron energy of 25 MeV for example
beam combinations used in Fig. 9 instead of the 36 in Fig. anterior-posterior almost parallel short opposed beams are
8 (at even 10° intervals). For this reason we have changed the only possibility owing to their short therapeutic range.
from P + (V) to plotting P + (V1, V2) as function of the two The other very striking result seen in Fig. 9 is that
angles of incidence V1 and V2 in the (V1, V2) plane with P + independent of radiation modality the highest peak is always
color coded using the Sucholov color table starting with at the same P + level close to 90%. However, the optimal
black as zero going through dark blue, green, yellow, orange, beam angle combination varies considerably with the radi-
red, pink and white as unity. ation modality. For protons the selection of beam directions
Acta Oncol Downloaded from informahealthcare.com by 177.155.212.60 on 05/20/14

In Fig. 9 six such phase space diagrams for 5 and 50 MV is not important, even parallel opposed beams work well due
photons, 25, 50 and 100 MeV electrons and 200 MeV protons to the low exit dose. All the electron and photon beams
are summarized. Some very interesting conclusions can be however have a few discrete regions where the incident beams
drawn from these diagrams. First of all, the diagonals of each should preferably be located to maximize P + . From this we
diagram in Fig. 9, where V1 V2, correspond strictly to may draw the conclusion that with biologically optimized
single beam treatments as they are the degenerate cases where intensity-modulated proton beams a rotary gantry is no more
the two beams coincide and are thus equal the corresponding necessary since we can always rotate the patient in the
modality in Fig. 8 (since P + (V) P + (V, V)). The color horizontal plane to get 90° and 270° beam portals. However,
For personal use only.

coding of diagonals in Fig. 9 thus corresponds to P + curves it should be pointed out that this study for reasons of
in Fig. 8 as seen by comparing the curves and colors for 50 treatment planning time consumption was made with a
MeV and 50 MV for example. It is clear that for most two-dimensional patient where every slice through the
modalities it is important to avoid the region around V1 : V2 patient is the same. Therefore, in some advanced treatments
since the resultant P + is decreased when the intensity-mod- where the ideal beam directions vary considerably from slice
ulating components become more and more aligned and to slice the extra flexibility of a rotary gantry may be desirable
dependent. From the small ‘side valleys’ parallel to the deep as seen from Fig. 8. However, the proton P + does not vary
central diagonal valley it is also seen that parallel opposed much with the single beam portal direction except possibly
photon beams (V2 =V1 9180°) should for the same reason for beams entering from the very sensitive rectal side
be avoided using intensity-modulated dose delivery. It is for (V :180° in Fig. 8).

Fig. 10. Schematic illustration of the clinical advantages of combining electron energy- and intensity-modulation for complex heteroge-
neous target volumes.
 Acta Oncologica 39 (2000) De6elopment of radiation therapy optimization 591
Acta Oncol Downloaded from informahealthcare.com by 177.155.212.60 on 05/20/14
For personal use only.

Fig. 11. Comparison of the different dose distributional properties of high- and low-energy electron and photon beams. A simplistic score
of counting only positive factors was used to compare the merits of different electron and photon beam combinations.

A way to get 360° coplanar rotation with a fixed hori- middle isodose diagram how the dose distribution in the
zontal proton beam port would be to treat the patient target has been improved and made more uniform. The
seated as commonly done in non-hospital based accelera- improvement is achieved partly by irradiating less sensitive
tors. It therefore is still worthwhile to consider using structures outside the target to get in scatter to the tumor.
radiobiologically optimized three-dimensional intensity- However, most of the hot and cold spots have been
modulated proton beams with fixed beam delivery to bring eliminated by the intensity-modulation as indicated by +
down the cost from, maybe 40–45 million USD for two and − in the middle panel of Fig. 10. In the last third of
rooms to maybe 25–30 million USD for three rooms Fig. 10 local energy- and intensity-modulation are used to
without rotary gantries. This would reduce the cost per enclose the target by a high uniform dose distribution. The
treatment room to about 40% of that with rotary gantries. lowest energy is predominantly used over the shallow
Some of the main reasons why high-energy electron and target portion to avoid irradiating normal tissues beyond
photon beams do almost as well as high-energy protons the tumor, whereas high energy is used to sharpen the dose
when intensity-modulation is used are illustrated in Fig. 10 distribution near the heterogeneities.
where a heterogeneous step phantom and target are irradi- In Fig. 11 the various dose distributional advantages of
ated by uniform but also energy and intensity-modulated combined high- and low-energy electron and photon beam
electron beams. This heterogeneity problem is also relevant treatments are compared schematically. A number of inter-
for proton beams which suffer similar multiple scatter esting modality combinations can be distinguished in this
interactions in the patient and to a lesser extent also to the figure primarily adopted to classical uniform beam deliv-
secondary electron and photon scatter in photon beams. ery. Starting with plain electron therapy, two regions can
The first third of the figure shows what happens with a be distinguished directly: the classical one, where a patch
uniform beam producing a cold spot under the ‘nose’ and work of different low-energy beams generally below 20
a severe hotspot just outside it due to out scatter. The MeV are used to save underlying tissues largely owing to
middle portion illustrates how the fluence profile of the the steep dose fall-off along the depth axis of such beams.
incident beam could be adjusted to correct for the dosimet- Normally the low-energy electrons are not really safe in
ric problems with uniform beams. It is clearly seen in the multiple non-parallel beam configurations in more shallow
 592 A. Brahme Acta Oncologica 39 (2000)

tumor regions due to the steep dose gradients at the end of lung tumors with a high density surrounded by low-density
their range. However, in the high-energy region the dose lung also parallel beam portals of electrons and photons
fall-off is less steep and the exit dose is still low so multiple have an interesting role to sharpen up the dose delivery at
beam combinations with either electrons or photons be- the periphery of the tumor. Obviously, many of these
come very useful. The narrower penumbra in high-energy advantages, most relevant for uniform dose delivery, will
electron beams is also useful in such cases as seen from the benefit considerably by combination with intensity-modu-
beam characteristics tabulated in Fig. 11. By combining lated electron and photon beams as illustrated in Fig. 10.
parallel or perpendicular electron and photon beams a Beside the main road of development of few field inten-
sharpening of the target volume dose distribution is also sity-modulated treatments that are safe and easy to verify
possible (49). (52) a large number of special techniques have been devel-
The low-energy photons have their greatest value for oped for various target locations (53). Elongated tumors
shallow to deep-seated tumors of small sizes. This is so following the lymph nodes or other extended organs of the
since a narrow penumbra is most important for small body are well treated by tomotherapy. For targets in the
targets and it is narrowest in low MV photon beams. When head the new dynamic g-knife using an automatic patient
the tumor diameter increases to about 5 cm the entrance positioning system and biological optimization of the mo-
dose becomes more and more of a problem for deep seated tion of the patient head opens new unprecedented possibil-
tumors requiring beams with broadest possible build-up ities for complex intra-cranial targets. In such devices the
Acta Oncol Downloaded from informahealthcare.com by 177.155.212.60 on 05/20/14

region and highest possible photon energy, cf. Söderström biological advantage of multiple concomitant beams on the
et al. (51). Deep therapy with combined high-energy elec- target, cf. Svensson (54), are likely to improve the cure of
tron and photon beams is most useful when the penumbra large complex lesions.
and entrance and exit dose regions can be optimally com- To illustrate the power of radiobiological treatment
bined e.g. using perpendicular beam configurations. For optimization with multiple conventional photon beams, a
For personal use only.

Fig. 12. A seven-field treatment plan of a prostate tumor where the intensity distribution and the angle of incidence of the beams have
been radiobiologically optimized using the universal optimization algorithm ORBIT within the Pinnacle dose planning system. The
perfect encloser of the target volume and significant sparing of organs at risk are striking.
 Acta Oncologica 39 (2000) De6elopment of radiation therapy optimization 593

treatment plan where the angles of incidence are optimized

together with their intensity-modulation, is shown in Fig.
12. This is a seven-field plan for a prostate tumor where the
initial suggestion was seven nearly equally spaced beam
portals. An almost 6% improvement in the complication
free cure could be achieved by fine adjusting the angles of
incidence so they better separate the dose delivery to the
prostate from the ‘spill-over’ to rectum and bladder. The
optimized beam directions are better aligned to the shape
of the tumor at the same time as they reduce unnecessary
irradiation of organs at risk as shown in Fig. 12. This ability
to radiobiologically optimize the directions of incidence is
very useful clinically as it performs a fine adjustment of the
angle of incidence to a degree, which is hard to achieve by
ocular inspection even by a trained clinician. The treatment
plan in Fig. 12 was developed using one of the most
advanced and flexible radiobiological optimization al-
Acta Oncol Downloaded from informahealthcare.com by 177.155.212.60 on 05/20/14

gorithms that is available today, called ORBIT (Optimiza-

tion of Radiation therapy Beams by Iterative Techniques
(25, 55)) here shown fully integrated in the Pinnacle treat-
Fig. 13. Schematic illustration of how different tumor (T), nodal
ment planning system. It allows optimum combination of (N) and metastatic (M) stages (Nnods and Nmets are the numbers of
different radiation modalities or beam energies as in Figs. positive nodes and discrete metastasis respectively), are most cost
8 and 9, but also optimization of the fractionation schedule effectively treated by standard radiation therapy and by the more
considering low dose hypersensitivity such that the optimal advanced new techniques including biological optimization of the
complication free cure. Obviously cases with discrete metastatic
dose level per treatment fraction is achieved in the late
spread may not be curative with radiation alone.
For personal use only.

responding organs at risk as discussed above. The ultimate

application of this type of algorithm would be to combine
prove the treatment outcome for many target sites. This
different types of low and high LET modalities such as
does not mean that we need to implement the most
protons, light and heavy ions, to really cure advanced
advanced form of intensity-modulated radiation therapy on
hypoxic tumors by optimized 3D dose delivery using the
all treatment units in every clinic. Most small tumors can
latest form of multi-particle synchrotrons (41).
be treated very effectively by standard treatment techniques
It is obvious that all patients do not really need the full
even though biologically optimization still may reduce late
power of efficient intensity-modulation. For example, many
morbidity in some of these tumors. However, when the new
of the early-diagnosed small T1–T2 tumors are already
treated very well with ordinary uniform beams of rectangu- optimized treatment techniques have been fully integrated
lar cross-section. As the tumor (T), nodal (N) and in modern equipment with full computer control and
metastatic (M) stages gradually deteriorate the need for biologically based treatment planning it will not be more
individual beam shaping and intensity-modulation increases complex to make advanced optimized intensity-modulated
rapidly. This is illustrated in Fig. 13 indicating the volumes treatment plans for all patients and deliver them using
in TNM-space where the different treatment modalities are computer controlled intensity-modulation. In fact, it will be
most cost effective. Depending on the tumor site different much simpler since most of the new methods are ideally
percentages of the patients will require more advanced suited for full automation using fully computer controlled
intensity-modulated treatments. However, even if uniform equipment in the clinic. In the transition period we now are
beams are sufficient for a curative outcome for some smaller living in costs will temporarily go up, since we have to
tumors, intensity-modulation may still be of interest, espe- improvise until most existing treatment units and planning
cially if the patient is young and has a considerable life systems are fully developed and well integrated. This obsta-
expectancy after a successful treatment. It is then important cle should be gradually resolved, hopefully some 5 years
to ensure the lowest possible risk for late adverse reactions from now. After that time it should be much easier to get
making advanced intensity-modulation the most attractive fully integrated equipment and equipment costs may also
method to maximize the long-term complication free cure. start to reduce slowly as the technology evolves.
In a 5- to 10-year perspective we will not only have well
functioning and integrated treatment units and planning
CONCLUSIONS AND FUTURE DEVELOPMENTS systems but also many rather accurate genetically based
It is absolutely clear that many of the new developments in assay techniques to better predict optimal treatment strate-
radiation therapy have the potential to considerably im- gies for the individual patient. In fact, the new biological
 594 A. Brahme Acta Oncologica 39 (2000)

optimization methods will become a central tool for the arguments will increase the use of advanced treatment
development of improved dose response models and pre- techniques and may be our best way to improve the
dictive assays. This is because the new biological models efficacy of radiation therapy. There is therefore no doubt
and optimization technique are the ideal tools for radia- that the new powerful tools of intensity-modulated radio-
tion biologists, radiation oncologists and radiation physi- therapy will be rapidly introduced in the first decade of the
cists, to work together for the development of our new millennium, not least for cost effectiveness reasons but
profession. Not only will they help optimizing the outcome also to speed up the systematic feedback of radiobiological
of a given treatment, they will also indicate those organs data from each patient treated to the benefit of future
and sites where the treatment is most likely to produce patients. Efficient intensity-modulated treatment units, ra-
adverse reactions to simplify follow-up and improve feed- diobiological treatment optimization, fast statistical analy-
back of response data to the models. In fact it is likely that sis and a continued search for molecular markers of
all the new optimization methods that are now becoming radiation sensitivity and tumor characterization are the
available may even alter the way randomized clinical trials key to the future.
are performed. The improved knowledge about the patient
and the substantially improved therapeutic modalities and
ACKNOWLEDGEMENTS
methods of follow-up will make the treatment of every
patient with advanced disease a radiobiological optimiza- The invaluable support of the Research Center for Radiation
Acta Oncol Downloaded from informahealthcare.com by 177.155.212.60 on 05/20/14

tion challenge. The treatment outcome will then be com- Therapy by NUTEK (The Swedish National Board for Industrial
and Technical Development) as well as continued stimulation by
pared with the accumulated historical experience which is
all its members and the researchers at the Department of Medical
contained in the predictions of the biological models, and Radiation Physics are gratefully acknowledged.
fewer and fewer patients are likely to be randomized to
sub-optimal treatment arms (56). The large historical data
sets (8–13) built into the biological models will ensure a REFERENCES
minimal contribution from patient individual variations, 1. Schulz M. The supervoltage story. Am J Roentgenol 1974;
which can not be avoided in classical randomized trials. 124: 541 – 59.
2. Allt WEC. Supervoltage radiation treatment in advanced
For personal use only.

Furthermore, model-based techniques will not only result

in optimized treatment plans for each patient, but also cancer of the uterine cervix. A preliminary report. Can Med
Assn J. 1969; 100: 792 – 7.
ensure the most efficient feedback for the development of 3. Bush RS. Management in carcinoma of cervix, uterus and
the radiobiological models and to the fastest possible ovary. London: Edward Arnold, 1979.
benefit of future patients. Recently a powerful Bayesian 4. Russell KJ, Caplan RJ, Laramore GE, et al. Photon versus
algorithm has been developed to rapidly feedback data to fast neutron external beam radiotherapy in the treatment of
the radiobiological models from all patients being treated locally advanced prostate cancer: results of a randomized
prospective trial. Int J Radiat Oncol Biol Phys 1993; 28:
at a clinic or group of clinics (56). This algorithm ensures 47 – 54.
a much faster and continuous feedback of clinical results 5. Holthusen H. Erfahrunger uber die Vertraglichkeitsgrenze fur
to the benefit of currently treated patients than waiting for Röntgenstrahlen und deren Nutzanwendung zur Verhutung
5–10 years to make a retrospective analysis. From this von Schade. Strahlenther 1936; 57: 254.
point of view it is important to have as large and uniform 6. Schulz RJ. Further improvements in dose distributions are
unlikely to affect cure rates. Med Phys 1999; 26: 1007 – 9.
patient recruitment population as possible to further speed 7. Cho KH, Khan FM, Levitt SH. Cost-benefit analysis of 3D
up data acquisition and improve the accuracy of follow-up conformal radiation therapy. Acta Oncol 1999; 38: 603 – 11.
in the interest of future patients. 8. A, gren-Cronqvist A, Källman P, Turesson I, et al. Volume
It is becoming increasingly clear that we have essentially and heterogeneity dependence of the dose-response relation-
one real chance to cure a patient, namely at the time of the ship for head and neck tumours. Acta Oncol 1995; 34:
851 – 60.
first treatment. Many tumors, where the initial treatment 9. Rubin P. A direction for clinical radiation pathology. The
was not radical enough, seem when they reoccur locally tolerance dose. In: Vaeth JM, ed. Front Radiation Therapy &
several years after the treatment, to have progressed in Oncology, vol. 6. Basel: S Karger, 1972: 1 – 16.
severity and often metastasized. Some theories explain this 10. Emami B, Lyman J, Brown A, et al. Tolerance of normal
to be owing to radiation induced genetic changes during tissue to therapeutic irradiation. Int J Radiat Oncol Biol Phys
1991; 21: 109 – 22.
the treatment (57) even though the probability for such 11. A, gren Cronqvist A-K. Quantification of the response of
positive events, from the tumors point of view, should be heterogeneous tumors and organized normal tissues to frac-
rather small. Independent of the real reason it is therefore tionated radiotherapy. Thesis, Stockholm University, ISBN
important, not least from a cost effectiveness point of 91-7153-371-0, 1995.
view, to be really radical at the first treatment and make 12. Brahme A. Optimized radiation therapy based on radiobio-
logical objectives. Semin Oncol 1999; 9: 35 – 47.
use of the tolerance of the patient and his own judgement 13. Brahme A. Biologically based treatment planning. In: Mus-
of acceptable side effects to really maximize the complica- takallio Centennial Symposium Helsinki. Acta Oncol 1999; 13
tion free cure and to some extent the quality of life. Such (Suppl): 61 – 8.
 Acta Oncologica 39 (2000) De6elopment of radiation therapy optimization 595

14. Brahme A. Dosimetric precision requirements in radiation 37. Withers HR, Taylor JMG, Maciejewski B. The hazard of
therapy. Acta Radiol Oncol 1984; 23: 379–91. accelerated tumor clonogen repopulation during radiotherapy.
15. Brahme A, A, gren A. On the optimal dose distribution for Acta Oncol 1988; 27: 131 – 46.
eradication of heterogeneous tumors. Acta Oncol 1987; 26: 38. Brahme A. Design principles, beam properties and clinical
377 – 85. possibilities of a new generation of radiation therapy equip-
16. Källman P, A, gren A, Brahme A. Tumor and normal tissue ment. Acta Oncol 1987; 26: 403 – 12.
responses to fractionated non uniform dose delivery. Int J Rad 39. Brahme A. Device for isocentric radiation therapy with heavy
Biol 1992; 62: 249–62. ions. Sw. Pat. 1985; 8501675-6: 1985.
17. Lyman JT. Complication probability as assessed from dose-vol- 40. Montelius A, Blomquist E, Naeser P, et al. The narrow proton
ume histograms. Radiat Res 1985; 104: S13–9. beam therapy unit at the The Svedberg Laboratory in Uppsala.
18. Wolbarst AB. Optimization of radiation therapy II: the critical- Acta Oncol 1991; 30: 739 – 45.
voxel model. Int J Radiat Oncol Biol Phys 1984; 10: 741 – 5. 41. Renner TR, Chu WT, Ludewigt BA. Advantages of beam
19. A, gren A, Brahme A, Turesson I. Optimization of uncompli- scanning and requirements of hadrontherapy facilities. In:
Amaldi U, Larsson B, eds. Hadrontherapy in Oncology,
cated control for head and neck tumors. Int J Rad Onc Biol
Excerpta Medica ICS 1077. Amsterdam: Elsevier, 1994: 453–
Phys 1990; 19: 1077–85.
61.
20. Aaltonen P, Brahme A, Lax I, et al. Specification of dose
42. Brahme A. Scanning system for charged and neutral particles.
delivery in radiation therapy. Recommendations by the NACP.
Sw. Pat. 7904360-0, 1979.
Acta Oncol 1997; 10 (Suppl): 1–32.
43. CROS (the Committee for Radiation Oncology Studies) Pro-
21. Timme TL, Moses RE. Review: Diseases with DNA damage
posal for a program in particle-beam radiation therapy in the
processing defects. Am J Med Sci 1988; 295: 40–8. United States. Cancer Clin Trials 1978; 1: 153 – 208.
Acta Oncol Downloaded from informahealthcare.com by 177.155.212.60 on 05/20/14

22. Löf J, Lind BK, Brahme A. Optimal radiation beam profiles 44. Tilikidis A, Brahme A. Microdosimetric description of beam
considering the stochastic process of patient positioning in quality and biological effectiveness in radiation therapy. Acta
fractionated radiation therapy. Inverse Problems 1995; 11: Oncol 1994; 33: 457 – 69.
1189 – 209. 45. Maor MH, Hussey DH, Fletcher GH, et al. Fast neutron
23. Löf J, Lind BK, Brahme A. An adaptive control algorithm for therapy for locally advanced head and neck tumors. Int J Rad
optimization of intensity-modulated radiotherapy considering Oncol Biol Phys 1981; 7: 155.
uncertainties in beam profiles, patient set-up, and internal 46. Brahme A. Physical and biologic aspects on the optimum choice
organ motion. Phys Med Biol 1998; 43: 1605–28. of radiation modality. Acta Radiol Oncol 1982; 21: 469 –79.
24. Gustafsson A, Lind BK, Brahme A, et al. A generalised pencil 47. Guichard M, Guelette J, Lublin G, et al. The comparative
beam algorithm for optimization of radiation therapy. Med response of human fibroblasts EMT6 and V79 cells to 50 MeV
For personal use only.

Phys 1994; 21: 343–56. neutrons. Int J Radiat Oncol Biol Phys 1978; 4: 621 – 7.
25. Löf J, Lind BK, Liander A, et al. Simultaneous beam orienta- 48. A, sell M, Hyödynmaa S, Söderström S, et al. Optimal electron
tion and intensity-modulation optimization using the new and combined electron and photon therapy in the phase space
universal radiotherapy optimization code ORBIT. Int J Radiat of complication free cure. Phys Med Biol 1999; 44: 235 –52.
Oncol, 2000; in press. 49. Korevaar EW, Heijmen BJM, Woudstra E, et al. Mixing
26. Turesson I, Brahme A. Clinical rationale for high precision intensity-modulated electron and photon beams: combining a
radiotherapy. Malmö: ESTRO, 1992. steep dose fall-off at depth with sharp and depth-independent
27. Wright KA, Proimos BS, Trump JG. Physical aspects of two penumbras and flat beam profiles. Phys Med Biol 1999; 44:
million volt X-ray therapy. Surg Clin North Am 1959; 39: 2171 – 81.
567 – 78. 50. Söderström S, Brahme A. Optimization of the dose delivery in
28. Takahashi S. Conformation radiotherapy, rotation techniques few field techniques using radiobiological objective functions.
as applied to radiography and radiotherapy. Acta Radiol Suppl Med Phys 1993; 20: 1201 – 10.
1965; 242. 51. Söderström S, Eklöf A, Brahme A. Aspects on the optimal
29. Hope CS, Orr HS. Computer optimization of 4 MeV treatment photon beam energy for radiation therapy. Acta Oncol 1999;
planning. Phys Med Biol 1965; 10: 365–73. 38: 179 – 87.
52. Brahme A. Optimization of radiation therapy and the develop-
30. Sterling TD, Perry H, Weinkam JJ. Automation of radiation
ment of multileaf collimation. Int J Rad Oncol Biol Phys 1993;
treatment planning. V. Calculation and visualisation of the
25: 373 – 5.
total treatment volume. Br J Radiol 1965; 38: 906–13.
53. Lind B, Brahme A. Development of treatment techniques for
31. Brahme A. Treatment optimization using physical and biolog-
radiotherapy optimization. Int J Imag Syst Technol 1995; 6:
ical objective functions. In: Smith A, ed. Radiation Therapy
33 – 42.
Physics. Berlin: Springer, 1995: 209–46.
54. Svensson R. Development of a compact high-energy treatment
32. Svensson R, Lind BK, Brahme A. Beam characteristics and unit combining narrow pencil beam scanning and multileaf
possibilities of a new compact treatment unit design combining collimation. Thesis, Stockholm University, ISBN 91-7153-740-
narrow pencil beam scanning and segmental multileaf collima- 6, 1998.
tion. Med Phys 1998; 25: 2358–69. 55. Löf J, Liander A, Kåver G, et al. ORBIT — A general object
33. Withers HR, Taylor JMG, Maciejewski B. Treatment volume oriented code for radiotherapy optimization. Radiother Oncol
and tissue tolerance. Int J Radiat Oncol Biol Phys 1988; 14: 1998; 48 (Suppl 1): S69.
751 – 9. 56. Maehle-Schmidt M, Palmgren J, Lind B, et al. A Bayesian
34. Zimmer KG. Studies on quantitative radiation biology. Lon- sequential model for updating radiobiological parameters in
don: Oliver & Boyd, 1961. radiation therapy. In: Second European Conf on Highly
35. Lambin P, Malaise EP, Joiner MC. The effect of very low Structured Stochastic Systems. Pavia, Book of Abstracts 1999;
radiation doses on the human bladder carcinoma cell line 180 – 1.
RT112. Radiother Oncol 1994; 32: 63–72. 57. Ling CC, Burman C, Chui CS, et al. Perspectives of multidi-
36. Turesson I, Joiner MC. Clinical evidence of hypersensitivity to mensional conformal radiation treatment. Radiother Oncol
low doses in radiotherapy. Radiother Oncol 1996; 40: 1 – 3. 1993; 29: 129 – 39.

View publication stats