0360.3016/92
Inr. J. Rodruriun Onlolo~v Biol. Ph.w. Vol. 22, PP. I I 17-I 124 Printed in the U.S.A. All rights reserved
$5.00 + .oO
copyright 0 1992 Pergamon Press Ltd.
??Technical Innovations and Notes
DOSIMETRY
AND DOSE SPECIFICATION FOR A NEW GYNECOLOGICAL BRACHYTHERAPY APPLICATOR
E. D. SLESSINGER,
M.S.,
C. A. PEREZ,
M.D.,
AND J. F. WILLIAMSON,
P. W. GRIGSBY,
M.D.,
PH.D.
Radiation Oncology Center, Mallinckrodt Institute of Radiology, Washington University School of Medicine, 5 10 South Kingshighway Blvd., St. Louis, MO 63 I 10 A new afterloadable gynecological intracavitary applicator has been designed and is now in use for the treatment of a wide range of vaginal, cervical, and endometrial cancers in a single application, with a dose distribution that more closely matches the prescribed treatment than previous methods. The plexiglass applicator consists of a bulbous section that is inserted up to the vaginal apex and loaded with right and left ovoid sources. The distal portion is cylindrical with a central channel for tandem sources. A straight tandem is used when the target volume extends to the vaginal apex, but an intrauterine tandem can also be used. Extensive dosimetric and radiographic evaluations were performed to guide design refinements and to validate the surface dose predictions of the brachytherapy treatment planning system. The final applicator design delivers 1 lo-120 cCy/hr to the vaginal apex,surface and 95-100 cGy/hr to the distal vaginal surfaces when loaded with a 144.6 U (20 mgRaEq) cesium tube ‘in each ovoid channel and 72.3, 72.3, and 144.6 U (10, 10, and 20 mgRaEq) cesium tubes in the vaginal cylinder channel. A system for treatment dose specification has been established that includes dose tables for manually calculating surface doses to reference points. A dose distribution comparison with a sequential colpostat-vaginal cylinder treatment demonstrates that dose delivery is more precise and uniform with this new applicator. Gynecological
malignancies,
Intracavitary irradiation, Brachytherapy,
Vaginal applicator, Dosimetry, Miralva.
treatments and achieve a uniform dose distribution or even quantify the total doses. Because of the limitations of the existing gynecological intracavitary applicators, a new gynecological brachytherapy applicator was introduced by us last year (5). This applicator can treat a wide range of vaginal, cervical, and endometrial cancer target volumes in a single application. The new applicator has been named MIRALVA, an acronym for the Mallinckrodt Institute of Radiology Afterloadable Vaginal Applicator. The Miralva design has been described in detail in an earlier paper (5). The plexiglass structure (shown in Fig. 1) is cylindrical in the lower f of the vagina and enlarges to a bulbous shape at the vaginal apex. There are three channels for the placement of Cesium- 137 sources. Two of the channels position one tube source each into the right and left ovoid locations in the bulbous part of the applicator. These sources are separated by 2 cm (center to center). The stainless steel tandem channel is concentrically located within the cylindrical portion and extends to the vaginal apex. Longer tandiems can also be used for intrauterine placement. Regardless of which tandem length is used, an inactive 2-cm length spacer in the
INTRODUCTION
The treatment of gynecological tumors that involve the vagina is often constrained by the limitations of the currently available applicator systems. This is due in part to the variability of the anatomic structures included in the target volume. An endocervical tumor with extension to the vagina requires an intrauterine tandem and vaginal ovoids to treat the paracervical region. If treatment to the middle or lower thirds of the vagina is also indicated, then a second application to treat the vagina is warranted. Vaginal cancer in the upper and middle vagina or an extensive endometrial cancer recurrence presents similar problems requiring two applications. For example, Fletcher colpostats may be used to treat the vaginal apex in the first procedure and Delclos cylinders in the second procedure to treat the distal vagina. It is undesirable to treat different portions of the target volume in a sequential approach. If the entire target volume is not encompassed during each treatment, disease in some locations may proliferate while being sterilized at other locations. It is also very difficult to match the two
Presented at the 32nd Meeting of the American Society for Therapeutic Radiology and Oncology, Miami, FL, 15- 19 October 1990.
Reprint requests to: Eric D. Slessinger. Accepted for publication 5 September 199 1. 1117
1. J. Radiation Oncology 0 Biology 0 Physics
Fig. I. The plexiglass Miralva applicator consists of a bulbous section that is inserted up to the vaginal apex and loaded with right and left ovoid sources. The distal portion is cylindrical with a central channel for tandem sources. A straight tandem is used when the target volume extends to the vaginal apex, but an intrauterine tandem can also be used as in the above configuration. Additional ovoid caps and cylindrical sleeves enable adjustment ofthe outer dimensions to maintain surface contact.
tandem, centered between the ovoid sources is always maintained. This is a safeguard to prevent an excessive dose due to intersecting ovoid and tandem sources. The outer dimensions and shape were designed to deliver a uniform dose distribution to the entire vaginal mucosa. Additional ovoid caps and cylindrical sleeves are also available for adjustment of the outer dimensions to maintain surface contact. The development of the Miralva device was an iterative process that began with certain clinical design criteria (5). In this paper, we present the results of dosimetric and radiographic studies that were performed to evaluate the dose distributions on the applicator surface. Adjustments to the applicator surfaces were made on the basis of these studies. The dosimetric study was also used to verify that the dose predictions of the treatment planning system were in good agreement with the measured doses. The planning system was used to develop a dose specification system and optimize source configurations that will deliver dose distributions according to certain dosimetric criteria. Also, a dose distribution comparison between the two step treatment approach and the Miralva approach is presented.
METHODS
AND MATERIALS
Dosimetric measurements and comparison with calculations The dosimetry evaluations involved the comparison of dose measurements using Lithium Fluoride TLD- 100 rods* and dose computations using the Modulex+ linear * Harshaw/Filtrol Partnership, 6801 Cochran Road, Solon, OH 44 139. + Computerized Medical Systems, Inc., 56 Worthington Drive, Maryland Heights, MO 63043.
Volume 22, Number 5, I992
source implant algorithm. This algorithm is based on the computational models of Batho and Young (12), which calculates dose factors on a reduced co-ordinate grid by evaluating a Sievert Integral. Tissue attenuation and buildup corrections are made using the third order polynomial developed by Shalek and Stovall (7). The effort was primarily directed to surface dosimetry but more distant locations were also analyzed. The TLD methodology of these studies utilizes TLD rods (1 mm diameter and 6mm long) that are organized into batches of 25. The relative sensitivity of each dosimeter was determined by irradiating all dosimeters to a uniform dose. A normalization factor (Ni) relates the response of the ith dosimeter to the average batch response. Each dosimeter is stored in an assigned location in the batch container. For dose determinations in experiments, all TL readings are corrected by multiplying with the corresponding normalization factors. Four to six dosimeters from each batch are used for calibration to establish a linear regression from the known doses and the corrected TL readings. Only one dosimeter is required for each calibration dose point because the normalization factor for each dosimeter has been predetermined. It has been shown that this approach to TL dosimetry can achieve an average precision of f 1.5% ( 10). A special lucite phantom was used to calibrate the TLD rods for the Miralva dose experiments. Lucite was used because it was readily available, and could be machined for precise positioning of a vertical cesium tubes (2 cm total length, 1.4 cm active length) and an array of vertical TLD rods 1 to 5 cm away from the perpendicular bisector of the source. Several different calibration doses were delivered simultaneously. The calibration doses were calculated as though the irradiation had been performed in water. Krishnaswamy Gamma Dose tables (3) were used for this purpose. The cesium dose data of Williamson (11) or Waggener et al. (8) could also have been used for this purpose. Although the energy dependence of the TL response was not evaluated explicitly, the dose calibrations incorporated the effects of the variations of energy spectra at the different depths. For all of the Miralva dose experiments the average agreement between the calibration doses calculated using the Krishnaswamy tables and the calibration doses determined from the linear regression was 3.2 + 2.6%. A 25 cm cubic water phantom was used for most of the dose evaluation experiments. A straight tandem that extended to the cephalad applicator surface as illustrated in Figure 2 was used in these experiments. The TLD rods on the Miralva were sealed in polyethylene wrap, numbered, and taped to multiple surface locations. The number of surface dose measurement points ranged from I8 * Medical Products Division/3M, MN 55112.
TCAAP 590, New Brighton,
Dose specification 0 E. D. SLEWNGER
et al.
1119
to 52. The number of measurement
points increased with each subsequent experiment to evaluate in greater detail the dose uniformity at the cephalad surface. The applicator was sealed with additional plastic wrap to prevent water leakage. With the loading end left open, the Miralva was then suspended in the water phantom and afterloaded. The sources were left in place for 1 hour. Source transfer time was approximately 30 seconds. Orthogonal radiographs of the Miralva, loaded with dummy cesium tubes and with solder wire fixed to the applicator surfaces, provided the localization information for the point dose computations. Dose measurements were also obtained with the applicator placed within a polystyrene phantom for dosimetric evaluations at points 1 cm away from the applicator surface and at multiple levels above and below the vaginal apex. Treatment dose specification To develop a system for Miralva treatment dose specification, the applicator surface was subdivided into reference dose regions. For each region, a series of points was selected on the midcoronal and midsagittal applicator surfaces. Using the Modulex” RTP system, the dose rate (cGy/hr) contribution from each source to each dose point was computed. The average dose rate contribution from each source to each region was calculated and tabulated. Dosimetric comparison with sequential treatment A dose analysis was undertaken to compare the sequential treatment scheme to the Miralva treatment. A treatment prescription of 6000 cGy to the surface of the vaginal apex and 4000 cGy to the distal vagina surface was selected for this comparison. The sequential treatment chosen for this study is representative of the approach that had been used at Mallinckrodt prior to the clinical introduction of the Miralva. The initial treatment in this study used 2 cm diameter ovoids, each loaded with 144.6 I-J”(20 mgRaEq) to deliver a surface dose from each source of 5000 cGy in 39.5 hr. This was followed by a second application using 2.5 cm diameter Delclos cylinders with a 2 cm spacer at the cephalad aspect and 6 cm active length configured as a 72.3, 72.3 and 144.6 U (cephalocaudal) source sequence to deliver 4000 cGy to the cylinder surface at the midpoint of the active length in 44.4 hr. The Miralva treatment would use 144.6 U sources in the ovoid channels and 36.2, 72.3 and 72.3 U loading in the tandem channel for 53 hr. Dose distributions were computed for each treatment and point doses were interpolated at 5 mm intervals along the midcoronal and midsagittal applicator surfaces starting at the center of the vaginal apex and moving distally from there. Corresponding dose data were also obtained at a depth of 5 mm from the applicator surfaces. For the sequential
‘IComputerized Medical Systems, Inc., 56 Worthington Drive, Maryland Heights, MO 63043.
Fig. 2. The original Miralva design (broken line) was modified by rounding the ovoid section (dotted line) to improve the ease of insertion. The ovoid section was enlarged for the final design (solid line) to achieve a dose rate of 120 cGy/hr to the vaginal apex surface with 144.6 U (20 mgRaEq) ceslum tubes loaded into the two ovoid channels.
treatment, two models were considered: one without vaginal retraction before the second application and the other model assumed that the 2 cm length of vaginal surface distal to the vaginal apex center was reduced to 1 cm at the time of the second application causing the tissues of the apex to recede away from the high dose region of the second application.
RESULTS Dosimetric measurements and comparisons with calculations Figure 2 illustrates how the Miralva design evolved. All of the modifications involved the sham and size of the ovoid section. Source spacing was never altered. The initial ovoid section was too rectangular. To improve the ease of insertion, the ovoid section was rounded as much as possible. Subsequently, the round shape was enlarged to achieve the surface dose rate of 1 lo- 120 cGy/hr to the vaginal apex when the ovoid positions were loaded with 144.6 U sources. The lateral separation was increased from 4.0 cm to 4.5 cm and the cephalad surface was increased in thickness from 1.O cm to 1.4 cm over the ovoid source axes, as shown in Figure 3. Figure 4 summarizes the dosimetry results that span the period from the initial to final Miralva stages. The results of the initial design appear on the left side. This is followed by the rounded apex
’ Unit of air-kerma strength (pGy - rn’. hr-‘); 7.23 U/mg Ra Eq.
1120
I. J. Radiation Oncology 0 Biology 0 Physics FINAL MIRALVA SHAPE AND DIMENSIONS
______ _11 _-___ Ls -_‘ti
1__-i__~ --_-__ _
--___--__+ _c__7_~
‘Nri _-_ u - _ _--__-
j
-n - -__-__ - ________
Fig. 3. The dimensions of the final Miralva design are shown along the midsagittal (upper view) and midcoronal (lower view) applicator planes. The dimensions are shown with and without the enlarging caps and sleeves. A 2 cm inactive spacer in the tandem channel is positioned between the ovoid sources to achieve the source spacing indicated in the upper view. This spacer also maintains a 1 cm space between the intrauterine tandem sources and the ovoid source plane.
design. The third experiment verified that symmetric source positioning with respect to the applicator surfaces had been achieved. Only measured data was obtained for that experiment. In the fourth experiment the ovoid source loadings in the apex were reduced to 108.5 U and the apex surface was built up to achieve 120 cGy/hr. In the final experiment the 144.6 U ovoid source loading was resumed, but now with thicker overlying plastic to obtain the dose distribution shown. In the second row of Figure 4 the average agreement between measured and computed
Volume 22, Number 5, 1992
dose results ranged from 0.90 f 0.10 to 0.98 + 0.08. This agreement was considered to be good because dosimetry in the sharp dose gradients adjacent to the applicator surface is very difficult due to dosimeter size (1 mm diameter and 6 mm length) and the uncertainties of dose point localization on orthogonal radiographs. An approximate estimate of dose variation due to dosimeter size could be calculated using the inverse square law from the closest source to a point on the surface 1.25 cm away. A range of *7% would be expected across the 1 mm diameter. The final dose experiment with 144.6 U cesium tubes in the ovoid positions achieved a surface dose rate of 120 cGy/hr at the top of the ovoid section and the overall average for the 52 measurement points was 109.6 + 9.6 cGy/hr. The third row of Figure 4 shows the dose uniformity to the vaginal apex. This region extends from the cephalad applicator surface to the plane formed by the apex sources. The dots indicate the point locations on the lateral, anterior, posterior and cephalad surfaces and the numbers shown are the average doses on the five surface regions. With each modification to the vaginal apex, the dosimetry evaluation was pursued in greater detail. The fourth row of Figure 4 shows the mean dose rate for all the vaginal apex measurement points.
ANTERIOR cGylhr #
0
??
-20
T3cm
30 40 2
EXPERIMENTAL
80
RESULTS
cGy/hr _
Fig. 4. The results of the surface dosimetry studies from the initial design (left) to the final design (right) are summarized. The top row indicates the source loading. The 2nd row compares the TLD results to the dose results computed using the treatment planning system for all surface dose points. The 3rd row pertains to the measured results at the vaginal apex, where all design modifications were made. The dots indicate measurement points and the numbers are the average doses on the anterior, posterior, lateral and cephalad surfaces. The mean vaginal apex dose is shown in the 4th row.
v
0
?? /
80
20
POSTERIOR Fig. 5. Point doses were measured in the plane of the ovoid sources as indicated by the solid dots. Isodose lines were interpolated from these data to evaluate the dose rate distribution in the transverse plane. These data supported the decision not to design internal shielding.
Dose specification 0 E. D. SLESSINGERet al.
The dosimetry results in the polystyrene phantom at 1 cm from the Miralva surface the doses were on average 45% of the corresponding surface points. The measured dose rate evaluation at levels above and below the vaginal apex sources is shown in Figure 5. From these results it was determined that at 5 mm and 10 mm from the midsag&al applicator surface the dose rates are 60% and 4 1% of the average surface dose rates, respectively. Because the dose rates in the midsagittal plane were on average 5% higher than those 1 cm off the midplane, the indication for internal shields to protect the bladder and rectum was not significant, especially since this applicator is required to deliver a uniform surface dose distribution. Treatment dose specljication The system for Miralva treatment dose specification uses five reference regions denoted by the Greek letters (Y,0, y, A, and Z shown in Figure 6. (Yrepresents the average of 19 surface points sampled along the midcoronal and midsagittal planes from the level of the ovoid sources to the cephalad aspect of the ovoid section. The other regions correspond to the average surface dose of four equally spaced circumferential points centered about each cylinder source. Table 1 shows the cGy/U-hr from each Miralva source location to each reference region. The upper section of Table 1 applies to the basic Miralva configuration while the lower section accounts for the Miralva with the 3 mm thick ovoid caps and 2.5 mm thick cylindrical sleeves. The dose rate to each reference region is calculated by summing the products of the air-kerma strength of each source with the corresponding cGy/U-hr values. The dose variation for each reference region is within + 10% except when the intrauterine tandem is used. The intrauterine tandem increases the dose gradient across the cephalad ovoid surface. Dosimetric comparison with sequential treatment The results of the dose comparisons between the Miralva treatment and the comparable sequential treatment are shown in Figure 7 (surface profiles) and Figure 8 (5 mm depth profiles). The Miralva surface dose profiles follow the treatment prescription the closest. The Miralva reference doses are between -3% and + 14% of the prescribed value, while doses from the sequential treatments range from +-25% of the prescribed dose. The dose gradients of the sequential treatments are sharper than the
Fig. 6. A set of reference regions were selected to facilitate surface dose specification and treatment prescription. Greek letters were chosen to avoid confusion with other gynecological brachytherapy reference points. The numbers specify the cesium source positions.
II21
Table 1. Miralva surface dose specification data cGy/U-hr Source
CY
P
Y
A
z
2.5 cm cylinder diameter/4.5 cm lateral diameter of ovoid portion 1 2 3 4 5 6 7 8 9
0.361 0.361 0.116 0.042 0.021 0.011 0.029 0.067 0.25 1
0.162 0.162 0.559 0.194 0.058 0.025 0.013 0.025 0.050
0.056 0.056 0.199 0.63 1 0.199 0.058 0.009 0.013 0.023
0.027 0.027 0.057 0.199 0.63 1 0.199 0.006 0.009 0.012
0.013 0.013 0.025 0.057 0.199 0.63 1 0.004 0.006 0.009
3.0 cm cylinder diameter/5.1 cm lateral diameter of ovoid portion
1 2 3 4 5 6 7 8 9
0.256 0.256 0.096 0.039 0.019 0.011 0.030 0.075 0.287
0.148 0.148 0.395 0.168 0.056 0.025 0.013 0.023 0.050
0.052 0.052 0.173 0.440 0.173 0.056 0.009 0.013 0.023
0.025 0.025 0.056 0.173 0.440 0.174 0.006 0.009 0.012
0.013 0.013 0.025 0.056 0.173 0.440 0.004 0.006 0.009
Miralva dose gradient. Along the midcoronal surface (Figure 7a) the non-retracted sequential model exceeds 7000 cGy where doses from the colpostat sources and the top cylinder source are substantial. If the 2 cm spacer had not been used at the top of the cylinders the maximum dose would be 9500 cGy. The retracted vagina model does not reach as a high a dose in this upper region because the retracted tissues have migrated toward the vaginal apex, away from the top cylinder source. In addition, the surface dose 4 to 6 cm from the vaginal apex is much lower than the prescribed dose. The midsagittal surface dose profiles (Fig. 7b) again demonstrate abrupt dose gradients with the sequential treatment schemes. The dose drops ranidly from 6000 cGy to 4500 and 4000 cGy for the nonretracted and retracted models, respectively. The sharp dose reductions are due to the lower doses on the anterior and posterior colpostat surfaces relative to the cephalaQ1and lateral colpostat surfaces. Use of the Miralva device leads to a more uniform distribution, and gives results closer to the dose prescription. The retracted vagina model shows the underdose 2 to 6 cm from the vaginal apex similar to that seen along the midcoronal surface. In P’igure 8 the dose profiles 5 mm away from the applicator surfaces have similar features to those seen on the applicator surfaces, but the dose variations are reduced. DISCUSSION The Miralva applicator provides the clinician with an important treatment device. It is similar in some respects
1122
I. J. Radiation Oncology 0 Biology 0 Physics MIDCORONAL VAGINA SURFACE DOSE PROFILES
3000-
- -
Trealmsnt
-
Miralva
e*.**
Ovoidr and Cylinder with Retracted VaOino
-UPPER
The Bloedorn applicator (4) with its two ovoid sources was influential in the Miralva design. All of these applicators were designed to treat the entire vaginal surface. The Miralva applicator uniquely combines several important design features: the option to use an intrauterine tandem, the bulbous ovoid section that distends the vaginal vault, surface contours aligned to the isodose distribution, and enlarging ovoid caps and cylinder sleeves that can be used independently. In addition, the dose distribution can be easily modified under the guidance of the Miralva reference point dose rate tables that are functionally similar to dose rate tables for other vaginal applicators (6). The clinician can quickly preplan the Miralva treatment by multiplying the proposed source activities by the cor-
Prascrmtion
144.6-144.6.
362-72.3-72.3
$
LOWER
Volume 22, Number 5, 1992
\ MIDCORONAL VAGINA 5mm MUCOSAL DOSE PROFILES cm FROM VAGINAL
APEX
MIDSAGITTAL VAGINA (Anterior] SURFACE DOSE PROFILES
-
Mmlvo (53 hr)
.-.
Ovadr 144.6-144.6 Cylinder Space -72 (44 hr)
.*.
144.6-144.6.
362-72.3-72 (39.5 3-72
-
Miralva (53 hrl
1446-144.6,
*-*
Omds 144.6-144.6 Cylinder Space -72 (44 hr)
362-72.3-72
3
(39.5 hr) 3-72.3-144.6
3
hrl 3-144.6
.aOvoid,
end CVl,ndcr with Retracted Vqma
UPPER
LOWER \
I.
0
I
.,,,,,,,,,,I,,,,,
2
4 6 a cm FROM VAGINAL APEX
IO
MIDSAGITTAL VAGINA (Anterior] Smm MUCOSAL DOSE PROFILES 0
I 2
.
a 4
,
1 6
,
1
a
,
-
( 10
Miralv.,446.1446.362-72
.-.
cm FROM VAGINAL APEX
3-72
3
(53 hrl
Ovoid, 144.6-144.6 (39.5 hrl Cylinder Spots -72 3-72.3-144.6
Fig. 7. Comparison of a Miralva treatment to the two step sequential treatment technique along the midcoronal (a) and midsag&al (b) vaginal surfaces. The prescribed treatment is 6000 cGy to the surface of the vaginal apex and 4000 cGy to the distal vagina surface. The Miralva reference doses ranged between -3% to + 14% of the prescribed dose while doses from the sequential treatment ranged from +25%. The sequential technique results in sharp dose gradient regions. According to this study, retraction of the vaginal apex between the first and second applications in
general causes lower doses to be delivered than the situation where no vaginal retraction occurs. to other improved vaginal applicators. For example, the two source positions in the right and left ovoid sections typify the emphasis on delivering the highest dose to the vaginal apex, where recurrence is usually most probable (2,4). The Delclos domed cylinders (1) use a single small source at the apex of the dome for that purpose. The Wang applicator (9) uses a perpendicular source at the apex.
0
2
4
6
a
10
cm FROM VAGINAL APEX
Fig. 8. Comparison of a Miralva treatment to the two step sequential treatment technique along the midcoronal (a) and midsag&al (b) applicator planes 5 mm away from the applicator surfaces demonstrates similar features to those seen on the applicator surfaces, but the dose variations are reduced.
Dose specification 0 E. D. SLESSINGER efal.
Table 2. Miralva dose rate values
Source configuration (2 cm. Cs tubes) 114.6 1144.6 72.3 144.6
144.6 1144.6 72.3 72.3 144.6
144.6 72.3 72.3 108.5 1108.5 72.3 144.6
144.6 72.3 72.3 108.5 1108.5 72.3 72.3 144.6
Reference regions and dose points
; Y RSD Rectum ; Y A RSD Rectum ; LSD Rectum PT A PT B PT P ; Y A RSD Rectum PT A PT B PT P
Dose rate (no caps) cGy/hr
Dose rate (with caps) cGy/hr
119 115 122 85 76 119 110 105 117 85 76 120 111 122 64 76 64 20 15 120 105 105 118 64 76 65 20 15
87 96 91 57 64 87 92 85 87 57 64 99 92 91 43 65 63 20 15 99 88 85 89 43 65 63 20 15
responding tabulated cGy/U-hr values for each reference point and summing the component dose rates to obtain the actual dose rate for each of the reference regions. This dose specification system is also convenient for treatment prescription and dose reporting. Table 2 gives the reference dose results for some common source configurations. A rectal dose point and the surface dose rate (RSD) are also listed. The rectal point is defined, relative to the applicator, to be 5 mm away from the midsagittal posterior surface at the midlevel of the ovoid sources. The RSD is the dose rate from only one ovoid source to the midpoint on the adjacent lateral surface. RSD has been a standard reference point for Fletcher colpostat and mini ovoid applications. When the intrauterine tandem is used the doses to points A, B, and P (pelvic sidewall 1 cm lateral to point B) are also reported. Additional preplans have been generated and are available for clinician reference. The treatment planning staff computes Miralva dose distributions in the midcoronal and midsagittal applicator planes for verification of the physician’s preplan and also to obtain dosimetry information away from the surface. Because of the fixed geometry of source positions and applicator defined reference points, the spatial coordinates of all source positions are stored in the treatment planning computer system and all reference dose coordinates are tabulated. Only the coordinates of the bladder dose point (posterior
1123
aspect of the Foley catheter balloon in the midsagittal plane) are determined from the orthogonal implant radiographs. The Miralva surface contours in the midcoronal and midsagittal planes have been digitized and stored for graphic display on the corresponding isodose distribution plots. The Bloedorn applicator was described by Delclos (1) as offering a reduction of dose to the bladder and rectum. This may be due to the self absorption of the ovoid sources in the direction toward those sensitive structures. It may also be due to the fact that a two-treatment approach might have been used if the Bloedorn applicator was not available. The two-treatment sequential technique would use separate vault colpostats and cylinder insertions. That approach introduces dose uncertainties due to the use of different applicators at different times and the difficulty of anatomically matching the two source configurations. Different shaped applicators also distend the superficial tissues differently. The issue is complicated further by the possibility of tumor reduction or vaginal retraction after the first application. This has been shown to result in underdosing of the target volume. As a result, doses from the sequential treatments deviate +25% from the prescribed dose in certain regions. The Miralva device has been designed to treat the entire target volume in each application and thereby avoids this large dose deviation. In addition, the reference dose system is used to design the source configuration that best accomplishes the dose prescription. SUMMARY A new gynecological brachytherapy applicator (Miralva) has been developed to treat a wide range of vaginal, cervical, and endometrial cancers in a single treatment. Extensive dosimetric and radiographic evaluations provided important design guidance for the specification and achievement of relatively uniform surface dose distributions. Because the Miralva device is intended to deliver a uniform surface dose, internal shields were not included in the design. However, a 2 cm inactive spacer is always maintained in the tandem between the ovoid sources to prevent excessive doses to the normal midline structures. Dose computations from the treatment planning system are used to design source configurations that achieve the dose distribution that most closely matches the prescription at the Miralva reference regions. A manual system for preplan design has also been developed based on dose rate tables for a set of Miralva reference dose regions. This system is helpful for treatment prescription and dose reporting. A comparison of the previous sequential treatment technique and the one-step Miralva treatment demonstrates that Miralva is superior because a treatment plan can be designed quickly based on the reference dose system, the source positioning is precise with respect to the target volume, and the design of the surface contours follows the shape of the isodose distribution.
1124
I. J. Radiation Oncology 0 Biology0 Physics
Volume 22, Number 5, 1992
REFERENCES 1. Delclos, L.; Fletcher, G. H.; Moore, E. B.; Sampiere, V. A.
2.
3. 4.
5.
6.
Mini-colpostats, dome cylinders, other additions and improvements of the Fletcher-Suit afterloadable system: indications and limitations of their use. Int. J. Radiat. Oncol. Biol. Phys. 6: 1195-1206; 1980. Korpivaara, E. T. A simple new intracavitary applicator for a fractionated course in an outpatient setting. AAMD J. 9(2): 24-26; 1984. Krishnaswamy, V. Dose distributions about 13’Cs sources in tissue. Radiology 105: 181-184; 1972. Perez, C. A.; Korba, A.; Sharma, S. Dosimetric considerations in irradiation of carcinoma of the vagina. Int. J. Radiat. Oncol. Biol. Phys. 2: 639-649; 1977. Perez, C. A.; Slessinger, E. D.; Grigsby, P. W. Design of an afterloading vaginal applicator (MIRALVA). Int. J. Radiat. Oncol. Biol. Phys. 18: 1503-I 508; 1990. Sharma, S. C., Gerbi, B.; Madoc-Jones, H. Dose rates for brachytherapy applicators using Cs- 137 sources. Int. J. Radiat. Oncol. Biol. Phys. 5: 1893-1897; 1979.
7. Stovall, M.; Shalek, R. J. A review of computer techniques for dosimetry and interstitial and intracavitary radiotherapy. Comp. Prog. Biomed. 2: 125-136; 1972. 8. Waggener, R.; Lange, J.; Feldmeier, P.; Eagen, P.; Martin, S. Cs- 137 dosimetry table for asymmetric source. Med. Phys. 16: 305-308; 1989. 9. Wang, C. C. An afterloading applicator for intracavitary vaginal irradiation. Radiology 17: 225; 1975. 10. Weller, M. K.; Slessinger, E. D.; Wong, J. W.; Van Dyke, J.; Leung, P. M. K. A practical method for precise thermoluminescent dosimetry. Treat. Plan. 8: 22-26; 1983. 11. Williamson, J. F. Monte Carlo an analytic calculation of absorbed dose near 137Cs intracavitary sources. Int. J. Radiat. Oncol. Biol. Phys. 15: 227-237; 1988. 12. Young, M. E. J.; Batho, H. F. Dose tables for linear radium sources calculated by an electronic computer. Brit. J. Radiol. 37: 38; 1964.
Inr. J. Rodruriun Onlolo~v Biol. Ph.w. Vol. 22, PP. I I 17-I 124 Printed in the U.S.A. All rights reserved
$5.00 + .oO
copyright 0 1992 Pergamon Press Ltd.
??Technical Innovations and Notes
DOSIMETRY
AND DOSE SPECIFICATION FOR A NEW GYNECOLOGICAL BRACHYTHERAPY APPLICATOR
E. D. SLESSINGER,
M.S.,
C. A. PEREZ,
M.D.,
AND J. F. WILLIAMSON,
P. W. GRIGSBY,
M.D.,
PH.D.
Radiation Oncology Center, Mallinckrodt Institute of Radiology, Washington University School of Medicine, 5 10 South Kingshighway Blvd., St. Louis, MO 63 I 10 A new afterloadable gynecological intracavitary applicator has been designed and is now in use for the treatment of a wide range of vaginal, cervical, and endometrial cancers in a single application, with a dose distribution that more closely matches the prescribed treatment than previous methods. The plexiglass applicator consists of a bulbous section that is inserted up to the vaginal apex and loaded with right and left ovoid sources. The distal portion is cylindrical with a central channel for tandem sources. A straight tandem is used when the target volume extends to the vaginal apex, but an intrauterine tandem can also be used. Extensive dosimetric and radiographic evaluations were performed to guide design refinements and to validate the surface dose predictions of the brachytherapy treatment planning system. The final applicator design delivers 1 lo-120 cCy/hr to the vaginal apex,surface and 95-100 cGy/hr to the distal vaginal surfaces when loaded with a 144.6 U (20 mgRaEq) cesium tube ‘in each ovoid channel and 72.3, 72.3, and 144.6 U (10, 10, and 20 mgRaEq) cesium tubes in the vaginal cylinder channel. A system for treatment dose specification has been established that includes dose tables for manually calculating surface doses to reference points. A dose distribution comparison with a sequential colpostat-vaginal cylinder treatment demonstrates that dose delivery is more precise and uniform with this new applicator. Gynecological
malignancies,
Intracavitary irradiation, Brachytherapy,
Vaginal applicator, Dosimetry, Miralva.
treatments and achieve a uniform dose distribution or even quantify the total doses. Because of the limitations of the existing gynecological intracavitary applicators, a new gynecological brachytherapy applicator was introduced by us last year (5). This applicator can treat a wide range of vaginal, cervical, and endometrial cancer target volumes in a single application. The new applicator has been named MIRALVA, an acronym for the Mallinckrodt Institute of Radiology Afterloadable Vaginal Applicator. The Miralva design has been described in detail in an earlier paper (5). The plexiglass structure (shown in Fig. 1) is cylindrical in the lower f of the vagina and enlarges to a bulbous shape at the vaginal apex. There are three channels for the placement of Cesium- 137 sources. Two of the channels position one tube source each into the right and left ovoid locations in the bulbous part of the applicator. These sources are separated by 2 cm (center to center). The stainless steel tandem channel is concentrically located within the cylindrical portion and extends to the vaginal apex. Longer tandiems can also be used for intrauterine placement. Regardless of which tandem length is used, an inactive 2-cm length spacer in the
INTRODUCTION
The treatment of gynecological tumors that involve the vagina is often constrained by the limitations of the currently available applicator systems. This is due in part to the variability of the anatomic structures included in the target volume. An endocervical tumor with extension to the vagina requires an intrauterine tandem and vaginal ovoids to treat the paracervical region. If treatment to the middle or lower thirds of the vagina is also indicated, then a second application to treat the vagina is warranted. Vaginal cancer in the upper and middle vagina or an extensive endometrial cancer recurrence presents similar problems requiring two applications. For example, Fletcher colpostats may be used to treat the vaginal apex in the first procedure and Delclos cylinders in the second procedure to treat the distal vagina. It is undesirable to treat different portions of the target volume in a sequential approach. If the entire target volume is not encompassed during each treatment, disease in some locations may proliferate while being sterilized at other locations. It is also very difficult to match the two
Presented at the 32nd Meeting of the American Society for Therapeutic Radiology and Oncology, Miami, FL, 15- 19 October 1990.
Reprint requests to: Eric D. Slessinger. Accepted for publication 5 September 199 1. 1117
1. J. Radiation Oncology 0 Biology 0 Physics
Fig. I. The plexiglass Miralva applicator consists of a bulbous section that is inserted up to the vaginal apex and loaded with right and left ovoid sources. The distal portion is cylindrical with a central channel for tandem sources. A straight tandem is used when the target volume extends to the vaginal apex, but an intrauterine tandem can also be used as in the above configuration. Additional ovoid caps and cylindrical sleeves enable adjustment ofthe outer dimensions to maintain surface contact.
tandem, centered between the ovoid sources is always maintained. This is a safeguard to prevent an excessive dose due to intersecting ovoid and tandem sources. The outer dimensions and shape were designed to deliver a uniform dose distribution to the entire vaginal mucosa. Additional ovoid caps and cylindrical sleeves are also available for adjustment of the outer dimensions to maintain surface contact. The development of the Miralva device was an iterative process that began with certain clinical design criteria (5). In this paper, we present the results of dosimetric and radiographic studies that were performed to evaluate the dose distributions on the applicator surface. Adjustments to the applicator surfaces were made on the basis of these studies. The dosimetric study was also used to verify that the dose predictions of the treatment planning system were in good agreement with the measured doses. The planning system was used to develop a dose specification system and optimize source configurations that will deliver dose distributions according to certain dosimetric criteria. Also, a dose distribution comparison between the two step treatment approach and the Miralva approach is presented.
METHODS
AND MATERIALS
Dosimetric measurements and comparison with calculations The dosimetry evaluations involved the comparison of dose measurements using Lithium Fluoride TLD- 100 rods* and dose computations using the Modulex+ linear * Harshaw/Filtrol Partnership, 6801 Cochran Road, Solon, OH 44 139. + Computerized Medical Systems, Inc., 56 Worthington Drive, Maryland Heights, MO 63043.
Volume 22, Number 5, I992
source implant algorithm. This algorithm is based on the computational models of Batho and Young (12), which calculates dose factors on a reduced co-ordinate grid by evaluating a Sievert Integral. Tissue attenuation and buildup corrections are made using the third order polynomial developed by Shalek and Stovall (7). The effort was primarily directed to surface dosimetry but more distant locations were also analyzed. The TLD methodology of these studies utilizes TLD rods (1 mm diameter and 6mm long) that are organized into batches of 25. The relative sensitivity of each dosimeter was determined by irradiating all dosimeters to a uniform dose. A normalization factor (Ni) relates the response of the ith dosimeter to the average batch response. Each dosimeter is stored in an assigned location in the batch container. For dose determinations in experiments, all TL readings are corrected by multiplying with the corresponding normalization factors. Four to six dosimeters from each batch are used for calibration to establish a linear regression from the known doses and the corrected TL readings. Only one dosimeter is required for each calibration dose point because the normalization factor for each dosimeter has been predetermined. It has been shown that this approach to TL dosimetry can achieve an average precision of f 1.5% ( 10). A special lucite phantom was used to calibrate the TLD rods for the Miralva dose experiments. Lucite was used because it was readily available, and could be machined for precise positioning of a vertical cesium tubes (2 cm total length, 1.4 cm active length) and an array of vertical TLD rods 1 to 5 cm away from the perpendicular bisector of the source. Several different calibration doses were delivered simultaneously. The calibration doses were calculated as though the irradiation had been performed in water. Krishnaswamy Gamma Dose tables (3) were used for this purpose. The cesium dose data of Williamson (11) or Waggener et al. (8) could also have been used for this purpose. Although the energy dependence of the TL response was not evaluated explicitly, the dose calibrations incorporated the effects of the variations of energy spectra at the different depths. For all of the Miralva dose experiments the average agreement between the calibration doses calculated using the Krishnaswamy tables and the calibration doses determined from the linear regression was 3.2 + 2.6%. A 25 cm cubic water phantom was used for most of the dose evaluation experiments. A straight tandem that extended to the cephalad applicator surface as illustrated in Figure 2 was used in these experiments. The TLD rods on the Miralva were sealed in polyethylene wrap, numbered, and taped to multiple surface locations. The number of surface dose measurement points ranged from I8 * Medical Products Division/3M, MN 55112.
TCAAP 590, New Brighton,
Dose specification 0 E. D. SLEWNGER
et al.
1119
to 52. The number of measurement
points increased with each subsequent experiment to evaluate in greater detail the dose uniformity at the cephalad surface. The applicator was sealed with additional plastic wrap to prevent water leakage. With the loading end left open, the Miralva was then suspended in the water phantom and afterloaded. The sources were left in place for 1 hour. Source transfer time was approximately 30 seconds. Orthogonal radiographs of the Miralva, loaded with dummy cesium tubes and with solder wire fixed to the applicator surfaces, provided the localization information for the point dose computations. Dose measurements were also obtained with the applicator placed within a polystyrene phantom for dosimetric evaluations at points 1 cm away from the applicator surface and at multiple levels above and below the vaginal apex. Treatment dose specification To develop a system for Miralva treatment dose specification, the applicator surface was subdivided into reference dose regions. For each region, a series of points was selected on the midcoronal and midsagittal applicator surfaces. Using the Modulex” RTP system, the dose rate (cGy/hr) contribution from each source to each dose point was computed. The average dose rate contribution from each source to each region was calculated and tabulated. Dosimetric comparison with sequential treatment A dose analysis was undertaken to compare the sequential treatment scheme to the Miralva treatment. A treatment prescription of 6000 cGy to the surface of the vaginal apex and 4000 cGy to the distal vagina surface was selected for this comparison. The sequential treatment chosen for this study is representative of the approach that had been used at Mallinckrodt prior to the clinical introduction of the Miralva. The initial treatment in this study used 2 cm diameter ovoids, each loaded with 144.6 I-J”(20 mgRaEq) to deliver a surface dose from each source of 5000 cGy in 39.5 hr. This was followed by a second application using 2.5 cm diameter Delclos cylinders with a 2 cm spacer at the cephalad aspect and 6 cm active length configured as a 72.3, 72.3 and 144.6 U (cephalocaudal) source sequence to deliver 4000 cGy to the cylinder surface at the midpoint of the active length in 44.4 hr. The Miralva treatment would use 144.6 U sources in the ovoid channels and 36.2, 72.3 and 72.3 U loading in the tandem channel for 53 hr. Dose distributions were computed for each treatment and point doses were interpolated at 5 mm intervals along the midcoronal and midsagittal applicator surfaces starting at the center of the vaginal apex and moving distally from there. Corresponding dose data were also obtained at a depth of 5 mm from the applicator surfaces. For the sequential
‘IComputerized Medical Systems, Inc., 56 Worthington Drive, Maryland Heights, MO 63043.
Fig. 2. The original Miralva design (broken line) was modified by rounding the ovoid section (dotted line) to improve the ease of insertion. The ovoid section was enlarged for the final design (solid line) to achieve a dose rate of 120 cGy/hr to the vaginal apex surface with 144.6 U (20 mgRaEq) ceslum tubes loaded into the two ovoid channels.
treatment, two models were considered: one without vaginal retraction before the second application and the other model assumed that the 2 cm length of vaginal surface distal to the vaginal apex center was reduced to 1 cm at the time of the second application causing the tissues of the apex to recede away from the high dose region of the second application.
RESULTS Dosimetric measurements and comparisons with calculations Figure 2 illustrates how the Miralva design evolved. All of the modifications involved the sham and size of the ovoid section. Source spacing was never altered. The initial ovoid section was too rectangular. To improve the ease of insertion, the ovoid section was rounded as much as possible. Subsequently, the round shape was enlarged to achieve the surface dose rate of 1 lo- 120 cGy/hr to the vaginal apex when the ovoid positions were loaded with 144.6 U sources. The lateral separation was increased from 4.0 cm to 4.5 cm and the cephalad surface was increased in thickness from 1.O cm to 1.4 cm over the ovoid source axes, as shown in Figure 3. Figure 4 summarizes the dosimetry results that span the period from the initial to final Miralva stages. The results of the initial design appear on the left side. This is followed by the rounded apex
’ Unit of air-kerma strength (pGy - rn’. hr-‘); 7.23 U/mg Ra Eq.
1120
I. J. Radiation Oncology 0 Biology 0 Physics FINAL MIRALVA SHAPE AND DIMENSIONS
______ _11 _-___ Ls -_‘ti
1__-i__~ --_-__ _
--___--__+ _c__7_~
‘Nri _-_ u - _ _--__-
j
-n - -__-__ - ________
Fig. 3. The dimensions of the final Miralva design are shown along the midsagittal (upper view) and midcoronal (lower view) applicator planes. The dimensions are shown with and without the enlarging caps and sleeves. A 2 cm inactive spacer in the tandem channel is positioned between the ovoid sources to achieve the source spacing indicated in the upper view. This spacer also maintains a 1 cm space between the intrauterine tandem sources and the ovoid source plane.
design. The third experiment verified that symmetric source positioning with respect to the applicator surfaces had been achieved. Only measured data was obtained for that experiment. In the fourth experiment the ovoid source loadings in the apex were reduced to 108.5 U and the apex surface was built up to achieve 120 cGy/hr. In the final experiment the 144.6 U ovoid source loading was resumed, but now with thicker overlying plastic to obtain the dose distribution shown. In the second row of Figure 4 the average agreement between measured and computed
Volume 22, Number 5, 1992
dose results ranged from 0.90 f 0.10 to 0.98 + 0.08. This agreement was considered to be good because dosimetry in the sharp dose gradients adjacent to the applicator surface is very difficult due to dosimeter size (1 mm diameter and 6 mm length) and the uncertainties of dose point localization on orthogonal radiographs. An approximate estimate of dose variation due to dosimeter size could be calculated using the inverse square law from the closest source to a point on the surface 1.25 cm away. A range of *7% would be expected across the 1 mm diameter. The final dose experiment with 144.6 U cesium tubes in the ovoid positions achieved a surface dose rate of 120 cGy/hr at the top of the ovoid section and the overall average for the 52 measurement points was 109.6 + 9.6 cGy/hr. The third row of Figure 4 shows the dose uniformity to the vaginal apex. This region extends from the cephalad applicator surface to the plane formed by the apex sources. The dots indicate the point locations on the lateral, anterior, posterior and cephalad surfaces and the numbers shown are the average doses on the five surface regions. With each modification to the vaginal apex, the dosimetry evaluation was pursued in greater detail. The fourth row of Figure 4 shows the mean dose rate for all the vaginal apex measurement points.
ANTERIOR cGylhr #
0
??
-20
T3cm
30 40 2
EXPERIMENTAL
80
RESULTS
cGy/hr _
Fig. 4. The results of the surface dosimetry studies from the initial design (left) to the final design (right) are summarized. The top row indicates the source loading. The 2nd row compares the TLD results to the dose results computed using the treatment planning system for all surface dose points. The 3rd row pertains to the measured results at the vaginal apex, where all design modifications were made. The dots indicate measurement points and the numbers are the average doses on the anterior, posterior, lateral and cephalad surfaces. The mean vaginal apex dose is shown in the 4th row.
v
0
?? /
80
20
POSTERIOR Fig. 5. Point doses were measured in the plane of the ovoid sources as indicated by the solid dots. Isodose lines were interpolated from these data to evaluate the dose rate distribution in the transverse plane. These data supported the decision not to design internal shielding.
Dose specification 0 E. D. SLESSINGERet al.
The dosimetry results in the polystyrene phantom at 1 cm from the Miralva surface the doses were on average 45% of the corresponding surface points. The measured dose rate evaluation at levels above and below the vaginal apex sources is shown in Figure 5. From these results it was determined that at 5 mm and 10 mm from the midsag&al applicator surface the dose rates are 60% and 4 1% of the average surface dose rates, respectively. Because the dose rates in the midsagittal plane were on average 5% higher than those 1 cm off the midplane, the indication for internal shields to protect the bladder and rectum was not significant, especially since this applicator is required to deliver a uniform surface dose distribution. Treatment dose specljication The system for Miralva treatment dose specification uses five reference regions denoted by the Greek letters (Y,0, y, A, and Z shown in Figure 6. (Yrepresents the average of 19 surface points sampled along the midcoronal and midsagittal planes from the level of the ovoid sources to the cephalad aspect of the ovoid section. The other regions correspond to the average surface dose of four equally spaced circumferential points centered about each cylinder source. Table 1 shows the cGy/U-hr from each Miralva source location to each reference region. The upper section of Table 1 applies to the basic Miralva configuration while the lower section accounts for the Miralva with the 3 mm thick ovoid caps and 2.5 mm thick cylindrical sleeves. The dose rate to each reference region is calculated by summing the products of the air-kerma strength of each source with the corresponding cGy/U-hr values. The dose variation for each reference region is within + 10% except when the intrauterine tandem is used. The intrauterine tandem increases the dose gradient across the cephalad ovoid surface. Dosimetric comparison with sequential treatment The results of the dose comparisons between the Miralva treatment and the comparable sequential treatment are shown in Figure 7 (surface profiles) and Figure 8 (5 mm depth profiles). The Miralva surface dose profiles follow the treatment prescription the closest. The Miralva reference doses are between -3% and + 14% of the prescribed value, while doses from the sequential treatments range from +-25% of the prescribed dose. The dose gradients of the sequential treatments are sharper than the
Fig. 6. A set of reference regions were selected to facilitate surface dose specification and treatment prescription. Greek letters were chosen to avoid confusion with other gynecological brachytherapy reference points. The numbers specify the cesium source positions.
II21
Table 1. Miralva surface dose specification data cGy/U-hr Source
CY
P
Y
A
z
2.5 cm cylinder diameter/4.5 cm lateral diameter of ovoid portion 1 2 3 4 5 6 7 8 9
0.361 0.361 0.116 0.042 0.021 0.011 0.029 0.067 0.25 1
0.162 0.162 0.559 0.194 0.058 0.025 0.013 0.025 0.050
0.056 0.056 0.199 0.63 1 0.199 0.058 0.009 0.013 0.023
0.027 0.027 0.057 0.199 0.63 1 0.199 0.006 0.009 0.012
0.013 0.013 0.025 0.057 0.199 0.63 1 0.004 0.006 0.009
3.0 cm cylinder diameter/5.1 cm lateral diameter of ovoid portion
1 2 3 4 5 6 7 8 9
0.256 0.256 0.096 0.039 0.019 0.011 0.030 0.075 0.287
0.148 0.148 0.395 0.168 0.056 0.025 0.013 0.023 0.050
0.052 0.052 0.173 0.440 0.173 0.056 0.009 0.013 0.023
0.025 0.025 0.056 0.173 0.440 0.174 0.006 0.009 0.012
0.013 0.013 0.025 0.056 0.173 0.440 0.004 0.006 0.009
Miralva dose gradient. Along the midcoronal surface (Figure 7a) the non-retracted sequential model exceeds 7000 cGy where doses from the colpostat sources and the top cylinder source are substantial. If the 2 cm spacer had not been used at the top of the cylinders the maximum dose would be 9500 cGy. The retracted vagina model does not reach as a high a dose in this upper region because the retracted tissues have migrated toward the vaginal apex, away from the top cylinder source. In addition, the surface dose 4 to 6 cm from the vaginal apex is much lower than the prescribed dose. The midsagittal surface dose profiles (Fig. 7b) again demonstrate abrupt dose gradients with the sequential treatment schemes. The dose drops ranidly from 6000 cGy to 4500 and 4000 cGy for the nonretracted and retracted models, respectively. The sharp dose reductions are due to the lower doses on the anterior and posterior colpostat surfaces relative to the cephalaQ1and lateral colpostat surfaces. Use of the Miralva device leads to a more uniform distribution, and gives results closer to the dose prescription. The retracted vagina model shows the underdose 2 to 6 cm from the vaginal apex similar to that seen along the midcoronal surface. In P’igure 8 the dose profiles 5 mm away from the applicator surfaces have similar features to those seen on the applicator surfaces, but the dose variations are reduced. DISCUSSION The Miralva applicator provides the clinician with an important treatment device. It is similar in some respects
1122
I. J. Radiation Oncology 0 Biology 0 Physics MIDCORONAL VAGINA SURFACE DOSE PROFILES
3000-
- -
Trealmsnt
-
Miralva
e*.**
Ovoidr and Cylinder with Retracted VaOino
-UPPER
The Bloedorn applicator (4) with its two ovoid sources was influential in the Miralva design. All of these applicators were designed to treat the entire vaginal surface. The Miralva applicator uniquely combines several important design features: the option to use an intrauterine tandem, the bulbous ovoid section that distends the vaginal vault, surface contours aligned to the isodose distribution, and enlarging ovoid caps and cylinder sleeves that can be used independently. In addition, the dose distribution can be easily modified under the guidance of the Miralva reference point dose rate tables that are functionally similar to dose rate tables for other vaginal applicators (6). The clinician can quickly preplan the Miralva treatment by multiplying the proposed source activities by the cor-
Prascrmtion
144.6-144.6.
362-72.3-72.3
$
LOWER
Volume 22, Number 5, 1992
\ MIDCORONAL VAGINA 5mm MUCOSAL DOSE PROFILES cm FROM VAGINAL
APEX
MIDSAGITTAL VAGINA (Anterior] SURFACE DOSE PROFILES
-
Mmlvo (53 hr)
.-.
Ovadr 144.6-144.6 Cylinder Space -72 (44 hr)
.*.
144.6-144.6.
362-72.3-72 (39.5 3-72
-
Miralva (53 hrl
1446-144.6,
*-*
Omds 144.6-144.6 Cylinder Space -72 (44 hr)
362-72.3-72
3
(39.5 hr) 3-72.3-144.6
3
hrl 3-144.6
.aOvoid,
end CVl,ndcr with Retracted Vqma
UPPER
LOWER \
I.
0
I
.,,,,,,,,,,I,,,,,
2
4 6 a cm FROM VAGINAL APEX
IO
MIDSAGITTAL VAGINA (Anterior] Smm MUCOSAL DOSE PROFILES 0
I 2
.
a 4
,
1 6
,
1
a
,
-
( 10
Miralv.,446.1446.362-72
.-.
cm FROM VAGINAL APEX
3-72
3
(53 hrl
Ovoid, 144.6-144.6 (39.5 hrl Cylinder Spots -72 3-72.3-144.6
Fig. 7. Comparison of a Miralva treatment to the two step sequential treatment technique along the midcoronal (a) and midsag&al (b) vaginal surfaces. The prescribed treatment is 6000 cGy to the surface of the vaginal apex and 4000 cGy to the distal vagina surface. The Miralva reference doses ranged between -3% to + 14% of the prescribed dose while doses from the sequential treatment ranged from +25%. The sequential technique results in sharp dose gradient regions. According to this study, retraction of the vaginal apex between the first and second applications in
general causes lower doses to be delivered than the situation where no vaginal retraction occurs. to other improved vaginal applicators. For example, the two source positions in the right and left ovoid sections typify the emphasis on delivering the highest dose to the vaginal apex, where recurrence is usually most probable (2,4). The Delclos domed cylinders (1) use a single small source at the apex of the dome for that purpose. The Wang applicator (9) uses a perpendicular source at the apex.
0
2
4
6
a
10
cm FROM VAGINAL APEX
Fig. 8. Comparison of a Miralva treatment to the two step sequential treatment technique along the midcoronal (a) and midsag&al (b) applicator planes 5 mm away from the applicator surfaces demonstrates similar features to those seen on the applicator surfaces, but the dose variations are reduced.
Dose specification 0 E. D. SLESSINGER efal.
Table 2. Miralva dose rate values
Source configuration (2 cm. Cs tubes) 114.6 1144.6 72.3 144.6
144.6 1144.6 72.3 72.3 144.6
144.6 72.3 72.3 108.5 1108.5 72.3 144.6
144.6 72.3 72.3 108.5 1108.5 72.3 72.3 144.6
Reference regions and dose points
; Y RSD Rectum ; Y A RSD Rectum ; LSD Rectum PT A PT B PT P ; Y A RSD Rectum PT A PT B PT P
Dose rate (no caps) cGy/hr
Dose rate (with caps) cGy/hr
119 115 122 85 76 119 110 105 117 85 76 120 111 122 64 76 64 20 15 120 105 105 118 64 76 65 20 15
87 96 91 57 64 87 92 85 87 57 64 99 92 91 43 65 63 20 15 99 88 85 89 43 65 63 20 15
responding tabulated cGy/U-hr values for each reference point and summing the component dose rates to obtain the actual dose rate for each of the reference regions. This dose specification system is also convenient for treatment prescription and dose reporting. Table 2 gives the reference dose results for some common source configurations. A rectal dose point and the surface dose rate (RSD) are also listed. The rectal point is defined, relative to the applicator, to be 5 mm away from the midsagittal posterior surface at the midlevel of the ovoid sources. The RSD is the dose rate from only one ovoid source to the midpoint on the adjacent lateral surface. RSD has been a standard reference point for Fletcher colpostat and mini ovoid applications. When the intrauterine tandem is used the doses to points A, B, and P (pelvic sidewall 1 cm lateral to point B) are also reported. Additional preplans have been generated and are available for clinician reference. The treatment planning staff computes Miralva dose distributions in the midcoronal and midsagittal applicator planes for verification of the physician’s preplan and also to obtain dosimetry information away from the surface. Because of the fixed geometry of source positions and applicator defined reference points, the spatial coordinates of all source positions are stored in the treatment planning computer system and all reference dose coordinates are tabulated. Only the coordinates of the bladder dose point (posterior
1123
aspect of the Foley catheter balloon in the midsagittal plane) are determined from the orthogonal implant radiographs. The Miralva surface contours in the midcoronal and midsagittal planes have been digitized and stored for graphic display on the corresponding isodose distribution plots. The Bloedorn applicator was described by Delclos (1) as offering a reduction of dose to the bladder and rectum. This may be due to the self absorption of the ovoid sources in the direction toward those sensitive structures. It may also be due to the fact that a two-treatment approach might have been used if the Bloedorn applicator was not available. The two-treatment sequential technique would use separate vault colpostats and cylinder insertions. That approach introduces dose uncertainties due to the use of different applicators at different times and the difficulty of anatomically matching the two source configurations. Different shaped applicators also distend the superficial tissues differently. The issue is complicated further by the possibility of tumor reduction or vaginal retraction after the first application. This has been shown to result in underdosing of the target volume. As a result, doses from the sequential treatments deviate +25% from the prescribed dose in certain regions. The Miralva device has been designed to treat the entire target volume in each application and thereby avoids this large dose deviation. In addition, the reference dose system is used to design the source configuration that best accomplishes the dose prescription. SUMMARY A new gynecological brachytherapy applicator (Miralva) has been developed to treat a wide range of vaginal, cervical, and endometrial cancers in a single treatment. Extensive dosimetric and radiographic evaluations provided important design guidance for the specification and achievement of relatively uniform surface dose distributions. Because the Miralva device is intended to deliver a uniform surface dose, internal shields were not included in the design. However, a 2 cm inactive spacer is always maintained in the tandem between the ovoid sources to prevent excessive doses to the normal midline structures. Dose computations from the treatment planning system are used to design source configurations that achieve the dose distribution that most closely matches the prescription at the Miralva reference regions. A manual system for preplan design has also been developed based on dose rate tables for a set of Miralva reference dose regions. This system is helpful for treatment prescription and dose reporting. A comparison of the previous sequential treatment technique and the one-step Miralva treatment demonstrates that Miralva is superior because a treatment plan can be designed quickly based on the reference dose system, the source positioning is precise with respect to the target volume, and the design of the surface contours follows the shape of the isodose distribution.
1124
I. J. Radiation Oncology 0 Biology0 Physics
Volume 22, Number 5, 1992
REFERENCES 1. Delclos, L.; Fletcher, G. H.; Moore, E. B.; Sampiere, V. A.
2.
3. 4.
5.
6.
Mini-colpostats, dome cylinders, other additions and improvements of the Fletcher-Suit afterloadable system: indications and limitations of their use. Int. J. Radiat. Oncol. Biol. Phys. 6: 1195-1206; 1980. Korpivaara, E. T. A simple new intracavitary applicator for a fractionated course in an outpatient setting. AAMD J. 9(2): 24-26; 1984. Krishnaswamy, V. Dose distributions about 13’Cs sources in tissue. Radiology 105: 181-184; 1972. Perez, C. A.; Korba, A.; Sharma, S. Dosimetric considerations in irradiation of carcinoma of the vagina. Int. J. Radiat. Oncol. Biol. Phys. 2: 639-649; 1977. Perez, C. A.; Slessinger, E. D.; Grigsby, P. W. Design of an afterloading vaginal applicator (MIRALVA). Int. J. Radiat. Oncol. Biol. Phys. 18: 1503-I 508; 1990. Sharma, S. C., Gerbi, B.; Madoc-Jones, H. Dose rates for brachytherapy applicators using Cs- 137 sources. Int. J. Radiat. Oncol. Biol. Phys. 5: 1893-1897; 1979.
7. Stovall, M.; Shalek, R. J. A review of computer techniques for dosimetry and interstitial and intracavitary radiotherapy. Comp. Prog. Biomed. 2: 125-136; 1972. 8. Waggener, R.; Lange, J.; Feldmeier, P.; Eagen, P.; Martin, S. Cs- 137 dosimetry table for asymmetric source. Med. Phys. 16: 305-308; 1989. 9. Wang, C. C. An afterloading applicator for intracavitary vaginal irradiation. Radiology 17: 225; 1975. 10. Weller, M. K.; Slessinger, E. D.; Wong, J. W.; Van Dyke, J.; Leung, P. M. K. A practical method for precise thermoluminescent dosimetry. Treat. Plan. 8: 22-26; 1983. 11. Williamson, J. F. Monte Carlo an analytic calculation of absorbed dose near 137Cs intracavitary sources. Int. J. Radiat. Oncol. Biol. Phys. 15: 227-237; 1988. 12. Young, M. E. J.; Batho, H. F. Dose tables for linear radium sources calculated by an electronic computer. Brit. J. Radiol. 37: 38; 1964.
