Int. J. Radiation Oncology Biol. Phys., Vol. 35, No. 4, pp. 721-730, 1996 Copyright 0 1996 Elsevier Science Inc. Printed in the USA. All rights reserved

0360-3016/96 $15.00 + .OO

ELSEVIER

l Physics Original Contribution

PII: SO360-3016(96)00174-5

CONFORMAL RADIATION TREATMENT OF PROSTATE CANCER USING INVERSELY-PLANNED INTENSITY-MODULATED PHOTON BEAMS

PRODUCED WITH DYNAMIC MULTILEAF COLLIMATION

C. CLIFTON LING, PH.D., CHANDRA BURMAN, PH.D., CHEN S. CHUI, PH.D., GERALD J. KUTCHER, PH.D., STEPHEN A. LEIBEL, M.D., THOMAS LOSASSO, PH.D.,

RADHE MOHAN, PH.D., THOMAS BORTFELD, PH.D.,* LARRY RBINSTBIN, PH.D,+ SPIRIDON SPIROU, PH.D., X. H. WANG, PH.D., QIIJWEN WV, PH.D.,

MICHAEL ZELEFSKY, M.D. AND ZVI FUKS, M.D.

Departments of Radiation Oncology and Medical Physics, Memorial Sloan Kettering Cancer Center, New York, NY

purpose: To implement radiotherapy with intensity-modulated beams, based on the inverse method of treatment deslgnd using a multileaf collimation system operating in the dynamic mode. Methods and Materials: An algorithm, based on tbe inverse technique, has been integrated into the radiotherapy treatment-planning computer system in our Center. This method of computer-assisted treatment design was used to derive intensity-modulated beams to optimize the boost portion of the treatment plan for a patient with a Tic cancer of the prostate. A dose of 72 Gy (in 40 fractions) was given with a six-field plan, and an additional 9 Gy (in five fractions) with six intensity-modulated beams. The intensity-modulated fields were delivered using dy- namic multileaf collimation’, that is, individual leaves were in motion during radiation delivery, with the treat- ment machine operating in the clinical mode. Exhaustive quality assurance measurement and monitoring were carried out to ensure safe and accurate implementation. Results: Dose distribution and dose-volume histogram of the “inverse method” boost plan and of the composite (72 Gy primary + 9 Gy boost) plan were judged clinically acceptable. Compared to a manually designed boost plan, the inverse treatment design gave improved conformality and increased dose homogeneity in the planning target volume. Flhn and ion chamber dosimetry, performed prior to the first treatment, indicated that each of the six intensity-modulated fields was accurately produced. Thermoluminescent doshneter (TLD) measurements performed on the patient confirmed that the intended dose was delivered in the treatment. In addition, computer- aided treatment-monitoring programs assured that the multileaf collimator (MLC) position llle was executed to the specified precision. In terms of the overall radiation treatment process, there will likely be labor savings in the planning and the treatment phases. Conclusions: We have placed into clinical use an integrated system of conformal radiation treatment that incor- porated the inverse method of treatment design and the use of dynamic multileaf collimation to deliver intensity- modulated beams. The system can provide better treatment design, which can he implemented reliably and safely. We are hopeful that improved treatment efficacy will result.

Intensity-modulated beams, Multllead collimation, Inverse technique.

INTRODUCTION

Conformal radiotherapy techniques have made it possible to safely increase the radiation doses to unprecedented levels in patients with adenocarcinoma of the prostate (11, 17, 18). The preliminary results of a dose-escalating study,

conducted at Memorial Sloan Kettering Cancer Center (MSKCC) between December 1988 and January 1994, have been recently reported (30). In total, 432 patients with Tic-T3 prostate cancer were treated with doses of 64.8-66.6 Gy (n = 89), 70.2 Gy (n = 199), 75.6 Gy (n = 98), and 8 1 .O Gy (n = 46). An improved prostate-spe-

*Visiting Investigator, June 1993. +Visiting Investigator, 1994. Reprint requests to: C. Clifton Ling, Ph.D., Memorial Sloan

Kettering Cancer Center, 1275 York Avenue, New York, New York 1002 1. Acknowledgements-This work was supported in part by Grant PO1 CA59017 from the National Institutes of Health, DHHS. We are grateful to Richard Levy, Timothy Guertin, Martin Kan-

des, Cal Hunzinger, Eric Lindquist, and William Minowitz of Varian Associates for their consultation and collaboration. The on-site engineering efforts of William Tsirakis (MSKCC), Scott Evan, Keith Burton, and John Au-Yeung (Varian Assocs.) are also acknowledged.

Accepted for publication 6 March 1996. ‘Varian Assoc., Palo Alto, CA.

721

722 I. J. Radiation Oncology 0 Biology 0 Physics Volume 35, Number 4, 1996

cific antigen (PSA) response was observed with the higher doses, whereas late treatment-related morbidity has been absent or minimal (Grade 0 or 1) in about 95% of the patients (17, 30). At the time of submission of this man- uscript, the medium follow-up times for the patients who received 75.6 and 81 Gy were 30 and 27 months, respectively.

We have previously described our approach to 3D con- formal radiation treatment of prostate carcinoma (17, 30). Briefly, six coplanar fields were applied, with apertures designed with Beam’s Eye View such that the planning target volume (PTV) included a 1 cm margin relative to the gross target volume (GTV), except in contiguity with the rectum, where a margin of 0.6 cm was assigned. This beam arrangement was used for all patients who received 75.6 Gy or less. Whereas 75.6 Gy given to the entire rec- tum would be above tolerance, the six field arrangement carried only 25-30% of the rectum to the prescription dose, and as discussed before, late rectal reaction has been minimal (30). For patients in the 81 Gy category, 72 Gy was delivered with the six-field plan, and the remaining 9 Gy with a boost plan consisting of six to eight lateral and posterior-lateral-wedged fields. For the boost fields, in- creased blocking was placed for the rectum to satisfy the constraint that the rectal dose-volume histogram (DVH) in the composite plan is similar to that of a six-field plan (without boost) delivering only 75.6 Gy.

Given the aim of our dose escalation study is to deter- mine the highest prescription dose with acceptable com- plication incidences, upon the completion of patient ac- crual at 81 Gy (in January 1994), we awaited sufficient follow-up time to observe late reactions before further in- crease in dose to 86 Gy. Concomitantly, technical ad- vances in treatment planning and delivery were investi- gated to improve the dose distribution conformality with a view to optimize treatment results. These developments included the inverse technique of treatment design and the use of multileaf collimators in the dynamic mode (DMLC) to deliver intensity-modulated beams (7, 8, 21, 23, 25- 28). Most recently, we have placed these techniques into clinical use in an integrated system of conformal radiation treatment. Radiation treatment using this approach for pa- tients with cancer of the prostate has begun.

METHODS AND MATERIALS

The inverse planning algorithm The inverse planning algorithm, originally designed and

developed by Bortfeld, was adapted for use in the MSKCC 3D conformal radiotherapy treatment design computer (1, 3,21). The characteristics of this methodology and its po- tential for designing conformal radiation treatment have been previously described (21). Several major enhance- ments have also been made, including: (a) the incorpora- tion of scattered radiation into the design of intensity mod- ulated beams; (b) the provision for adding celphalad and caudad margins to the CT-derived target volume, because

the original algorithm limited calculation to only those CT slices containing target contours; and (c) the ability to de- fine regions of overlap (due to organ motion or setup un- certainty) between the target volume and normal critical organ (e.g., the prostate and the rectum) and assign spe- cific dose constraints (21). A user interface has recently been designed and implemented. This interface facilitates the selection of structures of interests, the specifications of constraints, the output of documentations (e.g., beam profiles, port film aperture), and so on. The detail features of this program will be described in another publication that is under preparation. The dose calculation algorithm is based on a pencil beam convolution method developed by us and shown to be accurate and efficient for intensity- modulated beams and DMLC fields (7).

Patient data and treatment design The patient was a 60-year-old, with Stage Tic cancer

of the prostate. The seminal vesicles were included in the primary treatment, but not in the boost plan. The patient was immobilized in the prone position in a thermoplastic cast for computed tomography (CT) scanning, verification simulation and treatment. Thirty-one CT scan images, ob- tained on a GE9800 with slice thickness of 0.5 cm and variable separation of 0.5 and 1 cm, were transferred elec- tronically to the MSKCC treatment planning computer and used for treatment planning. The PTV, rectum, blad- der, and femoral head were contoured on all relevant CT slices. For the first phase of the treatment, a dose of 72 Gy was delivered using a six-field plan, with parallel-op- posed lateral fields and four obliquely angled beams at +45” relative to the horizontal. The beam apertures, ob- tained with Beam’s Eye View, were designed such that the entire PTV received >95% of the prescribed dose. As described previously, the PTV included a 1 cm margin relative to the CTV, except at the boundary of the rectum, where a margin of 0.6 cm was assigned (30). With this beam arrangement, about 25-30% of rectum received the prescription dose, which is acceptable based on our clin- ical data (30).

The boost of 9 Gy in five fractions was planned for a “cone-down” or boost PTV using two methods. First, as was previously performed for patients treated on our dose escalation protocol, it was planned manually by an expe- rienced clinical physicist using six static MLC fields with wedges, consisting of two laterals with 60” wedges, two 45” wedged fields at gantry angles of 60” and 300” (gantry angles in International Electrotechnical Commission con- vention), and two 30” wedged fields at gantry angles of 33” and 327”. This plan also served as the back-up for the DMLC treatment.

Second, the boost was planned with the inverse tech- nique, using the same six-field arrangement of the man- ually designed plan. This conservative approach was adopted so that the inverse planning and the DMLC treat- ment constitute only an incremental advance, and not a radical departure from our standard practice. The desired

RT using intensity-modulated beams and multileaf collimation 0 C. C. Lmc et al. 723

boost dose (1.8 Gy per fraction for five fractions) to the boost FTV and dose limit to the rectum (10% of boost PTV prescription dose, for instance, for a total of 0.9 Gy) were specified. In the inverse technique, the filtered back projection method was applied to obtain a “first guess” fluence distribution for the specified beams (2, 21). An optimization process was then implemented, changing the weight of each ray within each field iteratively, to mini- mize the difference between the desired dose and that which was achieved (1, 3, 21). In the region of overlap between the boost PTV and the rectum, priority was given to the rectum (over the PTV), and the relative penalties assessed for dose deviation were one for the FTV and three for the rectum, respectively. These optimization cri- teria recognized the importance of limiting the probability of rectal injury at the high prescription dose.

Because we have treated a large number of patients (about a hundred) with the standard six-field plan to a prescription dose of 75.6 Gy and have observed minimal late morbidity, we consider this treatment to be well tol- erated (17, 30). Thus, we generated the standard six-field plan for this patient and used the DVHs of the rectum as a reference for comparison with that of the composite plan (i.e., 72 Gy primary treatment plus the inversely planned boost of 9 Gy).

Simulation was then performed. The DMLC fields were superimposed onto the radiographs; this superposition was aided by digitally reconstructed radiographs (DRRs) on which the boundary of the DMLC beams was displayed.

Producing intensity-modulated beams using dynamic multileaf collimators

Spirou and Chui have developed a method to generate the desired intensity profiles using multileaf collimators in the dynamic mode, that is, individual leaves were in motion during the radiation delivery (25). In the so-called “sliding window” technique, the width of each window (as defined by a pair of leaves and which can be variable with time) and its “sliding” speed determined the radia- tion intensity at points underneath the trajectory of the leaves. Their approach, in forcing at least one of each pair of leaves to move at the maximum allowable speed, max- imizes the efficiency of treatment delivery, that is, the monitor units (MU) required for a given radiation dose are kept to a minimum. In addition, the algorithm takes into account the transmitted radiation through the leaves (25). The effect of the rounded leaf end (such that the edge does not represent 50% isodose) was accounted for by using an equivalent field edge 0.75 mm wider than the actual (12, 26). The variation of penumbra with leaf position was ignored because it was negligible for small field sizes. Chui et al. (7) and Wang et al. (26) have recently reported that this method can accurately produce the specified in- tensity profiles, both for arbitrarily designed dose patterns and for clinically applicable fields generated by the in- verse planning method.

The algorithm de&bed above was used to generate a file of MLC positions as a function of MU (the “DMLC file”) for each of the beams, according to the intensity profiles from the inverse plan (25). The DMLC files were then transferred via a “floppy” magnetic disk to the MLC control computer. The data in the files were used by the DMLC software to generate the appropriate MLC motions during radiation delivery, producing the intensity profiles for each of the six intensity modulated beams.

Peeormance evaluation and quality assurance A comprehensive program of evaluating the perfor-

mance of DMLC has been developed and recently re- ported (9,26). This includes the dosimetric verification of intensity modulated beams and a routine quality assurance check procedure (9,26). For this patient, dosimetric ver- ification was performed for all six intensity-modulated fields, the results of which will be described in the next section. The routine QA check consists of a series of tests that would produce predesigned dose patterns if the dy- namic MLC is functioning properly (9). The reproduction of these standard dose patterns during QA checks assures the constancy of leaf speed, the precision of leaf positions, and that the acceleration and deceleration of the leaves do not introduce anomalies (9). The detail results of these studies and checks, as reported in the quoted references (9,26), indicated that DMLC can be reliably and precisely implemented for the delivery of intensity-modulated fields.

Experimental ver@ication of desired intensity-modulation and in vivo dosimetry

The dose distribution of each of the six intensity-mod- ulated beams used in the boost treatment was measured with film at 5 cm depth (100 cm source-to-axis distance SAD) in a polystyrene phantom. In another experiment, an 0.5 cc cylindrical ionization chamber (0.9 cm inner diameter) was placed at 100 cm SAD and 5 cm depth in a polystyrene phantom to measure the dose along the cen- tral axis of the beam, and +2 cm from it in the direction of the MLC trajectory. The in-phantom dose distribution, at 100 cm SAD and 5 cm depth, was also calculated for each of the intensity-modulated beam and compared to the results measured with film and with an ionization chamber.

During the first treatment, TLD dosimeters were bo- lused and placed on the patient along the central axis of each oblique field. The measured entrance doses were then compared to those calculated with the treatment planning system for the same irradiation configuration.

Radiation treatment and assurance of DMLC pe$ormance

Radiation treatment was given with 15 MV photons from a computer-controlled treatment machine with dy-

724 I. J. Radiation Oncology 0 Biology 0 Physics Volume 35, Number 4, 1996

Fig. 1. Composite isodose contours, of the 72 Gy primary and the 9 Gy boost treatments, superimposed on the central transverse CT image. The PTV is delineated in white and the rectum in yellow. The isodose curves displayed are 86.5 (orange), 81 (blue), 75 (red), 50 (blue), 30 (yellow), 20 (green), and 10 (red) (in units of Gy). respectively.

namic MLC (DMLC) capability’. The projected leaf width of the MLC was 1 cm, at 100 cm from the source. The patient was immobilized and treated in the prone po- sition for the entire treatment course (17).

For each treatment session, after the patient has been properly set up, and the gantry angle set for the first treat- ment field, the radiotherapists exit the treatment room. The radiation machine is then programmed with the MU set- ting for the first field, and the corresponding DMLC file selected on the MLC control system. The MSKCC Record and Verify (RV) computer is interfaced with the acceler- ator and the MLC control systems. Prior to the delivery of each field, the RV system checks the gantry angle, the MU setting, the initial MLC postions, verifies that they are correct, and then permits “beam on.” During radiation delivery in the DMLC mode, the computer-controlled electromechanical system of the MLC moves the leaves to their specified positions as a function of MU. Every 55 ms during “beam on,” the leaf positions are assessed by redundant readouts, and an algorithm within the MLC control computer checks that the leaves are at their proper positions, to within a predefined tolerance (set to 0.1 mm for this patient). If the tolerance condition is satisfied, the radiation delivery continues. However, if one or more leaves were out of tolerance, the beam pulses would be interrupted momentarily (for at least the next 55 ms), to be resumed after the next check cycle-if the conditions become satisfied.

- ‘Clinac 21OOC, Varian Assoc., Palo Alto, CA.

Upon completion of the first DMLC field, the RV sys- tem records the MU delivered and the final positions of the MLC. Without reentering the room, the radiotherapists rotate the gantry to the next oreintation, program the ac- celerator and the MLC for the next field, and then deliver the next treatment. The process is repeated for each of the fields.

RESULTS

The composite dose distribution for the entire radio- therapy course, that is, the sum of the 72 Gy primary and the 9 Gy boost (with inverse planning) treatments, are given in Fig. 1 in the transverse CT plane, with the isodose values given in Gy. The important features are that 75 Gy encompasses the entire PTV volume, that a large portion of the PTV was given 81 Gy, and that the rectum is outside the 81 Gy region. Figure 1 shows clearly that the 81 Gy isodose curve conforms to the PTV and rectum boundary. Thus, the combined primary and the inverse-planned boost treatments produced satisfactory isodose distribu- tion in the central CT slice; dose-volume information will be discussed in the following paragraphs.

As alluded to previously, the rectum has been regarded as the dose-limiting critical normal tissue in our dose es- calation study. And, based on clinical experience to date (17, 30), the rectal DVH for our standard six-field plan with 75.6 Gy prescription dose has been adopted as being

RT using intensity-modulated beams and multileaf collimation l C. C. LING et nl. 725

acceptable. In Fig. 2 are presented the six-field plan DVH for this patient (dashed curve), the composite DVH for the 72 Gy primary treatment plus 9 Gy boost with inverse plan (solid curve), and the composite DVH for the 72 Gy pri- mary treatment plus the manually planned 9 Gy boost (dotted curve). The three DVHs are similar, with the 75.6 Gy treatment giving a larger volume in the high dose re- gion. Nevertheless, all three curves show that ~30% of the rectal volume receives the a dose of 75.6 Gy. The calculated normal tissue complication probability for this patient is 4% for the dashed curve, and about 3% for the other two curves.

The DVHs of the boost PTV for the composite plan (72 Gy primary + 9 Gy boost) are shown in Figure 3a; the two curves are for the two methods used in planning the boost, manual vs. inverse approach. In both cases, about 90% of the boost PTV receives the prescribed 81 Gy, and about 50% receives a dose of 85 Gy. To examine the dose-volume characteristics of the alternate planning methods, DVHs for the 9 Gy boost only are presented in Fig. 3b. The coverage of PTV is similar, although the manual plan engendered a larger volume receiving > 110% of the prescribed dose.

The Beam’s Eye View intensity profiles in the isocen- tric plane of the six intensity-modulated fields are pre- sented in Fig. 4. Each profile exhibits regions of fluence diminution, which correspond to the position of rectum. Consistent with anatomy of the patient in the prone po- sition, the region subtended by the rectum on the PTV is larger for the more posterior fields. Additional fluence variations are present in each field, the combined effect of which produce the optimized dose distribution.

As described previously, absorbed dose distribution for each intensity-modulated field was measured with film at a depth of 5 cm in a homogeneous polystyrene phantom, in a plane through the isocenter. The measured data for the right posterior oblique field (gantry angle-33”) is shown in Fig. 5 as dotted isodose contours. Also shown in Fig. 5, for comparison with the measured results, are calculated isodose curves generated by the treatment plan- ning system for the experimental conditions. Both the measured isodose curves (dotted lines in Fig. 5a) and the cross-field dose profile (dotted lines in Fig. 5b) agree well with the calculated results (solid curves). There is similar agreement for the other five intensity-modulated beams. A photograph of the exposed film gives a gray scale rep- resentation of the dose distribution in Fig. 5c.

As described in the Methods Section, point dose mea- surements were performed for each intensity-modulated beam in a phantom, with an ionization chamber placed at 5 cm depth and at selected points in the plane containing the isocenter. Point dose calculations using the treatment planning system for the experimental conditions were also performed. For all 12 measurement points located in regions of fairly uniform dose distribution, the comparison between measured and calculated dose yielded a mean deviation of 3% (range of O-7.3%). The remaining six

points were in regions of high dose gradients, and com- parison of the measured and calculated values involved much uncertainty and was, therefore, not meaningful.

Prior to the first boost treatment with intensity-modu- lated beams, portal films were obtained for each of the six fields. For each portal radiograph, the leaves were set to the extreme positions of the corresponding DMLC treat- ment field so as to include the entire area of radiation exposure. In other words, the trailing leaves (of the sliding window) were at their “start” positions, and the leading leaves at their “stop” positions. These portal films were reviewed by the attending physician in comparison to the simulation radiographs. No adjustment to the treatment field was deemed necessary.

In the actual treatment, after the patient had been ac- curately positioned, the gantry angle and the primary jaws for the first treatment field were appropriately set. The MLC leaves were then set to the “start” positions. Upon confirmation by the treatment machine control computer and by the MSKCC record and verify system that the start- ing parameters and MU settings were correct, radiation delivery was initiated. After completion of radiation de- livery, the final leaf positions were recorded by the MSKCC record and verify system. The process was then repeated for the second to the sixth fields, with the gantry orientation being set from the control console outside the treatment room. This procedure obviated the need for the radiotherapist to enter the treatment room between the in- dividual fields, and the entire radiation delivery was ac- complished in lo-12 min. (This treatment time will be reduced in the future with software changes, as detailed in the Discussion section.)

Entrance dose, measured with TLD during the first treatment along the central axis of the oblique intensity- modulated beams, was compared to the calculated values. The mean discrepancy in the four comparisons was 3.5%, with a range of 2.5-4%.

DISCUSSION

This article outlines the process that has been estab- lished at Memorial Sloan Kettering Cancer Center for im- plementing inverse planning and treatment with intensity- modulated beams, and describes our experience with the treatment of the first patient using this approach. In this stage of demonstrating the proof of principle, inverse plan- ning was applied only to the boost phase of the treatment. This approach is logical within the framework of our dose escalation study; for patients prescribed to 81 Gy, the first 72 Gy was delivered with the standard six-field plan, and the remaining 9 Gy with a boost plan with the constraint that the rectal DVH of the composite plan be similar to that of a six-field plan (without boost) delivering a pre- scription dose of 75.6 Gy. Subsequent to this phase, con- siderations will be given to use inverse planning for a larger portion of the 81 Gy treatment, and in due course for dose escalation to 86.4 Gy. Of importance in such

726 I. J. Radiation Oncology 0 Biology 0 Physics Volume 35, Number 4, 1996

s e

g 50

2

8 - 72 Py + Inv6r6a plan boost

25 .-‘--I- 72 Qy + Umrul plan booet

- - - - - 76.6 Qy - standard 6 fkld

0

20 40 50

Dose (Gy)

Fig. 2. DVH of the rectum for a standard six-field plan with 75.6 Gy prescription dose (dashed line), compared to those of composite plans of 72 Gy primary plus 9 Gy boost planned either with the inverse method (solid line) or manually (dotted line).

deliberation will be the radiobiophysical consequence of time-dose-fraction and dose heterogeneity for both the tar- get and normal critical structures.

The potential of the inverse method of treatment design, first suggested by Brahme and subsequently examined by many investigators, has been increasingly recognized (3- 5, 13, 21). These referenced studies have shown that this technique can produce dose distributions which conform to targets of irregular shapes (3-5, 13, 21). The inverse planning algorithm used in this study was that described by Bortfeld et al. (1,3) and significantly enhanced by Mo- han et al. (21). Other optimization methods and algorithms

KM ?ir E 3 >o 40

20

0 I? R % % 8

Dose (Gy) a

exist; their relative merits remain to be compared (6, 10, 14-16, 19,22,24,29). It is likely that improvements upon existing algorithms will be forthcoming, and that these will lead to even better treatment design. We emphasize that it is not the aim of this article to show the advantages of inverse planning, and the following only highlights the results for this case (an investigation aimed at comparing inverse and manual planning for a number of clinical cases is being described in another manuscript). Briefly, the dose distribution derived from the inverse method was improved relative to the planner’s design; it conformed better to the boost PTV in the lateral extent. it was more

100

- Inveme plan boost

Doee (%)

b

Fig. 3. (a) DVH of the boost FTV for the composite plan, with the 9 Gy boost planned either manually or with the inverse technique. (b) DVH of the boost PTV for the boost treatment only, planned either manually or with the inverse technique.

RT using intensity-modulated beams and multileaf collimation 0 C. C. Lmc et al. 127

(cl (4 5000

5 4000

f 3000 5000

!$ 2000 5 4000

1000 s

0 3

3000 60 2000

1000 100 0

woo 5 4000

3000 g 2000 60

1000

0

Fig. 4. The intensity profiles in the isocentric plane of the six fields derived from inverse planning. (a,b) Right posterior oblique at gantry angles of 33” and 60”, (c) right lateral; (e,f) left posterior oblique at gantry angles of 300” and 327”, (d) left lateral. Positive Y is in the cephalad direction. At 90”, positive x is towards the anterior; and at 270”, positive x is towards the posterior.

homogeneous, and the dose to the femoral head was re- duced. However, the improvement was probably con- strained by our conservative approach of adopting the planner’s gantry angles. In addition, because a large por- tion of the prescribed dose (72 Gy out of 81 Gy) was already given by the six-field standard treatment, we had to place severe constraint on the rectal dose for the in- verse-planned boost. We anticipate that if the inverse method is used for a larger portion of the prescribed dose, so as to lessen the impact of the rectal constraint on the dose to the PTV, the advantage will be increased.

In addition to the potential advantage of improved dose distribution, another important benefit of inverse planning is operational efficiency. The conventional ap- proach for planning complex cases, that of trial and er- ror and of manual iterative optimization, is labor inten- sive. The 9 Gy manual boost plan described in this article required approximately 2 to 3 person-days (this is because the primary treatment has already delivered 72 Gy to part of the rectum, and extreme care must be exercised to properly shape the dose distribution). Com- puter-aided optimization based on the inverse technique

728 I. J. Radiation Oncology l Biology l Physics

4-

2-

3 - 2 o- * _

-2 -

-4-

a...,...,...,...,... -6 -6 -4 -2

x & 2 4 6 6

a

Volume 35, Number 4, 1996

SQ~3OJkMS4_033;Dose on Y = 1.50

6al

60

. - calculated

. . . . . measured

* l...l.. * I..

-6 -4 -2 2 4 6

b

Fig. 5. Comparison of calculated dose distribution (solid lines) and film measurement data (dotted line) for the intensity-modulated field at 33” gantry angle. Both the calculation and measurement were performedfor the plane at isocenter, at a depth of 5 cm in a homogeneous phantom. (a) Isodose curves (in units of cGy); (b) dose profile at y = 0; (c) photograph of the exposed film for the intensity-modulated field at 33” gantry angle.

will most likely introduce cost-saving in the planning process. Based on the preliminary experience with six test cases, the estimated time required is 3 to 4 h per case, yielding time saving of about two person-days per case. In addition, improvement in our software and user interface will likely further shorten the time required. If this approach can be applied to both the primary and boost plans, the savings will be greater; if the primary and boost plan can be combined into one using a mod- ified schema, the savings will be greater still.

To implement the intensity-modulated beams pro- duced by the inverse planning algorithm, we have adopted the use of DMLC. Other methods may also be applied, for example, the use of beam madifying de- vices (compensators) or of multiple static MLG-shaped fields (3, 13, 20). In our institution, however, the ap- proach described in this article is the most suitable and efficacious. However, extensive dosimetric con&ma- tion and quality assurance were necessary prior to at- taining a high level of confidence in the delivery of the

treatment. These included verifying the agreement be- tween the calculated and the delivered dose distribu- tions, initial workup and routine quality assurance of DMLC performance, in viva dosimetry, and procedures for monitoring DMLC during radiation delivery and for record and verify. As was described, these components are now in place and functioning smoothly at our insti- tution. An additional quality assurance measure could be the use of electronic portal imaging devices, which is under consideration.

The clinical implementation of inverse planning and treatment delivery with DMLC is extremely complex and involves a substantial developmental effort. However, once accomplished, the process is efficient and capable of providing the dual benefits of improved dose distribution and cost savings. It is hoped that the first benefit will result in better clinical outcome of radiotherapy. The latter ad- vantage is based on our experience to date that both treat- ment planning and treatment delivery are rendered less labor intensive.

For the actual treatment for this case, the time (ex- clusive of patient setup) for delivering the six intensity- modulated fields was lo-12 min, comparable to the time needed to deliver six wedged fields. This was be- cause the time needed for room reentry (to change

wedges or their orientations) in the latter was balanced by an increase in beam-on-time required for DMLC. The longer beam-on-time was partly due to the in- creased MU needed for the intensity modulated beams, but also because the present DMLC control software imposed a precision tolerance of 0.1 mm on MLC leaf positions. This last constraint, which is overly restric- tive, severely limits the effective dose rate in the DMLC mode, to about 100 MU per minute. Whereas such pre- cision is prudent in the initial phase of implementation, it is likely that as we gain experience, the tolerance on MLC position can be less stringent so that radiation de- livery can be performed at a higher effective dose rate. This will result in a decrease in actual treatment time to about 5-8 min, depending on the complexity of the in- tensity modulation of the radiation fields. In April 1996, we began using five intensity-modulated fields to de- liver the entire 8 1 Gy. The actual treatment time for the five fields is 5-6 min.

In summary, conformal radiotherapy based on the in- verse method of planning and on the use of DMLC for delivery of intensity modulated beams is now a reality. We are hopeful that this approach will improve the clinical outcome of radiotherapy, and concomitantly provide in- creased operational efficiency.

RT using intensity-modulated beams and multileaf collimation 0 C. C. LING et d. 729

REFERENCES

1. Bortfeld, T.; Burkelbach, J.; Boesecke, R.; Schlegel, W. Methods of image reconstruction from projections applied to conformation radiotherapy. Phys. Med. Biol. 35:1423- 1434; 1990.

2. Bortfeld, T.; Burkelbach, J.; Boesecke, R.; Schlegel, W. Three-dimensional solution of the inverse problem in con- formation radiotherapy. In: Breit, A., ed. Advanced radiation therapy: Tumor response monitoring and treatment planning. Berlin: Springer Verlag; 1992:503-508.

3. Bortfeld, T.; Kahler, D. L.; Waldron, T. J.; Boyer, A. L. X- ray field compensation with multileaf collimators. Int. J. Ra- diat. Oncol. Biol. Phys. 28:723-730; 1994.

4. Brahme, A. Optimization of radiation therapy. Int. J. Radiat. Oncol. Biol. Phys. 28785-787; 1994.

5. Brahme, A. Optimization of stationary and moving beam ra- diation therapy techniques. Radiother. Oncol. 12129140; 1988.

6. Carol, M. P. Integrated 3-D conformal multi-vane intensity modulation delivery system for radiotherapy. In: XItb inter- national conference on the use of computers in radiation ther- apy. Manchester, UK: Christie Hospital NHS Trust, Man- chester; 1994:172-173.

7. Chui, C. S.; LoSasso, T.; Spirou, S. Dose calculations for photon beams with intensity modulation generated by dy- namic jaw or multi-leaf collimations. Med. Phys. 21: 1237- 1243; 1994.

8. Chui, C. S.; Mohan, R. Algorithms for dynamic jaws and multi-leaf collimators. II-Fluence distribution (Abstract). Med. Phys. 20:885; 1993.

9. Chui, C. S.; Spirou, S.; LoSasso, T. Testing of dynamic multi-leaf collimation. Med. Phys. 23:635-642; 1996.

10. Convery, D. J.; Rosenbloom, M. E. The generation of inten- sity-modulated fields for conformal radiotherapy by dynamic collimation. Phys. Med. Biol. 37:1359-1374; 1992.

11. Fuks, Z.; Leibel, S. A.; Kutcher, Cl. J.; Mohan, R.; Ling, C. C. Three dimensional conformal treatment. A new frontier in radiation therapy. In: DeVita, V. T.; Hell- man, S.; Rosenberg, S. A., eds. Important advances in on- cology. Philadelphia: J. B. Lippincott Company; 1991:151- 172.

12. Galvin, J. M.; Smith, A. R.; Lally, B. Characterization of a multi-leaf collimator system. Int. J. Radiat. Oncol. Biol. Phys. 25:181-192; 1993.

13. Galvin, J. M.; Xuan-Gen, C.; Smith, R. M. Combining mul- tileaf fields to modulate fluence distributions. Int. I. Radiat. Oncol. Biol. Phys. 27:697-705; 1993.

14. Jackson, A.; Wang, X.-H.; Mohan, R. Optimization of con- formal treatment planning and quadratic dose objectives. Med. Phys. 21:1006; 1994.

15. Kallman, P.; Lind, B. K.; Brahme, A. An algorithm for maximizing the probability of complication free tumor control in radiation therapy. Phys. Med. Biol. 37:871- 890; 1992.

16. Langer, M.; Brown, R.; Kijewski, P.; Ha, C. The reliability of optimization under dose-volume limits. Int. J. Radiat. Oncol. Biol. Phys. 26:529-538; 1993.

17. Leibel, S. A.; Heimann, R.; Kutcher, G. J.; Zelefsky, M. J.; Burman, C. M.; Melian, E.; Orazem, J. P.; Mohan, R.; Lo- sasso, T. J.; Lo, Y.; Wiseberg, J. A.; Chapman, D. S.; Ling, C. C.; Fuks, Z. Three dimensional conformal radiation ther- apy in locally advanced carcinoma of the prostate: Prelimi- nary results of a phase I dose escalation study. Int. J. Radiat. Oncol. Biol. Phys. 20:55-66; 1994.

18. Lichter, A. S.; Lawrence, T. S. Recent advances in radiation oncology. N. Engl. J. Med. 332:371-379; 1995.

19. Mageras, G. S.; Mohan, R. Application of fast simulated annealing to optimization of conformal radiation treatments. Med. Phys. 20:639*7; 1993.

730 I. J. Radiation Oncology l Biology 0 Physics Volume 35, Number 4, 1996

20. Mageras, G. S.; Mohan, R.; Burman, C.; Barest, G.; Kutcher, G. J. Compensators for three-dimensional treatment plan- ning. Med. Phys. l&133-140; 1991.

21. Mohan, R.; Wang, X.; Jackson, A.; Bortfeld, T.; Boyer, A. L.; Kutcher, G. J.; Leibel, S. A.; Fuks, Z.; Ling, C. C. The potential and limitations of the inverse radiotherapy technique. Radiother. Oncol. 32:232-248; 1994.

22. Niemierko, A.; Urie, M.; Goitein, M. Optimization of 3D radiation therapy with both physical and biological end points and constraints. Int. J. Radiat. Oncol. Biol. Phys. 23:99-108; 1992.

23. Reinstein, L. E.; Wang, X.; Burman, C.; Chen, Z.; Leibel, S. A.; Fuks, Z.; Mohan, R. 3D treatment plan optimization for cancer of the prostate using intensity modulated beams (abstract). Med. Phys. 21:927; 1994.

24. Saderstrom, S.; Gustafsson, A.; Brahme, A. optimization of the dose delivery in few field techniques using radiobiological ob- jective functions. Med. Phys. 20:1201-1210; 1993.

25. Spirou, S. V.; Chui, C. S. Generation of arbitrary fluence profiles by dynamic jaws or multileaf collimators. Med. Phys. 21:1031-1041; 1994.

26. Wang, X.; Spirou, S.; LoSasso, T.; Chui, C. S.; Mohan, R. Dosimetric verification of an intensity modulated treatment. Med. Phys. 23:317-327; 1996.

27. Wang, X. H.; Jackson, A.; Philips, T.; Mohan, R. Optimi- zation of Intensity Distributions for conformal treatment planning for prostate, brain and lung (Abstract). Med. Phys. 21:874; 1994.

28. Wang, X. H.; Mohan, R.; Jackson, A.; Leibel, S. A.; Fuks, Z.; Ling, C. C. Optimization of intensity modulated 3D conformal treatment plans based on biological indices. Radiother. Oncol. 37:87-180; 1995.

29. Webb, S. Optimization by simulated annealing of three-di- mensional conformal treatment planning for radiation fields defined by multi-leaf collimator: II. Inclusion of two-dimen- sional modulation of x-ray intensity. Phys. Med. Biol. 37: 1689-1704; 1992.

30. Zelefsky, M. J.; Leibel, S. A.; Kutcher, G. J.; Kelson. S.; Ling, C. C.; Fuks, Z. The feasibility of dose escalation with three-dimensional conformal radiotherapy with pa- tients with prostatic carcinoma. Cancer J. 142:142-150; 1995.