IMRT represents a major paradigm shift in radiation oncology and its clinical application is still evolving. As more institutions implement IMRT, decisions need to be made regarding the choice of equipment, optimization method, image feedback, organ deformation registration, and immobilization. We have identified limitations of existing IMRT systems and are attempting to address these with helical tomotherapy using online megavoltage computed tomography (MVCT) imaging, verification of patient positioning prior to and during treatment, verification of delivered dose, treatment modification to account for patient/organ displacement, and deformable reconstruction to account for organ motion and change in shape. These processes, which form the core concept of adaptive radiotherapy, allow for ongoing verification and continuous correction of treatment variations. The clinical application of these techniques opens up new clinical vistas for Radiation Oncology.

Optimized Treatment Delivery:
A Description of Helical Tomotherapy

Helical tomotherapy represents a new form of radiation treatment delivery, which has been pioneered at the University of Wisconsin and is now in use at various centers around North America (2). While standard radiotherapy is currently delivered using a few static fields, helical tomotherapy delivers treatment with a rotating, intensity-modulated fan beam. The patient is continuously translated through a ring gantry resulting in a helical source trajectory about the patient. The beam delivery is similar to that of helical (?spiral?) computed tomography (CT) and requires slip rings to transmit power and data. The ring gantry provides a stable and accurate platform to perform tomographic verification of both the patient setup and delivered dose.

Various systems have been developed to implement IMRT. In this regard, helical tomotherapy is most similar to the NOMOS Peacock™ system. The Peacock™ system also uses a fan beam delivered via an arcing gantry equipped with a multileaf collimator. However, it delivers treatment by ?translate-then-rotate? method rather than a continuous helical delivery because it is an attachment to a standard C-arm linear accelerator. The design of the helical tomotherapy unit allows for continuous delivery over 360 degree beam angles (3). In addition, this design minimizes the treatment time, which may hypothetically provide a radiobiologic advantage compared to other IMRT approaches (3, 4). The helical delivery minimizes the risk of significant high or low dose deposition in areas of overlap or junctioning (6). Assessments of sequential units presently in use, reveal that positioning errors as little as 1 mm can cause dose errors on the order of 10-20% in the abutment regions (7). In addition to full integration of IMRT delivery, an important advance with helical tomotherapy over the other current systems is the ability to provide accurate verification of radiation delivery via onboard tomographic imaging.

Adaptive Radiotherapy

Perhaps the most significant difference of the helical tomotherapy unit is the presence of an integrated online megavoltage CT (MVCT) unit. This permits verification of patient positioning, target tumor/organ registration to assess internal motion (including geometric shift, and shape/volume changes), and reconstruction of delivered dose. These capabilities offer the radiation oncology team the ability to verify and adjust the therapeutic plan as needed during the course of treatment. This concept is referred to as adaptive radiotherapy (8). These capabilities can be viewed as a closed-circuit loop, as illustrated in Figure 1. The integration of the MVCT and the treatment unit allows for options not possible with contemporary systems. For example, if a patient set-up is found to differ from the planned position, the current approach requires moving the patient to compensate for this positioning error. With the integrated helical unit, another option is having the patient remain in the ?incorrect? position and modifying the treatment delivery. The success of the modification is independent of the extent and direction of the offsets, within certain limits (9). Our preliminary results with MVCT in both phantoms and patients confirm its utility in verification of patient position and tumor localization. These MVCT images can be obtained at radiation doses of around 2 cGy, comparable to that of diagnostic CT imaging (10, 11) and lower than reported doses from low-dose megavoltage cone beam CT (65). Other methods of onboard imaging have been developed recently and are available clinically. Our group has recently acquired cone-beam kVCT technology and we are presently comparing this to helical tomotherapy MVCT capabilities.


Figure 1: Conceptual Flow Diagram of Adaptive Tomotherapy. As described in the text, each clinical project within our research strategy is uniquely designed to test one or more particular component of adaptive radiotherapy.

Delivery Modification: Dose Reconstruction

Unlike surgical oncology, where frozen section pathology allows rapid feedback, radiation oncologists are typically unable to rapidly assess and adjust their plans in light of their actual treatment. This is changing with the development of dose reconstruction tools, which offer the capability to determine the actual three-dimensional dose deposited. At the time of treatment on the tomotherapy unit, the incident energy fluence is computed from the signal detected at the exit detectors. An accurate, anatomically detailed, 3D representation of the patient is also obtained. A transfer matrix then converts this signal to incident energy fluence. In other words, the matrix allows one to infer from the signal at the detector the energy fluence issuing from the MLC. The integrated CT present in the tomotherapy unit provides details of the primary and scatter characteristics for every projection. Path-length and detector-to-patient distance are computed from the MVCT image. Leakage and transmission plus tongue and groove penumbra are also included in the calculation. The tongue-and-groove effect (TG) of the MLC leads to a variation in total fluence-per-leaf-opening with the number of adjacent open leaves. The planning system accounts for this effect using TG correction factors measured for each leaf. The method corrects for the limitations of other MLCs, which do not consider the latter factors, resulting in fluence errors as high as 20% in extreme situations (12). Finally, the treatment dose distribution is computed using the convolution/superposition algorithm. Kapatoes et al. (13) have demonstrated that the reconstructed distribution has excellent accuracy; the tolerances of the majority of voxels within the low and high dose gradient are 3% and 3 mm, respectively. Effectively this results in the generation of an accurate daily pictorial dose record, which can be fused to the treatment planning CT and compared directly with the planned dose-distribution.

Delivery Modification: Dose Comparison

The ability to accurately reconstruct the 3D dose distribution is a valuable addition to the radiation oncologist?s armamentarium. However, an efficient method of comparing desired and actual dose distributions is required. An example illustrating this issue is described in Figure 2. Desired isodose lines for a nasopharyngeal cancer helical tomotherapy optimization are depicted with the parotid glands and spinal cord areas to be avoided. This idealized dose distribution is the goal to be achieved during the actual treatment. To compare desired dose distributions with actually delivered dose distributions, a metric is required that is accurate, intuitive, easily usable, and provides the needed quantitative and qualitative information. One metric with these characteristics has been developed and is called ξ.


Figure 2: Nasopharyngeal helical tomotherapy optimization used for the preliminary tests of delivery modification and adaptive tomotherapy. In this case, the primary tumor and the regional field were identified as targets. The parotid glands and spinal cord were considered avoidance regions and, thus, the dose to these structures was minimized. This dose distribution can be considered as a gold standard that should be achieved during actual treatment delivery. A metric that is accurate, intuitive, simple to use, and provide quantitative and qualitative information has been developed and is called ξ that is useful in comparing three-dimensional IMRT dose distributions to this gold standard.

This method for comparing isodose distributions is based on the methods of Van Dyk et al. (14) and Low et al. (15). The two modes of comparison are dose difference (ΔD) and distance-to-agreement (DTA) analyses. For regions in which both the planned and measured distributions have high dose gradients, DTA comparisons are conducted. For all other cases, ΔD analyses are performed. Once the mode of comparison is decided, the ξ index can be computed by dividing the ΔD and DTA values by their respective tolerances:


The ξ value provides a measurement of quality for every voxel indicating if they are within the desired tolerance or how far they deviate from that tolerance. Typical tolerance values in IMRT are 3% and 3 mm, for ΔD and DTA, respectively. The smaller the ξ value, the more accurately the compared isodose distributions are aligned. To facilitate the spatial identification of problem areas, color-wash images of ξ maps can be displayed (Figure 3). The utility of this method is illustrated with an example below.


Figure 3: Examples of (a) successful and (b) unsuccessful delivery modifications following treatment given with a displacement from the intended patient position. The corresponding ξ figures are shown in (c) and (d) for the successful and unsuccessful delivery, respectively. Red indicates that the dose distributions are between 0-3% or 0-3 mm (within tolerance), green between 3-6% or 3­-6 mm, blue between 6-9% or 6-9 mm, and light blue >9% or >9 mm. Very small errors can be readily identified even in the case of a good delivery-modified dose distribution (panel c). Panel d demonstrates many different levels of error present with the incorrect delivery modification. From this, it appears that the ξ distributions will be very useful to detect errors and to analyze the optimal adaptive radiotherapy approach to be pursued.

Delivery Modification: Examples of Adaptive Radiotherapy

Suppose that a particular patient has an offset of 1 cm in the x direction. In order to compare the optimized and delivery-modified dose distributions, the ξ metric is used. Figure 3 (panels ?a? and ?b?) shows two sets of isodose lines for optimized and delivery-modified dose distributions, respectively, superimposed on the same CT image. One panel shows a successful (panel ?a?) and the other an unsuccessful (panel ?b?) treatment adaptation. It is quite difficult from panel ?b? to analyze precisely the impact of the delivery modification error from the isodose lines. It may be necessary to analyze many sets of isodose lines to accurately identify errors. Panels ?c? and ?d? represent the corresponding ξ distribution images for panels ?a? and ?b?, respectively. In this example, red indicates that the dose distributions are between 0-3% or 0-3 mm (within tolerance), green between 3-6% or 3­-6 mm, blue between 6-9% or 6-9 mm, and light blue >9% or >9 mm. Very small errors can readily be identified even in the case of a good delivery-modified dose distribution (panel ?c?).

This comparative information can be obtained during a single fraction or after several fractions. Figure 4 illustrates the ξ image comparing the planned dose and the reconstructed dose for the first week of treatment in which a systematic error was made. Several error regions appear on the targets and regions at risk, mainly in the high gradient regions. Panel 4b shows a gray scale image of the re-optimized dose distribution that could be delivered during the second week of treatment in order to correct for prior misalignment. In this dose image, a pattern appears that is very similar to the ξ image. This comparison is quite intuitive and, therefore, useful for checking by visual inspection.


Figure 4: Example of optimization to correct for one week of incorrectly delivered dose. Panel a is a ξ image of one week?s delivery with the patient shifted by 0.5 cm in the x and 0.5 cm in the y direction. Panel b is the gray scale image of the dose to be delivered in the second week designed to compensate for the previous week?s error.

Figure 5 illustrates the result of treatment with one week of treatment with a systematic error followed by a second week of either the original treatment plan given accurately (Figure 5a) or a re-optimized treatment plan designed to compensate for the errors of the first week (Figure 5b). Simply repositioning the patient and accurately delivering a week of the originally designed radiation plan (Figure 5a) will dilute the error incurred during the first week but cannot fully compensate for the error. Radiation delivery modifications are designed to compensate for the difference between actual and desired dose distributions. In the example presented, dose reconstruction and comparison reveal that excess dose is being deposited in the spinal cord and right parotid. In Figure 5b a modified plan specifically designed to compensate for the errors of week one is instituted in the second week and the dose reconstruction is performed again. Figure 5b shows the ξ image comparing the reconstructed dose delivery after two weeks of the incorrect and corrected treatments. Most of the errors within the tumor region are corrected. However, a small trade-off with dose to the parotid glands and spinal cord is necessary in order to rectify the errors induced during the first week of delivery. The specific thresholds for the trade-offs that should be accepted remain a matter of ongoing physics and clinical research.


Figure 5: Panel a compares the ξ image after the second week if no action is taken to correct for errors during the first week. Panel b shows the ξ image if action is taken to correct for the error.

With helical tomotherapy, MVCT images can be registered using a full mutual information algorithm, bone extracted feature fusion (EFF), and bone and tissue EFF algorithms, with uniform down-sampling of the MVCT images along the x and z axes (to provide a time-saving by a factor of up to 4) with and without the rotational registration components. These particular algorithms take into account any changes in patient anatomy between the reference image and fusion image when the image registration is carried out.

Conformal Avoidance IMRT

Current Paradigm & Pitfalls of Wide-Field Irradiation

A critical element in improving the therapeutic ratio in radiotherapy is the minimization of normal tissue irradiation. In many clinical situations, the difficulty in precisely identifying specific regions at risk for tumor invasion has led to the default paradigm of ?wide field irradiation? to include potential cancer-bearing sites with confidence. Clinical examples of this abound, and the treatment of head and neck cancer provides an instructive example.

Current Head and Neck IMRT Paradigm for Xerostomia

One of the more popular clinical applications of IMRT in the context of normal tissue sparing involves avoidance of major salivary glands during head and neck radiotherapy. Xerostomia, with its consequential taste impairment, difficulty chewing, speaking and swallowing, and increased dental caries and oral candidias, is a common toxicity experienced by head and neck cancer patients undergoing radiotherapy. These chronic complications represent a major source of quality of life deterioration for survivors, with nasopharyngeal cancer patients often experiencing notable effects due to the large volume radiation fields commonly employed (16).

Temporary symptomatic relief offered by moistening agents and saliva substitutes represents the only viable option for patients without residual salivary function. In patients with some residual salivary function, selected studies demonstrate that oral administration of pilocarpine can increase salivary flow and ameliorate symptoms (17) while other studies have contradicted these findings (18). The effectiveness of muscarinic cholinergic receptor agonists requires residual salivary function, which emphasizes the importance of sparing normal salivary tissue during irradiation (19). Based on a Phase III randomized trial, amifostine was approved for xerostomia prevention in the postoperative head and neck cancer setting. The incidence of clinically significant xerostomia was 57% in the control group and 34% in the study group (p = 0.002). Although statistically significant, this implies an overall absolute gain of only 23% (20).

Scintigraphic evaluation of post-radiation parotid function suggests that when bilateral whole parotid glands are irradiated, partial recovery is possible if the dose to the parotid is less than 52 Gy; however, recovery rarely occurs at doses exceeding 55 Gy. The 50% complication probability dose is less than 33 Gy for subacute xerostomia (<6 months) and 52.5 Gy for chronic xerostomia (>12 months) (21).

Eisbruch et al. (22) reported results using salivary gland sparing techniques. Forty-eight patients with unilateral treatments served as a comparison group. Treated parotid glands received an average dose of 55.2 Gy, while spared glands received 21.9 Gy. Unstimulated and stimulated parotid flow rates decreased dramatically in treated glands after the initiation of radiotherapy, remained low without detectable improvement, and were significantly lower at two years after radiotherapy compared with baseline. Conversely, parotid flow rates in spared glands underwent mild changes during radiotherapy and returned to baseline values at two years. Patients who had undergone unilateral treatment demonstrated a compensatory increase in salivary production by the spared glands. These results suggest that using conformal techniques to spare the parotids can enable meaningful improvement in xerostomia over time. Butler et al. (23) reported an attempt to minimize xerostomia using IMRT boost techniques. The mean dose delivered to the ipsilateral parotid was 23 Gy, and the contralateral parotid received 21 Gy; these mean doses were a function of a planned delivery of 50 Gy to the neck nodal regions. Often 60-70 Gy is required in these situations, and even with IMRT, the parotid dose may exceed the 50% tolerance dose of 33 Gy for subacute xerostomia. Although these studies are promising, they suggest that even with conventional IMRT, there is considerable room for improvement in parotid gland protection.

Conformal Avoidance: A New Paradigm Applied to Xerostomia

The current IMRT paradigm requires the clear delineation of target regions and organs at risk (24, 25). This paradigm requires meticulous and labor-intensive contouring of the target volume and at-risk lymph node regions. In addition, there remains a risk for geographic miss and consequential nodal failure. Due to the inability to precisely define and map tumor regions along with the need to prophylactically address nodal regions, large volumes of normal tissues are commonly still irradiated despite IMRT techniques (26). The concerns regarding geographic miss may result in mapping a larger region than necessary; thereby, eroding the advantages of IMRT dose conformality. The counterpart to this paradigm of target mapping is conformal avoidance (27). This strategy may prove easier to execute routinely than conformal targeting. It is anticipated that the routine definition and mapping of normal tissue structures in the head and neck (i.e., parotid glands, eyes, spinal cord, mandible) can be accomplished with greater ease, precision, and reproducibility across users than the process of delineating primary tumors with infiltrative soft tissue extension and associated regions of nodal spread as currently required for IMRT techniques. Conformal avoidance may offer the oncology team greater confidence that cancer-bearing regions will receive full dose radiation while defined normal tissues are specifically avoided (Figure 6).

Ecommerce Solutions
Online Payment Solutions