- Email: [email protected]
3-D Treatment Planning and Dose Delivery Verification Integrating a Variety of State-of-the-art techniques: A Case Report
Medrcal Prmted
Doswneir.v. Vol. 16, pp. 225-232 I” the U S.A. All rights reserved.
Copyright
0
0739-021 l/91 Ass~~mt~on of Medxal
1991 Amencan
$3.00 + .oO Dosimetnsts
3-D TREATMENT PLANNING AND DOSE DELIVERY VERIFICATION INTEGRATING A VARIETY OF STATE-OF-THE-ART TECHNIQUES: A CASE REPORT* FRANCA
T. KUCHNIR,
PH.D.,
CHESTER S. REP, Department
of Radiation
and Cellular
SUNAREE WATSON-BULLOCK,
PH.D., Oncology,
and DENNIS HALLAHAN, The University
of Chicago,
C.M.D., M.D.
Chicago,
IL 60637, U.S.A.
Abstract-A patient previously treated with radiation for base-of-tongue cancer presented with recurrent disease seven years later. The spinal cord had received tolerance dose. Using state-of-the-art treatment planning techniques, including beam’s-eye-view and volumetrics, dose-volume histograms, split field technique, mixed energies, and beam intensity modulation (with a compensator), we achieved uniform dose coverage of the target in 3-D. This was verified in vivo with thermoluminescence dosimeters positioned in the esophagus by means of a nasogastric tube that ran centrally through the target volume. The various techniques applied will be presented with a discussion of the rationale used in each step of plan optimization and verification. Key Words: Treatmentplanning,BEV,DVH.
INTRODUCI’ION Currently, we are conducting
an investigational
sented with recurrent disease. CT demonstrated a mass in the left pharynx extending down to the hypopharynx and displacing the larynx to the right. A mass in the left anterior neck was also noted.
pro-
tocol that combines radiation therapy with concomitant 5-m and hydroxyurea chemotherapy for the treatment of patients with recurrent head and neck carcinoma following radiotherapy. The potential for late complications resulting from this second course of radiation therapy is great. Therefore, extra care is exercised in the treatment planning and delivery in order to optimize and verify the dose distribution. Recently, a number of advanced treatment planning techniques have become available in the clinic. These include multilevel dose computations,’ beam’seye-view projections of contours derived from computerized tomography,* dose-volume histograms,3 and others. Here we describe in detail the sequential approach used in applying such techniques to improve homogeneity of dose distributions and spare a vital structure. We present the resulting optimized dose distributions and in vivo dose-point verification with thermoluminescence dosimeters.
Prescription A minimum tumor dose of 6300 cGy was prescribed in a split course (alternating week on/week off) combined with 5-FU and hydroxyurea over a total period of 14 weeks. Daily fraction size was 180 cGy. A target dose homogeneity of 57.5% and minimal dose to the spinal cord were requested.
Treatment plan
The patient was treated in 1983 for base of tongue cancer (T 1, N2) with opposed lateral fields to a total of 5200 cGy. The spinal cord received 4800 cGy. The primary tumor was boosted with gold seeds to 7700 cGy and the positive nodes with electrons to 6200 cGy. In September of 1990 this patient pre-
The patient was immobilized supine with a neutral chin extension using a light cast. A computerized tomography (CT) scan was obtained in this treatment position. Initially, 1800 cGy were delivered with lateral opposed anterior neck fields and an anterior supraclavicular field using 6 MV photons. A 9 MeV electron field treated the left posterior neck to the same total dose. The photon-electron field edges were matched on the skin. Half fields with independent jaws were used to eliminate overlap due to divergence of the noncoplanar photon beams. The match line was moved 1 cm superiorly at 900 cGy. Estimated cord dose for this part of the treatment was 250 cGy given in 10 fractions. To deliver the balance of the dose uniformly over the target volume, the plan described below was developed. A number of contours (external, target, and spi-
* Presented at the annual meeting of the American Association of Medical Dosimetrists in Seattle, Washington, June 3-7, 1991.
Corresponding Author: Franca T. Kuchnir, Dept. of Radiation Oncology, University of Chicago Medical Center, 5841 S. Maryland Ave., Box 442, Chicago, IL 60637.
MATERIALS
AND METHODS
History
225
226
Medical Dosimetry
nal cord) were generated on each of sixteen CT slices (no. 10 through no. 25). The strategy was to use op posed oblique fields that cover the target volume while minimizing cord dose. To visualize the relative positions of the various structures in three dimensions we routinely use beam’s-eye-view (BEV). This technique allows one to look at various anatomical structures (including the treatment target) from the perspective of the incident beam. Figure 1 shows a BEV plot from an LPO view for this case. The gantry angle for this view is 20” off lateral. The X, Y, and Z axes, shown in the figure, meet at the treatment isocenter. The CT slice numbers are indicated along the external contour (EXR); the spinal cord (CRR), and target (TAR) are also displayed. The coarseness of the contours is due to the fact that the CT slices are 1 cm thick. The bold line is the projection of the irregularly shaped treatment field framing the target with a 1 cm margin in all directions. At the end of the treatment planning procedure such BEV plots are used to design the irregularly shaped field defining blocks. Computer printouts are produced for a selected gantry angle with a magnification matching that of the simulation radiograph. The coarseness of the step-like contours is smoothed (with wax pencil) to facilitate transfer of the projected contours onto the film. To select the optimal angles for treatment we use BEV volumetrics.4 This involves framing the target
Volume 16, Number 4, I99 1
on all slices with a preselected margin (1 cm in this case), thus defining the borders of the irregularly shaped treatment field. We then compute the percent volume of the target and of the critical structure (spinal cord) intersected by the treatment field as a function of gantry angle. These quantities are plotted in Figure 2. The plot shows that the target is fully covered at all gantry angles (since the field shape is designed with this purpose) and that there are two broad minima at which the shaped field completely clears the spinal cord. However, the angles corresponding to opposed laterals (0” and 180”) are not within these minima due to tumor extension inferior to the shoulder. The next step in the treatment planning process was to split the tumor volume longitudinally and apply the BEV volumetric analysis to the upper and lower sections separately. The match line chosen is indicated in Fig. 1. The strategy was to select a pair of angles as close as possible to opposed laterals for the upper neck while for the lower neck we chose to go off lateral to clear the shoulders. Figure 3 displays the volumetrics results for the spinal cord only, with arrows indicating the angles selected for the upper and lower neck sections. Thus, by using a split field arrangement we achieved complete tumor coverage and spinal cord sparing as well as additional sparing of normal tissues in the shoulder by going off lateral as much as possible for the lower neck.
Fig. 1. LPO view reconstructed by BEV using 16 CT slices. Slice numbers are listed along the external contour (EXR). The bold line surrounding the treatment target (TAR) is the projection of the “frame,” which represents the field edges of the irregularly shaped radiation beam. The critical structure is the spinal cord @RR), which is difficult to visualize in an oblique projection due to interference of surrounding structures.
Treatment planning and dose delivery verification 0
F. T. KUCHNIR
et al
22-l
4 0
60
128
P.Pea65
IM
He
388
360
#CtEINECREES EUWJmCS
Fig. 2. BEV volumetric plot of the target (TAR) and spinal cord (CRR). The fraction of the volumes (%) of these two “structures,” which is within the irregularly shaped treatment field, is plotted as a function ofthe gantry angle.
The consequence of this optimization is that different plans and treatment angles are required for the upper and lower sections of the treatment volume. In practice such treatment is delivered by using the feature of independent jaws which allows treatment of both upper and lower sections without moving the patient or the isocenter. The use of a single isocenter
simplifies patient setup and facilitates accurate treatment. Due to the lack of divergence along the principal plane of the noncoplanar fields, the independent jaws also ensure an exact field match. We selected a central slice for the upper and the lower portions of the treatment volume. These slices, no. 15 and no. 23, are shown in Fig. 4 and Fig. 5,
BE UPPER
NECK
6B
Fig. 3. BEV volumetric plots for the “split” beam arrangement: (a) cephalad and (b) caudad sections of the treatment volume. The match line is between slices no. 20 and no. 2 1 (see Fig. 1 and Table 1).
Volume 16, Number 4, 1991
Medical Dosimetry
228 I”“““““““““““‘1
(a)
.x*.
I.,.*
I
\
I1
I
I
I
I
I
I
I.
I
I
I
I
.
I
I
*
I
,
Fig. 4. Isodose distributions on slice no. 15 (at the center of the upper portion of the treatment volume) obtained with opposed oblique 6 MV beams; (a) without, (b) with the use of a compensator for the LPO field. The gantry angle, G, and monitor units, W, are displayed in the upper right comers. Note that the monitor units are the same for both plans.
Treatment planning and dose delivery verification 0 F. T. KUCHNIR et al.
.O
CO”,
I
229
1.
(a)
,,.‘a
‘.._
‘.__ I._,
(b)
Fig. 5. Same as Fig. 4, but for slice no. 23 at the center of the lower section.
Medical Dosimetry
230
respectively. As exemplified by these two slices, which are 8 cm apart, there is significant variation of the external contour and the target along the longitudinal axis. In addition, the spinal cord shifts vertically, relative to the CT origin, which is represented by the large cross-hair. Optimal isodose plans were generated using the preselected gantry angles (Fig. 3). Due to the superficial nature of the target and concern for nodal involvement, the plan for the upper neck uses 6 MV, opposed oblique fields, 15’ off lateral. The plan for the lower neck uses dual energy: 6 MV for the RAO and 24 MV for the LPO fields. The oblique pair is 30” off lateral as determined by the BEV volumetric analysis. Our convention is to surround the target with the 100% isodose line, which is the minimum target dose (MTD). We strive to keep hot spots from exceeding 115% of the MTD, thus maintaining the dose uniformity in the target to within +7.5%. This goal was achieved on the two central slices (Fig. 4a and Fig. 5a), but when we examined the resulting dose distributions at other levels, we found that the target would be underdosed if we maintained the beam weights as determined on the two central slices. It was then necessary to adjust the relative beam contributions on each slice in order to obtain complete and uniform tumor coverage. Since, in practice, beam weights (monitor units) are fixed, this was achieved by using beam intensity modulation for one of the fields. Figure 4b and Fig. 5b show the dose distributions obtained on the two “central slices” with the use of a lead compensator for the LPO field. Table 1 summarizes the results for all slices. The slice numbers and distance from the match-line are listed in columns 1 and 2. Column 3 lists the percentage of overall dose maximum relative
Table 1. Overall maximum dose on each slice in percentage relative to the minimum target dose (MTD) for the plans without and with the use of a lead compensator of thickness t for the LPO field Compensator Slice
Distance
(#I
(cm)
11
9.5 8.5 1.5 6.5 5.5 4.5 3.5 2.5 1.5 0.5 0.0 = match line
12 13 14 15 16 17 18 19 20 21 22 23 24
-0.5 -1.5 -2.5 -3.5
t
Without % MTD
With % MTD
114 116 117 117 119 122 124 123 124 125
114 111 107 107 106 110 111 113 115 115
0
125 118 119 109
113 112 110 109
4.06 2.28 2.28 0
(mm) 1.14 2.28 2.28 2.86 2.86 2.86 2.28 2.28 2.28
Volume 16, Number 4, 1991
to the MTD without use of the compensator, the highest value being 125%. This is reduced to 115% (column 4) with the use of the compensator, which was fabricated from thin sheets of lead of thicknesses listed in column 5. Notice that there is no compensation required at the extreme field borders. This is due to the fall-off in the beam intensity as a result of lateral scatter. The apparently excessive compensation required on slice no. 2 1 is due to the rather abrupt variation in the external contour as one crosses the matchline, as well as the change in beam energy from 6 to 24 MV. Referring back to Fig. 4 and Fig. 5, we observe that for the same relative beam weights (shown in the upper right corner), the plans without the compensator have larger areas of inhomogeneities. With the use of the compensator, the area within the 110% isodose line is either zero or negligible, and the area within 105% line is greatly reduced. The 100% isodose completely surrounds the target in all cases. RESULTS
AND ANALYSIS
Dose-volume histograms A better way to demonstrate the effect of the compensator in improving dose homogeneity is through the use of dose-volume histograms (DVH) as shown in Fig. 6 for (a) the target and (b) the irradiated normal tissue, respectively. The first is defined by the contours drawn by the physician on the CT slices, the second is that part of normal tissue that falls within the geometric projection of the shaped treatment beams. Dose-volume histogram bar graphs show the percentage of volume receiving a dose within the values shown on the horizontal axis. The dose units are relative to the MTD. Thus, Fig. 6a shows that in both cases, with or without the use of a compensator, all of the target receives more than 100% ofthe MTD. However, while the plan using the compensator delivers no more than 115% of the MTD to the target volume, the plan without a compensator delivers between 115% and 120% of the MTD to 60% of the target volume and between 120% and 125% of the MTD to 20% of the target volume. Similarly, in Fig. 6b we see that a larger percentage of the normal tissue receives doses over 105% of the MTD when the compensator is not used; this trend is reversed below the 105% MTD level. Overall, the effect of the compensator is to redistribute the dose delivered so as to reduce the volume of hot spots. Dose-volume histograms give a quantitative representation of how well this is achieved.
In vivo verification using TLDs This case presents a special opportunity for in vivo verification of the dose delivered, since the target volume surrounds the esophagus. We placed a string of TLlOO lithium fluoride dosimeters in a pediatric nasogastric tube. The TLDs were rods with 1 mm2
Treatment planningand dose delivery verification 0 60 50
1
F. T. KUCHNIR et a/
23 I
f{ ;j; jj: j;: n
(a)
Percentage of Minimum Target Dose
60
1
(b)
El w/out camp
Percentage of Minimum Target Dose Fig. 6. Dose-volume histograms for (a) the target and (b) irradiated normal tissues. The percentage of volume receiving a relative dose between the values indicated is plotted in each column. The minimum target dose is 100% by design. Use of the compensator eliminates “hot spots” (regions receiving dose in excess of I 15%).
cross-section and 6 mm length. These were stacked in groups of five between 1 cm long plastic spacers containing lead markers. Prior to treatment, the nasogastric tube was placed in the esophagus, and an orthogonal pair of radiographs was taken on the simulator using the treatment isocenter with the patient in treatment position. The TLD position coordinates derived from these films were transferred to the corresponding CT slices, thus allowing us to do point-dose calculations at the location of the individual TLDs. At the beginning of treatments, ten TLDs were irradiated covering the length from slices no. 15 through no. 23 (see Fig. I), and a month later a new set of fifteen TLDs were irradiated, their locations ranging longitudinally from slice no. 11 through no. 20. Individual sensitivities were obtained for each dosimeter, and their identities were maintained. In this manner, uncertainties in dose measurements were of the order of &20/o. Ratios of measured to calculated dose were computed for each of the twenty-five TLDs. The aver-
age of the 10 data points from the first exposure was 0.99 & 0.03 and that of the 15 data points from the second exposure was 1.05 + 0.03. The higher ratio obtained in the second exposure is interpreted as being due to reduction of swelling and tumor regression. The standard deviation of t-3% is consistent with the t-2% spread in TLD dose response and +2% uncertainty in the point-dose calculations. Clinical course The patient tolerated the treatment well. Currently, six months after therapy, there is no evidence of disease within the area of irradiation, and there are no complications. The patient has, however, developed brain metastases. CONCLUSION We have presented the case of a patient with recurrent head and neck cancer who received radiation
232
Medical Dosimetry
therapy to the spinal cord and to the brain approximately seven years ago. What appeared to be an unsalvageable case proved successful through careful treatment planning using BEV, BEV volumetrics, dose-volume histograms, and beam intensity modulation. It was initially planned that the patient would receive 4500 cGy external beam followed by laryngopharyngectomy. However, after examination by the surgeon, the response from the combined modalities was felt to be very good, and radiation was continued to a total dose of 6300 cGy. The patient has since developed some distant metastasis but shows no evidence of local disease at six months after treatment.
Volume 16. Number 4. I99 1
REFERENCES I. Goitein, M.; Abrams, M. Multi-dimensional treatment planning: I. Delineation of anatomy. Int. J. Radiat. Oncol. Biol. Phys. 9:171-187; 1983. 2. Goitein, M.; Abrams, M. et al. Multidimensional treatment planning: II. Beam’s eye-view, back projection, and projection throueh CT sections. Int. J. Radiat. Oncol. Biol. Phvs. 9:789797; i$83. 3. Lyman, J.T.; Wolbarst, A.A. Optimization of radiation therapy, IV: A dose-volume histogram reduction algorithm. Int. J. Radiat. Oncol. Biol. Phys. 17~433-436; 1989. 4. Myrianthopoulos, L.C.; Chen, G.T.Y.; Spelbring, D.R. Treatment plan optimization from quantitative beam’s-eye-view volumetric considerations. Phys. Med. Biol. 33(Supplement 1):9 1; 1988.
Doswneir.v. Vol. 16, pp. 225-232 I” the U S.A. All rights reserved.
Copyright
0
0739-021 l/91 Ass~~mt~on of Medxal
1991 Amencan
$3.00 + .oO Dosimetnsts
3-D TREATMENT PLANNING AND DOSE DELIVERY VERIFICATION INTEGRATING A VARIETY OF STATE-OF-THE-ART TECHNIQUES: A CASE REPORT* FRANCA
T. KUCHNIR,
PH.D.,
CHESTER S. REP, Department
of Radiation
and Cellular
SUNAREE WATSON-BULLOCK,
PH.D., Oncology,
and DENNIS HALLAHAN, The University
of Chicago,
C.M.D., M.D.
Chicago,
IL 60637, U.S.A.
Abstract-A patient previously treated with radiation for base-of-tongue cancer presented with recurrent disease seven years later. The spinal cord had received tolerance dose. Using state-of-the-art treatment planning techniques, including beam’s-eye-view and volumetrics, dose-volume histograms, split field technique, mixed energies, and beam intensity modulation (with a compensator), we achieved uniform dose coverage of the target in 3-D. This was verified in vivo with thermoluminescence dosimeters positioned in the esophagus by means of a nasogastric tube that ran centrally through the target volume. The various techniques applied will be presented with a discussion of the rationale used in each step of plan optimization and verification. Key Words: Treatmentplanning,BEV,DVH.
INTRODUCI’ION Currently, we are conducting
an investigational
sented with recurrent disease. CT demonstrated a mass in the left pharynx extending down to the hypopharynx and displacing the larynx to the right. A mass in the left anterior neck was also noted.
pro-
tocol that combines radiation therapy with concomitant 5-m and hydroxyurea chemotherapy for the treatment of patients with recurrent head and neck carcinoma following radiotherapy. The potential for late complications resulting from this second course of radiation therapy is great. Therefore, extra care is exercised in the treatment planning and delivery in order to optimize and verify the dose distribution. Recently, a number of advanced treatment planning techniques have become available in the clinic. These include multilevel dose computations,’ beam’seye-view projections of contours derived from computerized tomography,* dose-volume histograms,3 and others. Here we describe in detail the sequential approach used in applying such techniques to improve homogeneity of dose distributions and spare a vital structure. We present the resulting optimized dose distributions and in vivo dose-point verification with thermoluminescence dosimeters.
Prescription A minimum tumor dose of 6300 cGy was prescribed in a split course (alternating week on/week off) combined with 5-FU and hydroxyurea over a total period of 14 weeks. Daily fraction size was 180 cGy. A target dose homogeneity of 57.5% and minimal dose to the spinal cord were requested.
Treatment plan
The patient was treated in 1983 for base of tongue cancer (T 1, N2) with opposed lateral fields to a total of 5200 cGy. The spinal cord received 4800 cGy. The primary tumor was boosted with gold seeds to 7700 cGy and the positive nodes with electrons to 6200 cGy. In September of 1990 this patient pre-
The patient was immobilized supine with a neutral chin extension using a light cast. A computerized tomography (CT) scan was obtained in this treatment position. Initially, 1800 cGy were delivered with lateral opposed anterior neck fields and an anterior supraclavicular field using 6 MV photons. A 9 MeV electron field treated the left posterior neck to the same total dose. The photon-electron field edges were matched on the skin. Half fields with independent jaws were used to eliminate overlap due to divergence of the noncoplanar photon beams. The match line was moved 1 cm superiorly at 900 cGy. Estimated cord dose for this part of the treatment was 250 cGy given in 10 fractions. To deliver the balance of the dose uniformly over the target volume, the plan described below was developed. A number of contours (external, target, and spi-
* Presented at the annual meeting of the American Association of Medical Dosimetrists in Seattle, Washington, June 3-7, 1991.
Corresponding Author: Franca T. Kuchnir, Dept. of Radiation Oncology, University of Chicago Medical Center, 5841 S. Maryland Ave., Box 442, Chicago, IL 60637.
MATERIALS
AND METHODS
History
225
226
Medical Dosimetry
nal cord) were generated on each of sixteen CT slices (no. 10 through no. 25). The strategy was to use op posed oblique fields that cover the target volume while minimizing cord dose. To visualize the relative positions of the various structures in three dimensions we routinely use beam’s-eye-view (BEV). This technique allows one to look at various anatomical structures (including the treatment target) from the perspective of the incident beam. Figure 1 shows a BEV plot from an LPO view for this case. The gantry angle for this view is 20” off lateral. The X, Y, and Z axes, shown in the figure, meet at the treatment isocenter. The CT slice numbers are indicated along the external contour (EXR); the spinal cord (CRR), and target (TAR) are also displayed. The coarseness of the contours is due to the fact that the CT slices are 1 cm thick. The bold line is the projection of the irregularly shaped treatment field framing the target with a 1 cm margin in all directions. At the end of the treatment planning procedure such BEV plots are used to design the irregularly shaped field defining blocks. Computer printouts are produced for a selected gantry angle with a magnification matching that of the simulation radiograph. The coarseness of the step-like contours is smoothed (with wax pencil) to facilitate transfer of the projected contours onto the film. To select the optimal angles for treatment we use BEV volumetrics.4 This involves framing the target
Volume 16, Number 4, I99 1
on all slices with a preselected margin (1 cm in this case), thus defining the borders of the irregularly shaped treatment field. We then compute the percent volume of the target and of the critical structure (spinal cord) intersected by the treatment field as a function of gantry angle. These quantities are plotted in Figure 2. The plot shows that the target is fully covered at all gantry angles (since the field shape is designed with this purpose) and that there are two broad minima at which the shaped field completely clears the spinal cord. However, the angles corresponding to opposed laterals (0” and 180”) are not within these minima due to tumor extension inferior to the shoulder. The next step in the treatment planning process was to split the tumor volume longitudinally and apply the BEV volumetric analysis to the upper and lower sections separately. The match line chosen is indicated in Fig. 1. The strategy was to select a pair of angles as close as possible to opposed laterals for the upper neck while for the lower neck we chose to go off lateral to clear the shoulders. Figure 3 displays the volumetrics results for the spinal cord only, with arrows indicating the angles selected for the upper and lower neck sections. Thus, by using a split field arrangement we achieved complete tumor coverage and spinal cord sparing as well as additional sparing of normal tissues in the shoulder by going off lateral as much as possible for the lower neck.
Fig. 1. LPO view reconstructed by BEV using 16 CT slices. Slice numbers are listed along the external contour (EXR). The bold line surrounding the treatment target (TAR) is the projection of the “frame,” which represents the field edges of the irregularly shaped radiation beam. The critical structure is the spinal cord @RR), which is difficult to visualize in an oblique projection due to interference of surrounding structures.
Treatment planning and dose delivery verification 0
F. T. KUCHNIR
et al
22-l
4 0
60
128
P.Pea65
IM
He
388
360
#CtEINECREES EUWJmCS
Fig. 2. BEV volumetric plot of the target (TAR) and spinal cord (CRR). The fraction of the volumes (%) of these two “structures,” which is within the irregularly shaped treatment field, is plotted as a function ofthe gantry angle.
The consequence of this optimization is that different plans and treatment angles are required for the upper and lower sections of the treatment volume. In practice such treatment is delivered by using the feature of independent jaws which allows treatment of both upper and lower sections without moving the patient or the isocenter. The use of a single isocenter
simplifies patient setup and facilitates accurate treatment. Due to the lack of divergence along the principal plane of the noncoplanar fields, the independent jaws also ensure an exact field match. We selected a central slice for the upper and the lower portions of the treatment volume. These slices, no. 15 and no. 23, are shown in Fig. 4 and Fig. 5,
BE UPPER
NECK
6B
Fig. 3. BEV volumetric plots for the “split” beam arrangement: (a) cephalad and (b) caudad sections of the treatment volume. The match line is between slices no. 20 and no. 2 1 (see Fig. 1 and Table 1).
Volume 16, Number 4, 1991
Medical Dosimetry
228 I”“““““““““““‘1
(a)
.x*.
I.,.*
I
\
I1
I
I
I
I
I
I
I.
I
I
I
I
.
I
I
*
I
,
Fig. 4. Isodose distributions on slice no. 15 (at the center of the upper portion of the treatment volume) obtained with opposed oblique 6 MV beams; (a) without, (b) with the use of a compensator for the LPO field. The gantry angle, G, and monitor units, W, are displayed in the upper right comers. Note that the monitor units are the same for both plans.
Treatment planning and dose delivery verification 0 F. T. KUCHNIR et al.
.O
CO”,
I
229
1.
(a)
,,.‘a
‘.._
‘.__ I._,
(b)
Fig. 5. Same as Fig. 4, but for slice no. 23 at the center of the lower section.
Medical Dosimetry
230
respectively. As exemplified by these two slices, which are 8 cm apart, there is significant variation of the external contour and the target along the longitudinal axis. In addition, the spinal cord shifts vertically, relative to the CT origin, which is represented by the large cross-hair. Optimal isodose plans were generated using the preselected gantry angles (Fig. 3). Due to the superficial nature of the target and concern for nodal involvement, the plan for the upper neck uses 6 MV, opposed oblique fields, 15’ off lateral. The plan for the lower neck uses dual energy: 6 MV for the RAO and 24 MV for the LPO fields. The oblique pair is 30” off lateral as determined by the BEV volumetric analysis. Our convention is to surround the target with the 100% isodose line, which is the minimum target dose (MTD). We strive to keep hot spots from exceeding 115% of the MTD, thus maintaining the dose uniformity in the target to within +7.5%. This goal was achieved on the two central slices (Fig. 4a and Fig. 5a), but when we examined the resulting dose distributions at other levels, we found that the target would be underdosed if we maintained the beam weights as determined on the two central slices. It was then necessary to adjust the relative beam contributions on each slice in order to obtain complete and uniform tumor coverage. Since, in practice, beam weights (monitor units) are fixed, this was achieved by using beam intensity modulation for one of the fields. Figure 4b and Fig. 5b show the dose distributions obtained on the two “central slices” with the use of a lead compensator for the LPO field. Table 1 summarizes the results for all slices. The slice numbers and distance from the match-line are listed in columns 1 and 2. Column 3 lists the percentage of overall dose maximum relative
Table 1. Overall maximum dose on each slice in percentage relative to the minimum target dose (MTD) for the plans without and with the use of a lead compensator of thickness t for the LPO field Compensator Slice
Distance
(#I
(cm)
11
9.5 8.5 1.5 6.5 5.5 4.5 3.5 2.5 1.5 0.5 0.0 = match line
12 13 14 15 16 17 18 19 20 21 22 23 24
-0.5 -1.5 -2.5 -3.5
t
Without % MTD
With % MTD
114 116 117 117 119 122 124 123 124 125
114 111 107 107 106 110 111 113 115 115
0
125 118 119 109
113 112 110 109
4.06 2.28 2.28 0
(mm) 1.14 2.28 2.28 2.86 2.86 2.86 2.28 2.28 2.28
Volume 16, Number 4, 1991
to the MTD without use of the compensator, the highest value being 125%. This is reduced to 115% (column 4) with the use of the compensator, which was fabricated from thin sheets of lead of thicknesses listed in column 5. Notice that there is no compensation required at the extreme field borders. This is due to the fall-off in the beam intensity as a result of lateral scatter. The apparently excessive compensation required on slice no. 2 1 is due to the rather abrupt variation in the external contour as one crosses the matchline, as well as the change in beam energy from 6 to 24 MV. Referring back to Fig. 4 and Fig. 5, we observe that for the same relative beam weights (shown in the upper right corner), the plans without the compensator have larger areas of inhomogeneities. With the use of the compensator, the area within the 110% isodose line is either zero or negligible, and the area within 105% line is greatly reduced. The 100% isodose completely surrounds the target in all cases. RESULTS
AND ANALYSIS
Dose-volume histograms A better way to demonstrate the effect of the compensator in improving dose homogeneity is through the use of dose-volume histograms (DVH) as shown in Fig. 6 for (a) the target and (b) the irradiated normal tissue, respectively. The first is defined by the contours drawn by the physician on the CT slices, the second is that part of normal tissue that falls within the geometric projection of the shaped treatment beams. Dose-volume histogram bar graphs show the percentage of volume receiving a dose within the values shown on the horizontal axis. The dose units are relative to the MTD. Thus, Fig. 6a shows that in both cases, with or without the use of a compensator, all of the target receives more than 100% ofthe MTD. However, while the plan using the compensator delivers no more than 115% of the MTD to the target volume, the plan without a compensator delivers between 115% and 120% of the MTD to 60% of the target volume and between 120% and 125% of the MTD to 20% of the target volume. Similarly, in Fig. 6b we see that a larger percentage of the normal tissue receives doses over 105% of the MTD when the compensator is not used; this trend is reversed below the 105% MTD level. Overall, the effect of the compensator is to redistribute the dose delivered so as to reduce the volume of hot spots. Dose-volume histograms give a quantitative representation of how well this is achieved.
In vivo verification using TLDs This case presents a special opportunity for in vivo verification of the dose delivered, since the target volume surrounds the esophagus. We placed a string of TLlOO lithium fluoride dosimeters in a pediatric nasogastric tube. The TLDs were rods with 1 mm2
Treatment planningand dose delivery verification 0 60 50
1
F. T. KUCHNIR et a/
23 I
f{ ;j; jj: j;: n
(a)
Percentage of Minimum Target Dose
60
1
(b)
El w/out camp
Percentage of Minimum Target Dose Fig. 6. Dose-volume histograms for (a) the target and (b) irradiated normal tissues. The percentage of volume receiving a relative dose between the values indicated is plotted in each column. The minimum target dose is 100% by design. Use of the compensator eliminates “hot spots” (regions receiving dose in excess of I 15%).
cross-section and 6 mm length. These were stacked in groups of five between 1 cm long plastic spacers containing lead markers. Prior to treatment, the nasogastric tube was placed in the esophagus, and an orthogonal pair of radiographs was taken on the simulator using the treatment isocenter with the patient in treatment position. The TLD position coordinates derived from these films were transferred to the corresponding CT slices, thus allowing us to do point-dose calculations at the location of the individual TLDs. At the beginning of treatments, ten TLDs were irradiated covering the length from slices no. 15 through no. 23 (see Fig. I), and a month later a new set of fifteen TLDs were irradiated, their locations ranging longitudinally from slice no. 11 through no. 20. Individual sensitivities were obtained for each dosimeter, and their identities were maintained. In this manner, uncertainties in dose measurements were of the order of &20/o. Ratios of measured to calculated dose were computed for each of the twenty-five TLDs. The aver-
age of the 10 data points from the first exposure was 0.99 & 0.03 and that of the 15 data points from the second exposure was 1.05 + 0.03. The higher ratio obtained in the second exposure is interpreted as being due to reduction of swelling and tumor regression. The standard deviation of t-3% is consistent with the t-2% spread in TLD dose response and +2% uncertainty in the point-dose calculations. Clinical course The patient tolerated the treatment well. Currently, six months after therapy, there is no evidence of disease within the area of irradiation, and there are no complications. The patient has, however, developed brain metastases. CONCLUSION We have presented the case of a patient with recurrent head and neck cancer who received radiation
232
Medical Dosimetry
therapy to the spinal cord and to the brain approximately seven years ago. What appeared to be an unsalvageable case proved successful through careful treatment planning using BEV, BEV volumetrics, dose-volume histograms, and beam intensity modulation. It was initially planned that the patient would receive 4500 cGy external beam followed by laryngopharyngectomy. However, after examination by the surgeon, the response from the combined modalities was felt to be very good, and radiation was continued to a total dose of 6300 cGy. The patient has since developed some distant metastasis but shows no evidence of local disease at six months after treatment.
Volume 16. Number 4. I99 1
REFERENCES I. Goitein, M.; Abrams, M. Multi-dimensional treatment planning: I. Delineation of anatomy. Int. J. Radiat. Oncol. Biol. Phys. 9:171-187; 1983. 2. Goitein, M.; Abrams, M. et al. Multidimensional treatment planning: II. Beam’s eye-view, back projection, and projection throueh CT sections. Int. J. Radiat. Oncol. Biol. Phvs. 9:789797; i$83. 3. Lyman, J.T.; Wolbarst, A.A. Optimization of radiation therapy, IV: A dose-volume histogram reduction algorithm. Int. J. Radiat. Oncol. Biol. Phys. 17~433-436; 1989. 4. Myrianthopoulos, L.C.; Chen, G.T.Y.; Spelbring, D.R. Treatment plan optimization from quantitative beam’s-eye-view volumetric considerations. Phys. Med. Biol. 33(Supplement 1):9 1; 1988.