Inr. J. Radiarion Oncology Biol. Phys Vol Punted m the IL.5 A. All nghts reserved

22. pp. 181-189 Copyright

??Technical Innovations

and Notes

COMPUTED TOMOGRAPHY IR-192 BRACHYTHERAPY JOHN K.

0360.3016/92 $5.00 + .W 0 1991 Pergamon Press plc

TREATMENT PLANNING IN IN THE HEAD AND NECK

HAYES, M.D. ,* JOHN H. MOELLER, M.S. R. KIM DAVIS, M.D.

,* DENNIS D. LEAVITT, PHD. ,*

,_t AND H. RIG HARNSBERGER, M.D.$

University of Utah Medical Center, Salt Lake City, UT 84132 Brachytherapy dose prescription and treatment planning lag behind the state-of-the-art for external beam therapy. As altered fractionation of external beam therapy improves patient outcome in head and neck cancer, there will be an increased need to compare the two radiotherapy techniques. Currently, implant techniques and dose prescription documentation are not uniform, dose prescription to a target volume is subjective, and implant quality is poorly understood and not routinely assessed. All contribute to a lack of scientifically rigorous brachytherapy clinical trials. Studies designed to combine tumor imaging and dosimetry data are important in the evolution of brachytherapy treatment planning. Head and neck implants, which often require nonparallel, arching, or looping source carriers for all but small tumors in order to encompass the target volume adequately, were used to evaluate the clinical utility and feasibility of computed tomography as a treatment planning tool in brachytherapy. Following placement of plastic afterloading tubes under general anesthesia, orthogonal radiographs with dummy sources in the afterloading tubes are obtained as customary for source localization. With the patient in the same position, axial CT scans are obtained with the dummy seeds still in place for treatment planning. The implant physician, using data from the pre-treatment diagnostic CT scan, outlines target areas on sequential images creating a J-dimensional target volume. By superimposing anatomic data with isodose curves one can objectively define implant parameters important in clinical trials analysis. These include minimum target absorbed dose, implant uniformity, and treatment to target volume ratio. The results of the first 10 patients are presented and implications of these data regarding the analysis of implant technique, implant quality, and implant optimization are discussed. The technique as performed is laborious but practicable in the clinical research setting of head and neck implant. Further research efforts should improve, simplify, and objectify brachytherapy and hasten the time when rigorous multi-institutional brachytherapy trials will be reality. Brachytherapy,

Computed

tomography,

Treatment

planning,

INTRODUCTION

Head and neck cancer, Iridium-192.

As hyperfractionated external beam radiation therapy has shown promise over standard fractionation in head and neck squamous cell carcinomas, traditionally accepted dogmas regarding interstitial implants have been questioned (7), and a clear need exists for scientifically rigorous clinical trials. Such trials in brachytherapy are burdened by the fact that uniformity of therapy has been more difficult to attain with implants compared to external beam therapy. This situation exists because implant techniques vary, dose prescription is more subjective, dose analysis more complex, and uniformly distributing sources throughout the target is far more difficult than designing external beam treatment ports. Ideally, it would be convenient and less costly to omit implant therapy in the head and neck cancer patients in

whom external beam therapy would render equally satisfactory results. However, the importance of local control with optimal functional integrity and minimum morbidity cannot be overemphasized in this disease, and the rationale for brachytherapy to meet these ends remains strong. Therefore, it is increasingly important to implement in the clinic methods to quantify and qualify head and neck brachytherapy treatment delivery more objectively, so that its practice can be optimized and then studied in prospective randomized trials (8). Effort has been made here to implement recommendations of the Interstitial Collaborative Working Group and the Physics Committee of the American Endocurietherapy Society (l-3) in regard to dose evaluation and implant parameters which further these objectives. Brachytherapy treatment planning which combines tumor imaging and dosimetry data, holds promise to make

presented at the 32nd Annual ASTRO Meeting, Miami Beach, FL, 15-20 October, 1990. *Division of Radiation Oncology. TDivision of Otolaryngology.

$Division of Diagnostic Radiology. Reprint requests to: John K. Hayes, M.D. Supported in part by the American Cancer Society Clinical Oncology Career Development Award No. 90-169. 181

182

I. J. Radiation Oncology 0 Biology 0 Physics

Volume 22, Number 1, 1992

Table 1. Clinical parameters Pt. no. I

2 3 4 5 6 7 8 9 10

66 61 58 62 63 58 72 43 62 55

Sex

Stage

TNM

Site

M F M M M M F M M M

IV

T4NOMO T 1NOM0 T4N2BMO T2N2CMO T3NOMO T3N2BMO T4N2AMO T3N2BMO T3NOMO Recurrent

Oral tongue Oral tongue Base tongue Base tongue Base tongue Base tongue Tonsillar fossa Tonsillar fossa Epiglottis Pyriform sinus

implant dose definition more objective and the analysis of implant parameters possible. These research goals were recognized as high priority by a brachytherapy consensus conference on technology (8). With appropriate clinical information, computed tomography (CT) has made it possible to more accurately define the submucosal extent of head and neck neoplasms and is a useful tool in defining the stage of the primary tumor and nodal metastases and, therefore, in treatment planning for radiotherapy. This study was done to assess CT as a tool in defining a clinically useful target volume in head and neck brachytherapy so that dose and implant parameters can be more objectively defined. The study was one of feasibility and the methods used in dose analysis approximate. METHODS

AND MATERIALS

Ten patients with a variety of head and neck squamous carcinomas were the subject of this report (Table 1). Treatment intent was curative in nine patients and palliative in one. Clinical outcome analysis, not the focus of this paper, is premature because of inadequate follow-up. Pre-treatment planning During simulation for external beam treatment, head position is rendered reproducible with a defined head position and directional marks that can be reproducibly aligned with lasers. A contrast enhanced CT scan using 3 or 5 mm slices at 5 mm intervals was done in treatment position using the laser lights in the scanner to reproduce head position. The tumor is delineated on the CT images with the aid of a radiologist with expertise in head and neck imaging taking into account information obtained at physical exam (Fig. la). The highest and lowest cuts with evident tumor are so marked and the information is used in designing external beam portals. Interstitial implants Implants were typically used as a boost following 50 Gy external irradiation (curative cases) using Henschke after-

IV IV III

IV IV IV III IV

loading technique (6, 9, 11, 15, 16, 18) to create a volume implant. The target was defined as the tumor without margin as visualized on pretreatment CT scan (before any external beam irradiation). Targets with margins of 0.5 and 1.0 cm around the tumor were evaluated in terms of implant parameters as discussed below. A dummy ribbon with seeds at 1 cm intervals was used in the operating room to measure the length of each afterloading tube and thereby estimate the number and positioning of seeds to cover the target best. Usually, the implant is directed to the primary site and metastatic disease in the neck is treated with external irradiation followed by neck dissection approximately 3 weeks after the implant, thus simplifying implant dose analysis. In two patients, however, the neck was implanted: in patient no. 7 in whom the upper right neck was implanted because of extensive carotid space invasion by tumor, and in patient no. 4, in whom both sides of the upper neck were boosted with the implant to control bilateral nodes. A tracheostomy was used in six patients and was technically necessary in order to perform the implant in three of these patients whose tumors involved or had extension to the supraglottic larynx and hypopharynx. Tracheostomies are not considered a routine part of the implant procedure since they add to cost and morbidity. Nine patients had afterloading tubes removed in the operating room under general anesthesia for airway control and prevention of patient discomfort and anxiety in the event of brisk bleeding. Brachytherapy treatment planning Following the implant, orthogonal radiographs are obtained for source localization with dummy seeds in ribbon in the afterloading tubes and the same head position as in the diagnostic CT scan (defined headrest and directional marks). Modification of seed position may be needed after visual analysis of these films. The dummy ribbons are taped in position and thin wires or polyethylene tubes filled with x-ray contrast medium are taped on the skin along the projection of the y-axis (craniocaudad dimension) of the source localization films on the right, left, and anterior of

“‘Jr brachytherapy 0 J. K.

(4

HAYES et al.

183

@I

Fig. 1. (a) Diagnostic CT scan, patient no. 8, tumor of the tonsillar fossa and soft palate. Target demarcated with dotted line. (b) Treatment planning CT scan, patient no. 8, same level as Fig. la. Wires on the skin relate to the source localization radiograph axis. Dummy sources in afterloading tubes are visualized as white spots. A nasogastric feeding tube is seen on the posterior pharyngeal wall. Sources in the tongue provide dose to the anterior and medial portion of the target. Sources in palate arch tubes (just medial to the mandible) provide dose to the lateral aspect of the target. Sources in the pharynx provide dose to the posterior part of the target.

the patient. The implant CT scan was done with the same head position using 3 or 5 mm slices at 5 mm spacing across the target so that the images corresponded as far as possible to the diagnostic images. Target areas from the corresponding diagnostic images are transposed to the implant images, the number depending on the craniocaudad dimension of the target (Fig. lb). Despite these efforts at patient positioning, implant films and diagnostic films did not always have the exact appearance (compare Figs. la and lb). This is not surprising for a number of reasons, including set-up uncertainty, patient movement, nutritional changes between the time of the scans, tumor shrinkage, edema, implant distortion of skin marks, and implant distortion of subcutaneous tissue and mucosa. Seed coordinates are digitized from orthogonal films and doses are calculated using a point source calculation model (Capintec-Tektronics treatment planning system with customized display) that includes corrections for radial dose falloff, dose buildup, and absorption in surrounding tissue. Doses to a given calculation point from each seed are

summed to achieve a composite dose. Coarse calculations which can be done with less computer time are done initially to identify hot spots (10) or cold spots within the target volume. These are regions of dose higher or lower than the desired dose rate in the target. Subsequent iterations using seeds of different activity are done in an effort to make the dose more uniform. In other words, areas within the treatment volume which by visual inspection had a dose rate considerably higher than the implant dose rate, and areas within the target which by visual inspection had a dose rate lower than the desired dose rate, were modified using differential loading of seed activity in trial and error calculations. Prior to loading, the activities of seeds received by the supplier are verified in a well ionization chamber. Once verified, higher resolution axial isodose plots are generated corresponding to each implant CT image (Fig. 2a) along with plots in coronal and sagittal planes. The correlation between CT and radiograph is based on the film isocenter which is clearly marked on the patient’s skin and with a small radio-opaque marker and

184

I.

J. Radiation Oncology 0 Biology 0 Physics

Volume 22, Number 1. 1992 15

CT cuts made relative to it. The implant CT images are enlarged by projection to the magnification of the source localization X rays (and isodose plots) and anatomic contours, target, and iridium seed positions are transferred to paper (Fig. 2b).

RIBBONS

WITH

95 Z-X

Ir-192

SEEDS

PLANE

I_

UUMC

Y=1.6 ACTIVITY mg. Fw.Eq TOTAL = 34.265 INDIVIDUAL SOURCES Activity t 0.3310 :

Implant analysis, definitions Pending an International Commission on Radiation Units and Measurements (ICRU) report on dose specification in brachytherapy, the ICRU definitions for dose specification in external beam radiation therapy were paralleled here for brachytherapy (10). Implant parameters for dose uniformity and treatment to target volume ratio were used according to interim recommendations from the Physics Committee of the American Endocurietherapy Society (1, 19). Similar parameters have been used to objectively compare the dosimetry, relative to a defined target, of stereotactic brain implants that vary by geometry and the type of radioisotope used (17).

o”::::o”

(4

Minimum target absorbed dose rate (~>r,,,) (10) is the highest dose rate curve which completely encompasses the target volume or, in other words, the lowest dose rate within the target volume. Choosing DT,min can be difficult when the target edge is near a periphery of the implant where the dose rate changes rapidly within millimeters. Dose specification to a b)7,mrn value in a region of high dose gradient can result in unacceptable overdosage if the bT,min is chosen from an isodose curve on the lower dose side of a steep gradient (14). Furthermore, the clinical significance of a small fraction of the target volume at the periphery of an implant receiving a low dose is unknown. Therefore, an average value was used in analysis of implant parameters instead of the absolute b T,min for the implant. The average bqrnin was calculated as a volume weighted average of the DT,min values at each level through the target, (b)

where Ai is the target area and h the distance between isodose planes. Minimum target absorbed dose (D,+) is rigorously defined as the highest value isodose curve which completely encompasses the target volume. That is, the lowest absorbed dose in the target volume (10). In the present analysis it was defined as is the mathematical product of DT,min and the duration of the implant. Maximum target absorbed dose, DTmM (lo), was not evaluated and will probably require more powerful techniques such as the contiguous volume analysis described by Neblett (14). Median target absorbed dose (DT,,_J is that dose wherein half of the target volume absorbs a dose greater

Fig. 2. (a) Isodose map, patient no. 8, same level as CT scans in Figs. la and lb. (b) Anatomic representation of implant CT scan showing seed locations and target (same level as Fig. 2a). Dose level 1: median target dose at this level. Dose level 2: minimum target dose at this level. Dose level 3: average minimum target dose for the implant.

and the other half a dose lower (10). In the present study, the median target dose rate (br,med) was defined as the dose encompassing a volume equal to 50% of the volume of the target and was calculated as a volume weighted average of the br.med values determined at each level through the target, a calculation similar to that shown above for

I5T.mm' Target volume (VT] (10) was the summation of the target areas on sequential CT cuts multiplied by the separation between them.

“‘Ir brachytherapy 0 J. K. HAYESer a[.

Treatment

volume (V&

(10) was the volume

enclosed

bY DT,rnin. Implant uniformity (UJ was expressed as a volume ratio, the numerator in the equation being the treatment volume less the volume enclosed by a dose that is 1.5 times &.min (1, 19).

u

=

I

vn- v1.5DTmin

185

Table 2. Interstitial implant technical features: 10 patients Technical feature

Mean

Range

No. seed ribbons No. 192-k seeds No. diff. act. seeds Total activity (gBq)

16.7 111 2.1 2.65

626 66-170 l-4 1.81-3.59

V Tr

In an ideal interstitial implant this ratio will approach unity because the volume receiving 150% of DT,min will be small. A related parameter also applied in this study was the minimum to median target dose ratio, b)T,min + b=,,,_,. Treatment to target volume ratio. (V,,+ VT) (1). In assessing the quality of an implant, one important issue is the extent to which the treatment dose extends to normal tissues. In an optimal hypothetical implant this should approach unity. If much greater than unity, an unnecessary volume of normal tissue was irradiated, and if less than unity, the implant volume was too small. Volume-dose analysis. Using a treatment planning light projection table, the CT image information (Fig. 2b without the isodose lines) is overlaid on the isodose plots with orientation made possible by matching isodose axes to skin wires (see Fig. 1b and Fig. 2b) and by matching seed marks on the isodose plot to the corresponding seed images on the scan (see Fig. 2a and 2b). The seeds provide multiple points of reference and usually the surest means of correspondence. When seeds are not visible, skin wires and anatomic relations with adjacent images provide orientation. Anatomic entities such as the spinal cord, mandible, and target should not have unexpected discontinuities when checked image by image. At each level the target area was digitized and its area calculated by computer. b)T.min at each level was objectively visualized. Areas of isodose curves within the target were digitized and computer calculated in order to determine, by extrapolative calculation, the dose whose area enclosed 50% of the target (used to determine b T,med). Critical doses such as to the spinal cord were readily identified. Having calculated target areas for each image, the target volume was then calculated and DT,rnin and fi~.,,,~d values for the implant were calculated as explained above.

RESULTS Implant technical features are listed in Table 2. The mean number of seeds per ribbon was 6.7 and was influenced by the frequent use of loops or arches which use more seeds than straight tubes. Seeds of different activity were used, usually two different activities per implant, in an effort to minimize hot and cold spots within the implant. Our current methods to do this are rudimentary at best and usually delay loading to the first postoperative day. An average of 6.0 planes were analyzed per implant,

between planes varying from 0.5 to 1.0 cm. An important issue pertaining to dose prescription is the relationship between DT,,min as determined by this type of dose analysis and ar,min determined by the usual method of superimposition of source localization radiographs on isodose plots. In the latter approach, anatomic information in three dimensions is compressed into two, and an accurate analysis of target coverage is beyond human abilities. Although this process may not objectively describe the implant in terms of implant parameters, it is important in the delivery of traditionally prescribed doses, which usually derive from the isodose line that best encompasses the volume implanted. In these 10 patients, BT,min determined by CT analysis was less than that determined clinically, the mean ratio of the two being 0.86 with a range of 0.58 to 1.02 (Table 3). As one analyzes a head and neck implant plane by plane, the importance of seed placement for complete geometric coverage of the entire target is readily appreciated. The mean absolute b)T,min value for the 10 implants was 0.33 Gylhr, which was 80% of BT,min and 69% of the i)T,min determined by superimposition of source localization radiographs on isodose plots. Our initial experience indicates that overlaying isodose plots on radiographs is not as sensitive a method as CT analysis for locating points of low dose in the target. The extreme example is patient no. 7 (Table 3). whose target included the tonsillar fossa and extracapsular nodal disease in the neck, in whom a, ,min determined by CT analysis was 0.26 Gy/hr and the b)T,min chosen by the usual method was 0.45 Gy/hr. The delivery of the predetermined dose of 30 Gy prescribed to DT,min would have resulted in an unacceptable overdosage, over 50 Gy to the volume enclosed by the 0.45 Gy/hr isodose line. This indicates that the geometry of the implant was poor (the target extended outside the geometric bounds of the implant), a fact that was not fully appreciated by overlaying isodose plots on radiographs. This is reflected in the parameters for uniformity, U, and Dr,,i,, + DT,med, which are low (Table 3), and illustrates the usefulness of these parameters in qualifying the geometric adequacy of the implant. The mean V, was 22.8 cm3 with a range of 7.0 to 47.5 cm3 (Table 3). The largest V, was in patient no. 7, in whom the target involved extranodal disease in the neck. The mean D,min was 0.41 Gylhr with a range of 0.265 to 0.510 Gyihr. The treatment volume was larger than the target in all cases, from 3.96 to 8.13 times, the mean V,, with spacing

186

I. J. Radiation

Oncology

0 Biology 0 Physics

Volume 22, Number

Table 3. Implant dose specification Dr., min T.P.* Ptno. 1

2 3 4 5 6 7 8 9 10 Mean *Minimum tMinimum

parameters

V,

D 7-Yln,n

D,, min c1in.i

(cm31

(GY)

@y/W

0.91 1.02 0.80 0.73 0.86 0.90 0.58 1.02 0.84 0.96 0.86

20.3 7.0 17.5 30.7 18.3 27.9 47.5 13.2 21.3 24.8 22.8

23.4 25.8 25.6 21.2 26.9 32.5 17.5 29.2 50.7 47.9 30.1

0.48 0.60 0.65 0.58 0.64 0.54 0.44 0.59 0.57 0.64 0.57

target absorbed dose determined target absorbed dose determined

0.36 0.51 0.48 0.36 0.43 0.41 0.26 0.46 0.38 0.48 0.41

b I-( med

by CT implant analysis. by superimposition of source localization

to VT ratio being 6.1. a7,med was on average 1.38 times BT,min. Mean U, for the 10 implants was 0.73 and was the same as a related parameter, the mean ~r,min+~T,med, a coincidence in that their method of calculation is different. Initially, because it was unclear what impact target size would have upon implant parameters and what the most appropriate target size would be, all parameters were calculated for the tumor volume and the tumor with 0.5 and 1.0 cm margins, those margins being somewhat variable with modification due to natural barriers to tumor spread, for example, mandible and air cavities. Margins make a large difference in VT as one would expect, the volume doubling and tripling at 0.5 and 1 .O cm margins (Fig. 3). Relatively, V,, increases less rapidly than VT and therefore, V,,+ VT decreases as the target margin increases (Fig. 3). G,min and DT,,,,~~ both decrease with target margin (Fig. 4), as one would expect, the latter being less target margin dependent. The two uniformity related parameters, U, and DT,min GD~,~_, also decrease with target margin, the latter less rapidly (Fig. 5).

1, 1992

116 51 104 217 134 111 330 77 110 111 136

radiograph

5.7 8.1 6.0 7.1 7.3 4.0 7.2 5.8 5.2 4.5 6.1

0.79 0.83 0.72 0.74 0.63 0.80 0.53 0.73 0.78 0.79 0.73

0.75 0.86 0.74 0.68 0.67 0.76 0.60 0.78 0.74 0.75 0.73

on isodose curves.

brachytherapists are also teletherapists by training, it seems reasonable to try to use the same terms for dose specification in both applications when possible. Terms relating to dose homogeneity which are important in brachytherapy can then be applied to external beam dosimetry as needed. An effort has been made here to more objectively and thoroughly quantitate brachytherapy treatment, without introducing new terminology foreign to external beam therapy. A very important clinical question relating to this study is the accuracy of CT in defining a target. Confidence in tumor definition clearly varies between patient scans, and we are currently studying the question prospectively. Our current hypothesis and clinical impression is that CT adds

200

150

DISCUSSION The language of brachytherapy dose specification has been less than clearly understood and uniformly applied. In head and neck implants, the concept of minimum tumor dose generally meant that dose which conforms approximately to the shape of a seed array without breaking apart within that volume. In practice this cannot be rigorously related to a target within that volume using plane radiographs. The 3-dimensional definition of a target, however, is important for the objective analysis of implant parameters important for studying outcome in carefully controlled clinical trials. Without objective implant parameters, that which constitutes a “good” implant is quantified only as an impression in the mind of the implanter and is nothing that can be documented in the record. Because

50

00 0.0

0.5

1.0

Target Margin Around Tumor (cm)

3. Target volume and treatment volume vs. margin around tumor. Treatment to target volume ratio vs. margin around tumor (scale at right).

Fig.

187

19’Irbrachytherapy0 J. K. HAYESer al. 0.8

0.6 1

1

0.3

o.22 0.0

0.4 0.5

1.0

Target Margin Around Tumor (cm)

1

! 0.0

1 0.5

1.0

Target Margin Around Tumor (cm)

Fig. 4. Minimum and median target dose rates vs. margin around tumor.

Fig. 5. Implant uniformity and minimum to median target dose ratio vs. margin around tumor.

considerably to the physical exam in definition of the deep tissue extent of tumor in head and neck cancer (4), and that it will be helpful in planning boost irradiation in brachytherapy. Clinical outcome analysis correlated with implant parameters should contribute to the understanding of the role of CT in target definition. Related to the accuracy of CT in target definition is the uncertainty in patient positioning between the pretreatment scan and the implant scan. Based on our initial experience, we believe that a more reliable method than skin marks and head positioning is important for correlating the diagnostic scan with the implant scan, and the isodose maps with the implant scan. Hopefully this can be achieved through non-invasive head immobilization techniques. When CT fails to show tumor adequately, MRI may be a useful alternative imaging modality. Another important question relates to the definition of the target for the implant. It is not always possible to add margin around a tumor in the head and neck due to anatomical barriers such as bone, the top of the tongue, mucosal surfaces, etc. It is not known, for the purposes of an interstitial implant boost, whether margin around tumor is necessary and appropriate, especially in the setting of tumor shrinkage after external beam therapy. Adding margin increases the technical difficulty in doing the implant and mandates substantially larger implants for adequate target coverage. We currently use the pretreatment tumor as the implant target. The data presented here indicate that implant quality parameters are target size dependent (Figs. 3, 5); and, uniform target definition is important in a controlled study. Analysis of tumor and normal tissue coverage affords

one the opportunity to locate regions of low dose and unnecessary dose and modify future implants of like kind accordingly. Thus, it becomes an aid for the improvement of implant technique. The data presented indicate the need on our part of technique improvement. Specifically, dose homogeneity, measured here as U, and brrni,, +bTrned, should be higher than a 0.73 average. With appropriate attention, subsequent implants have shown significant improvement in this parameter. Two features that repeatedly influenced homogeneity were a small region of low dose at a periphery of the target where dose falloff was rapid, usually an implant geometry problem, and small regions of dose significantly greater than ar.min within the target which were recalcitrant to our rudimentary efforts at trial and error optimization. Implant geometry problems occur when the edge of the target is too near or outside the geometric boundary of the implant array or when peripheral source ribbons are spaced too far apart. The parameters of dose homogeneity used in this study are closely related but differ in minor respects. U,, a ratio of treatment volumes, pertains more to homogeneity in the implanted volume as a whole. Dr,,j, GD~,,,~~, as calculated in this study, pertains more to homogeneity within the target. Both are readily calculated in this type of implant analysis. One need not use CT analysis to make clinical use of these parameters. For implants in which CT analysis is not used, these concepts are still easily applied, albeit with less rigor, and contribute to objectivity in implant documentation. For example, when i)rmi,, is determined by whatever means, such as overlaying isodose plots with radiographs, and VT is considered to be VT,., one is usually able to determine b)T,med to within a few cGy range

188

I. J. Radiation

Oncology ??Biology 0 Physics

by visual inspection alone on an isodose map printed in 5 to 10 cGy increments. One can also determine U, or &J?& + &?Wd directly from the implant dose-volume data. In our experience with head and neck implants, U, and b)T,min+ L$- med are very close to the same value if calculated from the dose-volume data. Routinely recorded, these very simply acquired parameters offer the implant physician a measure of dose uniformity that can be used to assess technique and to correlate with clinical outcome. These parameters are easier to apply clinically than dosevolume histograms and are probably adequate measures of dose homogeneity for implant optimization and clinical trials analysis. A mean VT,.+ VT of 6.1 of these implants indicates that too much normal tissue lies outside the target but within the treatment volume, and suggests that seed loading outside the target can be reduced. Although the optimal VT, + VT for maximum local control and minimum complications is unknown, it is probably safe to say that the values obtained here are not enviable standards. Our efforts to reduce the VT, + VT in the short term include more selective seed placement within afterloading tubes and more carefully designed implant geometry based on analysis of prior cases. Simply doing proportionately smaller implants, however, would clearly have detrimental effects on dose homogeneity, as most of these implants had at least one region where the tumor margin approached implant boundaries. Attention to this parameter must include judicious seed omission outside the target with the hoped for benefit of fewer complications and improved functional integrity.

Volume 22, Number

1, 1992

V,, + VT cannot be determined objectively without imaging techniques like CT. If the implant parameter data in this study are at all representative of complex head and neck implants in general, the implication is that significant improvement is possible with improved placement of implant tubes and with computer-optimized seed loading. Although a rigorous brachytherapy dosimetry analysis adds complexity to the treatment, it provides information for technique improvement that may contribute to improved local control and fewer complications. Technical advances in head and neck brachytherapy (5, 9, 16) can be studied more objectively by this type of analysis. In summary, we have applied CT treatment planning in the clinical setting of head and neck implants and shown that it holds promise for objective definition of a target, minimum target absorbed dose, and implant parameters necessary for optimization of technique and for clinical trials analysis. i)T,min + DT,med (or U,) and V,, + VT are very useful and easily understood parameters which derive from basic dosimetric terminology of teletherapy, and therefore do not require the introduction of new dosimetric concepts specific to brachytherapy. Application of these two parameters should facilitate rigorous scientific clinical trials in brachytherapy of the head and neck. Future efforts need to be directed to a more precise, less labor intensive process based on existing technologies (12, 13), one that optimizes source carrier loading to improve dose homogeneity throughout the target and minimizes dose to tissues outside of the target.

REFERENCES 1. Anderson, L. L. Dose specification in brachytherapy. American Endocurietherapy Society Physics Committee Communication, Dec. 20, 1988. 2. Anderson, L. L.; Wagner, L. K.; Schauer, T. H. Memorial Hospital methods of dose calculation for Ir-192. In: George, F. 0.) ed. Modem interstitial and intracavitary radiation cancer management. New York, NY: Masson;1981:1-7. 3. Anderson, L. L.; Weaver, K. A.; Nath, N.; Phillips, T. L.; Nori, D.; Yung, H. S. Dose contour evaluation and approximate methods. In: Interstitial Collaborative Working Group. Interstitial brachytherapy, physical, biological, and clinical considerations. New York, NY: Raven Press; 1990:283-291. 4. Bragg, D. G.; Hamsberger, H. R.; Thompson, W. Radiologic techniques in cancer of the head, neck and spine. In: DeVita, V. T., Hellman, S., Rosenberg, S. A., eds. Cancer principles and practice of oncology, Vol. 2, 3rd edition. Philadephia, PA: Lippincott, Co; 1989:448-450. 5. Clarke, D. H.; Edmundson, K. E.; Martinez, M.; Matter, R. C.; Vicini, F.; Sebastian, E. The clinical advantages of I-12.5 seeds as a substitute for h-192 seeds in temporary plastic tube implants. Int. J. Radiat. Oncol. Biol. Phys. 17:859863; 1989. 6. Crook, J.; Mazeron, J.; Marinello, G.; Martin, M.; Raynal, M.; Calitchi, E.; Faraldi, M.; Ganem, G.; LeBourgeois, J.; Pierquin, B. Combined external irradiation and interstitial implantation for Tl and T2 epidermoid carcinomas of base of tongue: the Creteil experience (1971-1981). Int. J. Ra-

diat. Oncol. Biol. Phys. 15:105-114; 1988. 7. Foote, R. L.; Parsons, J. T.; Mendenhall, W. M.; Million, R. R; Cassisi, N. J.; Stringer, S. P. Is interstitial implantation essential for successful radiotherapeutic treatment of base of tongue carcinoma? Int. J. Radiat. Oncol. Biol. Phys. 18:1293-1298; 1990. 8. Goffinet, D. R.; Cox, R. S.; Clarke, D. H.; Fu, K. K.; Hilaris, B.; Ling, C. C. Brachytherapy. Am. J. Clin. Oncol. 11:342-354; 1988. 9. Goffinet, D. R.; Fee, W. E.; Wells, .I.; Austin-Seymour, M.; Clarke, D.; Mariscal, J. M.; Goode, R. L. 19*Ir pharyngoepiglottic fold interstitial implants. Cancer 55:941-948; 1985. 10. International Commission on Radiation Units and Measurements. Dose specifications for reporting external beam therapy with photons and electrons. ICRU Report 29. Washington: ICRU; 1978. 11. Mazeron, J.; Marinello, G.; Crook, J.; Marin, L.; Mahot, P.; Raynal, M.; Calitchi, E.; Peynegre, R.; Ganem, G.; Faraldi, M.; Huart, J.; LeBourgeois, J.; Pierquin, B. Definitive radiation treatment for early stage carcinoma of the soft palate and uvula: the indications for iridium 192 implantation. Int. J. Radiat. Oncol. Biol. Phys. 13:1829-1837; 1987. 12. Meli, J. M.; Yung, H. S.; Huq, M. S. Computed tomography-based target volume dose calculations in brachytherapy (Abstr). Endocuriether. Hyperther. Oncol. 5:243; 1989. 13. Nasser, M.; DeWyngaert, J. K. Three-dimensional treatment

‘% brachytherapy 0 J. K. planning for brachytherapy (Abstr). Endocuriether. Hyperther. Oncol. 4:86; 1988. 14. Neblett, D. L.; Syed, A. M. N.; Puthawala, A. A.; Harrop, R.; Frey, H. S.; Hogan, S. E. An interstitial implant technique evaluated by contiguous volume analysis. Endocuriether. Hyperther. Oncol. 1:2 13-222; 1985. 15. Puthawala, A. A.; Syed, A. M. N.; Eads, D. L.; Gillin, L.; Gates, T. C. Limited external beam and interstitial “*iridium irradiation in the treatment of carcinoma of the base of the tongue: a ten year experience. Int. J. Radiat. Oncol. Biol. Phys. 14:839-848; 1988. 16. Puthawala, A. A.; Syed, A. M. N.; Gates, T. C. Iridium-

HAYES et al.

189

192 implants in the treatment of tonsillar region malignancies. Arch. Otolaryngol. 111:812-815; 1985. 17. Saw, C. B.; Suntharalingam, N.; Ayyangar, K. M.; Tupthong, L. Dosimetric considerations of stereotactic brain implants. Int. J. Radiat. Oncol. Biol. Phys. 17:887-891; 1989. 18. Syed, A. M. N.; Feder, B. H.; George, F. W.; Neblett, D. Iridium-192 afterloaded implant in the retreatment of head and neck cancers. Brit. J. Radiol. 51:814-820; 1978. 19. Wu, U. K.; Stemick, E. S. A dose homogeneity index for evaluating 19’Ir interstitial breast implants. Med. Phys. 15: 104-107; 1988.