42 04

measuring the leaf positioning accuracy with an ionization chamber for a single leaf pair, has been extended to two-dimensional, using the EPID (305). The measurements take 3 minutes, including the analyses. Errors in leaf positioning as small as 0.1 to 0.2 mm can be detected. Apart from the daily verification of the leaf motions, QA procedures are performed for each individual IMRT patient (228,304,306,328,329). These procedures aim at (a) verification of the final TPS dose calculation for the optimized treatment parameters such as the leaf trajectories, and (b) verification of the correct execution of the plan at the linac. Currently, the TPS dose is only verified by an independent dose calculation for a single or few points in the center of the tumor. A fully three-dimensional procedure is being developed. For verification of the correct fluence delivery at the linac, EPID dose measurements are performed both prior to the first treatment fraction (pretreatment verification [229,306]), and during treatment (“in vivo” verification [305]). For pretreatment verification, portal dose images (a two-dimensional dose distribution in the plane of the fluorescent screen of the EPID) measured with the EPID are compared with predictions. Differences point at errors in leaf sequencing, data transfer from the TPS to the linac, or to dosimetric/mechanical linac performance problems. Presently, portal dose image comparison (Fig. 42.45) has been fully integrated in the applied EPID software (Theraview NT, Cablon Medical, Heusden, The Netherlands); a method for automated image analysis also has been implemented. Images are only reviewed by a physicist in case of a failure to pass the automated test. Because of the high spatial resolution, EPIDs are suited for detection of tongue-and-groove underdosage effects. For a group of 270 IMRT patients, the pretreatment procedure has revealed four serious errors prior to the start of treatment (329). Recently, methods have been developed for back-projection of fluence profiles, measured with the EPID, in the planning CT scan or in an in-room acquired cone beam CT scan, allowing full three-dimensional analyses (328).
Deviations in in vivo measured PDIs may be due to errors in fluence delivery, but may also be caused by changes in patient anatomy or variations in patient setup. To discriminate between the two, the split IMRT field technique (305) has been developed, which is now routinely applied for all head and neck cancer patients.

Alternative IMRT Approaches
Beam Orientations
Instead of dedicated orientations, often a relatively large number of equiangular beams are used. Whereas some articles report techniques with seven equiangular beams for oropharynx tumors (62,239), others advocate nine beams (324). Generally, an odd number is used to avoid opposing beams.
Simultaneous Integrated Boost
For oropharyngeal cancer patients treated in the Erasmus MC, the boost is generally delivered with BT or the CyberKnife after 46 Gy. When using IMRT for full-dose delivery, a simultaneous


integrated boost technique may be applied (166,247,325). The involved simultaneous optimization of the large field and the boost technique does generally result in superior plans, compared with sequential optimization (211). With the simultaneous integrated boost technique, an enhanced fraction dose may be selected for the primary tumor, yielding two simultaneous opportunities for biologic dose escalation: a shortening of the total treatment duration and an increased LC as a result of the higher daily tumor dose. However, the possibilities for application of escalated fraction doses are limited by the risk of increased toxicity (166,325). Alternatively, to minimize complications, the fraction dose in the elective regions may be reduced. The current RTOG study H-0022 applies a simultaneous integrated boost technique, prescribing a GTV total dose of 66 Gy at 2.2 Gy/fraction, and a dose for the subclinical disease region of 54 Gy (1.8 Gy/fraction) (247).
Plan Optimization and Evaluation Using Radiobiologic Models
Instead of using dose- and dose-volume–based objectives and constraints, plan optimization and evaluation can, in principle, also be done using radiobiologic criteria such as the tumor control probability, normal tissue control probabilities, and the equivalent uniform dose for the tumor and organs at risk. For the head and neck region, several parameter sets for biologic models exist, derived from observed tumor control and toxicity data (76,77,259,269) (also used in tumor control probability/normal tissue control probabilities calculating modules [315]). Unfortunately, the results vary considerably with the applied parameter set: for a group of oropharynx cancer patients, van Vulpen et al. (310) reported predicted normal tissue control probabilities differences for the parotid glands ranging from -3% to +35% when applying different parameter sets. To our knowledge, articles describing a decisive role in clinical decision-making for the treatment of oropharyngeal cancer patients have not yet been published.
Step-and-Shoot or Segmental IMRT (SMLC)
Apart from the DMLC technique, intensity-modulated profiles can also be generated by sequential delivery of static field segments, with a variable shape and number of monitor units (SMLC). In the transition period from end of delivery of one segment to shaping of the next segment, the beam is switched off. DMLC allows for more precise realization of the optimized fluence profiles. However, some studies have concluded that the differences are of minor clinical importance (4,55,83). It has also been reported that DMLC treatments require more MUs and SMLC treatments take more time.
A leaf-sequencing algorithm for DMLC has been developed that fully prevents the occurrence of tongue-and-groove underdosage reported by van Santvoort et al. (266). Currently, we use the Cadplan TPS (Varian Medical Systems, Espoo, Finland) for inverse planning and leaf sequencing. It was demonstrated that for extreme profiles, tongue-and-groove underdosage of up to 30% may occur with this TPS (80). However, the protocol for pretreatment verification of the fluence profiles of each individual IMRT patient has never revealed a clinically relevant tongue-and-groove error. Also for SMLC, leaf-sequencing algorithms have been developed that reduce or prevent the occurrence of tongue-and-groove underdosage (61,182).

Risultati clinici
Some of the clinical results are presented in the section Xerostomia. Excellent reviews are presented by Puri et al. (243) and Lee et al. (168). Table 42.21 summarizes the preliminary clinical results of several studies of IMRT treatment for oropharyngeal carcinoma. The studies confirm the high rates for (loco-) regional control, distant metastases-free survival, disease-free survival, and OS in combination with reduced toxicity in comparison to conventional radiotherapy. Finally the multi-institutional RTOG study (H-0022) using IMRT for early-stage oropharyngeal cancer has completed accrual, and final results are to be expected shortly (247). IMRT allows dose to be concentrated in the tumor volume while sparing normal tissues. However, the downside to IMRT is the potential to increase the number of radiation-induced second cancers. The reasons for this potential are more MUs and, therefore, a larger total-body dose because of leakage radiation and, because IMRT involves more fields, a larger volume of normal tissue is exposed to lower radiation doses. In fact, Hall (114) calculated that IMRT may double the incidence of solid cancers in long-term survivals. In contrast to older patients, if balanced by an improvement of local tumor control, the use of IMRT might not be acceptable in children. An alternative might be to replace x-rays with protons in case of scanning pencil beams.
Alternative External Beam Approaches
Helical Tomotherapy
Apart from IMRT with linear accelerators, helical tomotherapy (186) (HiArt, TomoTherapy Inc., Madison, WI) can also be used for highly conformal dose delivery. Several articles report on dose distributions for head and neck cancer patients that might be superior to those obtained with linacs, regarding sparing of critical structures (82,230,310). Long-term clinical evaluations are not available as yet. Compared to linac-based IMRT, tomotherapy requires more MUs to deliver the same target dose, because of the applied fan-beam (212). This increases the whole-body dose equivalent, which may increase the risk for radiation-induced secondary malignancies (84). The clinical


implications of irradiating larger volumes to lower doses with tomotherapy, compared with smaller volumes with intermediate doses in linac IMRT, are unknown.

CyberKnife
The robotic CyberKnife system (Accuray Inc., Sunnyvale, CA) is another means of applying high dose of radiation with high accuracy (2). Some preliminary experience with the CyberKnife is available from the Erasmus MC for cancer in the oropharynx (Fig. 42.19 shows the protocol). This regards the delivery of a boost treatment of three fractions of 5.5 Gy on each consecutive day, prescribed at the 80% isodose. Patients are immobilized with the regular thermoplastic mask with a three-point fixation. Highly conformal plans with steep dose gradients are generated using 100 to 200 noncoplanar and nonisocentric coned beams. Figure 42.46 shows a typical dose distribution for a CyberKnife boost with the applied beam orientations. The CyberKnife image-guidance system and the patient skull are used for frequent measurement of the patient setup during treatment. Observed translations and rotations are used for immediate correction of the position and direction of the next beams. Because of these continuous adjustments, a PTV margin of only 2 mm was applied originally. Recently, the images obtained with the CyberKnife image-guidance system have been retrospectively analyzed, to quantify patient motion during delivery of a treatment fraction (135). For head and neck cancer and brain cancer patients, the maximum observed displacement in a 2-minute period was 2.8 mm in a single direction; a maximum rotation of 2.3 degrees was observed after 3 minutes. The overall systematic and random three-dimensional errors after 15 minutes are 1.3 and 1.2 mm (2 SD), respectively. With the CyberKnife image-guidance system, these in-fraction patient movements are automatically compensated using the robotic manipulator.
Cone-Beam CT
An important next step in image-guided RT for head and neck tumors may be the use of the recently introduced cone-beam CT scanners, integrated in linacs (Fig. 42.44) (146,172,202). In contrast to EPIDs, these systems allow for visualization of soft tissues. So far, the image quality is not as good as for modern diagnostic scanners. During the fractionated head and neck treatment, various processes may result in a gradual change of the patient anatomy, such as postoperative changes/edema, weight loss, and shrinking of the primary tumor and/or nodal masses (11,118). Large changes in the size of the GTV and the size and position of the parotid glands have been observed. These changes may result in suboptimal treatment as the dose delivery in all fractions is usually based on a treatment plan that is designed for the patient anatomy in the planning CT scan, which is acquired prior to the start of treatment. Studies have been performed to investigate the impact of replanning based on, or triggered by, anatomy changes observed in acquired cone beam CT scans (118,210). A major clinical question to be answered is the target definition in case of a shrunken gross target volume.
As part of the IMRT QA protocol, cone-beam CT scans may also be used to assess the “dose of the day” (328).
Future Technical Developments
Dose-Calculation Algorithms
It is well known that, especially in the presence of low-density inhomogeneities, significant dose-calculation errors may occur for single beams, even when using a modern commercial TPS. Such errors have also been observed for clinical, multibeam head and neck treatment plans (34,251,270). Improved accuracies can be obtained with Monte Carlo dose-calculation algorithms, and vendors of TPSs have started to offer this tool (129,251). However, to obtain clinically acceptable calculation times, approximations and simplifications are often used that could jeopardize the potential advantages of the full Monte Carlo technique. A comprehensive overview is provided by Reynaert et al. (252).



Paperless Electronic Records
In the previous section one is confronted with innovative, highly technological care, but also with clinical research regarding QOL issues. These processes will undoubtedly go on with virtually no limitations. From the organizational (data-retrieval) point of view, one could envisage that most departments of radiation oncology will eventually be structured as a “paperless office” (Fig. 42.47). Direct architecture changes the conventional workflow into a productive workspace environment; that is, with a click of a button it combines the ease of the use of Windows with access to all types of vendor applications, including record and verify, e-mail, IMRT-QA, and office applications. This server client architecture gives users the freedom to access their applications anywhere in the hospital or in the world for collaboration, consultation, or to access particular applications for personal use. Features like pen-enabled computing, centralized storage, and secure remote access of the applications via broadband are not new in the information technology arena, but definitely are not routine to radiation therapy. It is setting a stage for any type of new application to fit into the existing infrastructure without adding new workstations or PCs in the already fully taken workspace. The future generation of connectivity between radiotherapy applications is true flexibility at the physician's desk, a solution without constraints.