In, J Roduirnn Onc,,l,,y,~ Bioi PI?\, Vol Pnnted I” the US A. All nghts resened

IX. pp. 659-669

l Technical Innovations

DEVELOPMENT

0360.3016/90 $3.00 + .oo Copyright G 1990 Pergamon Press plc

and Notes

OF TOTAL-SKIN

R. S. Cox,

PH.D.,

of Radiation

Oncology,

Stanford

THERAPY

AT TWO ENERGIES

PH.D., P. FESSENDEN, PH.D.,

R. J. HECK,

C. J. KARZMARK, Department

ELECTRON

PH.D.

AND D.

University

C. RUST,

School of Medicine,

B.A. Stanford,

CA 94305, USA

Total-Skin Electron Therapy (TSET) modalities have been developed at two energies on a Varian Clinac 1800. The physical criteria for the beams were determined mainly from the requirement of continuing the Stanford treatment technique, which was 12 Total-Skin Electron Therapy portals combined in six pairs. The penetration of the lower energy mode matches that previously obtained at Stanford on the Varian Clinac 10, (about 4 mm for the 80% isodose contour in the 12-field treatment). The penetration of the higher energy mode is about 8 mm at the 80% contour. The Total-Skin Electron Therapy modes necessarily use electrons produced by the two standard electron-beam modes of lowest energy, nominally 6 and 9 MeV. Measurements to verify the beam specifications were carried out with diodes, a variety of ionization chambers, and a specially constructed circular phantom for film dosimetry. Initially, the penetration of the Total-Skin Electron Therapy beams was too large to match our criteria, so two methods of reducing it were explored: (a) the energies of the electron beams produced by the machine were reduced (which also reduced the energies of the corresponding standard electron modes) and (b) a large polymethylmethacrylate degrader (2.4 m X 1.2 m) 1 cm thick was placed just in front of the patient plane. Acceptable Total-Skin Electron Therapy beams could be produced by either method and the latter was finally used. The use of the standard dose monitoring system for the Total-Skin Electron Therapy modes considerably simplifies the daily treatment delivery as well as the implementation. However, the need for reasonable dose rates at the treatment plane (3.5 meters beyond the isocenter) requires dose rates of 24 Gy/min at the isocenter. Nevertheless, it is possible to use the internal dose monitor provided the problems associated with high dose rates (recombination and amplifier saturation) are addressed. Solutions to these problems involved switching the primary and back-up dose monitors, increasing the collecting voltage on the ion chambers, and calibrating the dose monitor so that 1 unit = I cGy at the patient rather than at the isocenter. Electron

therapy,

Total skin, Mycosis

fungoides.

In 1960, the original Stanford technique for TSET using four angled pairs was implemented on a 4.8 MeV linear accelerator that provided 2.5 MeV electrons at the patient (6). This technique was later modified to six angled pairs for better dose uniformity and a slightly more penetrating beam. nominally 4 MeV at the patient. In 1972 when the original treatment machine was replaced. TSET was redeveloped on a Varian* Clinac 10 (Cl-lo), which was used until 1986. The present Stanford technique for delivering TSET has been described fully in the literature (7). Briefly, it involves a combination of 12 electron portals in six pairs. The patient sequentially assumes six standing positions

INTRODUCTION

Over 30 years have passed since low-energy electron beams of large field size were first developed for the treatment of mycosis fungoides (9). Following the initial reports of clinical success. the ability to deliver Total-Skin Electron Therapy (TSET) has been independently developed on a variety of accelerators at many radiation therapy centers. The TSET technique and its dosimetry has recently been reviewed (2). Before this work was begun, no commercial manufacturer of medical accelerators offered TSET as a standard treatment option. and its development had to be carried out by the purchaser.

Presented at American Society for Therapeutic Radiology and Oncology, New Orleans, Louisiana, October 11, 1988. Reprint requests to: R. S. Cox. Ph.D. Ackno~~,led~eements-The authors wish to acknowledge the cooperation of Varian Associates for the engineering modifications which permitted the implementation of dual-mode TSET on the Cl- 1800. In particular, Drs. Robert Anderson and Stanley

Johnsen performed preliminary studies which were helpful in establishing the feasibility of our goals. Also, assistance with neutron measurements was provided by John A. Holmes of the Health Physics Department, Stanford University. Accepted for publication 19 July 1989. * Varian Assoc., Palo Alto. CA.

659

660

1. J. Radiation Oncology

0 Biology 0 Physics

in order to present the entire surface of the body to the beam at roughly 60” intervals about a vertical axis. In each position, the patient receives a pair of portals, one portal aimed toward the head and the other toward the feet (angled pair). This treatment method results in a dose distribution with acceptable uniformity over most of the patient’s skin. Recently, our clinicians expressed interest in having a second TSET modality significantly more penetrating than that originally developed. In response to this request, TSET modalities were developed at two different energies on a Varian Clinac 1800 (Cl- 1800), a modern medical linear accelerator, which has been installed in many radiotherapy facilities. Our goals from the technologists’ viewpoint were: a) simplicity of operation (the preparation of a TSET mode on the treatment machine should be similar to that of a standard electron mode), b) consistency of technique (apart from mode selection, the technique for treatment delivery should be identical for both modes). and c) adequate dose rates to keep treatment times to about 1 min. This paper describes the work done, presents the results obtained. and provides recommendations that may be of use to other centers intending to develop a similar TSET capability.

METHODS

AND

Electron beam characteristics

MATERIALS ftir

TSET

The physical considerations for TSET have been reviewed in detail by Holt and Perry (4). The specifications for the total treatment determine those for a single TSET beam and how single beams are to be combined to produce the total treatment. The electron beam emerges from the vacuum window ofthe bending magnet as a thin pencil about 3 mm in diameter. As it proceeds toward the patient the electrons are multiply scattered by the material in the head of the machine and the intervening air. Since the process is random, the current density, J(r), possesses circular symmetry about the beam axis and assumes. approximately, a Gaussian distribution in the transverse and radial dimension, r.

where I is the total electron current. The characteristic width of the Gaussian, w, is related to the half width at half maximum, h, by h=w\/lno

and varies slightly more rapidly than linearly with distance along the central axis, owing to scattering by the intervening air. The radial dose-rate distribution at the surface

March 1990. Volume 18, Number 3

of an absorber in any transverse proportional to J(r),

plane,

D(r), is roughly

where [S/p], is the mass collision stopping power and q is the charge on the electron. The first step in maximizing uniformity of dose over the skin is to obtain good uniformity over a region of the treatment plane roughly the size of a standing human ( 1.8 m vertically X 0.6 m horizontally). The dose distribution that results from the combination of a pair of horizontal parallel beams separated by 1.8h is shown in Figure 1. In practice, departures from a strict Gaussian dose distribution for a single beam and the use of angled rather than parallel beams require a slight modification in the separation of 1.8h to achieve a uniformity consistent with that shown in Figure 1. To produce a nearly uniform dose distribution on the patient’s skin six portals coming from fixed directions are used (7). Because of the curvature of the patient’s skin surface, electron displaced horizontally from the central axis enter the skin obliquely, so the number of angled pairs used affects the depth dose as well as the uniformity of the surface dose (3). In fact, the electron penetration for the full TSET treatment compared to that of a single beam at normal entry is substantially decreased.

la:

Rx

\,/, -02

1 b:

PLANE

00

HORIZONTAL

VERTICAL

CENTERLINE

,I/

/ 0.2

02

(h)

04

RELATIVE

06

08

/, 10

DOSE

Fig. 1. The combined isodose plot for two theoretical TSET beams with Gaussian dose distributions is shown in I a. Beams

are horizontal and the centers are displaced 1.8 times the half width, h. from each other. The dotted rectangle represents the treatment region (0.6h X 1Ah). Plots the vertical line perpendicular to the 1b. The contribution of the individual beams are shown. Dose is normalized the treatment region for the combined

of the relative dose along beam axes are shown in beams and the combined to 100% at the center of beams.

Total-skin

electron

Treatment machine The Cl-1800 provides seven standard therapy modes, two X ray and five small-field electron, the exact energies of which may be fixed by the user. Our choices for electron energies were 6, 9, 12, 16, and 20 MeV, with the expectation of deriving the TSET modes from (and perhaps changing) the two lowest. The Cl-1800 can reliably produce electron beams as low in energy as 4.5 MeV. Other relevant specifications include a maximum field size at the isocenter of 35 X 35 cm, which limits h for TSET beams to less than about 17 cm at the isocenter. Since h must be about I meter in the treatment plane (Fig. l), the corresponding dose rate will be decreased by about a factor 40 according to Eq (3) where h2 appears in the denominator. Thus, if 0.6 Gy/min is required on the central axis in the treatment plane, 24 Gy/min will be required on the central axis at the isocenter. The operating range for the standard dual dose monitor is 0 to 5 Gy/min, however, we hoped to make use of it for the TSET modes as well. The dose distributions for the photon and electron modes are adjusted by flattening filters and scattering foils, respectively, mounted on a carousel in the radiation head. The standard modes require five of the six positions available, leaving one position for a single internal scatterer for both TSET modes. Beryllium disks 1.25 mm thick were provided by Varian to use as the internal scatterer. Beryllium was chosen for its high melting point and low atomic number to limit bremsstrahlung production. The machine was reconfigured by Varian (according to Stanford’s instructions) to provide two TSET modes. An extra panel was provided at the console that has a key switch for activating the TSET modes and a third auxiliary counter. This counter is driven by the output of the primary dose monitor of the standard dual dosimetry system and has a four decimal place display rather than the usual three. The treatment is still set and terminated by the standard dosimeter circuitry, however, electrical connections are provided at the back of this panel for controlling the treatment externally. The dual ionization chamber with polyimide foil collecting plates, which are composed of lower Z materials than the mica design, is the sensitive element of the dose monitor. At the time of purchase, this chamber was standard on the Cl- 1800. Besides the TSET panel at the console, two additional “program” circuit boards were provided by Varian to set the parameters of the two TSET modes. A program board contains the printed circuitry for setting the machine parameters (e.g. gun current, magnet current) and the calibration of the dose monitor for a particular mode. Also, two empty wedge holders were provided to support the external scatters for the TSET modes. To operate a TSET mode, the corresponding holder must be inserted in the wedge slot, but the holder may be left empty if no external scatterer is required. Many of these changes are departures from the standard single-energy TSET capability that is now being offered by Varian.

therapy

661

0 R. S. COX et al.

Turning the TSET key switch and selecting one of the TSET energies causes the following to take place: 1. The auxiliary TSET panel and appropriate program board are activated. 2. A distinctive alarm sounds. 3. The carousel is rotated to the TSET position (formerly the spare). 4. The jaws are set to their full open position (35 cm X 35 cm). 5. Interlocks are activated that prevent the beam from coming on unless three further conditions are met: (a) the gantry is positioned at 90 f 20”; (b) the wedge slot in the head of the machine contains the TSET wedge holder corresponding to the mode selected; and (c) the wedge interlock at the console is set to match the code for the TSET mode selected. Thus, an alarm must be ignored and four simultaneous setting errors made (key switch plus 5a-c) to inadvertently select a TSET mode.

Parameters determining a TSET beam Four physical parameters need to be specified to define a single TSET treatment beam, namely, penetration, half width, dose rate, and x-ray contamination. Unfortunately, the beam parameters are not controlled directly and are influenced by the electron energy and beam current as well as other components along the beam axis (Fig. 2). The parameters actually set are: a) Gun Current, b) Beam Magnet Setting, c) Internal Scatter, d) External Scatterer, and e) Energy Degrader. None has a direct effect on one and only one beam parameter. For example, varying the magnet setting changes the electron beam energy, which affects all beam parameters. magnet

-vacuum window

intema scatte

/

ionization chambe

1.0 m

adjustable collimators external scatterer isocenter 2h = 35 cm -_i

@

1

k

1.

4.5 m

degrader

’ treatrnentpIane,/---

Fig. 2. Plan

2h=2m

view

7:’

of the TSET arrangement.

\I

662

I. J. Radiation

Oncology

0 Biology 0 Physics

March

1990. Volume

18. Number

3

Because of the high dose rates required at the isocenter for TSET, the response of the dose monitor must be investigated to ensure that the dose delivered is measured accurately. In particular, since the beam current on the central axis is being raised for TSET, the beam current density in and collecting voltage across the ion chambers must be such that recombination effects are negligible. The current density in the chamber is affected by the gun current, bending magnet setting. and thickness of the internal scatterer. Also, the gain of the amplifiers must be low enough to avoid saturating the final stages. Inslrumentution und measwemcu2ts To determine the values of the beam parameters, relative and absolute measurements were made with ionization chambers and electrometers. Three chambers were used for absolute measurements: a cylindrical 0.6 cc, the Markus+ parallel-plate, and a parallel-plate built at Stanford. For measurements in the treatment plane, extra shielding material was placed around the chamber cables to minimize leakage currents through the insulating materials that arise during irradiation. Also. measurements were taken with both polarities of collecting voltage and averaged to eliminate the remaining effects of leakage currents. Relative measurements at the isocenter were made with an automated dose scanner using both diodes and ionization chambers as radiation detectors. The xray contamination was measured with parallel-plate ionization chambers and confirmed with a condenser chamber. Neutron measurements were made with paired ‘LiF and 7LiF thermoluminescent dosimeters in a polyethylene moderator. The difference in the response of the dosimeters was attributed to the neutrons. Conversion of the readings to absorbed dose was made with reference to neurons from a PuBe source. Relative measurements were also made using photographic film,’ which is particularly useful for investigating the distributions obtained when multiple fields are combined. The phantom constructed for this purpose consisted of two circular polymethylmethacrylate (PMMA) slabs 2.5 cm in thickness and 25 cm in diameter between which film is placed (Fig. 3). Initially, the flat ends of the cylindrical pieces were covered with opaque cardboard to assure a light-tight seal at the edge of the phantom. The cardboard next to the film was found to distort the relative depth-dose function slightly at the larger depths, so measurements were later made without it. Since the penetration has been shown to be independent of radius of curvature for radii greater than the electron range (3). one phantom was enough for our purposes. In the darkroom the film was clamped between the cylinders and trimmed with a razor blade to match the circular cross section of the phantom. The edge of the

’ Nuclear

Assoc., Carle Place, NY.

Index Hole Index Arrow

Cardboard

83 ,..,,,.,,~,,,,,,,,,I,,,,

-10

-5

0

SCALE

Fig. 3. Exploded tographic film.

5

10

(cm)

view of the TSET phantom

for use with pho-

film was then covered with a single layer of black electrical tape (30 mg/cm”) to make a light-tight seal. In the treatment room, one of the index arrows (Fig. 3) was aligned with the central axis of the electron beam and an exposure was made. When the effects of multiple beams were studied. the phantom was rotated and exposures were made at each ofthe index arrows. The duration ofthe exposures were chosen to produce a maximum optical density in the range 1.O to 1.5. After all exposures were completed, the orientation of the beam axes was marked by inserting a scriber through the index holes and pricking the film. The film was developed and the optical density was measured using an automated photodensitometer. External slits were placed in the optical system of the photodensitometer to limit the spatial extent of the film sampled to about I mm in the direction scanned. Depth/ionization distributions were usually measured at the geometric center of the treatment plane. Measurements were carried out for both single horizontal beams (with the central axis aimed directly at the detector) and angled pairs of beams where the combined effect of two beams were studied. Similarly, the depth/optical-density distributions of both six horizontal beams and six angled pairs were determined with film and the cylindrical phantom. For ease of comparison, depth values plotted in the figures g/cm’.

are corrected Measurements

dose and normalized

t Eastman

for electron density and expressed in were converted to relative absorbed

so that the maximum

Kodak Co., Rochester,

is 100%. The

NY: XV-2 ready pack.

663

Total-skin electron therapy 0 R. S. Cox r/ al.

conversion of ionization data to absorbed dose was done according to reference (5). The absorbed dose delivered to the phantom by six angled pairs, D(6p), may be expressed in terms of the number of monitor units set for a single horizontal beam, M,,, as follows: D(6p) = M,,C,,R( 1p:l h)R(6p:lp).

(4)

The other three factors of Eq 4 are defined in terms of absorbed dose measurements made at the center of the treatment plane in phantom at the depth of maximum dose for each of the various beams. Ch is the calibration factor for a horizontal beam, that is, the absorbed dose per monitor unit delivered on the central axis. R( 1p: 1h) is the ratio of the dose delivered by one angled pair of beams to that delivered by one horizontal beam. Since the dose per monitor unit at the center of the treatment plane for each of the two angled beams is about half the dose per monitor unit on their central axes, R( lp: 1h) should have a value near unity. R(6p:lp) is the ratio of the dose delivered to a circular-cylindrical phantom by the full TSET treatment (six angled pairs) to that delivered by one angled pair. At any point around the circular phantom essentially three of the six angled pairs contribute all the dose. Thus, R(6p:lp) should have a value near 3. D(6p) may be taken as the dose delivered to the patient’s skin in the full TSET treatment, an assumption that we have verified with thermoluminescent dosimeters in many specific instances (8). Hence. for adequate patient dosimetry. the physicist must provide accurately measured values of Ch, R( 1p: 1h). and R(6p: 1p) for the particular TSET conditions finally established on the treatment machine. Absolute measurements were made to set the response of the dose monitor to Ch = 1 .OO. A parallel-plate ionization chamber was placed in a polystyrene block at the depth of maximum ionization and this assembly was placed in the center of the treatment plane. Measurements were made for both polarities of collecting voltage and averaged to remove the effects of leakage currents. Ionization was converted to absorbed dose according to the method given in reference (1). Ngas for the parallel-plate chamber was obtained by comparing its response in an electron beam to that of the 0.6 cc cylindrical chamber, for which Ngas is known.

RESULTS

plotted in Figure 4 under various conditions for the low and high energy modes. Consider the data for the 6 MeV mode in Figure 4a (crosses). As the dose rate is increased beyond the range normally used for small field electron operation (about 0.1 Gy/min in the TSET treatment plane), the slope of the response function shows two distinct breaks, one at 0.5 Gy/min and a steeper one at 1.5 Gy/min. The first comes at the onset of significant recombination in the ionization chamber and the second arises from saturation of the last stage of the amplifier circuit. The second break is really present in all the modes shown in Figure 4, but it occurs just beyond the last data point for the curves labeled “initial.” As seen in Figure 4, the effects of recombination were too large to allow accurate dose measurement with the configuration of the dose monitor furnished originally by Varian. Recombination in the ionization chamber may be decreased by lowering the current density in the chamber. This may be achieved by increasing the thickness of the internal scatterer. which spreads the beam more before it passes through the ion chamber. To maintain the same dose distribution in the treatment plane, any increase in the thickness of internal scatterer must be compensated by a decrease in the thickness of the external scatterer. Thus. this maneuver will not work if an external scatterer is not required or if a significant fraction of the beam is already intercepted by the collimator jaws. Another way to reduce the current density is to move the chamber farther away from the internal scatterer. The Cl- 1800 uses a dual dosimetry system wherein the ion chamber assembly houses two sequential transmission ion chambers. Normally, the proximal chamber, located 3 cm from the internal scatter, is used as the primary dose monitor. Additional distance (about 1 cm) can be obtained by using the distal chamber as the primary dose monitor. 4a:

cc

LOW

MODES

4b:

HIGH

ENERGY

MODES

MI 05

Performance of the dose monitor Medical linear accelerators are usually equipped with a dose monitoring system that consists of transmission ionization chambers, signal amplifiers, and appropriate metering. Ideally, the response of a dose monitor should be independent of the dose rate; this criterion, however, can only be approximated with practical components. The response function of the dose monitor of the Cl-1800 is

ENERGY

DOSE

1.0

15

RATE

2.0

25

(Gyimin)

0.5

DOSE

I.0

15

RATE

2.0

2.5

(Gyimin)

Fig. 4. Relative response of the Clinac 1800 dose monitor as a function of dose rate on the central axis at the TSET treatment plane. The triangles represent data taken initially at the beam energies and collecting voltages indicated. Other initial conditions were I .25 mm Be internal scatterer and the use of the proximal ion chamber. Crosses represent the responses obtained for the modes finally accepted, wherein 2.5 mm Be internal scatterer and the distal ionization chamber were used.

664

I. J. Radiation

Oncology

0 Biology 0 Physics

Recombination may also be reduced by increasing the collecting voltage on the ionization chamber. In the Cl1800, the collecting voltage applied is nominally 600 V. The effect of varying the collecting voltage may be seen in Figure 4. As the voltage is raised from 600 to 900 V, the first break in the response curve is moved to higher dose rates and the slope thereafter is reduced. The ion chambers of our machine have been run routinely at 750 V for over 2 years with no problems attributable to the overvoltage. The performance of similar ion chambers, however, may vary considerably. The need for the changes described above should be well documented with measurements and all modifications must be carried out in consultation with the manufacturer. After changes are made the response of the dose monitor should be carefully followed. By increasing the collecting voltage to 750 V, shifting to the distal ionization chamber for the primary dose monitor, and extracting as wide a beam as could be brought through the jaws without appreciable loss (h of 17 cm, about half the maximum field size for the lower energy mode, 14 cm for the higher), the variation in the recombination rate with dose rate was reduced to an acceptable level. As shown in Figure 4. the change in response for both modes was less than 3% for dose rates from 0.6 to 0.8 Gy/min, which was defined as the operating range. If the dose rate does not fall within these limits when the accelerator is turned on. treatments are halted and the automatic frequency control is retuned. The problem of amplifier saturation (arising from high dose rates at the isocenter) may be solved by decreasing the gain in the final stage. This, of course, changes the calibration factor. The convention normally used for accelerators is that 1 monitor unit corresponds to about 1 cGy at the isocenter and one must be willing to give up this relationship if the amplifier gain is changed. To enable adherence to the usual convention, Varian provided a third display counter for the TSET modes with precisely ten times the sensitivity of the standard counter. It was hoped that the standard counter could be calibrated to 10 cGy per unit at the isocenter and thus the TSET display counter would register 1 cGy per unit. Unfortunately, a factor of 10 is insufficient to avoid saturation, since the ratio of the dose rate at the isocenter to that at the treatment plane is about 40 for the lower energy mode. For the higher energy mode, the effect of the internal scatterer is reduced requiring, in addition, the use of an external scatterer. Here, the ratio of dose rates is even greater. Clearly, the convention of calibrating at the isocenter could not be maintained without more elaborate modifications to the electronics. The convention adopted was to calibrate the distal dose monitor for 1 cGy per monitor unit on the central axis in the treatment plane rather than at the isocenter. The required reduction in amplifier gain moves the dose rate at which the amplifier becomes saturated to well beyond 1 Gy/min in the treatment plane, even for the higher en-

March

1990. Volume

18, Number

3

ergy mode. The TSET display counter in the Stanford configuration provides no useful information and is ignored.

As mentioned above. the TSET electron beams emerge from the bending magnet and vacuum window at the same energies provided in the 6 or 9 MeV small-field electron modes. The beam is then degraded in energy by the beryllium scattering disk(s) (1.25 mm or 0.5 MeV per disk) and the material of the ion chamber. For the higher energy mode where additional scattering is needed, the beam is degraded by an external scatterer, a PMMA plate 4 mm (0.8 MeV) thick located at the wedge position. The beam is also degraded about 1.1 MeV by the intervening air over the 4.5 meter distance to the treatment plane. Our goal was to match the depth of the 80% isodose of the lower energy TSET beam on the Cl- 1800 to that of our Cl- IO. Starting with 6 MeV and 9 MeV electrons, the new TSET beams were much too penetrating to meet this specification. Two possible solutions that are easy to implement were attempted: (a) reduction ofthe initial energy of the two TSET modes to 4.5 and 8 MeV, respectively, by returning the beam energy and (b) placement of a large PMMA absorber 1 cm in thickness 20 cm in front of the treatment plane. The modes corresponding to solution a are called the “initial modes” and those to solution b are called the “final modes,” as the latter method was ultimately adopted. The central-axis depth-dose measurements obtained at the center of the treatment plane for both the initial and final modes are shown in Figure 5. The corresponding data for the Cl- 10 are also shown. Depths of penetration corresponding to a few specific dose levels are interpolated from these data and presented in Table 1. Uncertainties in the depth measurements are within 0.05 g/cm’. Data for a single horizontal beam or a single pair are shown in Figures 5a and 5c. The points represent data taken with an ionization chamber and the continuous curves represent data taken with film and the circular phantom. Data for six combined horizontal beams or angled pairs obtained with the circular phantom are shown in Figures 5b and 5d. The distribution of radiation along the radii of the phantom varies with angle. going through a cycle every 60”. A maximal extreme distribution occurs every 60” when the radius chosen is aligned with the vertical central plane of a single head-on or dual beam. A minimal extreme distribution is found along the six radii midway between those of the maximal distributions. For each mode plotted, two depth-dose curves are shown: one for the maximal and the other for the minimal distribution. The normalization is taken so that 100% corresponds to the maximum value of the average of the two extreme distributions. Note the generally good agreement obtained between the ionization chamber (points) and the film measurements (curves) shown in Figures 5a and 5c. For the 8

Total-skin INITIAL

MODES

FINAL

electron

MODES

Fig. 5. Penetration of the various TSET beams. Continuous curves represent relative dose obtained from optical-density measurements made with the phantom shown in Figure 3. Points appearing in 5a and SC represent relative dose computed from

measurements made with a parallel-plate chamber. For each mode shown in 5b and 5d, two curves were obtained: the curves with higher maxima were taken along a film radius coincident with a beam axis: and those with lower maxima were taken along a film radius midway between two such radii.

MeV mode in Figure 5a, where the disagreement is greatest, the film measurements were carried out with an external scatterer that was 0.8 mm thicker than those with the ionization chamber. Also, the measurements for the initial modes were carried out with cardboard adjacent to the film in the phantom. The lower density cardboard allowed some electrons to penetrate more deeply and the optical-density readings near the tails of the curves were increased (Fig. 5a). Had the cardboard been removed, as was the case for the final modes, the agreement would have been better (Fig. 5~). The match of the 4.5 MeV initial mode with the Cl- 10 beam at the 80% level is not good (Fig. 5b). A closer match was obtained with the 6 MeV final mode, which uses the degrader (Fig. 5d). The use of a degrader near the treatment plane offers several advantages, one being an almost independent control of the depth of penetration of the beam. If a degrader were placed in the head of the machine, the beam width and dose rate at the treatment plane would also be significantly affected. Measurements obtained for the final modes (with a degrader 1 cm in thickness) for one angled pair are shown in Figure 5c and for the combination of six angled pairs in Figure 5d. For the Cl- 10 beam, the shift from horizontal to angled beams produced a measurable decrease (about 1 mm) in the penetration. This may be seen by comparing the data for the Cl-10 beam in Figures 5a and 5b versus those in Figures SC and 5d, respectively. For the Cl- 1800 beams, this effect was considerably reduced and barely detectable

therapy

665

0 R. S. COX rl ul

in any of the modes. The reason for this difference probably arises from differences in the energy spread of the beams and in the thickness, composition, and positioning of the degraders and scatterers used. The Cl- 10 beam has initial energy near 8 MeV and emerges directly from the wave guide without magnetic energy analysis, Except for the intervening air, it is degraded to 4 MeV entirely by materials in the head of the machine. At the treatment plane, this beam is considerably wider than 1 meter (despite the shorter treatment distance of 3.5 m) and requires a larger angular separation (-t20”) than the Cl- 1800 (f 16” for both modes) to achieve acceptable uniformity. Since the penetration is the same for the horizontal and angled beams on the Cl- 1800, the ratio R( 1p: 1h), of Eq 4 may be determined from a ratio of ionization chamber readings. The values of R( 1p: 1h) were found to be 1.OO and 1.03 for the 6 MeV and 9 MeV modes. respectively. The ratio, R(6p: lp), of Eq 4 requires measurements in a circular phantom to obtain the dose for the full TSET treatment. so the film phantom was used. Then R(6p: lp) is the quotient of the optical density per monitor unit for six angled pairs divided by that for one angled pair. Repeated determinations of this ratio yielded an average value of 2.47 t- 0.10 for both modes. X TU)’and neulron contumination The x-ray contamination was measured izontal and angled beams with a variety

Table 1. Beam specifications

for both horof ionization

at the treatment

plane

CL- 1800 CL-IO 8

Accelerator nominal energy (MeV)

4.5

Single horizontal Depth of isodose % 80 (cm) 50 (cm) 20 (cm) Half width (cm) Dose rate (Gy/min) Bremsstrahlung Dose ratio (%) Meas. depth (g/cm2)

8

6*

9*

beam

0.90 1.30 1.75 145 0.5

1.05 1.35 1.70 93 0.7

2.00 2.45 2.90 95 0.7

0.45 0.70 1.10 94 0.7

1.25 1.70 2.05 96 0.7

2.0

0.1

0.4

0.6

0.9

3.5

2.7

4.4

3.5

7.0

1.45 2.10 2.60

0.35 0.65 0.95

0.80 1.30 I .90

Six angled pairs Depth of isodose % 80 (cm) 50 (cm) 20 (cm) Uniformity of dose at dmax (?%) Bremsstrahlung dose ratio+ (%)

0.40 0.80 1.30

0.80 1.15 1.55

9

9

1.0

0.3

14 1.0

tl

3 1.5

* With Lucite degrader I cm thick. + Dose ratio = bremsstrahlung dose at measurement electron dose at dmax.

2.3

depth/

666

I. J. Radiation

Oncology

0 Biology 0 Physics

chambers and was expressed as a percentage of the reading obtained at the depth of maximum ionization for the electrons. Measurements were carried out at depths near, but greater than, the extrapolated range of the electrons. Thus, the results should be thought of as upper limits of the x-ray contamination rather than the x-ray dose averaged over the whole patient. For the parallel-plate chambers, readings were obtained for both polarities of the collecting voltage; in fact, leakage currents were typically several times larger than the ionization currents arising from bremsstrahlung. From the Cl-10. the level of x-ray contamination decreased considerably with distance from the central axis. This can be seen in Figures 5c and 5d where the level of x-ray background for the angled beams is much reduced in comparison to that for the horizontal beams. Figures 5a and 5b. For the Cl- 1800. the same percentages (within errors) were obtained for angled and horizontal beams. This result is expected because most of the absorbers in front of the treatment plane consist of low-Z materials, so X rays produced in the phantom or degrader are the main component of the bremsstrahlung. Magnitudes for a single horizontal beam and the full TSET treatment are listed in Table 1. The depths at which the bremsstrahlung was measured are also listed. The contribution of the 1 cm PMMA degrader was investigated by taking readings with it both in and out of the beam. As expected, the presence ofthe degrader, which is roughly tissue equivalent. made no measurable difference in the level of x-ray contamination at depths greater than the electron range. For the full TSET treatment, we would expect the electron dose to be increased by about R(6d: Id) = 2.47 and the x-ray dose by about a factor of 6. Thus, the relative x-ray contamination should be increased by a factor of 2.4 and values obtained from film data for the full TSET treatment on the Cl-1800 are consistent with this computation (see Table 1). Neutron measurements were made at the isocenter and the treatment plane for the 9 MeV mode. Results were 0.02 and 0.0 1 neutron rem per electron rad, respectively.

Transverse distributions at isocenter for head-on beams at the various electron energies were measured with the automated dose scanner and ionization chambers. Scans were taken perpendicular to the beam axis and parallel to a collimator jaw. Isocentric geometry was used wherein the chamber was placed on the gantry axis (100 cm) and at the depth of maximum ionization in the phantom. The results are shown as the solid and dashed curves in Figures 6a and 6b. The dotted curves superimposed represent the Gaussian functions of Eq 3, with the same value of h as the experimental data. The agreement is good except for the widest beams, where the width is comparable to the collimator setting and many electrons are scattered. The data obtained for the initial modes are shown in

March

1900. Volume

18. Number

3

Figure 6a. The dashed curves were obtained when no internal scatterer was used and they have values for h of 14.7 and 9.3 cm, respectively. As discussed above, these beams were too narrow for optimal response of the ionization chamber. An internal scatterer consisting of I.25 mm of Be was placed in the head of the machine and the solid curves with values for h of 17.6 and 12.4 cm, respectively, were obtained. The Cl- 1800 penetration data of Figures 5a and 5b were obtained with the 1.25 mm internal scatterer in place. The data obtained for the final modes are shown in Figure 6b. To produce a comparable value of h = 17.1 cm (upper solid curve) for the 6 MeV mode, 2.5 mm of Be was required for the internal scatterer. With h = 17 cm at the isocenter, the natural divergence of the beam results in an h = 1 m at our treatment distance, which is 3.5 m beyond the isocenter. Thus, no additional external scatterer was required for the 6 MeV mode. For the 9 MeV mode, h = 13 cm at the isocenter (lower solid curve) and was considerably less than I meter at the treatment plane. To obtain h = 1 m at the treatment plane, an external scatterer consisting of 4 mm of PMMA was used and this increased h to 15 cm at the isocenter (dashed curve). This corresponds to the scattering conditions under which the Cl- 1800 penetration data of Figures 5c and 5d were obtained. Transverse distributions beyond the isocenter for both horizontal and angled beams were obtained point-by-point in air with both cylindrical and parallel-plate ionization chambers, to which build-up material of thickness equal to the depth of maximum ionization was added. The measurements are represented by the +‘s or X’s in Figures 6c and 6d. Dotted and dashed curves again represent the Gaussian functions of Eq 3. Measurements for the initial modes taken 2 meters beyond the isocenter are shown as points in Figure 6c. For the 4.5 MeV mode, the agreement with Eq 3 is considerably better here than it was at the isocenter. In fact. the only collimator effect evident is a slight (5%) variation in h with rotation, which is present in both modes; points on the dotted curve represent readings where the collimator axes were at 45”, whereas those on the dashed curve were taken with the axes horizontal and vertical. In the treatment plane (3.5 m beyond the isocenter), the 4.5 MeV beam spread further to h = I m as described (data not shown); for the 8 MeV mode. an external scatterer of 3 mm of PMMA was required to obtain an h = 1 m. When the external scatterer was used, the weak dependence of h on collimator angle seen in Figure 6c is decreased further. Measurements for the final modes taken in the treatment plane with the degrader in place may be seen in Figure 6d. Data (+‘s and X’s) were taken for both horizontal and angled beams and corrections for variations of the treatment distance with angle are indicated (squares and diamonds). Points in the treatment plane displaced vertically from the horizontal are at greater distances from

667

Total-skin electron therapy 0 R. S. Cox ef ul. BEAM

PROFILES

AT ISOCENTER

ea: INlTlPILMODES

,,.‘.!

-

NO External Scatterer

1.25 mm Be ht.

I

1

Sb: FINAL

I MODES

BEAM

I

I

scatterer

I

6~:

I

6d: FINAL

-15

DIST

-10

FROM

-5

BEYOND

MODES

1.25 mm Be ht. scanerer

AT 3 METERS

0

-50

ISOCENTER

50

100

2.5 mm Be ht. Scatterer

AT Rx PLANE

2.5 mm Be In,. Scatterer

- - - - 4 mm PMMA external no external

MODES

-100

%.:

Me’,*.”

z

Angled

n=1oocm..;

D

Horiz. Beam

-6

-

INITIAL

PROFILES

scatterer

\

Beam

‘,.,

-_---.J

0

‘i.

scatterer

0

CENTRAL

5

10

AXIS

15

(cm)

-150

VERTICAL

-100

DIST

-50

FROM

0

50

CENTRAL

100

AXIS

150

(cm)

Fig. 6. Transverse distributions. In 6a and 6b, the solid and dashed curves represent experimental data obtained at the isocenter with a scanner. The dotted curves superimposed represent the corresponding predictions of Eq. 3. In 6c and 6d, the X’s and +‘s represent measurements obtained with an ionization chamber and the continuous curves represent the predictions of Eq. 3. In 6d, the squares and diamonds indicate the effect of correcting the X’s and +‘s, respectively. for the effects of distance.

the electron source and, hence, have slightly decreased dose rates. This effect tends to narrow the measured width of a horizontal beam and increase that of an angled beam. However, if one “corrects” the measured data according to the inverse square law (squares and diamonds), most of the disagreement can be removed and the points can be shown to lie on the same Gaussian function (dotted curve). For the 9 MeV mode, the dose distribution of the horizontal beam (X’s) is best described when h = 96 cm and that of the angled beams (-t’s) by h = 108 cm. After correcting for distance, almost all points lie on a single curve with h = 102 cm. For the 6 MeV mode, which is scattered more heavily, slight discrepancies remain, particularly for those points lying near the edge of the degrader. For those points, electrons scattered toward and away from the detector by the degrader are less well balanced than is the case for the 9 MeV beam. The half widths reported in Table 1 are for horizontal beams without corrections. Measurements at the treatment plane (with the degrader in place) for final modes are also shown in Figure 7. Here the combined doses at the depth of maximum ionization resulting from an angled pair are noted at various rep-

resentative points in the treatment plane. The separation of the angled beams at the treatment plane is 2.00 m, corresponding to a half-angle of nearly 16”. The values of h are 1.06 and 1.08 m for 6 and 9 MeV, respectively. The uniformity of the radiation arising from the full TSET treatment may also be seen in Table 1 and the film data shown in Figures 5b and 5d. The differences between the maximal and minimal distributions are greatest near the surface and tend to disappear at depths beyond the 80% level. Also, the differences were largest (20% or more) for the Cl-10 and the initial modes of the Cl-l 800 (4.5 and 8 MeV), where a degrader at the treatment plane was not used. For the final modes, the differences was undetectable (< 1%) at 6 MeV and 7% at 9 MeV. DISCUSSION We have achieved a dual-energy TSET capability on a Cl-1800 that meets the general criteria set forth in the Introduction. The specifications of the TSET beams developed are described in the previous section and listed in Table 1. In particular, we were able to match the depth of the 80% isodose of the lower energy mode to that previously available on our Cl- 10 and provide a second mode

1. J. Radiation

668

7a:

Oncology

6 MeV

0 Biology 0 Physics

7b:

9 MeV

1.4 1.8

-1.8

0.5

-1.9 ':I

0.0

-3.9

-

1.8

-1.9

-

3.2

-0.8

1.4

-0.2

0.0

-0.2

0.2

(m)

HORIZONTAL

0.0

HORIZONTAL

0.2

(m)

Fig. 7. Dose distribution in the treatment plane for the final modes arising from an angled pair of beams. The numbers appearing above the points represent the departures of the dose rate from the central-axis value in percent.

of twice this penetration. two lowest

standard

This was achieved

electron

energies

by using the

and a large PMMA

degrader just in front on the treatment plane. We also explored the alternative of lowering the electron beam energies and omitting the degrader. Although we were not able to match our Cl-10 beam by this technique, the characteristics of the beams achieved are suitable for TSET, provided a somewhat greater penetration is considered desirable. Both techniques have advantages. By starting with lower energy electrons, less degrading is required and the level of x-ray contamination is reduced. On the other hand, the x-ray level with a degrader, although higher. remains acceptable and the machine runs with greater stability in the TSET mode at 6 MeV. Also the degrader diffuses the electrons as they pass through it and, when fields are combined in the full TSET treatment, produces a more uniform superficial dose distribution than would be obtained without it. However, none of these advantages is particularly compelling; for machines where only one TSET mode is available, a second less-penetrating mode could be obtained by adding a degrader. We also were able to meet our criteria for simplicity of operation. The two beams produced are of similar width at the treatment plane and use the same thickness of degrader. This allows the same treatment procedure to be used for both modes. Also, we found it possible to use the standard dual dose monitor for TSET after some minor modifications to the circuitry. Avoiding the devel-

March

1990. Volume

18, Number

3

opment and daily use of an external TSET dose monitor results in continuous saving of time for both physicists and technologists. There are, however, problems to solve when the standard dose monitor is used for TSET. Most noteworthy is that of recombination in the ionization chamber, which manifests itself as an increase in the calibration factor with an increase in the dose rate. We found that recombination effects could not be eliminated on the Cl- 1800 but, by making the changes described above, could be reduced to an acceptable level. In our configuration, the calibration factors for both modes stay constant within *l”/r, when the dose rate is held at 0.7 -t 0.1 Gy/min. Fortunately, 0.7 Gy/min is an acceptable dose rate and the treatment time for one ofthe I2 TSET beams is about 1 min. The Cl- 1800 is capable of delivering much higher dose rates, but the recombination problem prevents their utilization. Furthermore, the last stage of the dose-monitor amplifier saturates if one attempts to display the true dose rate at the isocenter. Our solution involved calibrating at the treatment plane instead; although we feel comfortable with this convention. care must be taken to assure that all technologists, dosimetrists. and physicists involved in delivering TSET to patients understand fully the dosimetric procedures finally adopted. The number of monitor units set for a typical TSET beam is near 80 and, hence. the total dose can be controlled to about 196.The metering of dose on the Cl- 1800 is to 0.0 I monitor unit. so knowledge of the actual monitor units delivered is considerably better than l%>. Having a third display counter with a calibration factor tied to that ofthe primary dose monitor was of no use. If a third counter is to show the dose delivered at the isocenter. its calibration must be independent of the calibration of the primary dose monitor. To develop a TSET modality on any accelerator, compromises must be made among the parameters of penetration, uniformity, dose rate, and x-ray contamination. Having explored several blind alleys in establishing our TSET modality on the Cl-1800. we offer a more direct procedure than the one actually followed. For the lowest energy TSET mode, adjust the thickness of the TSET internal scatterer to produce an h of about half the maximum field size at the isocenter. For the Cl- 1800, this requires about 2.5 mm of Be for the 6 MeV mode. Determine the amount of external scatterer needed to produce an h of about 1 m at the treament plane. For a treatment distance of 4.5 m (3.5 m beyond the isocenter) and the 6 MeV beam, an external scatterer may be unnecessary: for shorter treatment distances or higher energies some external scatterer may be required. In our case. 4 mm of PMMA was used for the 9 MeV mode. Measure the central axis depth dose at the treatment plane to determine the thickness of the degrader re-

Total-skin

electron

quired to produce the desired penetration. In our case, 1 cm of PMMA was used. 4. Check and adjust the dose monitor circuitry to minimize recombination and eliminate amplifier saturation effects. (Any modifications should be carried out after consultation with and approval by the manufacturer.) Then adjust the beam intensity to the maximum con-

therapy

0 R. S. COX el al.

sistent with tolerable dose monitor.

669

variation

in the response

of the

This procedure should lead to success in the fewest steps. However, the specific values recommended should be regarded as starting points and modified where necessary to obtain the desired results.

REFERENCES I. American Association of Physicists in Medicine Task Group

2.

3.

4.

5.

2 1. A protocol for the determination of absorbed dose from high energy photon and electron beams. Med. Phys. IO: 741-771: 1983. American Association of Physicists in Medicine, Report 23. Total skin electron therapy: technique and dosimetry. NY: Am. Inst. of Phys.; 1988. Bjarngard. B. E.; Chen, G. T. Y .; Piontek, R. W.; Svensson, G. K. Analysis ofdose distributions in whole body superficial electron therapy. Int. J. Radiat. Oncol. Biol. Phys. 2:319324, 1977. Holt. J. G.; Perry, D. J. Some physical considerations in whole skin electron beam therapy. Med. Phys. 9:769-776; 1982. International Commission on Radiological Units, Report

35. Radiation dosimetry: electron beams with energies between 1 and 50 MeV. Bethesda, MD: ICRU Publications; 1984. 6. Karzmark, C. J.; Loevinger, R.; Weissbluth, M. A technique for large-field electron therapy. Radiology 74:633-644; 1960. 7. Page, V.; Gardner, A.; Karzmark, C. J. Patient dosimetry in the treatment of large superficial lesions with electrons. Radiology 94:635-64 I; 1970. 8. Pales, B. B.; Fessenden, P. TL dosimetry for treatment of mycosis fungoides with 4 MeV electrons. (Abstr) Med. Phys. 9:618-619; 1982. 9. Trump, J. G.; Wright, K. A.; Evans, W. W.; Anson, J. H.; Hare, H. F.; Fromer, J. L.; Jacque, G.; Horn, K. W. High energy electrons for the treatment of extensive superficial lesions. Am. J. Roentgenol. 69:623-629: 1953.