- Email: [email protected]
Analysis of dose conformity and normal-tissue sparing using two different IMRT prescription methodologies for irregularly shaped CNS lesions irradiated with the Beak and 1-cm MIMiC collimators
Int. J. Radiation Oncology Biol. Phys., Vol. 59, No. 1, pp. 285–292, 2004 Copyright © 2004 Elsevier Inc. Printed in the USA. All rights reserved 0360-3016/04/$–see front matter
doi:10.1016/j.ijrobp.2004.01.018
PHYSICS CONTRIBUTION
ANALYSIS OF DOSE CONFORMITY AND NORMAL-TISSUE SPARING USING TWO DIFFERENT IMRT PRESCRIPTION METHODOLOGIES FOR IRREGULARLY SHAPED CNS LESIONS IRRADIATED WITH THE BEAK AND 1-CM MIMiC COLLIMATORS HEATHER D. ZINKIN, M.D., MARK J. RIVARD, PH.D., JOHN E. MIGNANO, M.D., PH.D., DAVID E. WAZER, M.D.
AND
Department of Radiation Oncology, Tufts–New England Medical Center, Tufts University School of Medicine, Boston, MA Purpose: To determine whether intensity modulated sequential tomotherapy using the NOMOS Beak provides superior dose conformity and organ sparing to the MIMiC “1-cm” mode, and if so, to identify a subset of patients most likely to benefit from Beak intensity modulated sequential tomotherapy. Methods and Materials: Twelve patients with irregularly shaped central nervous system tumors were selected for intensity modulated radiation therapy planning. Two treatment plans, one using the Beak collimator and the other using the 1-cm MIMiC collimator, were generated for each patient with identical anatomic contouring, prescriptions, and optimization algorithms. The Beak attaches to the MIMiC collimator and truncates the 1-cm MIMiC mode beamlet size from 1.00 ⴛ 0.85 cm2 to 1.00 ⴛ 0.39 cm2 at isocenter. Conformity indexes were calculated for each lesion using two different prescription methodologies, and mean doses to critical structures were recorded. Results: For the first prescription methodology using uniform prescribed isodose, mean conformity index was 2.19 (range, 1.33–3.90) for the Beak compared to 2.67 (range, 1.64 – 4.75) for the 1-cm mode (p ⴝ 0.0003). Mean doses to the brainstem, right orbit, and left optic nerve were significantly lower with the Beak than with the 1-cm mode (p ⴝ 0.0150, 0.0068, and 0.0284, respectively). For the second prescription methodology using uniform target volume coverage prescription, mean conformity index was 2.04 (range, 1.56 –2.70) for the Beak compared to 2.73 (range, 1.70 – 8.58) for the 1-cm mode (p ⴝ 0.07). Mean doses to the brain, brainstem, optic chiasm, right optic nerve, left optic nerve, and left orbit were significantly lower with the Beak than with the 1-cm mode (p ⴝ <0.0001, <0.0026, <0.0016, <0.0076, <0.0007, and <0.046, respectively). Conclusion: Beak intensity modulated sequential tomotherapy is superior to the 1-cm MIMiC mode for irregularly shaped central nervous system tumors, because it provides better conformity and critical organ sparing. These differences may allow for safer dose escalation and retreatment, so the method presents an alternative to gamma knife stereotactic radiosurgery. © 2004 Elsevier Inc. Intensity modulated radiation therapy, Beak, Conformity index, Central nervous system tumor.
There are several types of IMRT delivery techniques. Stepand-shoot employs multileaf collimators with stationary gantry angles, and intensity modulated sequential tomotherapy (IMST) uses the MIMiC collimator with continuous gantry rotation around the patient. During tomotherapy, the dose and rotation rates are usually kept constant during treatment delivery (7). The MIMiC collimator (NOMOS Corporation, Cranberry Township, PA) can administer IMST using 1 of 2 modes: (i) a 1-cm MIMiC collimator with a beamlet area of 1.00 ⫻ 0.85 cm2 at isocenter, and (ii) a 2-cm MIMiC collimator with a beamlet size of 1.00 ⫻ 1.70 cm2. For many tumors,
INTRODUCTION Many tumors, particularly those of the central nervous system (CNS), are challenging to treat aggressively with conventional radiation therapy (1, 2). Dose intensification for treatment of CNS lesions is limited by proximity of critical structures such as optic nerves, chiasm, spinal cord, and brainstem. Compared to conventional teletherapy, intensity modulated radiation therapy (IMRT) was developed to deliver more conformal isodose distributions and to better spare normal tissues (3–5). However, the efficacy of IMRT may be only as effective as the radiation beamlet size (6).
Acknowledgments—We extend our appreciation to Robin Ruthazer, M.P.H., for her assistance with statistical analyses, and to Drs. Jen-San Tsai and Mark J. Engler for their contributions. Received Jun 4, 2003, and in revised form Dec 12, 2003. Accepted for publication Jan 16, 2004.
Reprint requests to: Heather D. Zinkin, M.D., Department of Radiation Oncology (Box 359), Tufts–New England Medical Center, 750 Washington Street, Boston, MA 02111. Tel: (617) 6366161; Fax: (617) 636-6131; E-mail: [email protected] The NOMOS Corporation is acknowledged for manuscript review and funding publication of the color figures. 285
286
I. J. Radiation Oncology
● Biology ● Physics
Volume 59, Number 1, 2004
METHODS AND MATERIALS
Fig. 1. The MIMiC collimator is attached to the collimator head of a linear accelerator. The MIMiC is operated using a controller that delivers pneumatic pressure and electronic feedback via the tubes attached from the gantry. The Beak collimator is attached to the MIMiC and further reduces the field size along the direction of the gantry rotational axis.
these beamlets are small enough to provide conformal treatment plans. However, for tumors that are irregularly shaped with dimensions ⬍0.85 cm along the gantry rotational axis, treatment delivery with the 1-cm and 2-cm modes will not provide adequate tumor conformity; such tumors would benefit from smaller beamlets. Therefore, NOMOS has developed the Beak collimator as a treatment device that can create a more conformal treatment plan while sparing critical structures. The Beak (Fig. 1) attaches to the MIMiC collimator and truncates the 1-cm MIMiC mode beamlet size from 1.00 ⫻ 0.85 cm2 to 1.00 ⫻ 0.39 cm2 at isocenter (8). This therapy is referred to as Beak intensity modulated sequential tomotherapy (BIMST) and is simple for radiation therapists to employ on a daily basis. However, there is no validation mechanism or electronic sensor to assure its routine placement. Because use of the Beak is gaining in popularity, the purpose of this study was to compare BIMST with the 1-cm MIMiC collimator and to determine the methods’ effects on dose distributions, conformity index (CI), and relationship to target size and shape for patients with CNS tumors.
A research study was performed using pre-existing data for 15 patients previously treated with IMRT as definitive therapy or boost at the Tufts–New England Medical Center Department of Radiation Oncology. The lesions included the following: 5 pituitary adenomas, 4 meningiomas, 2 glioblastoma multiforme, 1 acoustic neuroma, 1 chondrosarcoma, 1 mixed glioma, and 1 metastatic melanoma. The patients were chosen based on the presence of CNS lesions, anatomic shape of the tumor, and the tumor proximity to normal tissues. The lesions represented relatively common diagnoses, and therefore results of this study may be of general interest. Patients were preferentially selected if their tumor was small, flattened (short dimension along the axis of gantry rotation), or had longitudinal irregularity along the gantry rotational axis, which was restricted to gantry rotational arcs ranging from 330° to 30° for this study. All patients were pathologically diagnosed with CNS tumors, and each patient was positioned with a custom-fit ␣-cradle head holder and aquaplast mask constructed for immobilization (9). Computed tomography images were obtained in the treatment position with 1–3-mm-thick slices. Axial images were transferred to the NOMOS CORVUS, version 5.0R1 IMRT treatment planning system. Planning tumor volume and normal structures, including the brainstem, brain, right and left optic nerves, optic chiasm, and right and left orbits, were contoured by a physician for each patient. Each image set was then duplicated so that 2 treatment plans could be generated, and prescriptions could be made with identical anatomic delineations for meaningful comparisons of 1-cm MIMiC vs. BIMST dose distributions. In total, 54 treatment plans were prepared for this study, and two distinct prescription methodologies were used to compare treatment plans. Uniform prescribed isodose The first prescription methodology included 12 patients, and identical prescription isodoses were selected for both plans of each patient, as shown in Table 1. In all but 1 case, the isodose line for the BIMST plan was chosen that covered at least 90% of the tumor volume. This same isodose line was then assigned to the 1-cm MIMiC plan and served as the variable through which plans were compared. These prescribed isodoses ranged from 73.0% to 89.0% (mean, 81.9%) for both the BIMST and 1-cm MIMiC plans. Target volumes ranged from 2.05 cm3 to 126.37 cm3 (mean, 23.09 cm3), as shown in Table 1; excluding the outlier lesion of 126.37 cm3, the mean target volume significantly decreased to 13.70 cm3. Numeric values were assigned for the target goal dose (prescription dose), percent volume below this goal, and the acceptable minimum and maximum target doses (Gy). The minimum acceptable critical structure dose was set at 0 Gy for all patients. Uniform target volume coverage prescription The second prescription methodology included 15 patients, and treatment plans were computed using 95% target
Dosimetric comparison of Beak and 1-cm MIMiC collimators
● H. D. ZINKIN et al.
287
Table 1. Patient, treatment parameters, and CI for the BIMST and 1-cm MIMiC collimators using the uniform prescribed isodose methodology
Patient
Diagnosis
A B C D E F G H I J K L
Glioblastoma multiforme Meningioma Pituitary adenoma Pituitary adenoma Pituitary adenoma Acoustic neuroma Chondrosarcoma Meningioma Acoustic neuroma Mixed glioma Metastatic tumor to orbit Meningioma
Target volume Target goal BIMST 1-cm MIMiC (cm3) (Gy) CI CI 18.13 24.87 2.66 3.79 17.00 22.77 32.76 12.96 3.47 126.37 2.05 10.26
25 45 16 45 45 50 66 45 50 14 30 45
1.33 1.65 1.65 2.66 1.69 1.83 2.25 2.03 3.88 1.42 3.90 1.95
BIMST and 1cm MIMiC Rx isodose (%)
1.76 2.06 2.00 2.99 1.87 2.63 2.36 3.26 4.48 1.64 4.75 2.27
82.0 79.8 73.0 87.0 83.3 82.0 88.0 84.0 89.0 80.5 75.0 79.0
BIMST 1-cm MIMiC % target % target coverage coverage 94.9 95.1 76.4 96.3 93.6 92.2 91.4 97.1 95.0 95.0 83.3 93.6
96.1 96.6 56.6 98.6 94.9 91.0 94.3 100.0 98.7 96.2 95.4 88.6
Abbreviations: CI ⫽ conformity index; BIMST ⫽ Beak intensity modulated sequential tomotherapy.
volume coverage prescriptions for plan normalization and comparison (10). This approach differed from the first prescription methodology in that target volume coverage was now affixed. For the BIMST plans, these prescribed isodoses ranged from 80.1% to 88.5% (mean, 84.4%) and ranged from 59.6% to 85.6% (mean, 81.0%) for the 1-cm MIMiC plans, as shown in Table 2. Target volumes also ranged from 2.05 cm3 to 126.37 cm3 (mean, 19.66 cm3), as shown in Table 1; excluding the outlier lesion of 126.37 cm3, the mean target volume significantly decreased to 12.03 cm3. Numeric values were assigned for the target goal dose (prescription dose), percent volume below this goal, and the acceptable minimum and maximum target doses (Gy). Although the first prescription methodology set the minimum acceptable critical structure dose at 0 Gy for all
patients, this practice may unnecessarily bias the optimizer. Therefore, the second prescription methodology set the minimum acceptable critical structure doses to more realistic values. For all patients, this dose was specified as ⱖ1 Gy for both the 1-cm MIMiC and BIMST plans. For both prescription methodologies, constraints were assigned for normal structure dose limits, volume above limit, and minimum and maximum acceptable doses. The optic nerve and optic chiasm dose limit specifications varied among patients and depended largely on proximity to the target; realistic expectations of dose were prescribed. If judged to be pertinent by the physician, target prescriptions were marked for prioritization in treatment planning. Based on the immobilization technique described by Tsai et al., positional and structure motion uncertainties were set to 1.0 mm in all directions (9). Except for 1 patient, all treatments
Table 2. Patient and treatment parameters for BIMST and 1-cm MIMiC intensity modulated sequential tomotherapy with uniform target volume coverage prescription Patient
Diagnosis
Target volume (cm3)
Target goal (Gy)
BIMST CI
1-cm MIMiC CI
BIMST Rx isodose (%)
1-cm MIMiC Rx isodose (%)
A B C D E F G H J K L M N O P
Glioblastoma multiforme Meningioma Pituitary adenoma Pituitary adenoma Pituitary adenoma Acoustic neuroma Chondrosarcoma Meningioma Mixed glioma Metastatic tumor to orbit Meningioma Metastatic melanoma Meningioma Glioblastoma multiforme Pituitary adenoma
18.13 24.87 2.66 3.79 17.00 22.77 32.76 12.96 126.37 2.05 10.26 2.14 3.56 9.86 5.66
25 45 16 45 45 50 66 45 14 30 45 25 45 28 45
1.91 1.92 2.65 2.04 1.64 1.88 1.61 2.31 1.61 2.70 2.20 2.62 2.26 1.56 1.72
2.50 2.09 8.58 2.58 1.73 2.04 1.77 2.97 1.70 2.18 2.39 3.23 3.00 2.16 2.04
85.2 85.0 81.0 87.7 85.1 81.0 86.6 87.7 83.3 84.4 82.9 80.1 83.2 84.2 88.5
83.5 82.1 59.6 84.4 85.6 82.3 85.1 80.1 84.3 85.4 85.2 79.2 79.5 75.5 82.6
Abbreviations: Same as for Table 1. Note: For all cases, the BIMST CI is compared to the 1-cm MIMiC CI using a 95% target volume coverage.
288
I. J. Radiation Oncology
● Biology ● Physics
were delivered with a single couch angle of 180°. Mean doses to critical structures were recorded for each plan, as well as the ratio of maximum to minimum dose received by the target. Furthermore, dose–volume histograms (DVHs) for the two modalities were also compared. Conformity index Conformity index was initially defined according to ICRU 62 (11) for use as a metric of tightness-of-fit of the planning target volume to the prescribed isodose volume. This definition was later modified and improved upon (12– 15). CI accounts for the total target volume (TV), prescription isodose surface volume (PV), and the ratio of target volume within the prescribed isodose surface to the total target volume (TVPV). CI is defined below (Eq. 1): PV/TV PV CI ⫽ , which may be rearranged as: TV PV/TV CI ⫽
PV · TV . TV PV2
(1)
A perfect treatment plan would set TVPV ⫽ TV ⫽ PV, yielding CI ⫽ unity (14); the closer CI is to 1, the more conformal the plan. CI values were calculated for all 54 plans (2 per patient). Statistical analysis In addition to CI, other outcome measures were analyzed, including: 1. ratio of maximum to minimum doses received by the target, used to model dose homogeneity, 2. number of treatment arcs, used to model overall treatment time, and 3. multivariate analysis to assess the relationship between target dose, target volume, and critical structure doses. Statistical analyses (SAS version 8.0) using paired t tests were used to determine whether significant differences between the two treatment modalities could be appreciated. Ninety-five percent confidence intervals were calculated for the difference between the mean doses to critical structures (95%CIdiff). Multivariate analyses with Spearman rank correlation coefficients were also performed to determine whether relationships existed between target volume, CI, and mean dose to critical structures. RESULTS For the uniform prescribed isodose methodology, mean doses to the brainstem, right orbit, and left optic nerve were significantly lower with the BIMST compared to the 1-cm MIMiC. Mean brainstem dose was 11.72 vs. 13.06 Gy for BIMST and 1-cm MIMiC, respectively (95%CIdiff 0.32 to 2.31, p ⫽ 0.0150). Mean right orbit dose was 5.39 Gy vs. 6.05 Gy for BIMST and 1-cm MIMiC, respectively (95%CIdiff 0.23 to 1.09, p ⫽ 0.0068). Left optic nerve mean
Volume 59, Number 1, 2004
dose was also lower with BIMST: 10.84 vs. 12.17 Gy (95%CIdiff 0.19 to 2.49, p ⫽ 0.0284). The CI was also significantly better with BIMST than with 1-cm MIMiC (2.19 vs. 2.67, p ⫽ 0.0003), as shown in Table 1. The second prescription methodology, uniform target volume coverage prescription, produced more significant differences in normal structure dose than the first prescription methodology. The mean doses to the (i) brain, (ii) brainstem, (iii) optic chiasm, (iv) right optic nerve, (v) left optic nerve, and (vi) left orbit were significantly lower with the BIMST compared to the 1-cm MIMiC plans. For the aforementioned structures, the mean doses were significantly lower with BIMST than with 1-cm MIMiC: i. 5.13 Gy vs. 5.95 Gy (95%CIdiff 0.54 to 1.09, p 0.0001), ii. 10.94 Gy vs. 12.28 Gy (95%CIdiff 0.57 to 2.11, p 0.003), iii. 20.95 Gy vs. 22.79 Gy (95%CIdiff 0.84 to 2.84, p 0.002), iv. 12.21 Gy vs. 13.14 Gy (95%CIdiff 0.29 to 1.56, p 0.008), v. 13.08 Gy vs. 14.42 Gy (95%CIdiff 0.68 to 2.01, p 0.007), and vi. 4.19 Gy vs. 4.58 Gy (95%CIdiff 0.01 to 0.76, p 0.046).
⬍ ⫽ ⫽ ⫽ ⫽ ⫽
Although the mean doses to the right orbit were lower with BIMST than with the 1-cm MIMiC (5.48 Gy vs. 5.95 Gy), this dosimetric improvement was less convincing (p ⫽ 0.08) than was observed for the 6 previously mentioned structures. For the second prescription methodology, the mean CI was also better (lower) for BIMST than for the 1-cm MIMiC (2.04 vs. 2.00, p ⫽ 0.07), as shown in Table 2, and for all but 1 patient, CI with BIMST was lower than with 1-cm MIMiC. Ratios of maximum to minimum dose received by the targets were used as a measure of target dose homogeneity and were not significantly different for the BIMST and the 1-cm MIMiC plans and for the two prescription methodologies. For the first prescription methodology, these dose ratios were 2.13 and 2.34 for the BIMST and the 1-cm MIMiC plans, respectively. For the second prescription methodology, these dose ratios were 1.47 and 1.46 for the BIMST and the 1-cm MIMiC plans, respectively. The number of treatment arcs for each plan was measured as a surrogate for treatment time. For BIMST vs. 1-cm MIMiC using the first prescription methodology, the mean number of arcs was 6.2 compared to 3.3, respectively (95%CIdiff from 1.4 to 4.3, p ⫽ 0.0015). For BIMST vs. 1-cm MIMiC using the second prescription methodology, the mean number of arcs was 4.8 compared to 2.7, respectively (95%CIdiff from 1.7 to 2.6, p ⱕ 0.0001). For both prescription methodologies, results of a multivariate analysis indicated that as target dose increased, the difference between the two modalities with regard to critical structures dose increased. When the BIMST treatment plans were compared with the 1-cm MIMiC plans, it was seen that
Dosimetric comparison of Beak and 1-cm MIMiC collimators
the DVHs improved when the Beak was used. For the second prescription methodology only, results indicated that as target volume increased, the difference between the BIMST and 1-cm MIMiC normal structure dose decreased, and the difference in measured CIs also decreased. Dose distributions and DVHs for target and normal tissue for a representative patient with the second prescription methodology are depicted in Fig. 2. For each critical structure, the BIMST plans produced significantly better DVHs than the 1-cm MIMiC plans. The V5 for the optic chiasm was 40% and 87%, the V10 for the left optic nerve was 41% and 20%, the V10 for the left orbit was 10% and 22%, and the V30 for the pituitary was 90% and 62%, respectively. DISCUSSION The results show that patients with irregularly shaped CNS lesions located adjacent to critical structures may benefit with IMRT using the Beak collimator. This study demonstrated a better CI and significant improvement in dose minimization to several critical structures for BIMST without sacrificing dose homogeneity. CI and tissue sparing were both assessed, because excellent conformity does not necessarily assure adequate dose falloff to surrounding structures; these results have shown that BIMST successfully accomplishes both. As expected, when BIMST vs. 1-cm MIMiC was used, small tumor volumes seemed to benefit significantly more than larger tumor volumes. This parameter and lesion shape seem to be appropriate indicators for determining whether a patient may benefit with treatment using the Beak collimator. Although lesion shape is not readily quantifiable, it appeared that irregularly shaped lesions achieved the greatest benefit with BIMST. The spherically shaped metastatic tumor to the orbit and the pituitary adenomas appeared to gain the least benefit in CI from the Beak, whereas the remaining irregularly shaped lesions gained a substantial improvement in their CI. Interestingly, tumors treated to higher doses seemed to benefit from the use of the Beak collimator more than those lesions treated to lower doses. This may be a factor to consider when attempting to optimize a treatment plan with IMRT. The differing results observed with bilateral organs were primarily due to lateralized tumor locations. Generally, conformity indexes were better with the first methodology, but admittedly, this may have been biased, as later discussed. Our CI values calculated with the second prescription methodology were compared with results by Kno¨ o¨ s et al. (12) and by Nakamura et al. (14). On average, Kno¨ o¨ s et al. (12) obtained CI values of 0.4, whereas Nakamura et al. (14). obtained values of 1.4. These values are slightly lower than those obtained in our study, with average CI values of 2.0 and 2.7 for the BIMST and 1-cm MIMiC plans, respectively. However, in the former study, the target volumes were substantially larger than in the current study, and for the latter study, the CI was calculated at 50% of the prescribed dose.
● H. D. ZINKIN et al.
289
Based on its smaller collimator width, the Beak should be ideal for lesions that are longer in the gantry axis rotational direction. When designing a BIMST plan for irregularly shaped lesions, it is helpful to stipulate that the gantry angle be parallel to the direction of the longest tumor dimension. A graphic depiction of the predicted advantage of Beak over the 1-cm MIMiC is presented in Fig. 3, where the black irregularity represents the clinical target volume. Complete tumor coverage requires each modality to configure a set of beamlets. The area of these beamlets is composed of healthy tissue (gray) and target (black). For BIMST, a total of 6 arcs is required, whereas 3 arcs are needed for the 1-cm MIMiC plan. The total irradiated area is 5.46 cm2 and 6.80 cm2 for BIMST and 1-cm MIMiC, respectively. The proportions of these irradiated areas subtended by target are 68% and 54% for the BIMST and 1-cm MIMiC, respectively. Thus, for BIMST, 32% of the irradiated volume is normal tissue, whereas for 1-cm MIMiC, 46% of the irradiated area is normal tissue. From this example, one can conclude that the BIMST treatment plan theoretically reduces toxicity. To perform a valid comparison of treatment plans, it was necessary to affix evaluation parameters and define a constant prescription methodology. Initially, the selected parameter was a uniform prescribed isodose line. Because isodose choice is subjective, the study could have been biased against the 1-cm MIMiC. Therefore, a second prescription methodology using uniform target volume coverage was employed, and treatment plans were recalculated and data recollected. With this second prescription methodology, DVHs were more meaningfully interpreted, and results seemed to have greater validity, because each plan was independently set to the 95% target volume coverage. With this second methodology, significant difference between the two modes (BIMST and 1-cm MIMiC) remained evident. Between the two different prescription methodologies, the second is preferred. One might argue that the increase in treatment time associated with this small collimator is not acceptable. For symmetric, round tumors and those located in other body sites that cannot be readily immobilized, we would agree. However, radiotherapy via IMRT with any size collimator usually entails a longer treatment time than three-dimensional conformal radiotherapy. If treatment with BIMST limits dose to, for example, the optic apparatus, thus preventing visual loss, we advocate that the additional treatment time is acceptable. Beak intensity modulated sequential tomotherapy may have advantages as definitive therapy, as a boost after previous irradiation, or as a retreatment method. In definitive therapy and for boosts, BIMST can apply more conformal doses to the tumor and deliver lower doses to normal structures than conventional teletherapy, which may decrease acute and late toxicity. For recurrent tumors, retreatment with BIMST may allow higher dose escalation than previously recognized (16, 17). Ten years ago, the introduction of IMRT was met with anticipation as a fractionated means for administering ra-
290
I. J. Radiation Oncology
● Biology ● Physics
Volume 59, Number 1, 2004
Fig. 2. For a representative patient (G), we show: (a) beak intensity modulated sequential tomotherapy (BIMST) axial view and (b) 1-cm MIMiC axial view with (c) BIMST and (d) 1-cm MIMiC sagittal views. The purple isodose line corresponds to the prescription line (84%), and the 90%, 70%, and 50% isodose lines are red, yellow, and green, respectively. In Figs. 2e and 2f, the BIMST and 1-cm MIMiC DVHs are presented for various clinical target volumes.
Dosimetric comparison of Beak and 1-cm MIMiC collimators
● H. D. ZINKIN et al.
291
Fig. 3. Geometric advantage of the beak intensity modulated sequential tomotherapy (BIMST) vs. 1-cm MIMiC for irradiation of irregularly shaped lesions, where the identical black region represents the clinical target volume for comparison between both modalities.
diosurgery. Beak IMST has been shown to be advantageous over Gamma Knife radiosurgery when a critical organ is adjacent to the tumor (18) and may be optimally implemented using radiobiologic modeling (19). Furthermore, prescribed doses used for single-fraction stereotactic radiosurgery treatments may be limited, because normal tissue tolerance is lower with 1 large fraction. Results of RTOG
90 – 05 using single treatment fractions indicated the maximum tolerable single dose for retreatment of CNS tumors was 27 Gy for tumors with a maximum length of 20 mm. For tumors with a maximum length between 31 and 40 mm, the maximum tolerable dose was 15 Gy (16). Unlike stereotactic radiosurgery, BIMST relies on multifractionated delivery and may permit safer dose escalation.
REFERENCES 1. Teh BS, Mai WY, Grant WH, et al. Intensity modulated radiotherapy (IMRT) decreases treatment-related morbidity and potentially enhances tumor control. Cancer Investig 2002; 20:437–451. 2. Larson DA, Wara WM. Radiotherapy of primary malignant brain tumors. Semin Surg Oncol 1998;14:34–42. 3. Pirzkall A, Carol M, Lohr F, Hoss A, Wannenmacher M, Debus J. Comparison of intensity-modulated radiotherapy with conventional conformal radiotherapy for complexshaped tumors. Int J Radiat Oncol Biol Phys 2000;48:1371– 1380. 4. Thilmann C, Zabel A, Grosser KH, Hoess A, Wannenmacher M, Debus J. Intensity-modulated radiotherapy with an integrated boost to the macroscopic tumor volume in the treatment of high-grade gliomas. Int J Cancer 2001;96:341–349. 5. Pirzkall A, Debus J, Haering P, et al. Intensity modulated radiotherapy (IMRT) for recurrent, residual, or untreated skull-base meningiomas: Preliminary clinical experience. Int J Radiat Oncol Biol Phys 2003;55:362–372. 6. Fiveash JB, Murshed H, Duan J, et al. Effect of multileaf collimator leaf width on physical dose distributions in the treatment of CNS and head and neck neoplasms with intensity modulated radiation therapy. Med Phys 2002;29:1116–1119. 7. Low D. Physics of intensity modulated radiation therapy for
8. 9.
10. 11.
12. 13. 14.
head and neck cancer. In: Chao KC, Ozyigit G, editors. Intensity modulated radiation therapy for head and neck cancer. St. Louis: Lippincott Williams and Wilkins, 2003: p. 1–15. Salter B. NOMOS Peacock IMRT utilizing the Beak post collimation device. Med Dosim 2001;26:37–45. Tsai J-S, Engler MJ, Ling MN, et al. A non-invasive immobilization system and related quality assurance for dynamic modulated radiation therapy of intracranial and head and neck disease. Int J Radiat Oncol Biol Phys 1999;43:455–467. Schulz RJ, Kagan AR. On the role of intensity-modulated radiation therapy in radiation oncology. Med Phys 2002;29: 1473–1482. ICRU Prescribing, recording and reporting photon beam therapy (Supplement to ICRU Report 50), International Commission of Radiation Units and Measurements (ICRU 62, Bethesda, MD1999). Kno¨ o¨ s T, Kristensen I, Nilsson P. Volumetric and dosimetric evaluation of radiation treatment plans: Radiation conformity index. Int J Radiat Oncol Biol Phys 1998;42:1169–1176. Paddick I. A simple scoring ratio to index the conformity of radiosurgical treatment plans. J Neurosurg 2000;93:219– 222. Nakamura JL, Verhey LJ, Smith V, et al. Dose conformity of
292
I. J. Radiation Oncology
● Biology ● Physics
Gamma Knife radiosurgery and risk factors for complications. Int J Radiat Oncol Biol Phys 2001;51:1313–1319. 15. Borden JA, Mahajan A, Tsai J-S. A quality factor to compare the dosimetry of gamma knife radiosurgery and intensitymodulated radiation therapy quantitatively as a function of target volume and shape. J Neurosurg 2000;93:228–232. 16. Shaw E, Scott C, Souhami L, et al. Radiosurgery for the treatment of previously irradiated recurrent primary brain tumors and brain metastases: Initial report of Radiation Therapy Oncology Group protocol 90 – 05. Int J Radiat Oncol Biol Phys 1996;34:647–654.
Volume 59, Number 1, 2004
17. Bauman GS, Sneed PK, Wara WM, et al. Reirradiation of primary CNS tumors. Int J Radiat Oncol Biol Phys 1996;36: 433–441. 18. Nakamura JL, Pirzkall A, Carol MP, et al. Comparison of intensity-modulated radiosurgery with gamma knife radiosurgery for challenging skull base lesions. Int J Radiat Oncol Biol Phys 2003;55:99–109. 19. Mignano JE, Engler MJ, Tsai J-S, Wazer DE. Comparison of radiobiologic modeling for one- and two-isocenter dose distributions applied to ellipsoidal radiosurgery targets. Int J Radiat Oncol Biol Phys 2001;49:833–837.
doi:10.1016/j.ijrobp.2004.01.018
PHYSICS CONTRIBUTION
ANALYSIS OF DOSE CONFORMITY AND NORMAL-TISSUE SPARING USING TWO DIFFERENT IMRT PRESCRIPTION METHODOLOGIES FOR IRREGULARLY SHAPED CNS LESIONS IRRADIATED WITH THE BEAK AND 1-CM MIMiC COLLIMATORS HEATHER D. ZINKIN, M.D., MARK J. RIVARD, PH.D., JOHN E. MIGNANO, M.D., PH.D., DAVID E. WAZER, M.D.
AND
Department of Radiation Oncology, Tufts–New England Medical Center, Tufts University School of Medicine, Boston, MA Purpose: To determine whether intensity modulated sequential tomotherapy using the NOMOS Beak provides superior dose conformity and organ sparing to the MIMiC “1-cm” mode, and if so, to identify a subset of patients most likely to benefit from Beak intensity modulated sequential tomotherapy. Methods and Materials: Twelve patients with irregularly shaped central nervous system tumors were selected for intensity modulated radiation therapy planning. Two treatment plans, one using the Beak collimator and the other using the 1-cm MIMiC collimator, were generated for each patient with identical anatomic contouring, prescriptions, and optimization algorithms. The Beak attaches to the MIMiC collimator and truncates the 1-cm MIMiC mode beamlet size from 1.00 ⴛ 0.85 cm2 to 1.00 ⴛ 0.39 cm2 at isocenter. Conformity indexes were calculated for each lesion using two different prescription methodologies, and mean doses to critical structures were recorded. Results: For the first prescription methodology using uniform prescribed isodose, mean conformity index was 2.19 (range, 1.33–3.90) for the Beak compared to 2.67 (range, 1.64 – 4.75) for the 1-cm mode (p ⴝ 0.0003). Mean doses to the brainstem, right orbit, and left optic nerve were significantly lower with the Beak than with the 1-cm mode (p ⴝ 0.0150, 0.0068, and 0.0284, respectively). For the second prescription methodology using uniform target volume coverage prescription, mean conformity index was 2.04 (range, 1.56 –2.70) for the Beak compared to 2.73 (range, 1.70 – 8.58) for the 1-cm mode (p ⴝ 0.07). Mean doses to the brain, brainstem, optic chiasm, right optic nerve, left optic nerve, and left orbit were significantly lower with the Beak than with the 1-cm mode (p ⴝ <0.0001, <0.0026, <0.0016, <0.0076, <0.0007, and <0.046, respectively). Conclusion: Beak intensity modulated sequential tomotherapy is superior to the 1-cm MIMiC mode for irregularly shaped central nervous system tumors, because it provides better conformity and critical organ sparing. These differences may allow for safer dose escalation and retreatment, so the method presents an alternative to gamma knife stereotactic radiosurgery. © 2004 Elsevier Inc. Intensity modulated radiation therapy, Beak, Conformity index, Central nervous system tumor.
There are several types of IMRT delivery techniques. Stepand-shoot employs multileaf collimators with stationary gantry angles, and intensity modulated sequential tomotherapy (IMST) uses the MIMiC collimator with continuous gantry rotation around the patient. During tomotherapy, the dose and rotation rates are usually kept constant during treatment delivery (7). The MIMiC collimator (NOMOS Corporation, Cranberry Township, PA) can administer IMST using 1 of 2 modes: (i) a 1-cm MIMiC collimator with a beamlet area of 1.00 ⫻ 0.85 cm2 at isocenter, and (ii) a 2-cm MIMiC collimator with a beamlet size of 1.00 ⫻ 1.70 cm2. For many tumors,
INTRODUCTION Many tumors, particularly those of the central nervous system (CNS), are challenging to treat aggressively with conventional radiation therapy (1, 2). Dose intensification for treatment of CNS lesions is limited by proximity of critical structures such as optic nerves, chiasm, spinal cord, and brainstem. Compared to conventional teletherapy, intensity modulated radiation therapy (IMRT) was developed to deliver more conformal isodose distributions and to better spare normal tissues (3–5). However, the efficacy of IMRT may be only as effective as the radiation beamlet size (6).
Acknowledgments—We extend our appreciation to Robin Ruthazer, M.P.H., for her assistance with statistical analyses, and to Drs. Jen-San Tsai and Mark J. Engler for their contributions. Received Jun 4, 2003, and in revised form Dec 12, 2003. Accepted for publication Jan 16, 2004.
Reprint requests to: Heather D. Zinkin, M.D., Department of Radiation Oncology (Box 359), Tufts–New England Medical Center, 750 Washington Street, Boston, MA 02111. Tel: (617) 6366161; Fax: (617) 636-6131; E-mail: [email protected] The NOMOS Corporation is acknowledged for manuscript review and funding publication of the color figures. 285
286
I. J. Radiation Oncology
● Biology ● Physics
Volume 59, Number 1, 2004
METHODS AND MATERIALS
Fig. 1. The MIMiC collimator is attached to the collimator head of a linear accelerator. The MIMiC is operated using a controller that delivers pneumatic pressure and electronic feedback via the tubes attached from the gantry. The Beak collimator is attached to the MIMiC and further reduces the field size along the direction of the gantry rotational axis.
these beamlets are small enough to provide conformal treatment plans. However, for tumors that are irregularly shaped with dimensions ⬍0.85 cm along the gantry rotational axis, treatment delivery with the 1-cm and 2-cm modes will not provide adequate tumor conformity; such tumors would benefit from smaller beamlets. Therefore, NOMOS has developed the Beak collimator as a treatment device that can create a more conformal treatment plan while sparing critical structures. The Beak (Fig. 1) attaches to the MIMiC collimator and truncates the 1-cm MIMiC mode beamlet size from 1.00 ⫻ 0.85 cm2 to 1.00 ⫻ 0.39 cm2 at isocenter (8). This therapy is referred to as Beak intensity modulated sequential tomotherapy (BIMST) and is simple for radiation therapists to employ on a daily basis. However, there is no validation mechanism or electronic sensor to assure its routine placement. Because use of the Beak is gaining in popularity, the purpose of this study was to compare BIMST with the 1-cm MIMiC collimator and to determine the methods’ effects on dose distributions, conformity index (CI), and relationship to target size and shape for patients with CNS tumors.
A research study was performed using pre-existing data for 15 patients previously treated with IMRT as definitive therapy or boost at the Tufts–New England Medical Center Department of Radiation Oncology. The lesions included the following: 5 pituitary adenomas, 4 meningiomas, 2 glioblastoma multiforme, 1 acoustic neuroma, 1 chondrosarcoma, 1 mixed glioma, and 1 metastatic melanoma. The patients were chosen based on the presence of CNS lesions, anatomic shape of the tumor, and the tumor proximity to normal tissues. The lesions represented relatively common diagnoses, and therefore results of this study may be of general interest. Patients were preferentially selected if their tumor was small, flattened (short dimension along the axis of gantry rotation), or had longitudinal irregularity along the gantry rotational axis, which was restricted to gantry rotational arcs ranging from 330° to 30° for this study. All patients were pathologically diagnosed with CNS tumors, and each patient was positioned with a custom-fit ␣-cradle head holder and aquaplast mask constructed for immobilization (9). Computed tomography images were obtained in the treatment position with 1–3-mm-thick slices. Axial images were transferred to the NOMOS CORVUS, version 5.0R1 IMRT treatment planning system. Planning tumor volume and normal structures, including the brainstem, brain, right and left optic nerves, optic chiasm, and right and left orbits, were contoured by a physician for each patient. Each image set was then duplicated so that 2 treatment plans could be generated, and prescriptions could be made with identical anatomic delineations for meaningful comparisons of 1-cm MIMiC vs. BIMST dose distributions. In total, 54 treatment plans were prepared for this study, and two distinct prescription methodologies were used to compare treatment plans. Uniform prescribed isodose The first prescription methodology included 12 patients, and identical prescription isodoses were selected for both plans of each patient, as shown in Table 1. In all but 1 case, the isodose line for the BIMST plan was chosen that covered at least 90% of the tumor volume. This same isodose line was then assigned to the 1-cm MIMiC plan and served as the variable through which plans were compared. These prescribed isodoses ranged from 73.0% to 89.0% (mean, 81.9%) for both the BIMST and 1-cm MIMiC plans. Target volumes ranged from 2.05 cm3 to 126.37 cm3 (mean, 23.09 cm3), as shown in Table 1; excluding the outlier lesion of 126.37 cm3, the mean target volume significantly decreased to 13.70 cm3. Numeric values were assigned for the target goal dose (prescription dose), percent volume below this goal, and the acceptable minimum and maximum target doses (Gy). The minimum acceptable critical structure dose was set at 0 Gy for all patients. Uniform target volume coverage prescription The second prescription methodology included 15 patients, and treatment plans were computed using 95% target
Dosimetric comparison of Beak and 1-cm MIMiC collimators
● H. D. ZINKIN et al.
287
Table 1. Patient, treatment parameters, and CI for the BIMST and 1-cm MIMiC collimators using the uniform prescribed isodose methodology
Patient
Diagnosis
A B C D E F G H I J K L
Glioblastoma multiforme Meningioma Pituitary adenoma Pituitary adenoma Pituitary adenoma Acoustic neuroma Chondrosarcoma Meningioma Acoustic neuroma Mixed glioma Metastatic tumor to orbit Meningioma
Target volume Target goal BIMST 1-cm MIMiC (cm3) (Gy) CI CI 18.13 24.87 2.66 3.79 17.00 22.77 32.76 12.96 3.47 126.37 2.05 10.26
25 45 16 45 45 50 66 45 50 14 30 45
1.33 1.65 1.65 2.66 1.69 1.83 2.25 2.03 3.88 1.42 3.90 1.95
BIMST and 1cm MIMiC Rx isodose (%)
1.76 2.06 2.00 2.99 1.87 2.63 2.36 3.26 4.48 1.64 4.75 2.27
82.0 79.8 73.0 87.0 83.3 82.0 88.0 84.0 89.0 80.5 75.0 79.0
BIMST 1-cm MIMiC % target % target coverage coverage 94.9 95.1 76.4 96.3 93.6 92.2 91.4 97.1 95.0 95.0 83.3 93.6
96.1 96.6 56.6 98.6 94.9 91.0 94.3 100.0 98.7 96.2 95.4 88.6
Abbreviations: CI ⫽ conformity index; BIMST ⫽ Beak intensity modulated sequential tomotherapy.
volume coverage prescriptions for plan normalization and comparison (10). This approach differed from the first prescription methodology in that target volume coverage was now affixed. For the BIMST plans, these prescribed isodoses ranged from 80.1% to 88.5% (mean, 84.4%) and ranged from 59.6% to 85.6% (mean, 81.0%) for the 1-cm MIMiC plans, as shown in Table 2. Target volumes also ranged from 2.05 cm3 to 126.37 cm3 (mean, 19.66 cm3), as shown in Table 1; excluding the outlier lesion of 126.37 cm3, the mean target volume significantly decreased to 12.03 cm3. Numeric values were assigned for the target goal dose (prescription dose), percent volume below this goal, and the acceptable minimum and maximum target doses (Gy). Although the first prescription methodology set the minimum acceptable critical structure dose at 0 Gy for all
patients, this practice may unnecessarily bias the optimizer. Therefore, the second prescription methodology set the minimum acceptable critical structure doses to more realistic values. For all patients, this dose was specified as ⱖ1 Gy for both the 1-cm MIMiC and BIMST plans. For both prescription methodologies, constraints were assigned for normal structure dose limits, volume above limit, and minimum and maximum acceptable doses. The optic nerve and optic chiasm dose limit specifications varied among patients and depended largely on proximity to the target; realistic expectations of dose were prescribed. If judged to be pertinent by the physician, target prescriptions were marked for prioritization in treatment planning. Based on the immobilization technique described by Tsai et al., positional and structure motion uncertainties were set to 1.0 mm in all directions (9). Except for 1 patient, all treatments
Table 2. Patient and treatment parameters for BIMST and 1-cm MIMiC intensity modulated sequential tomotherapy with uniform target volume coverage prescription Patient
Diagnosis
Target volume (cm3)
Target goal (Gy)
BIMST CI
1-cm MIMiC CI
BIMST Rx isodose (%)
1-cm MIMiC Rx isodose (%)
A B C D E F G H J K L M N O P
Glioblastoma multiforme Meningioma Pituitary adenoma Pituitary adenoma Pituitary adenoma Acoustic neuroma Chondrosarcoma Meningioma Mixed glioma Metastatic tumor to orbit Meningioma Metastatic melanoma Meningioma Glioblastoma multiforme Pituitary adenoma
18.13 24.87 2.66 3.79 17.00 22.77 32.76 12.96 126.37 2.05 10.26 2.14 3.56 9.86 5.66
25 45 16 45 45 50 66 45 14 30 45 25 45 28 45
1.91 1.92 2.65 2.04 1.64 1.88 1.61 2.31 1.61 2.70 2.20 2.62 2.26 1.56 1.72
2.50 2.09 8.58 2.58 1.73 2.04 1.77 2.97 1.70 2.18 2.39 3.23 3.00 2.16 2.04
85.2 85.0 81.0 87.7 85.1 81.0 86.6 87.7 83.3 84.4 82.9 80.1 83.2 84.2 88.5
83.5 82.1 59.6 84.4 85.6 82.3 85.1 80.1 84.3 85.4 85.2 79.2 79.5 75.5 82.6
Abbreviations: Same as for Table 1. Note: For all cases, the BIMST CI is compared to the 1-cm MIMiC CI using a 95% target volume coverage.
288
I. J. Radiation Oncology
● Biology ● Physics
were delivered with a single couch angle of 180°. Mean doses to critical structures were recorded for each plan, as well as the ratio of maximum to minimum dose received by the target. Furthermore, dose–volume histograms (DVHs) for the two modalities were also compared. Conformity index Conformity index was initially defined according to ICRU 62 (11) for use as a metric of tightness-of-fit of the planning target volume to the prescribed isodose volume. This definition was later modified and improved upon (12– 15). CI accounts for the total target volume (TV), prescription isodose surface volume (PV), and the ratio of target volume within the prescribed isodose surface to the total target volume (TVPV). CI is defined below (Eq. 1): PV/TV PV CI ⫽ , which may be rearranged as: TV PV/TV CI ⫽
PV · TV . TV PV2
(1)
A perfect treatment plan would set TVPV ⫽ TV ⫽ PV, yielding CI ⫽ unity (14); the closer CI is to 1, the more conformal the plan. CI values were calculated for all 54 plans (2 per patient). Statistical analysis In addition to CI, other outcome measures were analyzed, including: 1. ratio of maximum to minimum doses received by the target, used to model dose homogeneity, 2. number of treatment arcs, used to model overall treatment time, and 3. multivariate analysis to assess the relationship between target dose, target volume, and critical structure doses. Statistical analyses (SAS version 8.0) using paired t tests were used to determine whether significant differences between the two treatment modalities could be appreciated. Ninety-five percent confidence intervals were calculated for the difference between the mean doses to critical structures (95%CIdiff). Multivariate analyses with Spearman rank correlation coefficients were also performed to determine whether relationships existed between target volume, CI, and mean dose to critical structures. RESULTS For the uniform prescribed isodose methodology, mean doses to the brainstem, right orbit, and left optic nerve were significantly lower with the BIMST compared to the 1-cm MIMiC. Mean brainstem dose was 11.72 vs. 13.06 Gy for BIMST and 1-cm MIMiC, respectively (95%CIdiff 0.32 to 2.31, p ⫽ 0.0150). Mean right orbit dose was 5.39 Gy vs. 6.05 Gy for BIMST and 1-cm MIMiC, respectively (95%CIdiff 0.23 to 1.09, p ⫽ 0.0068). Left optic nerve mean
Volume 59, Number 1, 2004
dose was also lower with BIMST: 10.84 vs. 12.17 Gy (95%CIdiff 0.19 to 2.49, p ⫽ 0.0284). The CI was also significantly better with BIMST than with 1-cm MIMiC (2.19 vs. 2.67, p ⫽ 0.0003), as shown in Table 1. The second prescription methodology, uniform target volume coverage prescription, produced more significant differences in normal structure dose than the first prescription methodology. The mean doses to the (i) brain, (ii) brainstem, (iii) optic chiasm, (iv) right optic nerve, (v) left optic nerve, and (vi) left orbit were significantly lower with the BIMST compared to the 1-cm MIMiC plans. For the aforementioned structures, the mean doses were significantly lower with BIMST than with 1-cm MIMiC: i. 5.13 Gy vs. 5.95 Gy (95%CIdiff 0.54 to 1.09, p 0.0001), ii. 10.94 Gy vs. 12.28 Gy (95%CIdiff 0.57 to 2.11, p 0.003), iii. 20.95 Gy vs. 22.79 Gy (95%CIdiff 0.84 to 2.84, p 0.002), iv. 12.21 Gy vs. 13.14 Gy (95%CIdiff 0.29 to 1.56, p 0.008), v. 13.08 Gy vs. 14.42 Gy (95%CIdiff 0.68 to 2.01, p 0.007), and vi. 4.19 Gy vs. 4.58 Gy (95%CIdiff 0.01 to 0.76, p 0.046).
⬍ ⫽ ⫽ ⫽ ⫽ ⫽
Although the mean doses to the right orbit were lower with BIMST than with the 1-cm MIMiC (5.48 Gy vs. 5.95 Gy), this dosimetric improvement was less convincing (p ⫽ 0.08) than was observed for the 6 previously mentioned structures. For the second prescription methodology, the mean CI was also better (lower) for BIMST than for the 1-cm MIMiC (2.04 vs. 2.00, p ⫽ 0.07), as shown in Table 2, and for all but 1 patient, CI with BIMST was lower than with 1-cm MIMiC. Ratios of maximum to minimum dose received by the targets were used as a measure of target dose homogeneity and were not significantly different for the BIMST and the 1-cm MIMiC plans and for the two prescription methodologies. For the first prescription methodology, these dose ratios were 2.13 and 2.34 for the BIMST and the 1-cm MIMiC plans, respectively. For the second prescription methodology, these dose ratios were 1.47 and 1.46 for the BIMST and the 1-cm MIMiC plans, respectively. The number of treatment arcs for each plan was measured as a surrogate for treatment time. For BIMST vs. 1-cm MIMiC using the first prescription methodology, the mean number of arcs was 6.2 compared to 3.3, respectively (95%CIdiff from 1.4 to 4.3, p ⫽ 0.0015). For BIMST vs. 1-cm MIMiC using the second prescription methodology, the mean number of arcs was 4.8 compared to 2.7, respectively (95%CIdiff from 1.7 to 2.6, p ⱕ 0.0001). For both prescription methodologies, results of a multivariate analysis indicated that as target dose increased, the difference between the two modalities with regard to critical structures dose increased. When the BIMST treatment plans were compared with the 1-cm MIMiC plans, it was seen that
Dosimetric comparison of Beak and 1-cm MIMiC collimators
the DVHs improved when the Beak was used. For the second prescription methodology only, results indicated that as target volume increased, the difference between the BIMST and 1-cm MIMiC normal structure dose decreased, and the difference in measured CIs also decreased. Dose distributions and DVHs for target and normal tissue for a representative patient with the second prescription methodology are depicted in Fig. 2. For each critical structure, the BIMST plans produced significantly better DVHs than the 1-cm MIMiC plans. The V5 for the optic chiasm was 40% and 87%, the V10 for the left optic nerve was 41% and 20%, the V10 for the left orbit was 10% and 22%, and the V30 for the pituitary was 90% and 62%, respectively. DISCUSSION The results show that patients with irregularly shaped CNS lesions located adjacent to critical structures may benefit with IMRT using the Beak collimator. This study demonstrated a better CI and significant improvement in dose minimization to several critical structures for BIMST without sacrificing dose homogeneity. CI and tissue sparing were both assessed, because excellent conformity does not necessarily assure adequate dose falloff to surrounding structures; these results have shown that BIMST successfully accomplishes both. As expected, when BIMST vs. 1-cm MIMiC was used, small tumor volumes seemed to benefit significantly more than larger tumor volumes. This parameter and lesion shape seem to be appropriate indicators for determining whether a patient may benefit with treatment using the Beak collimator. Although lesion shape is not readily quantifiable, it appeared that irregularly shaped lesions achieved the greatest benefit with BIMST. The spherically shaped metastatic tumor to the orbit and the pituitary adenomas appeared to gain the least benefit in CI from the Beak, whereas the remaining irregularly shaped lesions gained a substantial improvement in their CI. Interestingly, tumors treated to higher doses seemed to benefit from the use of the Beak collimator more than those lesions treated to lower doses. This may be a factor to consider when attempting to optimize a treatment plan with IMRT. The differing results observed with bilateral organs were primarily due to lateralized tumor locations. Generally, conformity indexes were better with the first methodology, but admittedly, this may have been biased, as later discussed. Our CI values calculated with the second prescription methodology were compared with results by Kno¨ o¨ s et al. (12) and by Nakamura et al. (14). On average, Kno¨ o¨ s et al. (12) obtained CI values of 0.4, whereas Nakamura et al. (14). obtained values of 1.4. These values are slightly lower than those obtained in our study, with average CI values of 2.0 and 2.7 for the BIMST and 1-cm MIMiC plans, respectively. However, in the former study, the target volumes were substantially larger than in the current study, and for the latter study, the CI was calculated at 50% of the prescribed dose.
● H. D. ZINKIN et al.
289
Based on its smaller collimator width, the Beak should be ideal for lesions that are longer in the gantry axis rotational direction. When designing a BIMST plan for irregularly shaped lesions, it is helpful to stipulate that the gantry angle be parallel to the direction of the longest tumor dimension. A graphic depiction of the predicted advantage of Beak over the 1-cm MIMiC is presented in Fig. 3, where the black irregularity represents the clinical target volume. Complete tumor coverage requires each modality to configure a set of beamlets. The area of these beamlets is composed of healthy tissue (gray) and target (black). For BIMST, a total of 6 arcs is required, whereas 3 arcs are needed for the 1-cm MIMiC plan. The total irradiated area is 5.46 cm2 and 6.80 cm2 for BIMST and 1-cm MIMiC, respectively. The proportions of these irradiated areas subtended by target are 68% and 54% for the BIMST and 1-cm MIMiC, respectively. Thus, for BIMST, 32% of the irradiated volume is normal tissue, whereas for 1-cm MIMiC, 46% of the irradiated area is normal tissue. From this example, one can conclude that the BIMST treatment plan theoretically reduces toxicity. To perform a valid comparison of treatment plans, it was necessary to affix evaluation parameters and define a constant prescription methodology. Initially, the selected parameter was a uniform prescribed isodose line. Because isodose choice is subjective, the study could have been biased against the 1-cm MIMiC. Therefore, a second prescription methodology using uniform target volume coverage was employed, and treatment plans were recalculated and data recollected. With this second prescription methodology, DVHs were more meaningfully interpreted, and results seemed to have greater validity, because each plan was independently set to the 95% target volume coverage. With this second methodology, significant difference between the two modes (BIMST and 1-cm MIMiC) remained evident. Between the two different prescription methodologies, the second is preferred. One might argue that the increase in treatment time associated with this small collimator is not acceptable. For symmetric, round tumors and those located in other body sites that cannot be readily immobilized, we would agree. However, radiotherapy via IMRT with any size collimator usually entails a longer treatment time than three-dimensional conformal radiotherapy. If treatment with BIMST limits dose to, for example, the optic apparatus, thus preventing visual loss, we advocate that the additional treatment time is acceptable. Beak intensity modulated sequential tomotherapy may have advantages as definitive therapy, as a boost after previous irradiation, or as a retreatment method. In definitive therapy and for boosts, BIMST can apply more conformal doses to the tumor and deliver lower doses to normal structures than conventional teletherapy, which may decrease acute and late toxicity. For recurrent tumors, retreatment with BIMST may allow higher dose escalation than previously recognized (16, 17). Ten years ago, the introduction of IMRT was met with anticipation as a fractionated means for administering ra-
290
I. J. Radiation Oncology
● Biology ● Physics
Volume 59, Number 1, 2004
Fig. 2. For a representative patient (G), we show: (a) beak intensity modulated sequential tomotherapy (BIMST) axial view and (b) 1-cm MIMiC axial view with (c) BIMST and (d) 1-cm MIMiC sagittal views. The purple isodose line corresponds to the prescription line (84%), and the 90%, 70%, and 50% isodose lines are red, yellow, and green, respectively. In Figs. 2e and 2f, the BIMST and 1-cm MIMiC DVHs are presented for various clinical target volumes.
Dosimetric comparison of Beak and 1-cm MIMiC collimators
● H. D. ZINKIN et al.
291
Fig. 3. Geometric advantage of the beak intensity modulated sequential tomotherapy (BIMST) vs. 1-cm MIMiC for irradiation of irregularly shaped lesions, where the identical black region represents the clinical target volume for comparison between both modalities.
diosurgery. Beak IMST has been shown to be advantageous over Gamma Knife radiosurgery when a critical organ is adjacent to the tumor (18) and may be optimally implemented using radiobiologic modeling (19). Furthermore, prescribed doses used for single-fraction stereotactic radiosurgery treatments may be limited, because normal tissue tolerance is lower with 1 large fraction. Results of RTOG
90 – 05 using single treatment fractions indicated the maximum tolerable single dose for retreatment of CNS tumors was 27 Gy for tumors with a maximum length of 20 mm. For tumors with a maximum length between 31 and 40 mm, the maximum tolerable dose was 15 Gy (16). Unlike stereotactic radiosurgery, BIMST relies on multifractionated delivery and may permit safer dose escalation.
REFERENCES 1. Teh BS, Mai WY, Grant WH, et al. Intensity modulated radiotherapy (IMRT) decreases treatment-related morbidity and potentially enhances tumor control. Cancer Investig 2002; 20:437–451. 2. Larson DA, Wara WM. Radiotherapy of primary malignant brain tumors. Semin Surg Oncol 1998;14:34–42. 3. Pirzkall A, Carol M, Lohr F, Hoss A, Wannenmacher M, Debus J. Comparison of intensity-modulated radiotherapy with conventional conformal radiotherapy for complexshaped tumors. Int J Radiat Oncol Biol Phys 2000;48:1371– 1380. 4. Thilmann C, Zabel A, Grosser KH, Hoess A, Wannenmacher M, Debus J. Intensity-modulated radiotherapy with an integrated boost to the macroscopic tumor volume in the treatment of high-grade gliomas. Int J Cancer 2001;96:341–349. 5. Pirzkall A, Debus J, Haering P, et al. Intensity modulated radiotherapy (IMRT) for recurrent, residual, or untreated skull-base meningiomas: Preliminary clinical experience. Int J Radiat Oncol Biol Phys 2003;55:362–372. 6. Fiveash JB, Murshed H, Duan J, et al. Effect of multileaf collimator leaf width on physical dose distributions in the treatment of CNS and head and neck neoplasms with intensity modulated radiation therapy. Med Phys 2002;29:1116–1119. 7. Low D. Physics of intensity modulated radiation therapy for
8. 9.
10. 11.
12. 13. 14.
head and neck cancer. In: Chao KC, Ozyigit G, editors. Intensity modulated radiation therapy for head and neck cancer. St. Louis: Lippincott Williams and Wilkins, 2003: p. 1–15. Salter B. NOMOS Peacock IMRT utilizing the Beak post collimation device. Med Dosim 2001;26:37–45. Tsai J-S, Engler MJ, Ling MN, et al. A non-invasive immobilization system and related quality assurance for dynamic modulated radiation therapy of intracranial and head and neck disease. Int J Radiat Oncol Biol Phys 1999;43:455–467. Schulz RJ, Kagan AR. On the role of intensity-modulated radiation therapy in radiation oncology. Med Phys 2002;29: 1473–1482. ICRU Prescribing, recording and reporting photon beam therapy (Supplement to ICRU Report 50), International Commission of Radiation Units and Measurements (ICRU 62, Bethesda, MD1999). Kno¨ o¨ s T, Kristensen I, Nilsson P. Volumetric and dosimetric evaluation of radiation treatment plans: Radiation conformity index. Int J Radiat Oncol Biol Phys 1998;42:1169–1176. Paddick I. A simple scoring ratio to index the conformity of radiosurgical treatment plans. J Neurosurg 2000;93:219– 222. Nakamura JL, Verhey LJ, Smith V, et al. Dose conformity of
292
I. J. Radiation Oncology
● Biology ● Physics
Gamma Knife radiosurgery and risk factors for complications. Int J Radiat Oncol Biol Phys 2001;51:1313–1319. 15. Borden JA, Mahajan A, Tsai J-S. A quality factor to compare the dosimetry of gamma knife radiosurgery and intensitymodulated radiation therapy quantitatively as a function of target volume and shape. J Neurosurg 2000;93:228–232. 16. Shaw E, Scott C, Souhami L, et al. Radiosurgery for the treatment of previously irradiated recurrent primary brain tumors and brain metastases: Initial report of Radiation Therapy Oncology Group protocol 90 – 05. Int J Radiat Oncol Biol Phys 1996;34:647–654.
Volume 59, Number 1, 2004
17. Bauman GS, Sneed PK, Wara WM, et al. Reirradiation of primary CNS tumors. Int J Radiat Oncol Biol Phys 1996;36: 433–441. 18. Nakamura JL, Pirzkall A, Carol MP, et al. Comparison of intensity-modulated radiosurgery with gamma knife radiosurgery for challenging skull base lesions. Int J Radiat Oncol Biol Phys 2003;55:99–109. 19. Mignano JE, Engler MJ, Tsai J-S, Wazer DE. Comparison of radiobiologic modeling for one- and two-isocenter dose distributions applied to ellipsoidal radiosurgery targets. Int J Radiat Oncol Biol Phys 2001;49:833–837.