09_03
Impact of MLC Characteristics
Adjustments to leaf trajectories are required to account for the various effects associated with MLC characteristics, including the rounded leaf tips, tongue-and-groove leaf design, interleaf and intraleaf transmission, leaf scatter, and collimator scatter upstream from the MLC. The accuracy of dose delivered and the agreement between calculated and measured dose distributions depend on the adequate accounting of these effects. Approximate empirical corrections are applied for these effects by algorithms and software that convert optimized intensity distributions into leaf trajectories.
MLCs have an interlocking tongue-and-groove leaf design to minimize interleaf leakage. However, there is a difference in interleaf leakage and leakage through the leaves. This difference can become significant for beams that require large number of MUs and in portions of the beams that receive large fractions of their dose through leakage. Currently, this effect is ignored, although the use of Monte Carlo techniques to account for it is being investigated (63,113).
In addition, there are circumstances during creation of intensity profiles when a thin strip of the irradiated medium is shielded by the tongue of one leaf pair or the groove of the adjacent leaf pair rather than being completely exposed or completely blocked. van Santvoort and Heijmen (128) have demonstrated that this leads to an underdosage in the thin strip. They, and subsequently Webb et al. (135), also showed that this effect could be removed by the use of leaf motion-synchronizing techniques. However, such techniques result in an increase in the number of MUs. Furthermore, this effect is not considered to be of significant clinical consequence because of the smearing caused by multiple fields and the positioning and motion uncertainties. Using different collimator angles for each field can reduce this effect further.
Depending on the complexity (the frequency and amplitudes of peaks and valleys) of the intensity pattern, points within the field aperture may receive a substantial portion of the dose as a result of radiation transmitted through or scattered from the leaves when the points are in the shadow of the leaves. Points outside the leaf aperture receive their entire dose through these “indirect” sources. The complexity of intensity distributions produced by the IMRT optimization process depends on a combination of several clinical factors including the shapes, sizes, and relative locations of tumor and normal tissues; required tumor dose; dose homogeneity; and dose-volume limits of normal tissues. Intensity distributions for head and neck cases, for example, tend to be considerably more complex than for prostate cases. For beams with highly complex intensity patterns, the average window width to deliver the treatment tends to be small and, for the same dose received by the tumor, the treatment time (i.e., the number of MUs) is long. Consequently, the contribution of radiation transmitted through and scattered from the leaves may form a significant fraction of the total dose delivered. Because these contributions are accounted for approximately, the uncertainty in dose delivered is increased. In addition, the differences between interleaf and intraleaf transmissions may no longer be negligible. Another consequence of complex intensity patterns is that the lower limit of the deliverable intensity is high.
The deliverable dose distributions may be significantly different from the original optimized ones. There are different ways to overcome the difficulties resulting from the differences in desired and deliverable dose distributions. For example, if the deliverable dose to a particular normal structure is higher than the original optimized dose, the planner could modify the objective function to demand an appropriately lower dose. Alternatively, as shown in Figure 9.16, the optimization loop could include a pass through leaf sequence generation and calculation of deliverable dose distributions. The optimizer then adjusts ray weights based on deliverable dose distributions rather than the idealized ones. This scheme has been investigated by Siebers et al. (114).
P.253
QA for Intensity-Modulated Treatments
A number of QA steps unique to IMRT are needed to ensure the accuracy and safety of treatments. These include QA of the MLC in dynamic mode, dosimetric verification for each dynamic beam as well as for the composite treatment plans, portal imaging, treatment verification, in vivo dosimetry, and reduction in uncertainty associated with daily positioning and internal organ motion during irradiation.
When using conventional 3DCRT, MLC leaf position calibration errors influence the accuracy of the radiation distribution at the portal boundary. Because of PTV and beam penumbra margins, small errors in leaf calibration will have a minimal effect on the target volume dose. The accepted leaf calibration accuracy is 2 mm (1). However, for IMRT the MLCs are used to generate inhomogeneous fluence distributions. In the sliding-window technique, for instance, this is done by adjusting the velocity and width of leaf gaps during radiation delivery. If the MLC calibration is inaccurate, the delivered dose distribution will be in error. The error is a function of the ratio of leaf calibration error to the sliding-window width. For example, a 1-mm imprecision in the gap would result in a 10% error in dose if a uniform field were to be delivered using a sliding window of 1 cm. For step-and-shoot delivery, magnitudes of dose errors are greater (owing to the steep dose gradients near the MLC leaf edges), but they are confined to the subfield edges. Thus, it is important that the manufacturers of MLCs used for IMRT ensure that the leaves can be positioned with accuracy of better than 0.25 mm, and the physicists must ensure through routine QA procedures that such precise positioning is achieved and maintained. It is interesting that integral dose error is similar for both the step-and-shoot and sliding-window techniques, but the distribution of the error is different.
Because MLC leaf calibration and the accuracy of MLC operations influence the delivered dose distribution, new, more rigorous MLC QA procedures have been developed. Chui et al. (31) and LoSasso and Chui C-S, Ling (77), among others, have developed QA procedures specifically for MLCs used in dynamic mode. Periodic QA checks must ensure that the leaves of the MLC do indeed move to their designated positions at the specified values of MUs. Moreover, to ensure safe and accurate delivery of treatments with an MLC, the manufacturers must include redundant and independent sensors for the leaves of the MLC. Furthermore, in the event of treatment field interruption and resumption, there should be no perceptible change in dose delivered.
Another aspect of QA important for IMRT is the daily positioning uncertainty and motion during irradiation. IMRT is a highly conformal and highly precise form of radiotherapy frequently used to escalate dose. Dose distributions may have steep dose gradients between the target and the neighboring normal structures. Furthermore, margins may be much smaller than in conventional treatments. Patient positioning and immobilization requirements are more stringent than ever to ensure that the target volumes are covered adequately and the normal tissues are spared adequately. In fact, special immobilization devices and techniques are being developed to reproducibly and accurately position the target volume and normal anatomy. Many of these devices already are available commercially (e.g., rectal inserts to improve positioning for prostate IMRT).
Similarly, motion during treatment, mainly as a consequence of respiration, also can be a serious problem for IMRT of sites in the thorax and abdomen. Because IMRT is delivered dynamically, the moving target volume may move in and out of the instantaneous field of radiation. Some portions of the target volume may get more than the planned dose, whereas others may get less. A way to minimize effects of respiratory motion would be to use “gated treatments” in which radiation and leaf motion is turned on only during a specific, reproducible portion of the respiratory cycle or in an interval during which the patient's breath is voluntarily, or involuntarily, held (65,141). In any case, it is also important that imaging used for planning of treatments be gated in a similar way.
Dosimetric Verification of Intensity-Modulated Treatments
To implement a new treatment technology into routine clinical use, there are usually three distinct but closely-related phases: Acceptance tests: This is the initial set of tests that ensures the hardware and software meet the factory or customer-provided specifications. Usually, but not always, the written specifications contain the necessary instructions or guidelines for these tests (in order to avoid legal ambiguity in the measurements). It is also a good opportunity for the users to establish some performance baselines, especially for the hardware purchased. Commissioning tests: The IMRT commissioning is a process to implement IMRT treatments using the customer's hardware and beam data. Various groups have studied the general guidelines for commissioning a treatment planning system. The process usually starts with collection of essential beam data for beam modeling. The parameters of the dose-calculation algorithm are then tuned to provide the best performance for the user's beam. Additional tests should be performed to evaluate the limitations of the treatment-planning system and a solution or a work-around should be found if the problem is clearly identified. Then IMRT phantom measurements should be performed to test the accuracy of the delivery system and data connectivity. If the accuracy is judged to be acceptable, the system can be released to the clinic after the necessary user training and procedural implementations. It is recommended that a small (interdisciplinary) focus group should be assigned to lead the IMRT implementation in the clinic. The “train-the-trainer” approach has proven to be effective in translating new technology into routine clinical practice. On-going QA: After the system is released to the clinic, it is important to establish a routine QA program. The performance of various steps involved in performing IMRT treatments needs to be tracked so that the quality of the treatments can be maintained. The ongoing QA program can be separated into patient-specific QA and equipment QA, which will be described in more detail below.
Patient- and Equipment-Specific QA
Because of the complexity of irregular field shapes, small-field dosimetry, and time-dependent deliverable leaf sequences, it is recommended by the American Association of Physicists in Medicine and ASTRO, that patient-specific QA should be performed as a part of the IMRT management process and a requirement for billing for IMRT services. Figure 9.17 shows the general categories of patient- and equipment-specific QA which are detailed in the following.
Patient setup, although not specific to IMRT dosimetry, is considered a key step in ensuring accurate IMRT treatments. A variety of image-guided localization techniques have been proposed for use with IMRT treatments, from simple orthogonal portal films to the beam's eye view portal film with IMRT intensity pattern overlays (79), daily electronic portal imaging of implanted fiducials (48,137), daily ultrasound-guided localization (6,64,82), and to the most integrated tomotherapy solutions (87). The detailed discussion of these specific image-guided procedures is out of the scope of this chapter, but QA in patient positioning remains an important issue for IMRT.
The implementation of patient-specific QA depends highly on each institution. For example, dosimetric measurements of MU
P.254
settings can be verified for each beam individually (usually in a flat [slab] phantom geometry) or for the composite treatment plan (usually in a specially designed phantom, but it is also possible to use the simple slab phantom setup). Unlike single-beam verification in which the single-beam dose distribution can be significantly different from the original patient plan, the advantage of measuring the composite treatment plan in a phantom (regardless of the shape of the phantom) is that the composite dose distribution or the dose “pattern” generated in a phantom is usually similar to those in the original patient plan. This can be useful in selecting the measurement points or in visualizing potential dose errors. Absolute dosimetry is usually referred to as “MU verification” for IMRT. The traditional manual process for MU verification is virtually impossible to perform because of the large number of fields involved and the irregular shape and size of the treatment segments. Attempts have been made to verify MU settings in an IMRT plan using alternative calculation methods (102,112,129). However, these alternative calculation methods cannot predict the uncertainties during the actual delivery at the treatment machines and are also subject to limitations and approximations in their dose-calculation models. The most reliable and practical technique currently for IMRT MU verification is still the ion chamber-based point dose measurement in a phantom. Absolute dose measurement in a phantom is usually performed through a process called the hybrid phantom plan. In this plan, all beam angles and deliverable intensity patterns for a patient plan are transferred to the phantom, and doses in the phantom are computed for QA. The basic assumption in this process is that if the dose calculated in the phantom agrees with the measurement in the phantom, then the dose delivered to the patient agrees with the dose calculated in the patient. Relative dosimetry is usually performed using radiographic films or 2D array detectors. The process is similar to absolute dose measurement using the hybrid phantom plan technique. For film dosimetry, it is important to convert film density into relative dose using a film-calibration process. Because of the additional dimensionality, it becomes difficult to define good numerical criteria for evaluating relative 2D/3D measurements. Various numerical indicators (such as the distance to agreement, and gamma, or normalized agreement test) were proposed. In particular, the concept of gamma, combining the dose difference and distance to agreement, is appealing in evaluating 2D or 3D dose distributions. For clinical applications, the most reliable and practical way to evaluate 2D distribution is to overlay the measurement isodose lines with the calculated ones. Special attention should be made to the low-dose regions near critical structures in the original patient plan. Attention should also be made to the systematic shifts of isodose lines, which may reveal if the isocenter or any reference setup point may be off. The relative dosimetry verification for IMRT should be performed in conjunction with the absolute dose verification for IMRT. It would be useful if the relative dose distribution can be normalized to the absolute dose measurement point, which converts the relative dose measurement into absolute dose distributions (71).
Two-dimensional fluence verification of intensity patterns gained popularity with the invention of 2D array detectors and the necessary software (59,66,146). Fluence verification usually is performed for each IMRT beam at a fixed gantry angle with or without a flat phantom geometry. The purpose of fluence verification is to make sure the intensity patterns created in each IMRT plan can be faithfully delivered under ideal conditions (2D, beam's eye view). Fluence verification should be combined with other patient-specific and equipment QAs to make sure that IMRT treatments are executed accurately.
Figure 9.17 also illustrates equipment-specific QA procedures. In general, IMRT QA is a subset of general equipment QA processes. The technology of IMRT and techniques for QA are also evolving. It is strongly suggested that users of IMRT should attempt to attend national meetings and technology conferences or training courses so that their knowledge about the use of IMRT can be updated regularly.
IMRT has been variously termed as opaque, unintuitive, and nontransparent, partly because it is delivered using dynamic techniques. Many are skeptical about whether the dose distribution displayed on an IMRT plan is, in fact, delivered. Furthermore, because of the complexity of computations involved, there is no practical way to verify the MU settings by hand calculations as is done for conventional treatments. Moreover, because of the inherent nonuniformity of IMRT fields, it is important to know the dose accurately at every point within the beam. One way to check if the intended dose would be delivered to the patient at the time of the treatment is to conduct dosimetric verification measurements.
Two broad categories of IMRT treatment-plan verification approaches have been developed for MLC-based IMRT. First, the dose distribution from radiation fields are independently measured and evaluated. This often is accomplished by using a flat homogeneous water-equivalent phantom and irradiating each field independently. The film-measured dose distributions are compared against calculations conducted by the treatment-planning system under the same geometric conditions. The process is explained in Figure 9.18. For calculation of dose distributions, each field is transferred to a treatment plan with flat homogeneous phantom. A typical example for a sliding-window intensity-modulated beam dosimetric verification is shown in Figure 9.19. This technique has the advantage that discrepancies between the planned and delivered dose can be attributed to individual radiation portals. However, the total integrated dose distribution is not checked.
The second method uses a phantom that is irradiated by all beam portals, allowing the evaluation of the total dose distribution delivered (80,130). Typically, ionization chambers and radiographic film are the dosimeters used for these measurements. Although ionization chambers can be benchmark-quality dosimeters, they suffer from volume averaging and are inefficient for measuring multiple points. Because of the complexity of the dose distributions being measured, a 2D dosimeter is required for thorough evaluations of nonuniform dose distribution. Quantitative radiographic film measurements require careful dose calibrations using independently measured
P.255
sensitometric curves. The film optical densities are measured and converted to absolute dose using film-calibration data and compared with the predictions of the treatment-planning system (42).
In vivo dosimetry commonly is used to verify the dose delivered by conformal therapy radiation fields. The complex fluence distribution of IMRT fields makes quantitative use of in vivo dosimetry, specifically the use of skin-surface mounted dosimeters, difficult.
Film, thermoluminescent dosimeters, and diodes may not be sufficiently accurate; are laborious to use; and, in the case of thermoluminescent dosimeters and diodes, are incapable of providing detailed information. In the long run, the most efficient way to verify fixed intensity-modulated fields is expected to be with real-time 2D dosimetry systems using appropriately calibrated electronic portal imaging devices. Such devices could be used for dosimetric verification of IMRT beams before treatment delivery and for exit dosimetry using transmitted portal dose images (PDIs). For electronic portal imaging devices to be used for pretreatment dosimetric verification and exit dosimetry, they must operate in the integration mode to capture the transmitted radiation over the entire exposure of each beam. The result is a PDI that can be compared with an intensity-modulated digitally reconstructed PDI. For pretreatment dosimetric verification of a given beam, a PDI may be created using a 3D-treatment-planning system to compute dose deposited in the electronic portal imaging device detector. For exit dosimetry, the PDI may be calculated using the 3D CT image of the patient. In either case, for accurate dosimetric verification, the effect of scattered radiation and the variation in response of the detector with energy must be included. The former effect can be taken into account with dose-spread kernel superposition methods, but both can be accounted for using Monte Carlo techniques.
Treatment Setup and Delivery
Fixed-Gantry Intensity-Modulated Fields
As for conventional radiotherapy, for IMRT techniques using fixed intensity-modulated fields, it is necessary to verify the
P.256
patient alignment using portal images with beams used for actual treatment before the delivery of first treatment and then periodically thereafter. However, no beam apertures are required for IMRT. Therefore, special fields for portal imaging with apertures are created in which the shape of each aperture is defined by the terminal positions of the leading leaf tips and the starting positions of the trailing leaf tips.
Intensity-modulated treatments may be delivered remotely or automatically under computer control. The treatment machine computer may automatically set up the various components of the machine and switch on the radiation beam. For the sliding-window technique, it moves leaves during irradiation in the sequence specified in the leaf motion dataset. In the step-and-shoot mode, the radiation pauses while the leaves move. At the completion of the first field, the computer sets the machine for the next field and again goes through its leaf-motion sequence and irradiation. This process is repeated until all fields are delivered. The treatment times may vary somewhat and depend on the number of fields involved and the complexity of the fluence distribution. Current time estimates range from 5 to 20 minutes, excluding patient setup (76,77,131).
Setup and IMRT Delivery with Serial Tomotherapy
Because there are no specific beam directions or portals associated with serial tomotherapy beam delivery, the treatment QA concentrates on patient positioning and immobilization. The add-on multileaf collimator (MIMiC) is relatively heavy and its removal is time consuming, so portal films often are acquired with the MIMiC in place. This limits the portal fields to a roughly 3.4 × 20 cm2 field size. Therefore, the imaging of useful, immobile, bony anatomic landmarks is critical for each port film, meaning that the selection of the portal film locations is critical to the accurate determination of patient-positioning accuracy. The positions of the films do not need to coincide with the locations of the treated indices, but the digitally reconstructed radiograph that is used to compare against the portal film must be simulated at the same relative couch position as the portal film is acquired. Typically, anteroposterior and lateral films are acquired, and if the target is longer than 10 cm and is in a location where patient structures are flexible (e.g., in the neck), portal films may be required at multiple couch positions to assure the patient is in the correct orientation throughout the length of treatment.
Treatments are conducted by placing the patient on the couch and aligning the patient to the linear accelerator in the standard fashion. Once the patient is aligned (to a point analogous to isocenter for conventional treatments), the couch translation device (called CRANE) coordinates are set to zero and the couch is moved to the location of the first index. This position is determined by the treatment-planning system. The gantry is rotated to the starting arc position, and the patient treatment plan is loaded onto the MIMiC control computer. The linear accelerator is operated in normal arc mode, and the MIMiC control computer determines if the treatment can proceed. If the gantry speed is within acceptable limits, the MLC leaves are opened in their programmed sequence. The MIMiC communicates with the linear accelerator using the conventional door interlock. If the MIMiC control computer determines the treatment should not continue, the door interlock circuit is interrupted and the linear accelerator ceases operation just as if the door had been opened (the door interlock fault is tripped on the accelerator). Once the arc is delivered, the therapist enters the room to move the CRANE to the next couch position and reprograms the MIMiC controlling computer by following the screen prompts.
Tomotherapy Versus Fixed-Gantry IMRT
The physical and operational differences between tomotherapy and fixed-gantry IMRT lead to trade-offs when considering each system. The rotational beams used in tomotherapy could be a significant advantage until robust beam configuration optimization tools are developed, particularly those involving noncoplanar beams.
For serial tomotherapy delivery, one of the difficulties is the requirement of precisely moving the patient between successive arc deliveries (couch indexes). The dose-delivery error made for an incorrect junction move is similar to the errors in abutting conventional fields. Studies have shown that the maximum dose error is 25% mm-1 in the abutment region for errors in couch index movement or intrajunction patient motion (83). When conventional fields are abutted, feathering often is used to reduce the risk of systematic dose errors. A similar technique has been suggested for distributing the abutment regions for serial tomotherapy (41) by creating multiple treatment plans with modified target volumes to force a redistribution of indexes.
Even when perfectly abutted, there are dose heterogeneities within the abutment region caused by the divergent radiation fields, especially when arcs of less than 360 degrees are used. Low et al. (83) studied the abutment region dose distributions for arcs ranging from 180 degrees to 340 degrees and determined that the tumor doses can have significant cold spots when short arcs are used. These become more severe when the longer leaf setting (1.7 cm) is used. The accuracy of the treatment-planning system in predicting these heterogeneities was not evaluated, but the system tends to underestimate the severity of the heterogeneities. Although the divergence in the radiation beams is still present in helical tomotherapy, the helical path of the field edge distributes the diverging distribution such that dose errors caused by inaccuracies in couch motion, or by patient movement, are significantly smaller than with serial tomotherapy.
One of the advantages of fixed-gantry IMRT is the availability of noncoplanar directions. The commercial hardware device used to precisely move the couch between successive indexes also is produced in a model that attaches directly to the couch, allowing for couch rotations. Although limited noncoplanar dose delivery is possible when using serial tomotherapy, especially when treating the brain, this has not been widely adopted.
Gating for serial tomotherapy is impractical because of the use of conventional linear accelerators and the lack of shared information between the MLC and the linear accelerator control computers. Breath-hold techniques are also impractical because of the relatively long time to rotate the linear accelerator gantry. Because of the potentially large abutment-region dosimetry errors, it is important to consider the immobilization accuracy of targets and critical structures when selecting targets for serial tomotherapy. Gating for helical tomotherapy is possible by pausing the radiation beam and the couch motion when the gating circuitry dictates that no treatment should be delivered. However, there will be a delay in restarting the treatment after the gating signal has been restarted while waiting for the gantry to return to its position when the gating signal was interrupted.
Because the dose is delivered over many indexes or gantry rotations, there are many more MUs used when treating with tomotherapy than for conventional 3DCRT or MLC-based IMRT. The ratio of MUs can be as high as 10:1 even when compared with MLC-based IMRT (104). This increase in MUs leads to increases in whole-body dose that may yield a significant increase in secondary radiation-induced malignancies. The solution to
P.257
this is to improve the linear accelerator head shielding, the source of most of the whole-body dose in tomotherapy.
Another limitation of tomotherapy is the lack of electron beams. Electron beams (including energy and intensity-modulated electron beams), by themselves or in combination with intensity-modulated photon beams, are expected to play an important role in the treatment of certain sites (e.g., breast).
A major advantage of helical tomotherapy is that it is a dedicated IMRT device. However, MLC-based IMRT is likely to compete as a delivery mode resulting in part from the limitations of tomotherapy discussed earlier. Furthermore, the large base of MLC-mounted linear accelerators will mean that the adoption of tomotherapy for significant numbers of IMRT patient treatments will take many years.
Special Requirements of Facility Design for IMRT
The room-shielding design characteristics for IMRT delivery are different than those for conventional radiotherapy. Shielding requirements are determined separately for primary and scattered radiation barriers and for tomotherapy and MLC-based IMRT. For MLC-based IMRT, the total integrated radiation fluence remains similar to that used in conformal therapy, so no change in primary barrier thicknesses is expected. However, the increase in MUs of about a factor of 3 is expected to increase the required secondary shielding barrier attenuation, at least until the linear accelerator manufacturers improve the head leakage characteristics. For serial tomotherapy without a beam stopper, the same primary barrier is struck for each couch index, indicating that an increase in primary barrier thickness may be required. However, the use of a rotating beam, and the relatively small angle subtended by the MIMiC, reduces the effective use factor to the point that it almost exactly cancels the number of times the beam strikes the primary barrier. Increases in secondary shielding, however, may be greater than for IMRT because the total number of MUs is significantly greater.
Clinical Experience with IMRT
IMRT of Head and Neck Cancer
The first report of the application of IMRT to head and neck neoplasms was from Baylor College. Kuppersmith et al. (68) reported a decrease in dose to the parotid glands to less than 30 Gy in 28 patients treated with IMRT using serial tomotherapy. They also found the incidence of acute toxicity to be drastically lower than with conventional radiation therapy. Later, Butler et al. (18) implemented the “simultaneous modulated accelerated radiation therapy” (SMART) technique, an equivalent of the SIB technique, and found that 19 out 20 patients treated had complete response with acceptable toxicity. Low et al. (81) have described the application of the serial tomotherapy technique and QA practices for head and neck treatments at Washington University in St. Louis. Preliminary results of the use of these techniques for 17 patients have been reported by Chao et al. (23) and showed that the tumor control is promising with no severe adverse acute side effects. A prospective clinical study conducted by Chao et al. (22) also showed that the sparing of parotid glands translated into objective and subjective improvement of both xerostomia and quality-of-life scores in patients with head and neck cancers treated with IMRT.
In another publication, Chao et al. (24) also reported that the dosimetric advantage of IMRT did translate into significant reduction of late salivary toxicity in patients with oropharyngeal carcinoma and had no adverse impact on tumor control and disease-free survival. In this study, 430 patients with carcinoma of the oropharynx were treated. There were 260 patients with primary tumors of the tonsil and 170 patients with tumors arising from the base of the tongue. Twenty-four (6%) patients had stage I disease, 88 (20%) had stage II, 128 (30%) had stage III, and 190 (44%) had stage IV disease. Patients were divided into five groups. Group I consisted of 109 patients who received preoperative conventional radiation therapy. Group II consisted of 142 patients who received postoperative conventional radiation therapy. Group III consisted of 153 patients who received definitive conventional radiation therapy. Serial tomotherapy IMRT was used to treat 14 patients postoperatively (group IV) and 12 patients definitively without surgery (group V). With a median follow-up of 3.9 years, the 2-year locoregional control values for the five studied groups were 78%, 76%, 68%, 100%, and 88%, respectively. The 2-year disease-free survival values for the five studied groups were 68%, 74%, 58%, 92%, and 80%, respectively. IMRT significantly reduced the incidence of late xerostomia. The benefit of IMRT in nasopharyngeal carcinoma has been reported by Cheng et al. (28), who showed that target coverage of the primary tumor was maintained and nodal coverage was improved with IMRT in 17 nasopharyngeal carcinoma patients, as compared with conventional beam arrangements. Also, the ability of IMRT to spare parotid gland was considerably superior. Hunt et al. (56) reported similar results with 23 primary nasopharyngeal carcinoma patients. Lee et al. (72) reported data for 67 head and neck cancer patients treated with IMRT using three different techniques:
• Manually cut partial transmission blocks,
• Computer-controlled auto-sequencing multisegment approach, and
• Serial tomotherapy (MIMiC).
Fifty patients received concomitant cisplatinum and adjuvant cisplatinum and 5-fluorouracil chemotherapy according to the Intergroup 0099 trial. Twenty-six patients had fractionated high–dose-rate intracavitary brachytherapy boost and one patient had gamma-knife radiosurgery boost after external-beam radiotherapy. With a median follow-up of 31 months, the 4-year estimates of local progression-free, locoregional progression-free, and distant metastases-free rates were 97%, 98%, and 66%, respectively. IMRT provided excellent tumor target coverage and allowed the delivery of a high dose to the target leading to excellent locoregional control for nasopharyngeal carcinoma with significant sparing of the salivary glands and other nearby critical normal tissues.
IMRT also has been found to be suitable for patients with carcinoma of the paranasal sinuses. Claus et al. (32) observed no dry eye or other visual disturbances in 11 ethmoid sinus cancer patients, finding that the optic pathway dose can be reduced selectively by IMRT and that the IMRT has the potential to save binocular vision. In this study, however, no evaluation of retinopathy was done because the follow-up period was short. Similar significant advantages of IMRT were shown for six patients over conventional radiation therapy and 3DCRT for treatment of maxillary sinus carcinoma (2).
The Virginia Commonwealth University group has reported data on a dose-escalation trial of advanced head and neck squamous cell carcinomas (143). In their protocol, IMRT was designed and delivered using the SIB strategy. Primary dose levels of 68.1, 70.8, and 73.8 Gy, given in 30 fractions (biologically equivalent to 74, 79, and 85 Gy, respectively, if given in 2 Gy per fraction), were evaluated. Simultaneously, the subclinical disease and electively treated nodes were prescribed 60 and 54 Gy, respectively (biologically equivalent to 60 and 50 Gy, respectively, if given in 2 Gy fractions). For each prescribed dose level, the required cord and brainstem sparing and a variable degree of parotid gland sparing were achieved. Dose homogeneities for GTV were on average 6.2%, 8.3%, and 8.8% for the three dose
P.258
levels. An average of 25.8 Gy to the contralateral parotid gland and 40.8 Gy to the ipsilateral parotid gland were achieved.
In patients treated with IMRT, local recurrences have been mainly within the high-dose region. Chao et al. (26) reported the patterns of failure in 126 head and neck IMRT patients treated with serial tomotherapy. Seventeen locoregional failures were detected, and nine of those failures were inside the GTV. Two-year actuarial locoregional control rate was 85%, and 2-year ultimate locoregional control rate after salvage surgery was 89%. Dawson et al. (39) reported the patterns of failure analysis of 58 patients. Patients were treated with forward-planning IMRT and 3DCRT. The actuarial locoregional control rate was 79% with a median follow-up of 27 months. For 14 out 15 patients, the locoregional failures were in the region containing gross disease and in the adjacent soft tissue at risk.
IMRT of Prostate Cancer
Zelefsky et al. (149) reported the results of 171 patients with localized prostate cancer treated with IMRT using fixed-gantry beams and sliding-window techniques. They found that IMRT improved target coverage and significantly reduced the volumes of rectal and bladder walls exposed to high doses. Acute grade 1 to 2 and late grade 2 rectal toxicities were significantly lower in the IMRT group in comparison with the 3DCRT group despite the fact that the latter mostly used lower target doses. In another publication, Zelefsky et al. (150) reported the results of 772 patients with clinically localized prostate cancer treated with IMRT. A total of 698 patients (90%) were treated to 81 Gy, and 74 patients (10%) were treated to 86.4 Gy. The 3-year actuarial prostate-specific antigen relapse-free survival rates for favorable, intermediate, and unfavorable risk group patients were 92%, 86%, and 81%, respectively. Eleven patients (1.5%) developed late grade 2 rectal bleeding. Four patients (0.1%) experienced grade 3 rectal toxicity. No grade 4 rectal complications have been observed. The 3-year actuarial likelihood of late grade 2 rectal toxicity was 4%. Seventy-two patients (9%) experienced late grade 2 urinary toxicity, and five (0.5%) developed grade 3 urinary toxicity (urethral stricture). The 3-year actuarial likelihood of late grade 2 urinary toxicity was 15%. These data represented the largest compilation of patients treated with IMRT and demonstrated the feasibility of high-dose radiation delivery with IMRT for patients with localized prostate cancer. Acute and late rectal toxicities seemed to be significantly reduced compared with what has been observed with conventional 3DCRT. Short-term prostate-specific antigen control rates also seemed to be at least comparable to those achieved with 3DCRT at similar dose levels.
Pollack et al. (106) reported their preliminary results of acute toxicity on 100 patients treated on a randomized trial of 76 Gy in 38 fractions versus 70.2 Gy in 26 fractions. All patients were treated using IMRT. The hypofractionated regimen appeared to be well tolerated but did have slightly higher rates of acute GI toxicity. Kupelian et al. (67) reported the results of a single-arm hypofractionated IMRT protocol using 70 Gy in 28 fractions and found that biochemical relapse free survival was 97%, 88%, and 70%, respectively, for low, intermediate, and high-risk disease. Rates of grade 2 or 3 rectal toxicity at 5 years were 5%.
Other studies of prostate IMRT include one conducted at Royal Marsden Hospital. This study of 10 patients with prostate cancer showed that, when both prostate gland and pelvic lymph nodes are included in the target volume, a reduction in critical pelvic volume irradiated can be achieved with IMRT allowing a modest dose escalation with acceptable complication rates (105). Teh et al. (124) reported acceptable acute genitourinary toxicity with a mean dose of 69 Gy in 40 postprostatectomy patients. Using a rectal balloon for prostate immobilization, they further reported the treatment results of 100 patients with prostate cancer with definitive IMRT. They found that there was no grade III or IV acute gastrointestinal (GI) and genitourinary toxicity (125). Morr et al. (100) have evaluated the clinical feasibility of daily computer-assisted transabdominal ultrasonography for target position verification in the setting of IMRT for prostate cancer. Twenty-three patients with clinically localized prostate cancer were treated with serial tomotherapy and transabdominal ultrasonography. They found that this localization technique can be practically implemented as part of a daily setup routine and could be a useful tool in the clinical practice of IMRT for prostate cancers.
IMRT Experience with Other Cancer Sites
In addition to strong data supporting the use of IMRT in head and neck and prostate cancer, there are a number of preliminary studies reporting the feasibility and outcomes of IMRT of other cancers; many of these reports are theoretical dosimetric studies. Theoretically improved dosimetry alone probably does not serve as sufficient justification for the routine use of IMRT in these cases, and in the absence of robust clinical data regarding actual treatment outcomes, IMRT in these settings should be considered investigational.
IMRT of Breast Cancer
Commercial systems are now available that can autocontour the volume of breast tissue within conventionally designed tangential photon portals and then use an inverse planning algorithm to optimize dose homogeneity within these tangential portals (67). However, most commercial inverse planning systems could not handle the “skin flash” appropriately. Due to setup uncertainties and breathing motion, a portion of the breast (target) tissue may move outside the skin line as indicated by the treatment planning CT images of the patient. The traditional IMRT technique to overcome target motion uncertainty is to expand the PTV and optimize the dose coverage to the entire PTV. However, this strategy may not work because a portion of the PTV will be expanded into the air, which does not have the necessary mass to absorb the dose. Some treatment-planning systems ignore the regions outside the skin contour entirely. Therefore, it may be necessary to add “virtual” tissues in the PTV for the inverse planning system. Sometimes it may require users to manually open certain IMRT segments to take care of the skin flash effect.
Nevertheless, there has been interest in using IMRT for left-sided beast cancers in order to spare myocardium from the high-dose region of the radiotherapy fields. No robust data regarding clinical outcomes after IMRT for breast cancer exist; however, a variety of dosimetric studies (29,44,54,75,110,126,127) have suggested reductions in lung and myocardium doses when IMRT is compared to conventional radiotherapeutic techniques. Hurkmans et al. (57) used a normal tissue complication probability (NTCP) model to estimate the NTCP for cardiac and lung complications due to radiotherapy and found that IMRT did decrease the NTCP for late cardiac toxicity compared to more conventional radiotherapy techniques, but had a minimal effect on the NTCP for radiation pneumonitis.
Hong et al. (54) reported a dosimetric study of IMRT in 10 cases of intact breast cancer showing significant reduction of dose to the coronary arteries, ipsilateral lung, and surrounding soft tissues. It simultaneously improved dose homogeneity throughout the target volume. Li et al. (74) described a combined electron and IMRT technique for breast cancer treatment, which led to improvement over the conventional treatment technique using tangential fields with reduced dose to the ipsilateral
P.259
lung and the heart. Other studies (44,127) also confirmed that IMRT reduces the high-dose volume in tangential breast irradiation significantly and enables more complete cardiac sparing without compromising PTV coverage in some patients. Furthermore, IMRT creates a possibility to improve field matching in case of multiple field irradiations of the breast and lymph nodes (62,69,127). In addition, IMRT for tangential breast radiation therapy was found to be an effective and efficient method to achieve uniform dose throughout the breast. Preliminary findings reveal minimal or no acute skin reactions for patients with different breast sizes in 32 patients with early-stage breast cancer (62).
IMRT of Gynecologic Cancer
In regard to the targeting of pelvic lymphatics, with IMRT, Taylor et al. (123) mapped the pelvic lymphatics of 20 patients using MRI with the administration of iron oxide particles and found that a modified CTV margin of 7 mm around the iliac vessels resulted in adequate coverage of the pelvic lymphatics.
Ahamad et al. (3) analyzed the normal tissue-sparing effects of IMRT in the treatment of the pelvis after hysterectomy in patients with gynecologic cancers and found that although more small bowel, bladder, and rectum could be spared with IMRT compared to conventional radiotherapeutic techniques, these benefits rapidly diminished with even small expansions of the target volumes. D'Souza et al. (37) used the same dataset of patients as Ahamad et al. (3) and found that IMRT may allow higher doses of radiation (54 Gy) to be delivered safely to the node-bearing regions of the pelvis and the vaginal apex compared to conventional techniques that administer 50.4 Gy.
Salema et al. (111) reported 13 patients treated with extended field pelvic and paraaortic radiotherapy using IMRT and found that 2 patients experienced grade 3 or higher toxicity Both of these patients received concurrent cisplatin-based chemotherapy.
Portelance et al. (107) reported dosimetric comparison between 3DCRT and IMRT for 10 patients with cervical cancer. They demonstrated that, with similar target coverage, normal tissue-sparing was superior with IMRT. Mundt et al. (101) reported the clinical experience of 40 patients with gynecologic malignancy who underwent IMRT to the pelvis. Compared with 35 historic control patients who were treated with conventional techniques, patients treated with IMRT experienced fewer acute GI symptoms than those treated with conventional whole-pelvic radiotherapy.
IMRT of Gastrointestinal Cancer
Crane et al. (36) reported the results of a phase I dose-escalation study of radiotherapy with concurrent gemcitabine chemotherapy. The aim of this study was to alternate escalating the radiation dose by 3 Gy and the gemcitabine dose by 50 mg/m2. The starting dose of gemcitabine was 350 mg/m2 and 33 Gy per 11 fractions of IMRT to the regional lymphatics and primary disease. All three patients in the first cohort who were treated suffered dose-limiting toxicity, and the trial was ultimately closed because of excessive myelosuppression and upper GI toxicity.
Ben-Josef et al. (7) reported on 15 patients with pancreatic cancer treated with concurrent capecitabine and IMRT (45 to 55 Gy) and reported that only 1 patient had grade 3 GI toxicity, specifically GI ulceration that responded to medical management.
Brown et al. (17) performed a dosimetric analysis of 15 patients with pancreatic cancer and compared 3DCRT, IMRT with sequential boost, and IMRT with integrated boost, and found that IMRT with integrated boost allowed dose escalation up to 64.8 Gy to the primary tumor.
Guerrero-Urbano et al. (51) performed a dosimetric evaluation in five patients with locally advanced rectal cancer and found that IMRT with simultaneous integrated boost theoretically reduced the radiation dose to the small bowel compared to 3D conformal techniques. However, clinical data regarding outcomes of patients treated with IMRT for rectal cancer are lacking at this time.
Milano et al. (90) reported on 17 patients with squamous cell carcinomas of the anal canal treated with IMRT with whole pelvic radiation does of 45 Gy followed by boost to the anal canal. Thirteen patients received concurrent 5-fluorouracil and mitomycin-C chemotherapy. Treatment was well tolerated with no grade 3 or higher nonhematologic toxicity, and no required treatment breaks from skin or GI toxicity. However, one patient receiving mitomycin-C chemotherapy did experience grade 4 hematologic toxicity. Three patients who did not achieve a complete response required abdominoperineal resection and colostomy. With a mean follow-up of 20.3 months, there were no other local failures.
Milano et al. (89) also reported on seven patients with gastric cancer treated with IMRT to a dose of 50.4 Gy. No patient experienced grade 3 toxicity. The treated IMRT plans were compared to conventional AP:PA and three-field plans, and the IMRT plans were found to provide better coverage of the target volumes compared to conventional techniques, with better sparing of the liver and kidneys.
IMRT of Lung Cancer
Because of concerns regarding respiratory motion in radiotherapy of lung cancer, the use of IMRT in lung cancer requires some method to account for tumor and organ motion during treatment planning and delivery; these techniques include both respiratory gating (61) and 4D CT planning (4).
The poor local control rates of conventional radiotherapy doses in the treatment of lung cancer (19) have led to much interest in using IMRT to allow for dose escalation to improve local control. Holloway et al. (52) reported the initial results of five patients with unresectable stage II and III non–small cell carcinoma treated on a phase I dose-escalation trial using induction chemotherapy followed by IMRT to a dose of 84 Gy using 2.4 Gy daily fractions. PET CT was used to define target volumes. One patient developed lethal radiation pneumonitis and the trial was halted.
Murshed et al. (103) performed a dosimetric analysis of 41 patients initially treated with 3DCRT to a dose of 63 Gy. IMRT plans were then generated using these patients' initial planning CTs, and IMRT was found to decrease the volume of lung irradiated to both 10 and 20 Gy (V10 and V20). Target coverage was improved with IMRT, and the volumes of heart and esophagus irradiated were also reduced.
Grills et al. (50) performed a dosimetric comparison of four radiotherapy techniques in 18 patients with stage I-IIB lung cancer. The study compared IMRT, optimized multiple beam 3DCRT, two- to three-beam 3DCRT, and traditional wide-field radiotherapy with elective nodal irradiation. This study found that IMRT and optimized 3DCRT resulted in similar doses of radiotherapy to normal tissues in node-negative patients; however, in node-positive patients, IMRT resulted in a 15% decrease in the volume of lung treated to 20 Gy (V20) and a 30% decrease in the NTCP for radiation pneumonitis.
IMRT of Intracranial Malignancies
Iuchi et al. (58) from Japan published a retrospective report on 25 patients with malignant astrocytomas (World Health Organization grade III and IV) treated with IMRT using a hypofractionated regimen of 48 to 68 Gy in 8 fractions. Thirteen
P.260
patients were treated to 68 Gy and 12 patients received doses of 48 to 65 Gy. The IMRT group was compared to 60 patients treated with conventional techniques to doses of 40 to 60 Gy using 2 Gy daily fractions. The 2-year overall survival was significantly improved (p = .043) in patients treated with hypofractionated IMRT (55.6%) compared to those patient treated with conventional techniques (19.4%).
Floyd et al. (46) reported a pilot study of 20 patients treated with hypofractionated IMRT. Treatment was administered during 2 weeks (10 fractions), and 50 Gy (5 Gy/day) was prescribed to enhance disease or the surgical cavity and 30 Gy (3 Gy/day) was prescribed simultaneously to peritumoral edema. Three patients experienced grade 4 brain necrosis, requiring surgical re-excision. There was no mortality from brain necrosis; in fact, those patients manifesting brain necrosis had longer survival times.
Hwang et al. (55) reported on 15 patients with pediatric medulloblastoma treated with conventional craniospinal radiotherapy followed by a boost to the posterior fossa using IMRT. IMRT delivered much lower doses of radiation to the auditory apparatus while maintaining full doses to the desired target volume. Their findings suggested that, despite receiving higher doses of cisplatin and despite receiving radiotherapy before cisplatin therapy, IMRT can significantly decrease the rate of hearing loss in children treated for medulloblastoma.
In closing, IMRT clearly results in improved radiation dose distributions in a variety of cancers. In some cases, the superior dosimetry of IMRT has resulted in improved clinical outcomes for patients; however, the scientific evidence documenting these clinical improvements lags far behind the data documenting improved dosimetry. It is incumbent on radiation oncologists to continue to document improved clinical outcomes with IMRT in the peer-reviewed literature if we wish to justify the use of this expensive technology to our communities in an era of skyrocketing medical costs.
Adjustments to leaf trajectories are required to account for the various effects associated with MLC characteristics, including the rounded leaf tips, tongue-and-groove leaf design, interleaf and intraleaf transmission, leaf scatter, and collimator scatter upstream from the MLC. The accuracy of dose delivered and the agreement between calculated and measured dose distributions depend on the adequate accounting of these effects. Approximate empirical corrections are applied for these effects by algorithms and software that convert optimized intensity distributions into leaf trajectories.
MLCs have an interlocking tongue-and-groove leaf design to minimize interleaf leakage. However, there is a difference in interleaf leakage and leakage through the leaves. This difference can become significant for beams that require large number of MUs and in portions of the beams that receive large fractions of their dose through leakage. Currently, this effect is ignored, although the use of Monte Carlo techniques to account for it is being investigated (63,113).
In addition, there are circumstances during creation of intensity profiles when a thin strip of the irradiated medium is shielded by the tongue of one leaf pair or the groove of the adjacent leaf pair rather than being completely exposed or completely blocked. van Santvoort and Heijmen (128) have demonstrated that this leads to an underdosage in the thin strip. They, and subsequently Webb et al. (135), also showed that this effect could be removed by the use of leaf motion-synchronizing techniques. However, such techniques result in an increase in the number of MUs. Furthermore, this effect is not considered to be of significant clinical consequence because of the smearing caused by multiple fields and the positioning and motion uncertainties. Using different collimator angles for each field can reduce this effect further.
Depending on the complexity (the frequency and amplitudes of peaks and valleys) of the intensity pattern, points within the field aperture may receive a substantial portion of the dose as a result of radiation transmitted through or scattered from the leaves when the points are in the shadow of the leaves. Points outside the leaf aperture receive their entire dose through these “indirect” sources. The complexity of intensity distributions produced by the IMRT optimization process depends on a combination of several clinical factors including the shapes, sizes, and relative locations of tumor and normal tissues; required tumor dose; dose homogeneity; and dose-volume limits of normal tissues. Intensity distributions for head and neck cases, for example, tend to be considerably more complex than for prostate cases. For beams with highly complex intensity patterns, the average window width to deliver the treatment tends to be small and, for the same dose received by the tumor, the treatment time (i.e., the number of MUs) is long. Consequently, the contribution of radiation transmitted through and scattered from the leaves may form a significant fraction of the total dose delivered. Because these contributions are accounted for approximately, the uncertainty in dose delivered is increased. In addition, the differences between interleaf and intraleaf transmissions may no longer be negligible. Another consequence of complex intensity patterns is that the lower limit of the deliverable intensity is high.
The deliverable dose distributions may be significantly different from the original optimized ones. There are different ways to overcome the difficulties resulting from the differences in desired and deliverable dose distributions. For example, if the deliverable dose to a particular normal structure is higher than the original optimized dose, the planner could modify the objective function to demand an appropriately lower dose. Alternatively, as shown in Figure 9.16, the optimization loop could include a pass through leaf sequence generation and calculation of deliverable dose distributions. The optimizer then adjusts ray weights based on deliverable dose distributions rather than the idealized ones. This scheme has been investigated by Siebers et al. (114).
P.253
QA for Intensity-Modulated Treatments
A number of QA steps unique to IMRT are needed to ensure the accuracy and safety of treatments. These include QA of the MLC in dynamic mode, dosimetric verification for each dynamic beam as well as for the composite treatment plans, portal imaging, treatment verification, in vivo dosimetry, and reduction in uncertainty associated with daily positioning and internal organ motion during irradiation.
When using conventional 3DCRT, MLC leaf position calibration errors influence the accuracy of the radiation distribution at the portal boundary. Because of PTV and beam penumbra margins, small errors in leaf calibration will have a minimal effect on the target volume dose. The accepted leaf calibration accuracy is 2 mm (1). However, for IMRT the MLCs are used to generate inhomogeneous fluence distributions. In the sliding-window technique, for instance, this is done by adjusting the velocity and width of leaf gaps during radiation delivery. If the MLC calibration is inaccurate, the delivered dose distribution will be in error. The error is a function of the ratio of leaf calibration error to the sliding-window width. For example, a 1-mm imprecision in the gap would result in a 10% error in dose if a uniform field were to be delivered using a sliding window of 1 cm. For step-and-shoot delivery, magnitudes of dose errors are greater (owing to the steep dose gradients near the MLC leaf edges), but they are confined to the subfield edges. Thus, it is important that the manufacturers of MLCs used for IMRT ensure that the leaves can be positioned with accuracy of better than 0.25 mm, and the physicists must ensure through routine QA procedures that such precise positioning is achieved and maintained. It is interesting that integral dose error is similar for both the step-and-shoot and sliding-window techniques, but the distribution of the error is different.
Because MLC leaf calibration and the accuracy of MLC operations influence the delivered dose distribution, new, more rigorous MLC QA procedures have been developed. Chui et al. (31) and LoSasso and Chui C-S, Ling (77), among others, have developed QA procedures specifically for MLCs used in dynamic mode. Periodic QA checks must ensure that the leaves of the MLC do indeed move to their designated positions at the specified values of MUs. Moreover, to ensure safe and accurate delivery of treatments with an MLC, the manufacturers must include redundant and independent sensors for the leaves of the MLC. Furthermore, in the event of treatment field interruption and resumption, there should be no perceptible change in dose delivered.
Another aspect of QA important for IMRT is the daily positioning uncertainty and motion during irradiation. IMRT is a highly conformal and highly precise form of radiotherapy frequently used to escalate dose. Dose distributions may have steep dose gradients between the target and the neighboring normal structures. Furthermore, margins may be much smaller than in conventional treatments. Patient positioning and immobilization requirements are more stringent than ever to ensure that the target volumes are covered adequately and the normal tissues are spared adequately. In fact, special immobilization devices and techniques are being developed to reproducibly and accurately position the target volume and normal anatomy. Many of these devices already are available commercially (e.g., rectal inserts to improve positioning for prostate IMRT).
Similarly, motion during treatment, mainly as a consequence of respiration, also can be a serious problem for IMRT of sites in the thorax and abdomen. Because IMRT is delivered dynamically, the moving target volume may move in and out of the instantaneous field of radiation. Some portions of the target volume may get more than the planned dose, whereas others may get less. A way to minimize effects of respiratory motion would be to use “gated treatments” in which radiation and leaf motion is turned on only during a specific, reproducible portion of the respiratory cycle or in an interval during which the patient's breath is voluntarily, or involuntarily, held (65,141). In any case, it is also important that imaging used for planning of treatments be gated in a similar way.
Dosimetric Verification of Intensity-Modulated Treatments
To implement a new treatment technology into routine clinical use, there are usually three distinct but closely-related phases: Acceptance tests: This is the initial set of tests that ensures the hardware and software meet the factory or customer-provided specifications. Usually, but not always, the written specifications contain the necessary instructions or guidelines for these tests (in order to avoid legal ambiguity in the measurements). It is also a good opportunity for the users to establish some performance baselines, especially for the hardware purchased. Commissioning tests: The IMRT commissioning is a process to implement IMRT treatments using the customer's hardware and beam data. Various groups have studied the general guidelines for commissioning a treatment planning system. The process usually starts with collection of essential beam data for beam modeling. The parameters of the dose-calculation algorithm are then tuned to provide the best performance for the user's beam. Additional tests should be performed to evaluate the limitations of the treatment-planning system and a solution or a work-around should be found if the problem is clearly identified. Then IMRT phantom measurements should be performed to test the accuracy of the delivery system and data connectivity. If the accuracy is judged to be acceptable, the system can be released to the clinic after the necessary user training and procedural implementations. It is recommended that a small (interdisciplinary) focus group should be assigned to lead the IMRT implementation in the clinic. The “train-the-trainer” approach has proven to be effective in translating new technology into routine clinical practice. On-going QA: After the system is released to the clinic, it is important to establish a routine QA program. The performance of various steps involved in performing IMRT treatments needs to be tracked so that the quality of the treatments can be maintained. The ongoing QA program can be separated into patient-specific QA and equipment QA, which will be described in more detail below.
Patient- and Equipment-Specific QA
Because of the complexity of irregular field shapes, small-field dosimetry, and time-dependent deliverable leaf sequences, it is recommended by the American Association of Physicists in Medicine and ASTRO, that patient-specific QA should be performed as a part of the IMRT management process and a requirement for billing for IMRT services. Figure 9.17 shows the general categories of patient- and equipment-specific QA which are detailed in the following.
Patient setup, although not specific to IMRT dosimetry, is considered a key step in ensuring accurate IMRT treatments. A variety of image-guided localization techniques have been proposed for use with IMRT treatments, from simple orthogonal portal films to the beam's eye view portal film with IMRT intensity pattern overlays (79), daily electronic portal imaging of implanted fiducials (48,137), daily ultrasound-guided localization (6,64,82), and to the most integrated tomotherapy solutions (87). The detailed discussion of these specific image-guided procedures is out of the scope of this chapter, but QA in patient positioning remains an important issue for IMRT.
The implementation of patient-specific QA depends highly on each institution. For example, dosimetric measurements of MU
P.254
settings can be verified for each beam individually (usually in a flat [slab] phantom geometry) or for the composite treatment plan (usually in a specially designed phantom, but it is also possible to use the simple slab phantom setup). Unlike single-beam verification in which the single-beam dose distribution can be significantly different from the original patient plan, the advantage of measuring the composite treatment plan in a phantom (regardless of the shape of the phantom) is that the composite dose distribution or the dose “pattern” generated in a phantom is usually similar to those in the original patient plan. This can be useful in selecting the measurement points or in visualizing potential dose errors. Absolute dosimetry is usually referred to as “MU verification” for IMRT. The traditional manual process for MU verification is virtually impossible to perform because of the large number of fields involved and the irregular shape and size of the treatment segments. Attempts have been made to verify MU settings in an IMRT plan using alternative calculation methods (102,112,129). However, these alternative calculation methods cannot predict the uncertainties during the actual delivery at the treatment machines and are also subject to limitations and approximations in their dose-calculation models. The most reliable and practical technique currently for IMRT MU verification is still the ion chamber-based point dose measurement in a phantom. Absolute dose measurement in a phantom is usually performed through a process called the hybrid phantom plan. In this plan, all beam angles and deliverable intensity patterns for a patient plan are transferred to the phantom, and doses in the phantom are computed for QA. The basic assumption in this process is that if the dose calculated in the phantom agrees with the measurement in the phantom, then the dose delivered to the patient agrees with the dose calculated in the patient. Relative dosimetry is usually performed using radiographic films or 2D array detectors. The process is similar to absolute dose measurement using the hybrid phantom plan technique. For film dosimetry, it is important to convert film density into relative dose using a film-calibration process. Because of the additional dimensionality, it becomes difficult to define good numerical criteria for evaluating relative 2D/3D measurements. Various numerical indicators (such as the distance to agreement, and gamma, or normalized agreement test) were proposed. In particular, the concept of gamma, combining the dose difference and distance to agreement, is appealing in evaluating 2D or 3D dose distributions. For clinical applications, the most reliable and practical way to evaluate 2D distribution is to overlay the measurement isodose lines with the calculated ones. Special attention should be made to the low-dose regions near critical structures in the original patient plan. Attention should also be made to the systematic shifts of isodose lines, which may reveal if the isocenter or any reference setup point may be off. The relative dosimetry verification for IMRT should be performed in conjunction with the absolute dose verification for IMRT. It would be useful if the relative dose distribution can be normalized to the absolute dose measurement point, which converts the relative dose measurement into absolute dose distributions (71).
Two-dimensional fluence verification of intensity patterns gained popularity with the invention of 2D array detectors and the necessary software (59,66,146). Fluence verification usually is performed for each IMRT beam at a fixed gantry angle with or without a flat phantom geometry. The purpose of fluence verification is to make sure the intensity patterns created in each IMRT plan can be faithfully delivered under ideal conditions (2D, beam's eye view). Fluence verification should be combined with other patient-specific and equipment QAs to make sure that IMRT treatments are executed accurately.
Figure 9.17 also illustrates equipment-specific QA procedures. In general, IMRT QA is a subset of general equipment QA processes. The technology of IMRT and techniques for QA are also evolving. It is strongly suggested that users of IMRT should attempt to attend national meetings and technology conferences or training courses so that their knowledge about the use of IMRT can be updated regularly.
IMRT has been variously termed as opaque, unintuitive, and nontransparent, partly because it is delivered using dynamic techniques. Many are skeptical about whether the dose distribution displayed on an IMRT plan is, in fact, delivered. Furthermore, because of the complexity of computations involved, there is no practical way to verify the MU settings by hand calculations as is done for conventional treatments. Moreover, because of the inherent nonuniformity of IMRT fields, it is important to know the dose accurately at every point within the beam. One way to check if the intended dose would be delivered to the patient at the time of the treatment is to conduct dosimetric verification measurements.
Two broad categories of IMRT treatment-plan verification approaches have been developed for MLC-based IMRT. First, the dose distribution from radiation fields are independently measured and evaluated. This often is accomplished by using a flat homogeneous water-equivalent phantom and irradiating each field independently. The film-measured dose distributions are compared against calculations conducted by the treatment-planning system under the same geometric conditions. The process is explained in Figure 9.18. For calculation of dose distributions, each field is transferred to a treatment plan with flat homogeneous phantom. A typical example for a sliding-window intensity-modulated beam dosimetric verification is shown in Figure 9.19. This technique has the advantage that discrepancies between the planned and delivered dose can be attributed to individual radiation portals. However, the total integrated dose distribution is not checked.
The second method uses a phantom that is irradiated by all beam portals, allowing the evaluation of the total dose distribution delivered (80,130). Typically, ionization chambers and radiographic film are the dosimeters used for these measurements. Although ionization chambers can be benchmark-quality dosimeters, they suffer from volume averaging and are inefficient for measuring multiple points. Because of the complexity of the dose distributions being measured, a 2D dosimeter is required for thorough evaluations of nonuniform dose distribution. Quantitative radiographic film measurements require careful dose calibrations using independently measured
P.255
sensitometric curves. The film optical densities are measured and converted to absolute dose using film-calibration data and compared with the predictions of the treatment-planning system (42).
In vivo dosimetry commonly is used to verify the dose delivered by conformal therapy radiation fields. The complex fluence distribution of IMRT fields makes quantitative use of in vivo dosimetry, specifically the use of skin-surface mounted dosimeters, difficult.
Film, thermoluminescent dosimeters, and diodes may not be sufficiently accurate; are laborious to use; and, in the case of thermoluminescent dosimeters and diodes, are incapable of providing detailed information. In the long run, the most efficient way to verify fixed intensity-modulated fields is expected to be with real-time 2D dosimetry systems using appropriately calibrated electronic portal imaging devices. Such devices could be used for dosimetric verification of IMRT beams before treatment delivery and for exit dosimetry using transmitted portal dose images (PDIs). For electronic portal imaging devices to be used for pretreatment dosimetric verification and exit dosimetry, they must operate in the integration mode to capture the transmitted radiation over the entire exposure of each beam. The result is a PDI that can be compared with an intensity-modulated digitally reconstructed PDI. For pretreatment dosimetric verification of a given beam, a PDI may be created using a 3D-treatment-planning system to compute dose deposited in the electronic portal imaging device detector. For exit dosimetry, the PDI may be calculated using the 3D CT image of the patient. In either case, for accurate dosimetric verification, the effect of scattered radiation and the variation in response of the detector with energy must be included. The former effect can be taken into account with dose-spread kernel superposition methods, but both can be accounted for using Monte Carlo techniques.
Treatment Setup and Delivery
Fixed-Gantry Intensity-Modulated Fields
As for conventional radiotherapy, for IMRT techniques using fixed intensity-modulated fields, it is necessary to verify the
P.256
patient alignment using portal images with beams used for actual treatment before the delivery of first treatment and then periodically thereafter. However, no beam apertures are required for IMRT. Therefore, special fields for portal imaging with apertures are created in which the shape of each aperture is defined by the terminal positions of the leading leaf tips and the starting positions of the trailing leaf tips.
Intensity-modulated treatments may be delivered remotely or automatically under computer control. The treatment machine computer may automatically set up the various components of the machine and switch on the radiation beam. For the sliding-window technique, it moves leaves during irradiation in the sequence specified in the leaf motion dataset. In the step-and-shoot mode, the radiation pauses while the leaves move. At the completion of the first field, the computer sets the machine for the next field and again goes through its leaf-motion sequence and irradiation. This process is repeated until all fields are delivered. The treatment times may vary somewhat and depend on the number of fields involved and the complexity of the fluence distribution. Current time estimates range from 5 to 20 minutes, excluding patient setup (76,77,131).
Setup and IMRT Delivery with Serial Tomotherapy
Because there are no specific beam directions or portals associated with serial tomotherapy beam delivery, the treatment QA concentrates on patient positioning and immobilization. The add-on multileaf collimator (MIMiC) is relatively heavy and its removal is time consuming, so portal films often are acquired with the MIMiC in place. This limits the portal fields to a roughly 3.4 × 20 cm2 field size. Therefore, the imaging of useful, immobile, bony anatomic landmarks is critical for each port film, meaning that the selection of the portal film locations is critical to the accurate determination of patient-positioning accuracy. The positions of the films do not need to coincide with the locations of the treated indices, but the digitally reconstructed radiograph that is used to compare against the portal film must be simulated at the same relative couch position as the portal film is acquired. Typically, anteroposterior and lateral films are acquired, and if the target is longer than 10 cm and is in a location where patient structures are flexible (e.g., in the neck), portal films may be required at multiple couch positions to assure the patient is in the correct orientation throughout the length of treatment.
Treatments are conducted by placing the patient on the couch and aligning the patient to the linear accelerator in the standard fashion. Once the patient is aligned (to a point analogous to isocenter for conventional treatments), the couch translation device (called CRANE) coordinates are set to zero and the couch is moved to the location of the first index. This position is determined by the treatment-planning system. The gantry is rotated to the starting arc position, and the patient treatment plan is loaded onto the MIMiC control computer. The linear accelerator is operated in normal arc mode, and the MIMiC control computer determines if the treatment can proceed. If the gantry speed is within acceptable limits, the MLC leaves are opened in their programmed sequence. The MIMiC communicates with the linear accelerator using the conventional door interlock. If the MIMiC control computer determines the treatment should not continue, the door interlock circuit is interrupted and the linear accelerator ceases operation just as if the door had been opened (the door interlock fault is tripped on the accelerator). Once the arc is delivered, the therapist enters the room to move the CRANE to the next couch position and reprograms the MIMiC controlling computer by following the screen prompts.
Tomotherapy Versus Fixed-Gantry IMRT
The physical and operational differences between tomotherapy and fixed-gantry IMRT lead to trade-offs when considering each system. The rotational beams used in tomotherapy could be a significant advantage until robust beam configuration optimization tools are developed, particularly those involving noncoplanar beams.
For serial tomotherapy delivery, one of the difficulties is the requirement of precisely moving the patient between successive arc deliveries (couch indexes). The dose-delivery error made for an incorrect junction move is similar to the errors in abutting conventional fields. Studies have shown that the maximum dose error is 25% mm-1 in the abutment region for errors in couch index movement or intrajunction patient motion (83). When conventional fields are abutted, feathering often is used to reduce the risk of systematic dose errors. A similar technique has been suggested for distributing the abutment regions for serial tomotherapy (41) by creating multiple treatment plans with modified target volumes to force a redistribution of indexes.
Even when perfectly abutted, there are dose heterogeneities within the abutment region caused by the divergent radiation fields, especially when arcs of less than 360 degrees are used. Low et al. (83) studied the abutment region dose distributions for arcs ranging from 180 degrees to 340 degrees and determined that the tumor doses can have significant cold spots when short arcs are used. These become more severe when the longer leaf setting (1.7 cm) is used. The accuracy of the treatment-planning system in predicting these heterogeneities was not evaluated, but the system tends to underestimate the severity of the heterogeneities. Although the divergence in the radiation beams is still present in helical tomotherapy, the helical path of the field edge distributes the diverging distribution such that dose errors caused by inaccuracies in couch motion, or by patient movement, are significantly smaller than with serial tomotherapy.
One of the advantages of fixed-gantry IMRT is the availability of noncoplanar directions. The commercial hardware device used to precisely move the couch between successive indexes also is produced in a model that attaches directly to the couch, allowing for couch rotations. Although limited noncoplanar dose delivery is possible when using serial tomotherapy, especially when treating the brain, this has not been widely adopted.
Gating for serial tomotherapy is impractical because of the use of conventional linear accelerators and the lack of shared information between the MLC and the linear accelerator control computers. Breath-hold techniques are also impractical because of the relatively long time to rotate the linear accelerator gantry. Because of the potentially large abutment-region dosimetry errors, it is important to consider the immobilization accuracy of targets and critical structures when selecting targets for serial tomotherapy. Gating for helical tomotherapy is possible by pausing the radiation beam and the couch motion when the gating circuitry dictates that no treatment should be delivered. However, there will be a delay in restarting the treatment after the gating signal has been restarted while waiting for the gantry to return to its position when the gating signal was interrupted.
Because the dose is delivered over many indexes or gantry rotations, there are many more MUs used when treating with tomotherapy than for conventional 3DCRT or MLC-based IMRT. The ratio of MUs can be as high as 10:1 even when compared with MLC-based IMRT (104). This increase in MUs leads to increases in whole-body dose that may yield a significant increase in secondary radiation-induced malignancies. The solution to
P.257
this is to improve the linear accelerator head shielding, the source of most of the whole-body dose in tomotherapy.
Another limitation of tomotherapy is the lack of electron beams. Electron beams (including energy and intensity-modulated electron beams), by themselves or in combination with intensity-modulated photon beams, are expected to play an important role in the treatment of certain sites (e.g., breast).
A major advantage of helical tomotherapy is that it is a dedicated IMRT device. However, MLC-based IMRT is likely to compete as a delivery mode resulting in part from the limitations of tomotherapy discussed earlier. Furthermore, the large base of MLC-mounted linear accelerators will mean that the adoption of tomotherapy for significant numbers of IMRT patient treatments will take many years.
Special Requirements of Facility Design for IMRT
The room-shielding design characteristics for IMRT delivery are different than those for conventional radiotherapy. Shielding requirements are determined separately for primary and scattered radiation barriers and for tomotherapy and MLC-based IMRT. For MLC-based IMRT, the total integrated radiation fluence remains similar to that used in conformal therapy, so no change in primary barrier thicknesses is expected. However, the increase in MUs of about a factor of 3 is expected to increase the required secondary shielding barrier attenuation, at least until the linear accelerator manufacturers improve the head leakage characteristics. For serial tomotherapy without a beam stopper, the same primary barrier is struck for each couch index, indicating that an increase in primary barrier thickness may be required. However, the use of a rotating beam, and the relatively small angle subtended by the MIMiC, reduces the effective use factor to the point that it almost exactly cancels the number of times the beam strikes the primary barrier. Increases in secondary shielding, however, may be greater than for IMRT because the total number of MUs is significantly greater.
Clinical Experience with IMRT
IMRT of Head and Neck Cancer
The first report of the application of IMRT to head and neck neoplasms was from Baylor College. Kuppersmith et al. (68) reported a decrease in dose to the parotid glands to less than 30 Gy in 28 patients treated with IMRT using serial tomotherapy. They also found the incidence of acute toxicity to be drastically lower than with conventional radiation therapy. Later, Butler et al. (18) implemented the “simultaneous modulated accelerated radiation therapy” (SMART) technique, an equivalent of the SIB technique, and found that 19 out 20 patients treated had complete response with acceptable toxicity. Low et al. (81) have described the application of the serial tomotherapy technique and QA practices for head and neck treatments at Washington University in St. Louis. Preliminary results of the use of these techniques for 17 patients have been reported by Chao et al. (23) and showed that the tumor control is promising with no severe adverse acute side effects. A prospective clinical study conducted by Chao et al. (22) also showed that the sparing of parotid glands translated into objective and subjective improvement of both xerostomia and quality-of-life scores in patients with head and neck cancers treated with IMRT.
In another publication, Chao et al. (24) also reported that the dosimetric advantage of IMRT did translate into significant reduction of late salivary toxicity in patients with oropharyngeal carcinoma and had no adverse impact on tumor control and disease-free survival. In this study, 430 patients with carcinoma of the oropharynx were treated. There were 260 patients with primary tumors of the tonsil and 170 patients with tumors arising from the base of the tongue. Twenty-four (6%) patients had stage I disease, 88 (20%) had stage II, 128 (30%) had stage III, and 190 (44%) had stage IV disease. Patients were divided into five groups. Group I consisted of 109 patients who received preoperative conventional radiation therapy. Group II consisted of 142 patients who received postoperative conventional radiation therapy. Group III consisted of 153 patients who received definitive conventional radiation therapy. Serial tomotherapy IMRT was used to treat 14 patients postoperatively (group IV) and 12 patients definitively without surgery (group V). With a median follow-up of 3.9 years, the 2-year locoregional control values for the five studied groups were 78%, 76%, 68%, 100%, and 88%, respectively. The 2-year disease-free survival values for the five studied groups were 68%, 74%, 58%, 92%, and 80%, respectively. IMRT significantly reduced the incidence of late xerostomia. The benefit of IMRT in nasopharyngeal carcinoma has been reported by Cheng et al. (28), who showed that target coverage of the primary tumor was maintained and nodal coverage was improved with IMRT in 17 nasopharyngeal carcinoma patients, as compared with conventional beam arrangements. Also, the ability of IMRT to spare parotid gland was considerably superior. Hunt et al. (56) reported similar results with 23 primary nasopharyngeal carcinoma patients. Lee et al. (72) reported data for 67 head and neck cancer patients treated with IMRT using three different techniques:
• Manually cut partial transmission blocks,
• Computer-controlled auto-sequencing multisegment approach, and
• Serial tomotherapy (MIMiC).
Fifty patients received concomitant cisplatinum and adjuvant cisplatinum and 5-fluorouracil chemotherapy according to the Intergroup 0099 trial. Twenty-six patients had fractionated high–dose-rate intracavitary brachytherapy boost and one patient had gamma-knife radiosurgery boost after external-beam radiotherapy. With a median follow-up of 31 months, the 4-year estimates of local progression-free, locoregional progression-free, and distant metastases-free rates were 97%, 98%, and 66%, respectively. IMRT provided excellent tumor target coverage and allowed the delivery of a high dose to the target leading to excellent locoregional control for nasopharyngeal carcinoma with significant sparing of the salivary glands and other nearby critical normal tissues.
IMRT also has been found to be suitable for patients with carcinoma of the paranasal sinuses. Claus et al. (32) observed no dry eye or other visual disturbances in 11 ethmoid sinus cancer patients, finding that the optic pathway dose can be reduced selectively by IMRT and that the IMRT has the potential to save binocular vision. In this study, however, no evaluation of retinopathy was done because the follow-up period was short. Similar significant advantages of IMRT were shown for six patients over conventional radiation therapy and 3DCRT for treatment of maxillary sinus carcinoma (2).
The Virginia Commonwealth University group has reported data on a dose-escalation trial of advanced head and neck squamous cell carcinomas (143). In their protocol, IMRT was designed and delivered using the SIB strategy. Primary dose levels of 68.1, 70.8, and 73.8 Gy, given in 30 fractions (biologically equivalent to 74, 79, and 85 Gy, respectively, if given in 2 Gy per fraction), were evaluated. Simultaneously, the subclinical disease and electively treated nodes were prescribed 60 and 54 Gy, respectively (biologically equivalent to 60 and 50 Gy, respectively, if given in 2 Gy fractions). For each prescribed dose level, the required cord and brainstem sparing and a variable degree of parotid gland sparing were achieved. Dose homogeneities for GTV were on average 6.2%, 8.3%, and 8.8% for the three dose
P.258
levels. An average of 25.8 Gy to the contralateral parotid gland and 40.8 Gy to the ipsilateral parotid gland were achieved.
In patients treated with IMRT, local recurrences have been mainly within the high-dose region. Chao et al. (26) reported the patterns of failure in 126 head and neck IMRT patients treated with serial tomotherapy. Seventeen locoregional failures were detected, and nine of those failures were inside the GTV. Two-year actuarial locoregional control rate was 85%, and 2-year ultimate locoregional control rate after salvage surgery was 89%. Dawson et al. (39) reported the patterns of failure analysis of 58 patients. Patients were treated with forward-planning IMRT and 3DCRT. The actuarial locoregional control rate was 79% with a median follow-up of 27 months. For 14 out 15 patients, the locoregional failures were in the region containing gross disease and in the adjacent soft tissue at risk.
IMRT of Prostate Cancer
Zelefsky et al. (149) reported the results of 171 patients with localized prostate cancer treated with IMRT using fixed-gantry beams and sliding-window techniques. They found that IMRT improved target coverage and significantly reduced the volumes of rectal and bladder walls exposed to high doses. Acute grade 1 to 2 and late grade 2 rectal toxicities were significantly lower in the IMRT group in comparison with the 3DCRT group despite the fact that the latter mostly used lower target doses. In another publication, Zelefsky et al. (150) reported the results of 772 patients with clinically localized prostate cancer treated with IMRT. A total of 698 patients (90%) were treated to 81 Gy, and 74 patients (10%) were treated to 86.4 Gy. The 3-year actuarial prostate-specific antigen relapse-free survival rates for favorable, intermediate, and unfavorable risk group patients were 92%, 86%, and 81%, respectively. Eleven patients (1.5%) developed late grade 2 rectal bleeding. Four patients (0.1%) experienced grade 3 rectal toxicity. No grade 4 rectal complications have been observed. The 3-year actuarial likelihood of late grade 2 rectal toxicity was 4%. Seventy-two patients (9%) experienced late grade 2 urinary toxicity, and five (0.5%) developed grade 3 urinary toxicity (urethral stricture). The 3-year actuarial likelihood of late grade 2 urinary toxicity was 15%. These data represented the largest compilation of patients treated with IMRT and demonstrated the feasibility of high-dose radiation delivery with IMRT for patients with localized prostate cancer. Acute and late rectal toxicities seemed to be significantly reduced compared with what has been observed with conventional 3DCRT. Short-term prostate-specific antigen control rates also seemed to be at least comparable to those achieved with 3DCRT at similar dose levels.
Pollack et al. (106) reported their preliminary results of acute toxicity on 100 patients treated on a randomized trial of 76 Gy in 38 fractions versus 70.2 Gy in 26 fractions. All patients were treated using IMRT. The hypofractionated regimen appeared to be well tolerated but did have slightly higher rates of acute GI toxicity. Kupelian et al. (67) reported the results of a single-arm hypofractionated IMRT protocol using 70 Gy in 28 fractions and found that biochemical relapse free survival was 97%, 88%, and 70%, respectively, for low, intermediate, and high-risk disease. Rates of grade 2 or 3 rectal toxicity at 5 years were 5%.
Other studies of prostate IMRT include one conducted at Royal Marsden Hospital. This study of 10 patients with prostate cancer showed that, when both prostate gland and pelvic lymph nodes are included in the target volume, a reduction in critical pelvic volume irradiated can be achieved with IMRT allowing a modest dose escalation with acceptable complication rates (105). Teh et al. (124) reported acceptable acute genitourinary toxicity with a mean dose of 69 Gy in 40 postprostatectomy patients. Using a rectal balloon for prostate immobilization, they further reported the treatment results of 100 patients with prostate cancer with definitive IMRT. They found that there was no grade III or IV acute gastrointestinal (GI) and genitourinary toxicity (125). Morr et al. (100) have evaluated the clinical feasibility of daily computer-assisted transabdominal ultrasonography for target position verification in the setting of IMRT for prostate cancer. Twenty-three patients with clinically localized prostate cancer were treated with serial tomotherapy and transabdominal ultrasonography. They found that this localization technique can be practically implemented as part of a daily setup routine and could be a useful tool in the clinical practice of IMRT for prostate cancers.
IMRT Experience with Other Cancer Sites
In addition to strong data supporting the use of IMRT in head and neck and prostate cancer, there are a number of preliminary studies reporting the feasibility and outcomes of IMRT of other cancers; many of these reports are theoretical dosimetric studies. Theoretically improved dosimetry alone probably does not serve as sufficient justification for the routine use of IMRT in these cases, and in the absence of robust clinical data regarding actual treatment outcomes, IMRT in these settings should be considered investigational.
IMRT of Breast Cancer
Commercial systems are now available that can autocontour the volume of breast tissue within conventionally designed tangential photon portals and then use an inverse planning algorithm to optimize dose homogeneity within these tangential portals (67). However, most commercial inverse planning systems could not handle the “skin flash” appropriately. Due to setup uncertainties and breathing motion, a portion of the breast (target) tissue may move outside the skin line as indicated by the treatment planning CT images of the patient. The traditional IMRT technique to overcome target motion uncertainty is to expand the PTV and optimize the dose coverage to the entire PTV. However, this strategy may not work because a portion of the PTV will be expanded into the air, which does not have the necessary mass to absorb the dose. Some treatment-planning systems ignore the regions outside the skin contour entirely. Therefore, it may be necessary to add “virtual” tissues in the PTV for the inverse planning system. Sometimes it may require users to manually open certain IMRT segments to take care of the skin flash effect.
Nevertheless, there has been interest in using IMRT for left-sided beast cancers in order to spare myocardium from the high-dose region of the radiotherapy fields. No robust data regarding clinical outcomes after IMRT for breast cancer exist; however, a variety of dosimetric studies (29,44,54,75,110,126,127) have suggested reductions in lung and myocardium doses when IMRT is compared to conventional radiotherapeutic techniques. Hurkmans et al. (57) used a normal tissue complication probability (NTCP) model to estimate the NTCP for cardiac and lung complications due to radiotherapy and found that IMRT did decrease the NTCP for late cardiac toxicity compared to more conventional radiotherapy techniques, but had a minimal effect on the NTCP for radiation pneumonitis.
Hong et al. (54) reported a dosimetric study of IMRT in 10 cases of intact breast cancer showing significant reduction of dose to the coronary arteries, ipsilateral lung, and surrounding soft tissues. It simultaneously improved dose homogeneity throughout the target volume. Li et al. (74) described a combined electron and IMRT technique for breast cancer treatment, which led to improvement over the conventional treatment technique using tangential fields with reduced dose to the ipsilateral
P.259
lung and the heart. Other studies (44,127) also confirmed that IMRT reduces the high-dose volume in tangential breast irradiation significantly and enables more complete cardiac sparing without compromising PTV coverage in some patients. Furthermore, IMRT creates a possibility to improve field matching in case of multiple field irradiations of the breast and lymph nodes (62,69,127). In addition, IMRT for tangential breast radiation therapy was found to be an effective and efficient method to achieve uniform dose throughout the breast. Preliminary findings reveal minimal or no acute skin reactions for patients with different breast sizes in 32 patients with early-stage breast cancer (62).
IMRT of Gynecologic Cancer
In regard to the targeting of pelvic lymphatics, with IMRT, Taylor et al. (123) mapped the pelvic lymphatics of 20 patients using MRI with the administration of iron oxide particles and found that a modified CTV margin of 7 mm around the iliac vessels resulted in adequate coverage of the pelvic lymphatics.
Ahamad et al. (3) analyzed the normal tissue-sparing effects of IMRT in the treatment of the pelvis after hysterectomy in patients with gynecologic cancers and found that although more small bowel, bladder, and rectum could be spared with IMRT compared to conventional radiotherapeutic techniques, these benefits rapidly diminished with even small expansions of the target volumes. D'Souza et al. (37) used the same dataset of patients as Ahamad et al. (3) and found that IMRT may allow higher doses of radiation (54 Gy) to be delivered safely to the node-bearing regions of the pelvis and the vaginal apex compared to conventional techniques that administer 50.4 Gy.
Salema et al. (111) reported 13 patients treated with extended field pelvic and paraaortic radiotherapy using IMRT and found that 2 patients experienced grade 3 or higher toxicity Both of these patients received concurrent cisplatin-based chemotherapy.
Portelance et al. (107) reported dosimetric comparison between 3DCRT and IMRT for 10 patients with cervical cancer. They demonstrated that, with similar target coverage, normal tissue-sparing was superior with IMRT. Mundt et al. (101) reported the clinical experience of 40 patients with gynecologic malignancy who underwent IMRT to the pelvis. Compared with 35 historic control patients who were treated with conventional techniques, patients treated with IMRT experienced fewer acute GI symptoms than those treated with conventional whole-pelvic radiotherapy.
IMRT of Gastrointestinal Cancer
Crane et al. (36) reported the results of a phase I dose-escalation study of radiotherapy with concurrent gemcitabine chemotherapy. The aim of this study was to alternate escalating the radiation dose by 3 Gy and the gemcitabine dose by 50 mg/m2. The starting dose of gemcitabine was 350 mg/m2 and 33 Gy per 11 fractions of IMRT to the regional lymphatics and primary disease. All three patients in the first cohort who were treated suffered dose-limiting toxicity, and the trial was ultimately closed because of excessive myelosuppression and upper GI toxicity.
Ben-Josef et al. (7) reported on 15 patients with pancreatic cancer treated with concurrent capecitabine and IMRT (45 to 55 Gy) and reported that only 1 patient had grade 3 GI toxicity, specifically GI ulceration that responded to medical management.
Brown et al. (17) performed a dosimetric analysis of 15 patients with pancreatic cancer and compared 3DCRT, IMRT with sequential boost, and IMRT with integrated boost, and found that IMRT with integrated boost allowed dose escalation up to 64.8 Gy to the primary tumor.
Guerrero-Urbano et al. (51) performed a dosimetric evaluation in five patients with locally advanced rectal cancer and found that IMRT with simultaneous integrated boost theoretically reduced the radiation dose to the small bowel compared to 3D conformal techniques. However, clinical data regarding outcomes of patients treated with IMRT for rectal cancer are lacking at this time.
Milano et al. (90) reported on 17 patients with squamous cell carcinomas of the anal canal treated with IMRT with whole pelvic radiation does of 45 Gy followed by boost to the anal canal. Thirteen patients received concurrent 5-fluorouracil and mitomycin-C chemotherapy. Treatment was well tolerated with no grade 3 or higher nonhematologic toxicity, and no required treatment breaks from skin or GI toxicity. However, one patient receiving mitomycin-C chemotherapy did experience grade 4 hematologic toxicity. Three patients who did not achieve a complete response required abdominoperineal resection and colostomy. With a mean follow-up of 20.3 months, there were no other local failures.
Milano et al. (89) also reported on seven patients with gastric cancer treated with IMRT to a dose of 50.4 Gy. No patient experienced grade 3 toxicity. The treated IMRT plans were compared to conventional AP:PA and three-field plans, and the IMRT plans were found to provide better coverage of the target volumes compared to conventional techniques, with better sparing of the liver and kidneys.
IMRT of Lung Cancer
Because of concerns regarding respiratory motion in radiotherapy of lung cancer, the use of IMRT in lung cancer requires some method to account for tumor and organ motion during treatment planning and delivery; these techniques include both respiratory gating (61) and 4D CT planning (4).
The poor local control rates of conventional radiotherapy doses in the treatment of lung cancer (19) have led to much interest in using IMRT to allow for dose escalation to improve local control. Holloway et al. (52) reported the initial results of five patients with unresectable stage II and III non–small cell carcinoma treated on a phase I dose-escalation trial using induction chemotherapy followed by IMRT to a dose of 84 Gy using 2.4 Gy daily fractions. PET CT was used to define target volumes. One patient developed lethal radiation pneumonitis and the trial was halted.
Murshed et al. (103) performed a dosimetric analysis of 41 patients initially treated with 3DCRT to a dose of 63 Gy. IMRT plans were then generated using these patients' initial planning CTs, and IMRT was found to decrease the volume of lung irradiated to both 10 and 20 Gy (V10 and V20). Target coverage was improved with IMRT, and the volumes of heart and esophagus irradiated were also reduced.
Grills et al. (50) performed a dosimetric comparison of four radiotherapy techniques in 18 patients with stage I-IIB lung cancer. The study compared IMRT, optimized multiple beam 3DCRT, two- to three-beam 3DCRT, and traditional wide-field radiotherapy with elective nodal irradiation. This study found that IMRT and optimized 3DCRT resulted in similar doses of radiotherapy to normal tissues in node-negative patients; however, in node-positive patients, IMRT resulted in a 15% decrease in the volume of lung treated to 20 Gy (V20) and a 30% decrease in the NTCP for radiation pneumonitis.
IMRT of Intracranial Malignancies
Iuchi et al. (58) from Japan published a retrospective report on 25 patients with malignant astrocytomas (World Health Organization grade III and IV) treated with IMRT using a hypofractionated regimen of 48 to 68 Gy in 8 fractions. Thirteen
P.260
patients were treated to 68 Gy and 12 patients received doses of 48 to 65 Gy. The IMRT group was compared to 60 patients treated with conventional techniques to doses of 40 to 60 Gy using 2 Gy daily fractions. The 2-year overall survival was significantly improved (p = .043) in patients treated with hypofractionated IMRT (55.6%) compared to those patient treated with conventional techniques (19.4%).
Floyd et al. (46) reported a pilot study of 20 patients treated with hypofractionated IMRT. Treatment was administered during 2 weeks (10 fractions), and 50 Gy (5 Gy/day) was prescribed to enhance disease or the surgical cavity and 30 Gy (3 Gy/day) was prescribed simultaneously to peritumoral edema. Three patients experienced grade 4 brain necrosis, requiring surgical re-excision. There was no mortality from brain necrosis; in fact, those patients manifesting brain necrosis had longer survival times.
Hwang et al. (55) reported on 15 patients with pediatric medulloblastoma treated with conventional craniospinal radiotherapy followed by a boost to the posterior fossa using IMRT. IMRT delivered much lower doses of radiation to the auditory apparatus while maintaining full doses to the desired target volume. Their findings suggested that, despite receiving higher doses of cisplatin and despite receiving radiotherapy before cisplatin therapy, IMRT can significantly decrease the rate of hearing loss in children treated for medulloblastoma.
In closing, IMRT clearly results in improved radiation dose distributions in a variety of cancers. In some cases, the superior dosimetry of IMRT has resulted in improved clinical outcomes for patients; however, the scientific evidence documenting these clinical improvements lags far behind the data documenting improved dosimetry. It is incumbent on radiation oncologists to continue to document improved clinical outcomes with IMRT in the peer-reviewed literature if we wish to justify the use of this expensive technology to our communities in an era of skyrocketing medical costs.
posted by stefanoxx at 5:35 AM
0 Comments:
Post a Comment
<< Home