11.1 Intensity-Modulated Radiation Therapy for Lung Cancer

(1)

11.1 Intensity-Modulated Radiation Therapy for Lung Cancer

Wilfried De Neve, Marie Chaltin, K. Vandecasteele, Werner De Gersem

W. De Neve, M.D., Ph.D.; M. Chaltin, M.D.; K. Vandecasteele, M.D., W. De Gersem, Ir.

Department of Radiotherapy, Ghent University Hospital, De Pintelaan 185, 9000 Gent, Belgium

Acknowledgement: The project “Conformal Radio- therapy Ghent University Hospital” is supported by the Belgische Federatie tegen Kanker and by grants from the Fonds voor Wetenschappelijk Onderzoek Vlaanderen (grants FWO G.0049.98, G.0039.97), the Ghent University (GOA 12050401, BOF 01112300, 011V0497, 011B3300), and the Centrum voor Studie en Behandeling van Gezwelziekten.

11.1.1. Introduction

In the 1980s, Brahme (1982) demonstrated the unique potential of intensity-modulated (IM) beams to create homogeneous concave dose distributions.

Inside IM beams, the radiation fl uence (intensity) was not equal at all sites inside the beam i.e. the beam was not fl at (unmodulated) but had a value that was function of its location across the fi eld (Lax and Brahme 1982). Brahme (1988) also proposed the concept of inverse planning as a possible strategy to make the design of IM beams feasible. Intensity- modulated radiation therapy (IMRT) remained a re-

search topic in physics laboratories until 1993, when Carol et al. (1996) proposed a novel planning and delivery system (NOMOS MiMiC) as a comprehen- sive solution for clinical IM tomotherapy. Since 1993, the three major vendors of linear accelerators have developed multileaf collimator (MLC) technology capable of delivering IMRT, and smaller companies have developed micro-MLCs with IMRT capability.

IMRT research is intense, and clinical results have been published for various tumour sites, including the prostate, head and neck, and base of the skull.

A PubMed search on 25 March 2004 using “inten- sity modulated lung cancer” as keywords yielded 45 publications, most of which were on physics issues.

None reported on the clinical outcome of IMRT in lung cancer.

Against this background, a chapter on the use of IMRT in lung cancer remains largely speculative. Our aims are to formulate the clinical objectives of IMRT to treat lung cancer; to discuss the anatomical chal- lenges of IMRT, the choice of beam directions, and the potential of intensity-modulated beams to spare lung, oesophagus, and spinal cord; to describe the potential clinical benefi t of biological image-guided IMRT optimisation; to discuss specifi c planning is- sues, including the problem of heterogeneities in tissue density for IMRT optimisation; and fi nally to discuss the implementation and quality assurance problems that have delayed clinical trials.

11.1.2 Clinical Objectives

In limited-disease (LD) small cell lung cancer (SCLC), randomised trials comparing early versus late accel- erated radiation therapy concurrent with chemother- apy showed a signiﬁ cant increase in 5-year survival from 13-20% for the late arm to 22-30% for the early arm (Jeremic et al. 1997; Takada et al. 2002; Murray et al. 1993). A large difference in survival between early and late thoracic radiation as well as survival

CONTENTS

11.1.1. Introduction 423 11.1.2 Clinical Objectives 423 11.1.3 Challenges Related to Anatomy

and Preservation of Organ Function 424 11.1.4 Selection of Beam Directions 425 11.1.5 Increasing Dose and/or Dose Intensity

Selectively to Tumour 427

11.1.6 Dose Computation for IMRT Planning 428 11.1.7 Quality Assurance for Clinical Trials 431 11.1.8 Conclusions 432

References 432

(2)

rates above 20% were seen in randomised trials us- ing a dose intensity of about 15 Gy/week (Perry et al. 1998; Work et al. 1997) instead of the standard of 9-10 Gy/week. Using early thoracic radiation, a ran- domised trial (Turrisi et al. 1999) comparing 45 Gy in 3 weeks (group 1) with 45 Gy in 5 weeks (group 2) conﬁ rmed the advantage of a high dose intensity, with a 26% 5-year survival rate for group 1 and 16%

for group 2 (p=0.04). A 50-66% local control rate that was achieved with the best schedules (Knoos et al.

1995; Murray et al. 1993; Turrisi et al. 1999) using 40-54 Gy in 3-3.5 weeks indicates the existence of a window for improvement with more efﬁ cient local treatment. A phase I dose and dose-intensity escala- tion study showed that the maximum tolerated radia- tion dose intensity is limited by acute oesophageal toxicity at 45 Gy in 30 fractions over 3 weeks (Choi et al. 1998). An analysis of patients with LD-SCLC treated with doses ≥ 50 Gy suggests further increase of dose response above 50 Gy (Roof et al. 2003).

These results direct us to a design of IMRT studies with further dose and dose-intensity escalation at the tumour, respecting isotoxicity at the oesophagus by selective underdosage. The hypothesis that such use of IMRT can improve the therapeutic result should be tested.

In patients with locally advanced (LA) non-small cell lung cancer (NSCLC), combined treatment with radiotherapy and second-generation chemotherapy drugs was extensively studied over the past 20 years, and it became the standard over radiotherapy alone in patients with good performance status. Cisplatin seems the drug of choice but results in signiﬁ cant increase of oesophageal toxicity. In LA-NSCLC, the maximum dose of radiotherapy with or without con- current chemotherapy is most often restricted by pulmonary toxicity (radiation pneumonitis). For fur- ther improvement in survival, the two components of the treatment need to be improved. An effective treat- ment of micrometastatic disease through full-dose delivery of cytotoxic drugs could be obtained by add- ing at least one more active drug in conjunction with cisplatin. To further improve loco-regional control of the disease, radiotherapy dose escalation seems a logical strategy. Clinical data regarding the magni- tude of dose escalation that can be achieved by IMRT are inexistent. In planning studies, the Rotterdam Oncological Study Group (Van Sornsen de Koste et al. 2001) showed a reduction of 20.3% in the mean lung dose using three-dimensional (3D) missing tis- sue compensators, as well as a reduction in the total lung volume exceeding 20 Gy (V20). Derycke et al.

(1997) compared a three- or four-beams conventional

3D technique (3D-CRT) and two techniques involv- ing, respectively, seven and ﬁ ve non-coplanar beam incidences with intensity modulation and showed an improvement both in tumour control probability (TCP) and lung normal tissue complication probabil- ity (NTCP) for the IMRT plans, with a window for 20- 30% dose escalation. Marnitz et al. (2002) showed a reduction of the irradiated lung volume using non- coplanar IMRT ﬁ elds.

Randomised trials have shown an improved out- come of combined radiation therapy and chemother- apy over radiotherapy alone, with the concurrent ra- dio-chemotherapy schedules being the most efﬁ cient (Lara et al. 2002). Accelerated radiotherapy sched- ules were shown to be superior to schedules using conventional fractionation (Saunders et al. 1996).

The design objectives of IMRT for LA-NSCLC seem very similar to those of IMRT for LD-SCLC, namely to obtain an accelerated radiation treatment that can be delivered simultaneously with chemotherapy. For both pathologies, IMRT needs to address at least three objectives: limiting oesophageal toxicity, limit- ing the risk of radiation pneumonitis, and increas- ing dose and dose intensity selectively to the tumour.

Dose intensity escalation seems to be more important than physical dose escalation for LD-SCLC, whereas both types of dose escalation seem important for LA- NSCLC. As a result of improved survival and enhanced local control, most of the present radiochemotherapy studies show a signiﬁ cant increase in the incidence of brain metastases (Reboul 2004). Addressing the question of prophylactic cranial irradiation might be a 4th objective in future IMRT trials.

11.1.3 Challenges Related to Anatomy and Preservation of Organ Function

Safe delivery of high doses to lung tumours is pro-

hibited most often by risk of toxicity to lung, spinal

cord, and oesophagus. Lung can be considered as an

organ that consists of functional units organised in a

parallel architecture. The probability of life-threaten-

ing radiation pneumonitis can be estimated from the

percentage volume of lung irradiated above a critical

dose – for example, 20 Gy (V20) (Graham et al. 1999)

– or from the mean (biological) lung dose (MLD)

(Kwa et al. 1998; Seppenwoolde et al. 2003). With a

ﬁ xed constraint on V20 or MLD, the maximum pre-

scription dose decreases for larger planning target

volumes (PTVs) and is dependent on the location of

the PTV. For equal PTV size and doses above 50 Gy

(3)

in 2-Gy fractions, often used as the maximum dose that can be safely delivered to the spinal cord, a PTV with a more peripheral or more cranial location can be planned to higher doses than can a PTV with a more central or caudal location. The MLD was shown to be a strong predictor of the risk of life-threatening radiation pneumonitis. Mathematically, MLD = 1/

V.∫D.dV where V is the total lung volume and D the biologically normalised dose at the volume element dV. By synchronising irradiation with deep inspira- tion breath hold, V is increased (Rosenzweig et al.

2000) and the MLD decreases. Breathing control is discussed elsewhere in this book.

The second term, ∫D.dV, is the lung integral (bio- logical) dose that can be lowered by decreasing beam aperture and by applying beams with a shorter path length through lung. Brugmans et al. (1999) and Dirkx et al. (1997) have shown that a speciﬁ c form of intensity modulation involving the creation of sharp intensity peaks near the beam edges allows the application of smaller beam apertures. The inﬂ uence of beam energy on the integral dose to lung is con- troversial. Some authors advocate the use of beams of less than 10 MV (Brugmans et al. 1999). Liu et al.

(2004) compared IMRT plans using 6 MV and 18 MV beams. The use of 18 MV beams showed no notice- able difference in the quality of the IMRT plans. In our experience, replacement of 18 MV beams with 6 MV beams often decreases the quality of the plans (De Gersem, unpublished).

11.1.4 Selection of Beam Directions

For most lung tumours, the PTV prescription dose is limited by lung, spinal cord, and oesophagus.

Because lung tumours have a poor prognosis and cardiac toxicity is a late event, larger volumes of heart irradiated at high doses are usually allowed in these patients than in patients with breast can- cer or lymphomas. When lung tumours are located close to the diaphragm, the dose-volume integral of radiation to liver and kidneys may be of concern, especially when set-ups with nontransverse plane beams are used. The use of appropriate beam direc- tions is as important in IMRT as in conventional ra- diation techniques, and beam directions should be optimised. However, optimisation of the number of beams and their orientations in three dimensions is still a research challenge and is not routinely avail- able in IMRT planning systems. In daily practice,

beam directions are imposed by a class solution or are chosen by the planner.

The basic principles of choosing beam directions are very similar for photon IMRT as for ﬂ at-beam conformal treatments. First, the best beam direc- tions are those that feature the smallest aperture, which is especially important if the beam trajec- tory passes through organs of parallel functional unit architecture, such as lung. Second, the loca- tion and magnitude of the intended dose gradients determine the choice of beam directions. Where the PTV extends close to organs at risk with serial functional unit architecture, such as spinal cord or oesophagus, steep dose gradients are needed. The steepness of dose gradients is limited by the penum- bra width achieved by the beam collimating system.

Beam directions orthogonal to the desired gradient vector yield the steepest dose gradients. The choice of beam directions is, however, limited by physical constraints imposed by the gantry in relation to the table couch and patient, and by concerns on dosi- metric uncertainty (beam entrance through the pa- tients’ arms).

For the centrally (close to the midsagittal plane) located tumour shown in Fig. 11.1.1a, beam 1 seems the best choice to spare lung, for two reasons: 1) it exhibits the smallest aperture, and 2) the beam axis is aligned with the long axis of the tumour. Beam 1 irradiates the smallest area of lung, but its beam weight and thus its contribution to the PTV prescrip- tion dose will be limited by the spinal cord tolerance.

Other beam directions will be needed to increase the minimum PTV dose above the spinal cord toler- ance, irrespective of the use of intensity modulation.

Two candidate beams (beams 2 and 3 in Fig. 11.1.1a) have equal angular separations to beam 1 and have the same aperture, and both can create the required dose difference. However, beam 2 is a better choice than beam 3 because the former irradiates less lung volume. Fig. 11.1.1a illustrates the beneﬁ ts of using parasagittal beams (i.e. beams that make small angles with the sagittal plane) for the treatment of centrally located tumours. Parasagittal beams can be used to deposit entrance- and exit-dose in the mediastinum rather than in lung. More lateral beams with gantry angles around 90° or 270° are obviously poor choices to spare lung.

For peripherally located tumours, tangential beams can be used to limit the irradiated lung vol- ume (Fig. 11.1.1b).

A centrally located PTV with its largest axis in

the laterolateral direction forms one of the biggest

planning challenges (Fig. 11.1.1c). The advantages of

(4)

parasagittal beams are reduced because these beams also expose a large lung volume because of their wide beam apertures that are needed to cover the PTV.

Lateral beams use smaller apertures and are needed to create a dose gradient between PTV and spinal cord or oesophagus, but they feature long trajecto- ries through lung.

Nontransverse beams may provide additional possibilities for creating dose gradients between the PTV and spinal cord or oesophagus, as illustrated by Fig. 11.1.2a. The beam set-up shown in Fig. 11.1.2b also enables beam entrance above the heart for PTV locations at the bottom of the lungs.

Fig. 11.1.3 shows a dose distribution of a clini- cal IMRT plan for lung cancer in a transverse slice.

The PTV volume was 1,101 cc. The optimisation of segment weights and leaf positions was performed using a biophysical objective function. All beams used 18 MV photons. This slice demonstrates the possibilities of a non-coplanar beam set-up to de- posit exit doses in the mediastinum instead of in- side the lungs. The largest part of the exit dose is deposited outside the slice shown in the ﬁ gure. With a coplanar beam setup, it is not feasible to spare the homolateral as well as the contralateral lung in this

slice to this amount. In this planning, 75% of the lung volume receives less than 20 Gy. The ﬁ gure also displays the avoidance of high doses to the spinal cord. The treatment was well tolerated, and tumour regression was visible on portal images taken for pa- tient setup.

Fig. 11.1.4 shows a dose distribution in a coronal slice of a clinical IMRT planning for a treatment with two dose levels (70/50 Gy) administered in one phase.

The close conformity of isodose lines to the PTV in a coronal view is typical for a parasagittal beam setup. The PTV volume for this patient was 810 cc, and the volume of the dose grid inside the patient was 54,049 cc. The planning for this patient was com- plex due to the patient’s obesity and to the extent of the PTV and its location close to the spinal cord and extending over almost the whole craniocaudal range of the lungs. In order to respect the clinically applied dose constraint to the lungs (V20<30%), the leaf po- sition optimisation eroded the dose distribution at the edge of the PTV. The ﬁ gure also shows the pos- sibility of using a beam with a long path through the PTV with entrance through the patient’s left shoulder (at the right side on the ﬁ gure and tilted anteriorly with regard to the coronal slice).

Fig. 11.1.1 a The volume of lung irradiated by the entrance and exit paths increases with increasing hinge angles to the sagittal plane of the beams for a central PTV. b “Tangential” beams limit the volume of irradiated lung for a peripheral PTV. c Centrally located tumour with its long axis in the laterolateral direction. All beams in the transverse plane irradiate large lung volumes: a parasagittal beam (beam 1) because of its wide aperture, and a lateral beam (beam 2) because of its long path length through lung

c

a b

(5)

11.1.5 Increasing Dose and/or Dose Intensity Selectively to Tumour

At the risk of oversimpliﬁ cation, we could state that larger tumours need higher doses for cure (Bradley et al. 2002). Especially in many patients with LA- NSCLC, the volumes are too large for a strategy of dose escalation when the aim is a homogeneously

irradiated PTV. Dose escalation focused to small sub- volumes of the PTV may be the option. The potential of inhomogeneous dose distributions for dose escala- tion has been demonstrated previously (De Gersem et al. 1999). The safety of substantial dose escalation to small volumes is illustrated by studies conducted at the University of Michigan (Hayman et al. 2001).

In their study design, the maximum prescription dose was limited by the predicted risk of severe radiation

Fig. 11.1.2 a A lateral beam (1) allows the creation of a steep dose gradient in the anteroposterior direction of the patient; for example, between the PTV and oesophagus or spinal cord at the expense of a long trajectory through lung. By isocentric rotation of the couch (beam 2), the trajectory through lung can be shortened because the beam’s exit path tends to leave the thorax through the mediastinum with or without a shorter path through the heterolateral lung. With further couch rotation (beam 3), lung sparing improves as an increasingly large volume of the exit as well as the entrance trajectories of the beam traverse the mediastinum instead of lung, but the risk of collision between the collimator and the patient’s head increases. By changing the gantry angle, collision can be avoided at the expense of a decreased steepness of the dose gradient in the anteroposterior direction. A compromise between lung sparing and the steepness of the anteroposterior dose gradient can be made using beams that enter the patient through the shoulders. To use such beams, a patient position with the arms alongside the body is suitable. A couch design with an Ω-shaped head-end allows anterior as well as posterior beam entrance through the shoulders.

b The anterior part of the beam set that is used as a class solution at Ghent University Hospital. Planning is now started with a set of nine beam directions using three couch isocentre rotation angles. Six of the seven anterior-side beams are shown. At couch rotation angles of 45° and −45°, the set consists of beams with gantry angles of 60°, 30°, and −45° and −60°, −30°, and 45°, respectively. Not shown are the beams at couch rotation angle of 0°, namely the anterior-side beam at gantry 0° and the two posterior-side beams with gantry angles of 155° and −155°

Fig. 11.1.3 Dose distribution of an IMRT planning for lung cancer in a transverse slice. The clinical target volume (CTV) is drawn in purple, the PTV (5-mm expansion of the CTV) in red, the 5-mm expansion of spinal cord in green, a 10-mm expansion of the spinal cord in light blue, and the oesophagus in green

a b

Ganthy angles Ganthy angles

(6)

pneumonitis. Doses over 100 Gy could be delivered to small PTV volumes. The maximum tolerated dose to the structural elements of lung (bronchi, blood vessels) had not been reached. For larger tumours, the unwanted dose to lung becomes too large, and such high doses could not be attempted because of an unacceptably high risk of severe radiation pneu- monitis. Considering the size of the PTV in most patients with LA-NSCLC, it seems unlikely that es- calation to doses around 100 Gy will be possible by IMRT. Therefore, it may be preferable to direct the foci of dose escalation to the regions inside the tu- mour that are supposed to be the most radiation-re- sistant. Novel biological imaging techniques, mostly based on positron-emission computed tomography (PET), magnetic resonance imaging (MRI), and mag- netic resonance spectroscopy (MRS), may have the potential to construct 3D maps of radiobiologically relevant parameters (Bentzen et al. 2002; Van De Wiele et al. 2003). These maps can be fused with high-resolution computed tomography (CT) and MRI for treatment design and optimisation with a strategy of small-volume focused dose escalation. At Ghent University Hospital, the strategy for clinical trials of focused dose escalation involves the fl ow of procedures given in Fig. 11.1.5. Anatomical (CT) information on CT (Fig. 11.1.5a) remains the basis for conventional PTV defi nition. Biological (PET) imaging (Fig. 11.1.5b) provides radiobiological information as signal intensi- ties (SI) to voxels, related to radiobiological parameters such as hypoxia, proliferation, and intrinsic radiation sensitivity. Fusing provides an image (Fig. 11.1.5c) in which each voxel has a Hounsfi eld value for computa- tion of absorbed dose and an SI for intratumour guid-

ance of the dose distribution. Bioimage-guided-IMRT optimisation requires the development of a transfor- mation engine (Fig. 11.1.5c and Fig. 11.1.5d) that se- cures a spatial dose variation in the anatomical PTV (Fig. 11.1.5e) as a function of SI in the PET imaging.

For the design of early clinical implementation studies, we refer to Fig. 11.1.5d and Fig. 11.1.5e. The D-base in Fig. 11.1.5d means a conventionally applied dose level that encompasses the anatomical PTV. Dose escalation (D+, D++, Dmax) is limited to intra-PTV regions as a function of SI values.

11.1.6 Dose Computation for IMRT Planning

The low density of lung tissue (typically 0.3 g/cm

³

) considerably complicates the computation of the dose distribution in the human thorax and deteriorates the accuracy of all conventional computation algorithms.

Especially when the beams cross inhomogeneities as air cavities (trachea, bronchi) and tumour tissue in lung, dose planning system calculations using ana- lytical approximations are inadequate (Knoos et al.

1995; Mohan and Antolak 2001).

The lower attenuation of radiation in lung gives rise to a higher dose in the tissues downstream from the lung volume. This effect is adequately taken into ac- count by most dose planning systems. Three counter- acting effects, however, are not well modelled in con- ventional dose calculation algorithms. They are all due to a loss of electron equilibrium: absorbed electrons are not balanced in number by the produced (leaving)

Fig. 11.1.4 Coronal view of the dose distribution of an IMRT planning for lung cancer. The PTV (CTV + 5 mm) is red, and the purple contour delineates the part of the PTV with a prescription dose of 70 Gy

(7)

electrons. In addition, the secondary electrons after single and multiple scattering can deposit their energy at a relatively large distance, i.e. their path lengths are longer in low-density tissues such as lung (Fig. 11.1.6a).

The three particular effects are the following:

1. A local dose decrease in regions where the beam reenters the soft tissue (rebuild-up). This rebuild-up is caused by the higher production of secondary elec- trons in tissue outside the lungs and can be important for beams that traverse lung tissue before hitting the (soft-tissue equivalent) tumour (Fig. 11.1.6b). In the case of small beam width, the underdosage in the re- build-up region is deepened by the loss of secondary electrons outside the beam’s boundaries (Martens et al. 2002).

2. Lateral dose spread in lung tissue beyond the geo- metrically expected beam boundaries (Fig. 11.1.6c).

The reason is that even for modest photon energies, the electron path length in lung tissue is in the order of centimetres. This implies that the beam edges be- come dosimetrically blurred and that larger volumes of lung are exposed to signiﬁ cant doses (Dirkx et al.

1997; Miller et al. 1998).

3. Underdosage in regions where the tumour ﬂ anks air-like tissue at the beam edges because more electrons leave the tumour interface zone than arrive from the air-like tissues (Fig. 11.1.6d) (Dirkx et al.

1997; Miller et al. 1998).

The conventional dose computation algorithms lead to deviations larger than 10% from measure- ments at lung tissue or bone tissue interfaces and in build-up regions behind air cavities (Mijnheer et al.

1988; Werner et al. 1987). More recent convolution/

superposition methods using point spread functions or kernels may provide more accurate dose distribu- tions, dependent on the speciﬁ c implementation of tissue inhomogeneity corrections (path length cor- rections and adaptations of point spread functions or kernels in regions with high electron density in- homogeneities).

In most IMRT planning systems, conventional computation algorithms are used during the optimis- ation process. Inaccuracies in dose computation may lead to erroneous adaptations of beamlet intensities during inverse planning optimisation. The term con- vergence error (Jeraj and Keall 2000) has been used to describe the error in the result of an optimisation algorithm that was misled by inaccurate dose com- putation. Computer performance limits the possibil- ity to incorporate more accurate dose computation based on convolution-superposition or Monte Carlo algorithms in the optimisation process. Inaccuracies in dose computation and the subsequent conver- gence error in optimisation have hampered the clini- cal implementation of IMRT for lung tumours. Two interesting approaches to reduce the convergence er-

Fig. 11.1.5 Biological image guided intensity modulated radiation therapy (BIG-IMRT). a CT scanning provides the anatomical information, resulting in the anatomical PTV. b Biological images (e.g. PET) provide information on radiobiological parameters.

c Image fusion is performed to position the biological on the anatomical information. d IMRT is used to irradiate the anatomical PTV to a minimal dose level (Dbase), and to increase the dose selectively within the PTV, dependant on the signal intensities of the biological images (panel e).

a

fusion

c d

b

e

(8)

ror have been presented, one by Hong et al. (2002), applicable to ﬂ uence optimisation-based inverse planning, and the other by De Gersem et al. (2001b), applicable to direct segment outline (aperture) and weight optimisation.

The method described by Hong et al. (2002) was devised to take into account the scattered dose com- ponent during ﬂ uence optimisation for the large ﬁ elds used for whole abdominal irradiation. The iterative process they used (only) included scatter from within a 2-mm radius of a pencil beam kernel.

At the end of each optimisation cycle, the dose dis- tribution was recomputed with full scatter contri- butions. The difference between the accurately com- puted dose distribution and the dose distribution computed with restricted scatter contribution was used as a correction in the next optimisation cycle, and the process was iterated until further improve- ment became minimal. The principle the authors de- scribed could be used to account for loss of electron equilibrium during optimisation of lung tumours.

After each optimisation cycle, leaf sequencing should be performed and the dose distribution re- computed with an appropriate dose algorithm such as convolution/superposition or Monte Carlo.

The method of De Gersem et al. to incorporate accurate dose computation in direct segment outline and weight optimisation is shown in Fig. 11.1.7.

For each chosen incidence, an anatomy-based segmentation tool (ABST) created segments (De Gersem et al. 2001a). By the use of ABST, a start- ing set of segments is created. For each segment, a 0.4×0.4×0.4-cm

³

dose grid is computed. A start- ing set of weights is obtained using SWOT, a seg- ment weight optimisation tool previously described (De Gersem et al. 1999). Subsequently, the method of direct segment aperture and weight optimisation (SOWAT, segment outline and weight adapting tool), was applied to optimise the plans (De Gersem et al. 2001b). SOWAT is built to use dose grids com- puted by an external dose computation engine, as shown in Fig. 11.1.7. The Philips-Pinnacle (Philips Medical Systems, Eindhoven, The Netherlands) convolution-superposition algorithm was used as the external engine. This algorithm allows relatively accurate computing of the dose delivered to lung tissue. Both the penumbra broadening in lung and rebuild-up downstream from lung are well repro- duced (Ahnesjo and Aspradakis 1999; Martens et al. 2002). Inside SOWAT, a predeﬁ ned set of MLC leaf repositioning values is tested according to the algorithm drawn in Fig. 11.1.7. The default set of repositioning values spanned a range of ±1–8-mm (positive values indicate an opening leaf position change; negative values indicate leaf closing). After each leaf position adaptation cycle, monitor units are optimised, and a new repositioning value is set

Fig. 11.1.6 a The path length of scattered electrons is much longer in lung than in other tissues of the thorax. b Lower density of ionisation in lung leads to less scattered electrons which, in turn, causes rebuild-up at the lung-tumour interface. c Degradation of the beam penumbra in lung leading to underdosage close to the beam edges and dose deposition in lung outside the beam edges. d Underdosage at the surface of the tumour close to the beam edge

a c

d b

(9)

to execute the next cycle. When all repositioning values have been tested, SOWAT sends beam seg- ments with optimised apertures and weights to the external dose engine to compute the dose grids.

As shown in Fig. 11.1.7, the cycle with passage through SOWAT and the external dose engine is to be reiterated until the plan acceptance criteria are fulﬁ lled. Then, segment sequencing by the CRASH (combine, reorder and step and shoot) tool results in a prescription ﬁ le for the linear accelerator (De Neve et al. 1999).

11.1.7 Quality Assurance for Clinical Trials

Many of the difficulties regarding the implemen- tation of IMRT in lung cancer clinical trials are being solved. Solutions exist for accurate dose computation in lung and across interfaces be- tween lung and other tissues during optimisation.

Respiratory gating techniques become feasible for clinical practice. Accurate delineation of critical organs and pretreatment analysis of toxicity-pre- dicting factors allow for safer application of high-

dose schedules. Considering the complexity of the chain of procedures that involves imaging, plan- ning, and optimisation and that finally leads to the instruction files for the linear accelerator, a test system to evaluate whether the execution of the instruction files leads to the calculated dose distribution would be welcome. Polymer gel do- simetry has been used for this purpose (Vergote et al. 2003) and has the advantage of providing 3D quantitative information. However, the cost of gel dosimetry is prohibitive for testing each individual IMRT plan before it is delivered to the patient.

More economical systems need to be developed.

In a future European Organization for Research and Treatment of Cancer (EORTC) trial, a multi- purpose phantom that allows for studying the ef- fects of tissue inhomogeneities on dose deposition will be used (Swinnen et al. 2002). This phantom allows studying key discrepancies between cal- culations and measurements for each individual intensity-modulated beam. Considering the large variability in IMRT techniques and procedures, practical validation systems of patient-individual treatments will be required for clinical trials of IMRT in lung cancer patients, especially in a mul- ticentre setting.

Fig. 11.1.7 Use of a convolution-superposition dose algorithm in IMRT optimisation. Description of the algorithm can be found in the text section 11.1.6 on dose computation for IMRT planning

(10)

11.1.8 Conclusions

IMRT may become an important element of future strategies to improve local control and survival in lung cancer. For LD-SCLC as well as for LA-NSCLC, concurrent dose-intensive radiation and chemo- therapy seem to be the paradigm. In such sched- ules, safe delivery of radiation will involve multiple technical improvements including (1) a decrease of the internal margin of the PTV by breathing control techniques, (2) a decrease of the external margin of the PTV by online imaging and correction, (3) use of dose computation algorithms during IMRT op- timisation that accurately model electron nonequi- librium, (4) a decrease of the beam aperture by a rind-boost technique, (5) a focused dose escalation to subvolumes, determined by biological imaging, inside the PTV, (6) a better dose prescription and constraint deﬁ nition to decrease ambiguity in clinical protocols, (7) the development of class solutions for routine clinical implementation, and (8) the develop- ment of quality assurance for clinical trials of IMRT in lung cancer.

References

Ahnesjo A and Aspradakis MM (1999) Dose calculations for external photon beams in radiotherapy. Phys Med Biol 44:

R99–155

Bentzen SM, Saunders MI, Dische S (2002) From CHART to CHARTWEL in non-small cell lung cancer: clinical radiobiological modelling of the expected change in outcome.

Clin Oncol (R Coll Radiol) 14:372–381

Bradley JD, Ieumwananonthachai N, Purdy JA, Wasserman TH, Lockett MA, Graham MV, Perez CA (2002) Gross tumor volume, critical prognostic factor in patients treated with three-dimensional conformal radiation therapy for non- small-cell lung carcinoma. Int J Radiat Oncol Biol Phys 52:49–57

Brahme A (1988) Optimization of stationary and moving beam radiation therapy techniques. Radiother Oncol 12:129–140 Brahme A, Roos JE, Lax I (1982) Solution of an integral

equation encountered in rotation therapy. Phys Med Biol 27:1221–1229

Brugmans MJ, van der Horst A, Lebesque JV, Mijnheer BJ (1999) Beam intensity modulation to reduce the ﬁ eld sizes for conformal irradiation of lung tumors: a dosimetric study. Int J Radiat Oncol Biol Phys 43:893–904

Carol M, Grant III WH, Pavord D, Eddy P, Targovnik HS, Butler B, Woo S, Figura J, Onufrey V, Grossman R, Selkar R (1996) Initial clinical experience with the Peacock intensity modulation of a 3-D conformal radiation therapy system.

Stereotact Funct Neurosurg 66:30–34

Choi NC, Herndon JE, Rosenman J, Carey RW, Chung CT, Ber-

nard S, Leone L, Seagren S, Green M (1998) Phase I study to determine the maximum-tolerated dose of radiation in standard daily and hyperfractionated-accelerated twice- daily radiation schedules with concurrent chemotherapy for limited-stage small-cell lung cancer. J Clin Oncol 16:3528–3536

De Gersem W, Claus F, de Wagter C, De Neve W (2001a) An anatomy-based beam segmentation tool for intensity-modulated radiation therapy and its application to head-and- neck cancer. Int J Radiat Oncol Biol Phys 51:849–859 De Gersem W, Claus F, de Wagter C, Van Duyse B, De Neve

W, (2001b) Leaf position optimization for step-and-shoot IMRT. Int J Radiat Oncol Biol Phys 51:1371–1388

De Gersem, WR, Derycke S, Colle CO, de Wagter C, De Neve WJ (1999) Inhomogeneous target-dose distributions: a dimen- sion more for optimization? Int J Radiat Oncol Biol Phys 44:461–468

De Neve W, De Gersem W, Derycke S, De Meerleer G, Moerman M, Bate MT, Van Duyse B, Vakaet L, De Deene Y, Mersseman B, De Wagter C, De Waeter C (1999) Clinical delivery of intensity modulated conformal radiotherapy for relapsed or second-primary head and neck cancer using a multileaf collimator with dynamic control. Radiother Oncol 50:301–314

Derycke S, Van Duyse B, De Gersem W, de Wagter C, De Neve W (1997) Non-coplanar beam intensity modulation allows large dose escalation in stage III lung cancer. Radiother Oncol 45:253–261

Dirkx ML, Heijmen,BJ, Korevaar GA, van Os MJ, Stroom JC, Koper PC, Levendag PC (1997) Field margin reduction using intensity-modulated x-ray beams formed with a multileaf collimator. Int J Radiat Oncol Biol Phys 38:1123–1129 Graham MV, Purdy JA, Emami B, Harms W, Bosch W, Lock-

ett MA, Perez CA (1999) Clinical dose-volume histogram analysis for pneumonitis after 3D treatment for non-small cell lung cancer (NSCLC). Int J Radiat Oncol Biol Phys 45:323–329

Hayman JA, Martel MK, Ten Haken RK, Normolle DP, Todd III RF, Littles JF, Sullivan MA, Possert PW, Turrisi AT, Lichter, AS (2001) Dose escalation in non-small-cell lung cancer using three-dimensional conformal radiation therapy:

update of a phase I trial. J Clin Oncol 19:127–136 Hong L, Alektiar K, Chui C, LoSasso T, Hunt M, Spirou S, Yang

J, Amols H., Ling C, Fuks Z, Leibel S (2002) IMRT of large ﬁ elds: whole-abdomen irradiation. Int J Radiat Oncol Biol Phys 54:278–289

Jeraj R Keall P (2000) The effect of statistical uncertainty on inverse treatment planning based on Monte Carlo dose calculation. Phys Med Biol 45:3601–3613

Jeremic B, Shibamoto Y, Acimovic, L, Milisavljevic S (1997) Ini- tial versus delayed accelerated hyperfractionated radiation therapy and concurrent chemotherapy in limited small-cell lung cancer: a randomized study. J Clin Oncol 15:893–900 Knoos T, Ahnesjo A, Nilsson P, Weber L (1995) Limitations of a

pencil beam approach to photon dose calculations in lung tissue. Phys Med Biol 40:1411–1420

Kwa SL, Lebesque JV, Theuws JC, Marks LB, Munley MT, Bentel G, Oetzel D, Spahn U, Graham MV, Drzymala RE, Purdy JA, Lichter AS, Martel MK, Ten Haken RK (1998) Radiation pneumonitis as a function of mean lung dose: an analysis of pooled data of 540 patients. Int J Radiat Oncol Biol Phys 42:1–9

Lara PN, Jr, Goldberg Z, Davies A, Lau DH, Gandara DR (2002

(11)

Concurrent chemoradiation strategies in the management of unresectable stage III non-small-cell lung cancer. Clin Lung Cancer 3(suppl 2):S42–S48

Lax I, Brahme A (1982) Rotation therapy using a novel high- gradient ﬁ lter. Radiology 145:473–478

Liu HH, Wang X, Dong L, Wu Q, Liao Z, Stevens CW, Guerrero TM, Komaki R, Cox JD, Mohan R (2004) Feasibility of sparing lung and other thoracic structures with intensity- modulated radiotherapy for non-small-cell lung cancer. Int J Radiat Oncol Biol Phys 58:1268–1279

Marnitz S, Stuschke M, Bohsung J, Moys A, Reng I, Wurm R, Budach V (2002) Intraindividual comparison of conventional three-dimensional radiotherapy and intensity modulated radiotherapy in the therapy of locally advanced non-small cell lung cancer a planning study. Strahlenther Onkol 178:651–658

Martens C, Reynaert N, de Wagter C, Nilsson P, Coghe M, Palmans H, Thierens H, De Neve W (2002) Underdos- age of the upper-airway mucosa for small ﬁ elds as used in intensity-modulated radiation therapy: a comparison between radiochromic ﬁ lm measurements, Monte Carlo simulations, and collapsed cone convolution calculations.

Med Phys 29:1528–1535

Mijnheer BJ, Rice RK, Chin LM (1988) Lead-polystyrene transition zone dosimetry in high-energy photon beams.

Radiother Oncol 11:379–386

Miller RC, Bonner JA, Kline R (1998) Impact of beam energy and ﬁ eld margin on penumbra at lung tumor-lung paren- chyma interfaces. Int J Radiat Oncol Biol Phys 41:707–713 Mohan R, Antolak J (2001 ) Monte Carlo techniques should

replace analytical methods for estimating dose distributions in radiotherapy treatment planning. Med Phys 28:123–126

Murray N, Coy P, Pater JL, Hodson I, Arnold A, Zee BC, Payne D, Kostashuk EC, Evans WK, Dixon P, et al. (1993) Impor- tance of timing for thoracic irradiation in the combined modality treatment of limited-stage small-cell lung cancer.

The National Cancer Institute of Canada Clinical Trials Group. J Clin Oncol 11:336–344

Perry MC, Herndon III JE, Eaton WL, Green MR (1998) Thoracic radiation therapy added to chemotherapy for small-cell lung cancer: an update of Cancer and Leukemia Group B Study 8083. J Clin Oncol 16:2466–2467

Reboul FL (2004) Radiotherapy and chemotherapy in locally advanced non-small cell lung cancer: preclinical and early clinical data. Hematol Oncol Clin N Am 18:41–53

Roof KS, Fidias P, Lynch TJ, Ancukiewicz M, Choi NC (2003) Radiation dose escalation in limited-stage small-cell lung cancer. Int J Radiat Oncol Biol Phys 57:701–708

Rosenzweig KE, Hanley J, Mah D, Mageras G, Hunt M, Toner S, Burman C, Ling CC, Mychalczak B, Fuks Z, Leibel SA

(2000) The deep inspiration breath-hold technique in the treatment of inoperable non-small-cell lung cancer. Int J Radiat Oncol Biol Phys 48:81–87

Saunders MI, Dische S, Barrett A, Parmar MK, Harvey A, Gibson D (1996) Randomised multicentre trials of CHART vs conventional radiotherapy in head and neck and non- small-cell lung cancer: an interim report. CHART Steering Committee. Br J Cancer 73:1455–1462

Seppenwoolde Y, Lebesque JV, de Jaeger K, Belderbos JS, Boersma LJ, Schilstra C, Henning GT, Hayman J., Martel MK, Ten Haken RK (2003 Comparing different NTCP models that predict the incidence of radiation pneumonitis. Normal tissue complication probability. Int J Radiat Oncol Biol Phys 55:724–735

Swinnen A, Verstraete J, Huyskens D (2002) The use of a mul- tipurpose phantom for mailed dosimetry checks of therapeutic photon beams: ‘OPERA’ (operational phantom for external radiotherapy audit). Radiother Oncol 64:317–326 Takada M, Fukuoka M, Kawahara M, Sugiura T, Yokoyama

A, Yokota S, Nishiwaki Y, Watanabe K, Noda K, Tamura T, Fukuda H, Saijo N (2002) Phase III study of concurrent versus sequential thoracic radiotherapy in combination with cisplatin and etoposide for limited-stage small-cell lung cancer: results of the Japan Clinical Oncology Group Study 9104. J Clin Oncol 20:3054–3060

Turrisi III AT, Kim K, Blum R, Sause WT, Livingston RB, Komaki R, Wagner H, Aisner S, Johnson DH (1999) Twice- daily compared with once-daily thoracic radiotherapy in limited small-cell lung cancer treated concurrently with cisplatin and etoposide. N Engl J Med 340:265–271 Van de Wiele C, Lahorte C, Oyen W, Boerman O, Goethals I,

Slegers G, Dierckx RA (2003) Nuclear medicine imaging to predict response to radiotherapy: a review. Int J Radiat Oncol Biol Phys 55:5–15

Van Sornsen de Koste J, Voet P, Dirkx M, van Meerbeeck J, Senan S (2001) An evaluation of two techniques for beam intensity modulation in patients irradiated for stage III non-small cell lung cancer. Lung Cancer 32:145–153 Vergote K, De Deene Y, Claus F, De Gersem W, Van Duyse B,

Paelinck L, Achten E, De Neve W, De Wagter C (2003) Appli- cation of monomer/polymer gel dosimetry to study the effects of tissue inhomogeneities on intensity-modulated radiation therapy (IMRT) dose distributions. Radiother Oncol 67:119–128

Werner BL, Das IJ, Khan FM. Meigooni AS (1987) Dose perturba- tions at interfaces in photon beams. Med Phys 14:585–595 Work E, Nielsen OS, Bentzen SM, Fode K, Palshof T (1997)

Randomized study of initial versus late chest irradiation combined with chemotherapy in limited-stage small-cell lung cancer. Aarhus Lung Cancer Group. J Clin Oncol 15:3030–3037