The integral biologically effective dose to predict brain stem toxicity of hypofractionated stereotactic radiotherapy

Int. J. Radiation

Oncology

Biol.

Phys., Vol. 40, No. 3, pp. 667-675. 1998 Copyright 0 1998 Elsevier Science Inc. Printed in the USA. All rights reserved 0360-3016/98 $19.00 + .OO

PI1 SO360-3016(97)00734-7

l

Clinical

Investigation THE

INTEGRAL BIOLOGICALLY BRAIN STEM TOXICITY STEREOTACTIC

EFFECTIVE DOSE TO PREDICT OF HYPOFRACTIONATED RADIOTHERAPY

BRENDA G. CLARK, PH.D.,* LUIS SOUHAMI, M.D.,* CONRADO FLA, PH.D.,* ABDULLAH S. AL-AMRO, M.D.,* JEAN-PAUL BAHARY, M.D.,* JEAN-GUY VILLEMURE, M.D.,+ JEAN-LOUIS CARON, M.D.,+ AND& OLIVIER, M.D.+ AND ERVIN B. PODGORSAK, PH.D.* Departmentsof *Oncology, Division of RadiationOncology, and +Neurosurgery, McGill University, Montrkal, Canada Objective: The aim of this work was to develop a parameter for use during fractionated stereotactic radiotherapy treatment planning to aid in the determination of the appropriate treatment volume and fractionation regimen that will minimize risk of late damage to normal tissue. Materials & Methods: We have used the linear quadratic model to assess the biologically effective dose at the periphery of stereotactic radiotherapy treatment volumes that impinge on the brain stem. This paper reports a retrospective study of 77 patients with malignant and benign intracranial lesions, treated between 1987 and 1995, with the dynamic rotation technique in 6 fractions over a period of 2 weeks, to a total dose of 42 Gy prescribed at the 90% isodose surface. From differential dose-volume histograms, we evaluated biologically effective dose-volume histograms and obtained an integral biologically-effective dose (IBED) in each case. Results: Of the 77 patients in the study, 36 had target volumes positioned so that the brain stem received more than 1% of the prescribed dose, and 4 of these, all treated for meningioma, developed serious late damage involving the brain stem. Other than type of lesion, the only significant variable was the volume of brain stem exposed. An analysis of the IBEDs received by these 36 patients shows evidence of a threshold value for late damage to the brain stem consistent with similar thresholds that have been determined for external beam radiotherapy. Conclusion: We have introduced a new parameter, the IBED, that may be used to represent the fractional effective dose to structures such as the brain stem that are partially irradiated with stereotactic dose distributions. The IBED is easily calculated prior to treatment and may be used to determine appropriate treatment volumes and fractionation regimens minimizing possible toxicity to normal tissue. 0 1998 Elsevier Science Inc. Fractionated

stereotactic

radiosurgery,

Late complications,

INTRODUCTION The rationale for the choice of fractionated stereotactic irradiation (stereotactic radiotherapy) over external beam radiotherapy to treat selected small cerebral tumors is to take advantage of highly-focused radiation dose distributions to deliver a tumoricidal dose to the target volume while sparing surrounding normal tissue. With relatively few centers reporting late sequellaefrom fractionated stereotactic treatments (12, 13, 17, 27), little human doseresponsedata is available for this type of treatment, and prediction of late damageis problematical. One of the many factors remaining unresolved is the threshold dose-volume relationship for radiation-induced damage to structures at risk. Although the steepdose fall-off at the edge of stereotactic radiotherapy treatment volumes provides a measureof protection for surrounding tissue, the corresponding dose inhomogeneity in this peripheral region requires accurate dose-volume analysis to assessclinical tolerance. In situaReprintrequeststo: BrendaG. Clark, Ph.D., FCCPM, Medical Physics,B. C. CancerAgency, 600West 10fhAvenue,Vancouver,

Linear

quadratic

model.

tions where the treatment volume is very close to or in contact with sensitive structures,relatively small portions of these structures may receive high doses,with the remaining volume being exposedto much lower or insignificant doses. This highly inhomogeneous irradiation of small volumes within structures could result in a different clinical effect than the relatively uniform irradiation obtained with standard external beam therapy. Although it has long been recognized that dose-volume effects could play a significant role in tissue tolerance, it is only recently that accurate methods of analysis using differential and cumulative dose-volume histograms, evaluated on a pixel-by-pixel basisfrom computed tomography (CT) images,have become practical. This work describesa method of dose-volume analysis derived from the linear quadratic model that can be used to assessbrain-stem tolerance in fractionated stereotactic radiotherapy regime for lesions impinging on the brain stem. We have used this BC V5Z 4E6 Canada.E-mail: [email protected] Acceptedfor publication22 August 1997.

668

I. J. Radiation Oncology l Biology l Physics

method to retrospectively analyze clinical results from 77 patients treated with fractionated stereotactic radiotherapy, and to develop a parameter for use during treatment planning to aid in the determination of the appropriate treatment volume and fractionation regimen that will minimize risk of late damageto normal tissue in the brain stem. MATERIALS

Volume 40, Number 3, 1998 Table 1. Patient characteristics

Age (years) Gender Diagnosis

AND METHODS

The linear quadratic model has been applied extensively to fractionated radiotherapy, as reviewed, for example, by Fowler (9) and Hall (lo), to provide a simple and convenient method of assessingfractionation schemesin terms of the difference in responsebetween early- and late-reacting tissues. The model describes an effective dose-response relationship for a multifraction regimen using an equation defining the biological effect, E, (for example, log cell kill) resulting from a number, n, of high dose-rate fractions of dosed each, asE = n(ad + /3d2),where the two coefficients CYand p are constant. The biologically-effective doseBED is the quantity by which the different fractionation regimens are intercompared, and is defined by the ratio Ela given by:

Tumor volume (cc)

Median Men Women Astrocytoma Meningioma Pineal tumor Low grade glioma Craniopharyngioma Pituitary adenoma Other Median

42 (range 8-82) 34 43 24 20 6 5 4 14

5 (rangel-26)

(di/(o/P))(AvJv)] in a similar mannerto the modification of the total dose, nd, by the factor [l + (d/(a/fl))] in the definition of BED (equation 1). The summationof theseincremental BED values over the structure gives an integral BED (ZBED) characterizing the effective dose to the structure and defined by the equation:

(1) This equation can also be expressedas BED = (total dose) X (relative effectiveness) because the term [ 1+(d/(alfi))] has the effect of modifying the total dose, nd. The ratio (Y/P has the dimensionsof dose and is characteristic of a particular tissue. Clinical and laboratory data suggestthat there is a consistent difference between earlyand late-responding tissuesin their responseto changing fractionation patterns that is reflected in differing values of the ratio (Y/P (7, 28). For early-reacting tissues,including many types of tumor, alp has beenmeasuredto be 10 Gy or higher, whereas for late-responding tissues, c~/p is low, between 1.5 Gy and approximately 5 Gy (8). We have applied this model to stereotactic radiotherapy using differential dose-volume histograms, as described below. Each delineated structure in the treatment volume was divided into i dose-bandsof width 1% where 100% is the value of the maximum dose delivered to the target volume and di is the dose delivered to the ith band. The volume within each dose-band, AU, was calculated and expressedas a fraction of the total volume, V, of the structure giving a partial volume for the ith doseband of Av,lV. An incremental BED, ABED, was calculated for each dose value, d;, resulting from n fractions using the equation:

From this equation, it can be seenthat this incremental BED representsthe modification of the total doseexperienced by the ith volume increment, given by nd, by a factor [l +

The patient population in this study consists of 77 patients with small, malignant, or benign intracranial lesions (Table 1) treated with hypofractionated stereotactic radiotherapy. The treatments were delivered between May 1987 and May 1995 with the stereotactic dynamic rotation technique (21), in a typical fractionation protocol of 6 fractions of 7 Gy each, given on alternate days, with all treatments being delivered over a period of 2 weeks (3, 22). The treatments were delivered using a single isocenter with dosesprescribedat 90% of the maximum target dose,except for 5 patients who were treated using two isocentersand 1 patient who was treated using three isocenters, with doses prescribed between 50% and 65% of the maximum target dosefor these5 patients. Deviations from this protocol were 6 X 6 Gy (10 patients), 10 X 4 Gy (2 patients), 3 X 6 Gy, 8 X 3.5 Gy, 10 X 2.5 Gy and 9 X 2.2 Gy (1 patient each). For accurate assessment of the dosedelivered to the brain stem, the dose-distributions for each patient were reconstructed retrospectively from the axial CT or magnetic resonance (MR) imaging studies used for the treatment planning. The analysis was carried out with new inhouse treatment-planning software (20) that was not available when the original treatment planning was performed. On each axial image, the tumor and healthy structures, such as the brain stem, optic chiasm, and optic nerve were delineated. For the brain stem, only the midbrain and the pons were included, the medulla extended below the target volume and also, generally, below the volume included in the target-localization image study and was, therefore, excluded. Differential dose-volume histograms were calculated for each patient and an ZBED evaluated individually, as described by Eq. 3.

Hypofractionated

stereotactic

radiotherapy

dosage

l

B. G. CLARK

et al.

669

RESULTS Of the 77 patients in the study, 36 had treatment volumes positioned so that some dose was delivered to the brain stem; 4 of these patients developed serious late neurological complications attributable to irradiation of the brain stem. These 4 patients (3 women, 1 man of ages 52, 58, 66, and 81 years) were all treated for meningioma, with lesions impinging on the midbrain for 2 patients and the pons for the other 2 patients. All treatments for these 4 patients were prescribed to the isodose volume defined by 90% of the maximum target dose with one isocenter, and delivered in 6 fractions of 7 Gy each. The median expression time for the damage was 9 months. Other than type of lesion, the only significant variable was the volume of brain stem exposed. Also, 3 patients developed damage to the optic pathway. The tumor volume, target volume, and volume of normal tissue within the target volume for each patient are shown in Fig. la, b, c. Figure 2 shows a comparison between planning axial CT images (left) and follow-up MR images (right) taken for 1 of the 4 patients who developed a neurological deficit. The CT images show the tumor, brain stem, and optic nerve delineated and the reconstructed treatment plan is superimposed on the image with isodose lines depicting 90%, 50%, 30%, and 10% of the maximum dose. The prescribed dose was delivered to the 90% isodose surface. The MR images shown on the right were taken 17 months post-treatment after the development of neurological deficit. These Tlweighted images show a right parasellar lesion with inhomogeneous enhancement. There are also areas of enhancement in the right aspect of the pons. T2-weighted images (not shown) demonstrated the pons lesion to be hyperintense with mild enhancement after gadolinium injection; changes that are consistent with radiation necrosis. This damage appears to follow the periphery of the treatment volume, as described by the isodose distribution. Figure 3a, b shows integral dose-volume histograms describing the distribution of dose to the brain stem for each of the 4 patients who developed late sequellae, together with those from 4 patients arbitrarily selected from the rest of the patient population for comparison. The fractional volume of the brain stem irradiated at the high end of the dose range for the first group of patients is demonstrably higher than that for the second group of patients. For each of the 4 patients who developed late sequellae, the volume irradiated to the prescribed dose (target volume) formed part of a sphere encroaching on the brain stem from the side, and the fractional volumes of brain stem within the target volume were 18%, 20%, 21%, and 35%. For comparison, the brainstem volume included in the target volume for the rest of the 36 patients was in the range 0 to 10% with a median of 2%. Thus,

the volume

of brain

stem included

in the treatment

target volume was a significant factor in the development of damage. Using an crlp ratio of 2.5 as representative of the doseresponseof brain-stem tissue, we calculated IBEDs for the

10

“0

20 Patient

30 40 Identification

50

60

40 ,

I *

n

(b)

Cl

-i

30

.

t

-1

i

>

2$10 zFB 20

0

0

0

00 0

10

0

0

00

0 Oo

20 Patient

0

A

30 40 Identification

0

.O

n0

0

50

1

60

Fig. 1. Scatter diagrams showing (a) the total tumor volume delineated on the CT slices, (b) the treated target volume (i.e., the volume defined by the prescription isodose surface), and (c) the absolute volume of normal tissue within the treated target volume.

patients from the population who had tumors positioned so that somedosagewas delivered to the brain stem(36 of 77). Of the patients, 32 had ZBEDs calculated at less than 50 GYM and suffered no late effects, whereas all patients whose ZBED was greater than 70 Gy,,, developed complications. The histogram in Fig. 4 shows the distribution of ZBED for these36 patients and demonstratesa clear distinction between the 4 patients who developed late complications and the remaining patient population. None of the patients in our study had an IBED to the brain stem evaluated between 50 and 70 Gy,.,. DISCUSSION An editorial by Ling et al. (14) discussingthe estimation of risks of radiation damage in late-effect normal tissues identifies the need for improved knowledge of the partial

670

I. .I. Radiation Oncology

l

Biology 0 Physics

Fig. 2. Comparison between axial CT images used for treatment planning with isodose distributions superimposed (left) and corresponding follow-up axial MR images (right) at 17 months posttreatment. The isodose lines on the CT images represent 90%, 50%, 30%, and 10% of the maximum dose.

volume effect of inhomogeneously irradiated organs. Several approaches have already been described, including dose-volume histogram reduction schemesthat use a power-law relationship to reduce an inhomogeneousdose-volume distribution to a homogeneousone. This hasbeen done in two ways: by determining an effective volume to which a normal tissue complication probability may be applied (11) and by applying an effective complication probability to each dose step to determine the single-step histogram volume required to match the complication probability of the original (15). The shortcomings of these two models, and also the one presentedhere, is that they reflect inadequate biological and radiobiological understanding and a paucity of experimental data (14). However, the advantage of our approach is that it does not rely on normal tissuecomplication probability data that may or may not be applicable to inhomogeneously partially irradiated structures. The only parameterused in our analysis is the ratio of o/p. Also, we suggestthat the model proposedhere is usedin the sameway asthe linear quadratic model is recommendedfor use (i.e., for comparison of dose-fractionation schedules rather than for absolute dose-responsedetermination).

Volume 40, Number 3, 1998

To date, analysis of damageafter stereotactic irradiation has centered around either isoeffect curves, plotting minimum dose representing tissue tolerance as a function of entire dose distribution (5), or correlation of damagewith treatment volume (6, 18). Although the volume of normal tissue irradiated is undoubtedly highly correlated with the development of late sequellae,our data, as shown in Fig. 1, reveals that, considered alone, healthy tissue volume irradiated may not be the only factor in the development of late damage.Of the 11 patients treated with a prescription isodose (target) volume greater than or equal to 20 cm3, 6 did not show any late damage(Fig. lb). Although most of our patients were treated with a target volume that included a small absolute volume of healthy brain tissue, 6 of the 11 patients whose target volume included 10 cm3 or more of healthy tissue did not develop complications (Fig. lc). Thus, an analysisbasedon isoeffect curves of dosesolely as a function of treated volume may not be sufficient to characterize the results from these patients. Because the four complications seen in our group of patients were directly attributable to damageto the brain stem, and correlated well with the volume of brain stem included in the treatment volume, it was appropriate to examine carefully the dosevolume relationship within the brain stem rather than integrate over the whole treated volume. Although the differences in the dose-volume histograms in Fig. 3 are qualitatively apparent, a quantitative estimate of the potential for damageis not immediately obvious from these distributions. How much dose could be tolerated and to what fraction of the volume are questionsthat currently remain unresolved for highly inhomogeneousirradiation. During treatment planning for lesions impinging on the brain stem, where it is currently very difficult to avoid irradiating the structure completely, the precise evaluation of dose effects to partial volumes is vital. Dose-effect relationships frequently allow for the use of only a single value for the tissuetolerance dose,and it is unclear whether the most appropriate value is the minimum, the mean,or the modal dose. It has been shown that consideration of some representative volume only, often the region of highest dose,and to assumethat it is only the tolerance of the tissues in this representative volume that is of importance, is frequently an unsatisfactory approximation, and that the use of someform of integration over the entire irradiated volume is neededto provide an appropriate solution (19). The value of an assessmentusing the biologically-effective dose is demonstrated in Fig. 5a, b, which shows a dose-volume histogram describing the incremental volume within each dose band, and the corresponding histogram describing the distribution of incremental BED. This comparison illustrates the quadratic nature of the relationship between dose and BED, demonstrating that a relatively small fractional volume irradiated at the high end of the dosescale(for example, the volume receiving between 70% and 100% of the prescription dose) will contribute substantially to the effective dose to the structure at risk. The use of the linear quadratic model to characterize the

Hypofractionated stereotactic radiotherapy dosage l B.

G. CLARK

671

et al.

, 0.47cq

0.28~~1

4.42~~

1.75cc

.I

5011

601

5oa

OS

i

0.63~~

1.92cc

501

85s

501

1001

W Fig. 3. Differential dose-volume histograms in 1% dose steps for (a) the 4 patients who developed complications, and (b) for 4 other arbitrarily chosen patients. In each histogram, the vertical axis is absolute volume and the horizontal axis is dose in percentage of the maximum dose.

dosedistributions asdescribedhere enablesus to investigate fractionation options and compare our results with other published data. Although IBED data between 50 Gy,., and 70 Gy,., is not available from this study, there is an indication of a possible threshold ZBED for late damageto the brain stemwithin this rangefor irradiation schedulessimilar to the one used for our group of patients. Earlier work,

analyzing the follow-up results from patients treated with external beam irradiation using the nominal standard dose (NSD) model, has establishedthat the incidence of necrosis in the brain is rarely seen after a total dose biologically equivalent to 52-54 Gy delivered in 1.7-2.0 Gy fractions over a period of about 42 days, but the risk of necrosis increases rapidly above these levels (16, 26). Using the

672

I. J. Radiation Oncology l Biology l Physics 20.

5

Oo-~-

20

40

60

Integral

-so-

100

120

BED (GY~,~>

Fig. 4. The distribution of calculated values of integral biologically-effective dose to the brain stem for the 36 patients whose tumor was positioned so that part of the brain stem received some of the treatment dose.

linear quadratic model, these values correspond to a threshold BED for homogeneousirradiation in the range 84-93 Gy,,,, which correlates well with the value obtained in the current work for inhomogeneousirradiation. With an incidence of 100% (1 patient) in the range 75- 80 Gy,.,, further conclusions from our study would be difficult to justify. Direct comparison of our results with these figures would require the addition of a time factor to the model, to allow for the difference in the elapsedtime for treatment delivery between our regimen of 12 days and the standardextemalbeam therapy regimen. This adds several stagesof complexity and of potential error (8), with the introduction of variables allowing for an effective cell doubling time, proliferation time, and a value of (Y. However, this procedure would result in a reduction of the values associatedwith the longer time intervals relative to those of the shorter intervals, bringing the threshold values even closer. Stereotactic radiotherapy has been carried out with fractionation schemesranging from “conventional” daily fractions of 1.8-2.0 Gylday over 5 weeks (4, 13,25) to various accelerated schemesof 2-10 fractions delivered in l-2 weeks (1, 3, 12, 23, 24). It was noted by Brenner and Hall (2) that there is no justification for extending the overall treatment time to reduce early normal-tissue sequellaebecausethe focused dosedistributions, intrinsic to stereotactic radiotherapy treatment, provide sufficient sparing of tissues, such as skin and mucosa, that are most often the limiting factor in radiotherapy. They give a rationale for short overall treatment times, l-2 weeks, using a large number of fractions, 5-20, delivered within that time period to limit tumor repopulation. For tumors where repopulation is not an issue, for example, meningioma, acoustic neuroma, or craniopharyngioma, extending the treatment to 5 weeks to reduce late effects may be preferable, depending on the particular

repeat-fixation

device

used. However,

there are

Volume 40, Number 3, 1998

frequently circumstanceswhere a shorter treatment period is preferred, and could be safely used. The impact of a reduction in dose per fraction with a corresponding increase in number of fractions while still restricting the overall length of treatment to 12 days is illustrated in Fig. 6. Our standard schedulefor tumors not compromising sensitive structures was 6 fractions of 7 Gy, each delivered over a period of 12 days to a total doseof 42 Gy. For the 4 patients who developed complications, a simulation was carried out by calculating the IBED for an increasing number of fractions to a total of 20. This is the maximum number of fractions that could be delivered on a twice-daily schedulein a period of 12 days (2 weeks with a rest at the weekend). The doseper fraction for each schedule was determined according to Eq. 1 to keep the effective dose prescribed to the tumor constant. For example, if the number of fractions had been increased to 10, the IBED would have been reduced on average by 11% (Fig. 6). This simulation indicates that, for patients at risk of brain-stem damage, it is possible to substantially reduce the effective doseto the brain stem by increasingthe number of fractions

10

20

40

60

80

100

Dose (% of maximum) 6, Q @

I

5

W

0

20

40

I

60

80

1

100

BED (% of maximum) Fig. 5. An example of (a) a brain stem dose-volume histogram and (b) the corresponding incremental biologically-effective dosehistogram determined by calculating the BED for each volume increment at each dose value.

Hypofractionated stereotactic radiotherapy dosage

d jg u % z? 3

B. G. CLARK er al.

Tumor

-5 c .r(

l

-10 -15 -20 -25 -30 L

6

I

8

I

10

I

I

/

12 14 16 Number of Fractions

1

I

18

5

20

Fig. 6. The calculated reduction in integral biologically-effective dose with increasing number of fractions averaged over the 4 patients who developed late complications.

with a corresponding reduction in dose per fraction, the length of treatment time, and the effective doseto the tumor remaining the same. A further example of the use of this parameter is the determination of the optimal plan for a volume impinging directly on the brain stem.Figure 7 showsone axial CT slice

Fig. 7. Axial CT image of a patient with a tumor situated between the brain stem and the optic chiasm, showing delineated structures and 90%, SO%, 30%. and 10% isodose lines.

0 -10 -20 -30 2

3

-40

Displacement (mm)

Fig. 8. Percentage change in integral biologically-effective dose to the tumor and the brain stem for the plan shown in Fig. 7, with displacements of the isocenter posteriorly (positive direction) and anteriorly (negative direction).

of the dose distribution calculated for a patient with a suprasellarcraniopharyngioma. On this slice, the tumor is located between the brain stem and the optic chiasm, presenting a considerable treatment-planning challenge. The prescription isodose is the innermost (black) 90% isodose line. A determination of the IBED for this plan using our standardfractionation schemegave a value of 33 Gy,,,, well below the threshold indicated by the values in the histogram of Fig. 3. This could be reduced further by increasing the number of fractions, as demonstrated above. In view of the proximity of this tumor to the brain stem, an analysis was carried out to assessthe effect of uncertainties in isocenter positioning. The calculations were repeated with the isocenter shifted anteriorly and posteriorly by several mm and the results are plotted in Fig. 8. Our quality-assurance program has been developed to ensure an accuracy of isocenter positioning of less than 2 mm. These calculations confirm that, even with an uncertainty on the order of 3 mm, the brain stem would receive considerably less than the threshold IBED of approximately 60 Gy2.s. With the advantage of hindsight, it is evident that the planning performed for the patient shown in Fig. 1 was not optimal, and this example stressesthe vital importance of the treatment-planning systemsusedfor stereotactic irradiation. Tumors, such as this one, that are irregularly shaped throughout their volume require a sophisticatedtreatmentplanning systemto determine treatment parametersthat will minimize irradiation of healthy tissueand, consequently, the risk of late effects. Specialized planning systemshave been developed during the last 10 years that offer features for dose optimization that were not accessibleat the time that the majority of the patients in this study were treated. Our present system (20) includes: image zoom, delineation of tumor and normal structures on each axial slice, automatic isocenter determination according to tumor volume, color presentation of the isodose lines and delineated structures, fast flexible dose calculations for multiple isocenter treatments, cumulative dose-volume histograms for tumor and normal tissues,and differential dose-volume histogramsfor

I. J. Radiation Oncology

614

l

Biology

l

Physics

tumor and normal tissues, and differential dose-volume histograms for the tumor and all delineated normal structures. If these features had been available at the time this treatment was planned, the shape of the dose distribution delivered to this patient would have been more conformal to the tumor, with the result that the late damage may have been prevented. CONCLUSIONS We have described the determination of a single number parameter, an integral biologically-effective dose (IBED), that may be used to represent effective dosageto structures, such as the brain stem, that are irradiated with inhomogeneous dose distributions having steep dose gradients. Our clinical results have demonstrated a threshold value in the region of 60 Gy,,, for effective dosageto the brain stem. In the absence of other clinical indicators, we

Volume 40, Number 3, 1998

now use this value in the planning of treatments where some irradiation of the brain stem is unavoidable. Most stereotactic treatment-planning algorithms now calculate dose-volume histograms as a routine for treatments of this nature and could easily be programmed to include a calculation of ZBED for delineated structures, giving an immediate quantitative indication of the possible toxicity of the plan. We have demonstrated that an assessmentof IBED during treatment planning can be used to determine the appropriate treatment volume and fractionation regimen that will minimize toxicity to surrounding structures, such as the brain stem. A note of caution, however. There is no indication that an analysisof this kind could be usedfor linear structures,such asthe optic pathway. Our experience showsthat a high dose delivered to a small section of the structure will result in seriousdamage,which doesnot seemto be the casefor the brain stem.

REFERENCES 1. Brada, M.; Laing, R. W.; Graham, J.; Wanington,

A. P.; Hines, F. Fractionated stereotactic radiotherapy in the treatment of recurrent high grade glioma-dose escalation study. Acta Neurochir. 122: 151; 1993. 2. Brenner, D. J.; Hall, E. J. Stereotactic radiotherapy of intracranial tumors-An ideal candidate for accelerated treatment. Int. J. Radiat. Oncol. Biol. Phys. 28:1039-1041; 1994. 3. Clark, B. G.; Podgorsak, E. B.; Souhami, L.; Olivier, A.; Sixel, K. E.; Caron, J-L. A halo-ring technique for fractionated stereotactic radiotherapy. Brit. J. Radiol. 66522-527; 1993. 4. Dunbar, S. F.; Tarbell, N.; Kooy, H. M.; Black, P. M.; Alexander, E. A.: Loeffler, J. S. Early experience with stereotactic radiation therapy in the management of intracranial lesions: The first 1000 treatments.Acta Neurochir. 122:170-171;

1993. 5. Flickinger, J. C.; Schell, M. C.; Larson, D. A. Estimation of complications for linear accelerator radiosurgery with the integrated logistic formula. Int. J. Radiat. Oncol. Biol. Phys. 19:143-148; 1990. 6. Flickinger, J. C.; Lunsford, L. D.; Kondziolka, D.; Maitz, A. H.; Epstein, A. H.; Simons, S. R.; Wu. A. Radiosurgery and brain tolerance: An analysis of neurodiagnostic imaging changes following gamma knife radiosurgery for arteriovenous malformations. Int. J. Radiat. Oncol. Biol. Phys. 23: 19-26; 1992. 7. Fowler, J. F. The second Klaas Breur memorial lecture. La Ronde-radiation sciences and medical radiology. Radiother. Oncol. l:l-22; 1983. 8. Fowler, J. F. Brief summary of radiobiological principles in fractionated radiotherapy. Semin. Radiat. Oncol. 2: 16-21; 1992. 9. Fowler, J. F. Intercomparisons of new and old schedules in fractionated radiotherapy. Semin. Radiat. Oncol. 2:67-72; 1992. 10. Hall, E. The Janeway Lecture 1992. Nine decades of radiobiology: is radiation therapy any the better for it? Cancer 71: 3753-3766; 1993. 11. Kutcher, G. J.; Burman, C. Calculation of complication probability factors for non-uniform normal tissue irradiation: The effective volume method. Int. J. Radiat. Oncol. Biol. Phys. 16:1623-1630; 1989. 12. Laing, R. W.; Warrington, A. P.; Graham, J.; Britton, J.; Hines, F.; Brada, M. Efficacy and toxicity of fractionated

13.

14.

15. 16. 17.

18.

19. 20.

21. 22.

23.

stereotactic radiotherapy in the treatment of recurrent gliomas (phase I/II study). Radiother. Oncol. 27:22-29; 1993. Landy, H. J.; Schwade, J. G.: Houdek, P. V.; Markoe, A. M.; Feun, L. Long-term follow-up of gliomas treated with fractionated stereotactic irradiation. (Abstr.) Acta Neurochir. 62(Suppl):67-71; 1994. Ling, C. C.; Kutcher, G. J.; Mohan, R. Understanding radiation damage in late effect normal tissues: Learning to negotiate the dose-volume-complication terrain or waiting for Godot? Int. J. Radiat. Oncol. Biol. Phys. 31:1357-1360; 1995. Lyman, J. T.; Wolbarst, A. B. Optimization of radiation therapy, IV: A dose-volume histogram reduction algorithm. Int. J. Radiat. Oncol. Biol. Phys. 17:433-436; 1989. Marks, J. E., Wong, J. The risk of cerebral radionecrosis in relation to dose, time and fractionation: a follow-up study. Prog. Exp. Tumor Res. 29:210-218; 1985. McKenzie, M. R.; Souhami L.; Caron J-L.; Olivier A.; Villemure J-G.; Podgorsak E. B. Early and late complications following dynamic stereotactic radiosurgery and fractionated stereotactic radiotherapy. Can. J. Neurol. Sci. 20:279-285; 1993. Nedzi, L. A.; Kooy, L. I.; Alexander, E.; Gelman, R. S.; Loeffler, J. S. Variables associated with the development of complications from radiosurgery of intracranial tumors. Int. J. Radiat. Oncol. Biol. Phys. 21:591-599; 1991. Orton, C. G. A unifiedapproachto dose-effectrelationships in radiotherapy. II: Inhomogeneous dose distributions. Int. J. Radiat. Oncol. Biol. Phys. 14:557-560; 1988. Pla, C. A Macintosh based treatment planning system for brachytherapy and stereotactic radiosurgery. Proceedings of the XI International Conference on Use of Computers in Radiotherapy, Manchester, UK, 20-24 March 1994, pp. 138139, 1994. Podgorsak, E. B.; Olivier, A.; Pla, M.; Lefebvre, P-L.; Hazel, J. Dynamic stereotactic radiosurgery. Int. J. Radiat. Oncol. Biol. Phys. 14:115-125; 1988. Podgorsak, E. B.; Souhami L.; Caron J-L.; Pla, M.; Clark, B. G.; Pla, C.: Cadman, P. A technique for fractionated stereotactic radiotherapy in the treatment of intracranial lesions. Int. J. Radiat. Oncol. Biol. Phys. 27:1225-1230; 1993. Pozza, F.; Colombo, F.; Chierego, G.; Avanzo, R. C.; Marchetti, C.; Benedetti, A.; Casentini, L.; Danieli, D. Low grade

Hypofractionated stereotactic radiotherapy dosage l B. G. CLARK et al. astrocytomas: Treatment with unconventional fractionated stereotactic radiation therapy. Radiology 1715655569; 1989. 24. Rieke, J. W.; Jones, D.; Hafermann, M. D. Fractionated stereotactic radiosurgery of brain tumours: Results and toxicity in 41 patients. Acta Neurochir. 122:170; 1993. 25. Schwade, J. G.; Houdek, P. V.; Landy, H. J.; Bujnoski, J. L.; Lewin, A. A.; Abitol, A. A.; Serago, C. F.; Pisciotta, V. J. Small-field stereotactic external-beam radiation therapy of intracranial lesions: Fractionated treatment with a fixed-halo immobilization device. Radiology 176563-565; 1990.

615

26. Sheline, G. E.; Wara, W. W.; Smith, V. Therapeutic irradiation and brain injury. Int. J. Radiat. Oncol. Biol. Phys. 6: 12151228; 1980. 27. Souhami, L.; Olivier. A.; Podgorsak, E. B.; Villemure, J-G.: Pla, M.; Sadikot, A. F. Fractionated stereotactic radiation therapy for intracranial tumors. Cancer 68:2101-2108; 1991. 28. Withers, H. R., Thames, H. D., Peters, L. J. Differences in the fractionation response of acute and late responding tissues. In: Karcher K. H., Kogelmik, H. D., Reinartz, G.. eds. Progress in radio-oncology II. New York: Raven; 1982:287-296.