62_04
Beam Selection and Planning
3DCRT Conformal Plans.
The hallmark of 3D planning is the use of a multifield beam arrangement with field apertures designed using Bbeam's-eye-view projections that are conformal to the shape of the PTV, thereby shielding the normal tissues. At MSKCC, the standard 3D conformal beam arrangement consisted of six coplanar fields, including two lateral, two anterior and two oblique beams (237). Conformal apertures were drawn around the PTV adding a margin of approximately 5 to 6 mm in the axial directions to account for beam penumbra. This margin was sufficient dosimetrically in the axial plane because of the effect of the overlapping beams, whereas in the superior and inferior directions, a margin of 1 cm was typically necessary. For the beam shaping, multileaf collimation was used, which has effectively eliminated the handling of lead-cadmium alloy blocks. Several other beam arrangements have been proposed for 3DCRT, with the most common being the conformal four-field “box” (235,248,316,401).
Once the treatment fields and aperture shapes were defined, the dose distribution was calculated for a few representative planes, typically transverse, coronal, and sagittal planes through the isocenter. Dose-volume histograms were generated for the PTV, femoral heads, and rectal and bladder walls. If the bowel was located near the prostate and seminal vesicles, a dose calculation for the bowel was also done. For the MSKCC six-field plan, the two lateral beams typically delivered approximately half of the dose to the isocenter with the four oblique beams contributing the rest. The beam weights of the anterior oblique and posterior oblique beams were adjusted to obtain a uniform dose within the PTV and to place the hot spots away from the rectum. The plan was normalized so that the prescription isodose (100%) covered the PTV with a hot spot of 6% to 9% within the PTV. Although the portion of the rectal wall enclosed within the PTV was expected to receive the prescription dose, or slightly higher, the rectal wall volume receiving 75.6 Gy or more did not exceed 30%. Other normal tissue dose limits for these 3DCRT plans included limiting the maximum dose to the femurs to ≤68 Gy (90%), maximum dose to large bowel to ≤60 Gy (79%), and the maximum dose to small bowel to ≤50 Gy (66%).
Intensity-Modulated Radiation Therapy.
Unlike 3DCRT treatment planning in which the planner defines the shape as well as the amount of radiation to be delivered from each treatment field, planners define dose “constraints” or “objectives” for the target and normal tissues, which describe the desired dose distribution in IMRT planning. These constraints typically consist of maximum or minimum dose limits on targets and dose and
P.1460
dose-volume limits on normal tissues. The planner specifies as many individual constraints for a specific target or normal tissue as desired, giving each its own weight or “penalty,” reflective of its clinical importance. Using special computer software, these constraints drive a mathematical optimization of the radiation intensities of many small “beamlets” within each treatment field. The result of this optimization is a set of intensity patterns for the treatment fields and a dose distribution with characteristics as close as possible to the constraints entered by the planner. The MSKCC IMRT optimization algorithm was developed by Spirou and Chui (378) and relies on a conjugate gradient minimization to find the optimal intensity patterns and dose distribution. The IMRT intensity profiles are then converted to leaf motion using the algorithm of Spirou and Chui (379) for use during the dynamic MLC delivery. A typical intensity pattern for a posterior IMRT field for prostate treatment is shown in Figure 62.11.
Because normal tissue shielding can be accomplished by modulating the beam intensity, IMRT beam directions are often somewhat nonintuitive and can differ significantly from those typically chosen for 3DCRT. Determination of appropriate target and normal tissue constraints can be tedious, and planners often repeat the optimization process multiple times, evaluating the dose distributions after each iteration and making small adjustments to the constraints before obtaining a final, acceptable plan. Most institutions performing IMRT planning set up templates that specify both the clinical goals of the dose distribution and initial target and normal tissue constraints for optimization. The MSKCC template for prostate IMRT planning to 81 Gy is shown in Table 62.8. Dose and dose-volume constraints for the PTV, PTV overlap with the rectum, and rectal and bladder walls are listed. It should be noted that constraint templates vary significantly between treatment planning systems; therefore, these constraints should be used only after a thorough evaluation on the user's system.
A variety of beam arrangements have been proposed for prostate IMRT treatment including multifield axial or noncoplanar arrangements in addition to intensity modulated arc therapy (51,326,437). At MSKCC, a standard arrangement of five 15-MV beams directed from the posterior, posterior oblique, and anterior oblique directions is used (Fig. 62.12). Primarily because of concern about increased risk of secondary cancers from higher neutron doses associated with IMRT treatment at high energies (165,166,215,216), some groups have proposed 6-MV IMRT techniques for the treatment of prostate cancer. As shown by Pirzkall et al. (318), however, a larger number of treatment fields may be necessary to achieve a dose distribution similar to that observed with 15-MV x-rays.
P.1461
The MSKCC clinical goals used to evaluate the IMRT dose distributions and dose-volume histograms for prostate patients are outlined in Table 62.8. These dosimetric guidelines defining acceptable target coverage, dose uniformity, and normal tissue doses have grown out of our 3DCRT and IMRT planning experience during the past 20 years and include several refinements resulting from retrospective outcome and toxicity analyses from our institution. Most notably, studies by Skwarchuk et al. (369) and Jackson et al. (196) retrospectively evaluating the rectal wall dose-volume histograms for patients treated to 70.2 and 75.6 Gy using 3DCRT techniques found that, on average, patients with late rectal bleeding had significantly higher rectal dose-volume histograms than patients who did not bleed. Both high- and intermediate-dose levels were found to be independently correlated with rectal bleeding. As a result of these studies, two rectal wall dose-volume historgram limits were implemented and are routinely enforced at MSKCC when treating prostate cancer: no more than 30% of the rectal wall may receive more than 75.6 Gy and no more than 53% of the rectal wall can receive more than 47 Gy.
Typical dose distributions and dose-volume histograms for an 81 Gy IMRT plan are shown in Figures 62.13 and 62.14, respectively. The physician should carefully review the treatment plan and dose-volume histograms of the target and normal tissue structures to select the optimal treatment plan for the patient. Target coverage should be carefully assessed, as well the dose inhomogeneity and location of hot spots. In addition, careful attention should be given to determining if the treatment plan adequately meets acceptable dose constraints for the rectum, bladder, and bowel.
Standard Prescription Doses for 3DCRT and IMRT
Based on the results of a randomized trial from the M.D. Anderson Hospital (322), a radiation dose of 78 Gy (prescribed to the isocenter) would be appropriate for patients with PSA >10 ng/mL. Other institutions have demonstrated improved outcomes with higher radiation doses for all prognostic risk groups, including favorable-risk, early-stage disease. At MSKCC, 81 to 86 Gy are delivered to favorable-risk and intermediate/unfavorable-risk patients using IMRT. Daily fractions of 1.8 to 2 Gy, five fractions per week, are routinely used; however, others have reported encouraging results with a hypofractionated scheme delivering 70 Gy with fractions of 2.5 Gy (222,320). At MSKCC, the dose is prescribed to an isodose line that encompasses as much of the PTV as possible while still respecting the target and normal tissue goals listed in Table 62.8. Typically, at least 90% to 95% of the PTV receives the prescription dose.
The usual dose delivered to the pelvic lymph nodes (when these lymph nodes are to be irradiated) is 45 Gy, with a subsequent boost to the prostate through reduced fields. 3DCRT and IMRT techniques have been used at MSKCC and elsewhere when pelvic lymph node irradiation is necessary, and observations indicate a significant reduction in bowel dose and improvement in the overall tolerance of therapy (21,66,300).
Treatment Delivery and Organ Motion Concerns
Movement of the prostate during treatment or between treatment fractions has long been a concern for prostate radiotherapy. A large number of studies investigating inter- and intrafractional motion of the prostate and seminal vesicles have been reported (4,15,23,28,34,77,93,105,186,210,227,263,294,345,350,355,357,410,411,429,431,448). Most groups have measured prostate motion relative to bony landmarks through repeated imaging of implanted radio-opaque markers or serial CT studies. Although the reported magnitude of motion has varied, relatively little motion in the lateral direction and potentially significant movement in the anterior-posterior and superior-inferior directions has been consistently reported. Many studies have also observed a correlation between prostate and seminal vesicle motion and rectal or bladder filling.
Interfractional prostate motion was studied in approximately 50 patients by Crook et al. (93). Gold seeds implanted in the prostate were visualized on kilovolt radiographs taken at the simulation and approximately midway through treatment. Minimal prostate motion was observed in the lateral directions (0.1 to 0.5 cm) but inferior displacements of 0.5 and 1.0 cm or more were observed in 43% and 11% of their patients. Average displacement in the posterior direction was 0.72, 0.62, and 0.46 cm for seeds placed at the seminal vesicles, posterior aspect of the prostate, or apex of the gland, respectively; 60% of patients showed more than 0.5 cm posterior displacement of the prostate base, and 30% showed more than 1 cm.
Roeske et al. (345) evaluated 42 CT scans taken during the course of treatment on 10 patients. The standard deviations of the center of mass motion of the prostate were 0.7, 3.9, and 3.2 mm in the left-right, anterior-posterior, and superior-inferior directions, respectively. Corresponding data for center of mass motion of the seminal vesicles was 3.2, 7.3, and 4.1 mm. Motion of the prostate and seminal vesicles in the anteroposterior direction was strongly related to changes in rectal volume but only weakly related to bladder volume changes. For the prostate, a margin twice the patient group standard deviation (1.4 mm left-right, 7.8 mm anterior-posterior, and 6.4 mm superior-inferior)
P.1462
would have encompassed more than 95% of the center-of-mass motion.
Zelefsky and colleagues (448) obtained four serial CT studies for 50 patients (planning scan and three additional scans during the course of therapy). Prostate displacements in the anteroposterior and superoinferior directions were most frequently observed. The mean prostate motion in the anteroposterior, superoinferior, and left–right directions was 1.2, 0.5, and 0.6 mm, respectively. Anteroposterior movements were correlated with changes in rectal volume. Patients with large rectal volumes (>60 cm3) and large bladder volumes (>40 cm3) on the planning scan experienced a higher likelihood of having >3 mm systematic displacement of the prostate and seminal vesicles, leading the authors to conclude that these patients may require more generous PTV margins to ensure adequate CTV coverage. However, among patients without larger bladder and rectal volumes, a 1-cm margin around the CTV with a 6-mm margin at the prostate–rectal interface enclosed the posterior, anterior, superoinferior and left-right aspects of the CTV within the prescription dose level with a probability of 90%, 100%, 99% and 100%, respectively, indicating that the MSKCC margins provided adequate CTV coverage for most patients.
Intrafractional prostate motion was studied in 20 patients by Huang et al. (186) using pre- and posttreatment B-mode rectal ultrasound evaluations. Although the intrafractional motion was relatively insignificant in all directions, the predominant directions of motion were in the anterior and superior directions. Standard deviations of 0.4, 1.3, and 1.0 mm were observed in the lateral, anterior, and superior directions, respectively. A combination of real-time tumor tracking technology, fluoroscopy, and implanted gold markers were used by Kitamura et al. (210) to quantify intrafractional motion in 10 patients. Data were obtained with the patients in both supine and prone positions. The amplitude of the observed 3D motion was as much as 2.7 mm in the supine position and 24 mm in the prone position, indicating a great impact from respiration and bowel movement in the prone position.
Several methods have been developed to reduce uncertainty due to interfractional organ motion and thereby improve treatment delivery using computer-assisted transabdominal ultrasonography, radio-opaque marker tracking, or CT image-guidance. With the ultrasound system (BAT, Nomos Corporation, Sewickley, PA), patients are instructed to maintain a full bladder and are initially set up based on their tattoos in the supine position. The system is attached to the accelerator collimator and imports the coordinates of the isocenter as well as the target contours from the planning CT. Computer software facilitates 3D matching of the target contours with the prostate visualized on ultrasonography and determination of the necessary patient position modification. This system may not be reliable for patients who cannot maintain a full bladder and those with a large body habitus or other anatomic constraints because of anticipated poor image quality. It has also been noted that BAT measurements can be associated with a 2- to 3-mm error, which is sometimes in the range of the required shifts. Nevertheless, this system has been used by several investigators to verify the prostate position and results are consistent with improved accuracy of the daily treatment delivery (70,123,129,130,186,217,225,228,229,240,282,363). One such study, by Morr et al. (282), evaluated the BAT system on 23 patients undergoing IMRT treatment to the prostate. Once users of the system were sufficiently experienced, pretreatment ultrasound procedures could be performed in an average of about 5 minutes. Positional corrections averaged 2.6 ± 2.1 mm, 4.7 ± 2.7 mm, and 4.2 ± 2.8 mm in the lateral, anterior-posterior, and superior-inferior directions, respectively. The authors concluded that daily system use was feasible and resulted in clinically significant adjustments that would not have been possible using other conventional verification methods.
Recent technological advances have opened up the possibility of acquiring pre- or posttreatment megavoltage or kilovoltage CT images directly on the linear accelerator with the patient in the treatment position. One example of such a device, known as a tomotherapy unit (TomoTherapy Hi-ART, Madison, WI), consists of a 6-MV accelerator mounted within a CT-type gantry. Megavoltage CT images can be obtained prior to treatment, registered with the planning CT study, and the resulting positional corrections can then be applied prior to treatment. IMRT treatment is delivered through synchronized circular motion of the accelerator, couch translational motion, and multileaf collimation. Langen et al. (226) compared three methods of registering megavoltage CT images from a tomotherapy unit with the kilovoltage planning CT images and found that manual registration performed using implanted fiducial markers exhibited the least interobserver variability and agreed best with automatic registration computed from the center of mass of the three implanted fiducial markers.
Linac-based kilovoltage image guidance systems have recently become commercially available. These systems are comprised of a kilovoltage x-ray tube mounted 90 degrees from the accelerator head and a kilovotage imaging plate mounted 90 degrees from the standard megavoltage imaging device. They possess capabilities for kilovoltage two-dimensional projection imaging (radiographs), fluoroscopy, and 3D cone-beam CT, and are thus ideally suited for monitoring of inter- and intrafractional motion. Although little clinical experience using these kilovoltage gantry-mounted systems has been obtained to date, several groups have extensive experience monitoring and correcting for changes due to patient setup and organ motion with either two-dimensional electronic megavoltage portal imaging (EPID) or conventional CT systems (185,238,250,373,374,432,434,435). As a result, several methods for fast prostate localization on CT images, appropriate for either off-line or on-line, image-guided radiotherapy have already been described. Smitsmans et al. (373,374) have developed an automated 3D gray scale registration method that they have applied to both conventional and 3D cone-beam CT images. Collimating the field of view during the cone-beam CT acquisition significantly improves the cone-beam CT image quality and hence the registration success rate. Cone-beam CT artifacts caused by gas pockets moving during the CT acquisition are the main cause of unsuccessful registration.
P.1463
Hua et al. (185) have developed a semiautomatic method for localizing the prostate on pretreatment CT images based on manual identification of the posterior, anterior, left, and right extents of the prostate on the CT slices. The prostate displacement relative to the planning scan is then estimated through a simultaneous fitting of these “extents” to a finely spaced contour template from the planning scan. Identification of the prostatic extents on five pretreatment CT slices was found to be sufficient for reliable determination of the prostatic displacement. The approach of Yan et al. (435) has been to acquire an initial sequence of daily CT scans (typically 5 to 10 serial scans) from which the organ motion and patient setup inaccuracy can be reliably estimated. Based on these data, a confidence-limited PTV can be constructed that ensures, to within a defined statistical limit, that the CTV will receive a dose within a predefined tolerance. For example, the authors determined that a confidence-limited PTV constructed from daily CT scans obtained during the first week of 3DCRT treatment was sufficient to achieve a maximum dose reduction of ≤2% in the CTV for at least 80% of the patients or a 4.5% reduction for 95% of the patients. IMRT treatment required 2 weeks of CT data to achieve the same level of dosimetric coverage. Referred to as adaptive radiotherapy, this off-line correction strategy is capable of excluding systematic error and compensating for random uncertainties. It requires serial scans, as previously outlined, and a single plan modification after the first or second week of therapy.
3DCRT Conformal Plans.
The hallmark of 3D planning is the use of a multifield beam arrangement with field apertures designed using Bbeam's-eye-view projections that are conformal to the shape of the PTV, thereby shielding the normal tissues. At MSKCC, the standard 3D conformal beam arrangement consisted of six coplanar fields, including two lateral, two anterior and two oblique beams (237). Conformal apertures were drawn around the PTV adding a margin of approximately 5 to 6 mm in the axial directions to account for beam penumbra. This margin was sufficient dosimetrically in the axial plane because of the effect of the overlapping beams, whereas in the superior and inferior directions, a margin of 1 cm was typically necessary. For the beam shaping, multileaf collimation was used, which has effectively eliminated the handling of lead-cadmium alloy blocks. Several other beam arrangements have been proposed for 3DCRT, with the most common being the conformal four-field “box” (235,248,316,401).
Once the treatment fields and aperture shapes were defined, the dose distribution was calculated for a few representative planes, typically transverse, coronal, and sagittal planes through the isocenter. Dose-volume histograms were generated for the PTV, femoral heads, and rectal and bladder walls. If the bowel was located near the prostate and seminal vesicles, a dose calculation for the bowel was also done. For the MSKCC six-field plan, the two lateral beams typically delivered approximately half of the dose to the isocenter with the four oblique beams contributing the rest. The beam weights of the anterior oblique and posterior oblique beams were adjusted to obtain a uniform dose within the PTV and to place the hot spots away from the rectum. The plan was normalized so that the prescription isodose (100%) covered the PTV with a hot spot of 6% to 9% within the PTV. Although the portion of the rectal wall enclosed within the PTV was expected to receive the prescription dose, or slightly higher, the rectal wall volume receiving 75.6 Gy or more did not exceed 30%. Other normal tissue dose limits for these 3DCRT plans included limiting the maximum dose to the femurs to ≤68 Gy (90%), maximum dose to large bowel to ≤60 Gy (79%), and the maximum dose to small bowel to ≤50 Gy (66%).
Intensity-Modulated Radiation Therapy.
Unlike 3DCRT treatment planning in which the planner defines the shape as well as the amount of radiation to be delivered from each treatment field, planners define dose “constraints” or “objectives” for the target and normal tissues, which describe the desired dose distribution in IMRT planning. These constraints typically consist of maximum or minimum dose limits on targets and dose and
P.1460
dose-volume limits on normal tissues. The planner specifies as many individual constraints for a specific target or normal tissue as desired, giving each its own weight or “penalty,” reflective of its clinical importance. Using special computer software, these constraints drive a mathematical optimization of the radiation intensities of many small “beamlets” within each treatment field. The result of this optimization is a set of intensity patterns for the treatment fields and a dose distribution with characteristics as close as possible to the constraints entered by the planner. The MSKCC IMRT optimization algorithm was developed by Spirou and Chui (378) and relies on a conjugate gradient minimization to find the optimal intensity patterns and dose distribution. The IMRT intensity profiles are then converted to leaf motion using the algorithm of Spirou and Chui (379) for use during the dynamic MLC delivery. A typical intensity pattern for a posterior IMRT field for prostate treatment is shown in Figure 62.11.
Because normal tissue shielding can be accomplished by modulating the beam intensity, IMRT beam directions are often somewhat nonintuitive and can differ significantly from those typically chosen for 3DCRT. Determination of appropriate target and normal tissue constraints can be tedious, and planners often repeat the optimization process multiple times, evaluating the dose distributions after each iteration and making small adjustments to the constraints before obtaining a final, acceptable plan. Most institutions performing IMRT planning set up templates that specify both the clinical goals of the dose distribution and initial target and normal tissue constraints for optimization. The MSKCC template for prostate IMRT planning to 81 Gy is shown in Table 62.8. Dose and dose-volume constraints for the PTV, PTV overlap with the rectum, and rectal and bladder walls are listed. It should be noted that constraint templates vary significantly between treatment planning systems; therefore, these constraints should be used only after a thorough evaluation on the user's system.
A variety of beam arrangements have been proposed for prostate IMRT treatment including multifield axial or noncoplanar arrangements in addition to intensity modulated arc therapy (51,326,437). At MSKCC, a standard arrangement of five 15-MV beams directed from the posterior, posterior oblique, and anterior oblique directions is used (Fig. 62.12). Primarily because of concern about increased risk of secondary cancers from higher neutron doses associated with IMRT treatment at high energies (165,166,215,216), some groups have proposed 6-MV IMRT techniques for the treatment of prostate cancer. As shown by Pirzkall et al. (318), however, a larger number of treatment fields may be necessary to achieve a dose distribution similar to that observed with 15-MV x-rays.
P.1461
The MSKCC clinical goals used to evaluate the IMRT dose distributions and dose-volume histograms for prostate patients are outlined in Table 62.8. These dosimetric guidelines defining acceptable target coverage, dose uniformity, and normal tissue doses have grown out of our 3DCRT and IMRT planning experience during the past 20 years and include several refinements resulting from retrospective outcome and toxicity analyses from our institution. Most notably, studies by Skwarchuk et al. (369) and Jackson et al. (196) retrospectively evaluating the rectal wall dose-volume histograms for patients treated to 70.2 and 75.6 Gy using 3DCRT techniques found that, on average, patients with late rectal bleeding had significantly higher rectal dose-volume histograms than patients who did not bleed. Both high- and intermediate-dose levels were found to be independently correlated with rectal bleeding. As a result of these studies, two rectal wall dose-volume historgram limits were implemented and are routinely enforced at MSKCC when treating prostate cancer: no more than 30% of the rectal wall may receive more than 75.6 Gy and no more than 53% of the rectal wall can receive more than 47 Gy.
Typical dose distributions and dose-volume histograms for an 81 Gy IMRT plan are shown in Figures 62.13 and 62.14, respectively. The physician should carefully review the treatment plan and dose-volume histograms of the target and normal tissue structures to select the optimal treatment plan for the patient. Target coverage should be carefully assessed, as well the dose inhomogeneity and location of hot spots. In addition, careful attention should be given to determining if the treatment plan adequately meets acceptable dose constraints for the rectum, bladder, and bowel.
Standard Prescription Doses for 3DCRT and IMRT
Based on the results of a randomized trial from the M.D. Anderson Hospital (322), a radiation dose of 78 Gy (prescribed to the isocenter) would be appropriate for patients with PSA >10 ng/mL. Other institutions have demonstrated improved outcomes with higher radiation doses for all prognostic risk groups, including favorable-risk, early-stage disease. At MSKCC, 81 to 86 Gy are delivered to favorable-risk and intermediate/unfavorable-risk patients using IMRT. Daily fractions of 1.8 to 2 Gy, five fractions per week, are routinely used; however, others have reported encouraging results with a hypofractionated scheme delivering 70 Gy with fractions of 2.5 Gy (222,320). At MSKCC, the dose is prescribed to an isodose line that encompasses as much of the PTV as possible while still respecting the target and normal tissue goals listed in Table 62.8. Typically, at least 90% to 95% of the PTV receives the prescription dose.
The usual dose delivered to the pelvic lymph nodes (when these lymph nodes are to be irradiated) is 45 Gy, with a subsequent boost to the prostate through reduced fields. 3DCRT and IMRT techniques have been used at MSKCC and elsewhere when pelvic lymph node irradiation is necessary, and observations indicate a significant reduction in bowel dose and improvement in the overall tolerance of therapy (21,66,300).
Treatment Delivery and Organ Motion Concerns
Movement of the prostate during treatment or between treatment fractions has long been a concern for prostate radiotherapy. A large number of studies investigating inter- and intrafractional motion of the prostate and seminal vesicles have been reported (4,15,23,28,34,77,93,105,186,210,227,263,294,345,350,355,357,410,411,429,431,448). Most groups have measured prostate motion relative to bony landmarks through repeated imaging of implanted radio-opaque markers or serial CT studies. Although the reported magnitude of motion has varied, relatively little motion in the lateral direction and potentially significant movement in the anterior-posterior and superior-inferior directions has been consistently reported. Many studies have also observed a correlation between prostate and seminal vesicle motion and rectal or bladder filling.
Interfractional prostate motion was studied in approximately 50 patients by Crook et al. (93). Gold seeds implanted in the prostate were visualized on kilovolt radiographs taken at the simulation and approximately midway through treatment. Minimal prostate motion was observed in the lateral directions (0.1 to 0.5 cm) but inferior displacements of 0.5 and 1.0 cm or more were observed in 43% and 11% of their patients. Average displacement in the posterior direction was 0.72, 0.62, and 0.46 cm for seeds placed at the seminal vesicles, posterior aspect of the prostate, or apex of the gland, respectively; 60% of patients showed more than 0.5 cm posterior displacement of the prostate base, and 30% showed more than 1 cm.
Roeske et al. (345) evaluated 42 CT scans taken during the course of treatment on 10 patients. The standard deviations of the center of mass motion of the prostate were 0.7, 3.9, and 3.2 mm in the left-right, anterior-posterior, and superior-inferior directions, respectively. Corresponding data for center of mass motion of the seminal vesicles was 3.2, 7.3, and 4.1 mm. Motion of the prostate and seminal vesicles in the anteroposterior direction was strongly related to changes in rectal volume but only weakly related to bladder volume changes. For the prostate, a margin twice the patient group standard deviation (1.4 mm left-right, 7.8 mm anterior-posterior, and 6.4 mm superior-inferior)
P.1462
would have encompassed more than 95% of the center-of-mass motion.
Zelefsky and colleagues (448) obtained four serial CT studies for 50 patients (planning scan and three additional scans during the course of therapy). Prostate displacements in the anteroposterior and superoinferior directions were most frequently observed. The mean prostate motion in the anteroposterior, superoinferior, and left–right directions was 1.2, 0.5, and 0.6 mm, respectively. Anteroposterior movements were correlated with changes in rectal volume. Patients with large rectal volumes (>60 cm3) and large bladder volumes (>40 cm3) on the planning scan experienced a higher likelihood of having >3 mm systematic displacement of the prostate and seminal vesicles, leading the authors to conclude that these patients may require more generous PTV margins to ensure adequate CTV coverage. However, among patients without larger bladder and rectal volumes, a 1-cm margin around the CTV with a 6-mm margin at the prostate–rectal interface enclosed the posterior, anterior, superoinferior and left-right aspects of the CTV within the prescription dose level with a probability of 90%, 100%, 99% and 100%, respectively, indicating that the MSKCC margins provided adequate CTV coverage for most patients.
Intrafractional prostate motion was studied in 20 patients by Huang et al. (186) using pre- and posttreatment B-mode rectal ultrasound evaluations. Although the intrafractional motion was relatively insignificant in all directions, the predominant directions of motion were in the anterior and superior directions. Standard deviations of 0.4, 1.3, and 1.0 mm were observed in the lateral, anterior, and superior directions, respectively. A combination of real-time tumor tracking technology, fluoroscopy, and implanted gold markers were used by Kitamura et al. (210) to quantify intrafractional motion in 10 patients. Data were obtained with the patients in both supine and prone positions. The amplitude of the observed 3D motion was as much as 2.7 mm in the supine position and 24 mm in the prone position, indicating a great impact from respiration and bowel movement in the prone position.
Several methods have been developed to reduce uncertainty due to interfractional organ motion and thereby improve treatment delivery using computer-assisted transabdominal ultrasonography, radio-opaque marker tracking, or CT image-guidance. With the ultrasound system (BAT, Nomos Corporation, Sewickley, PA), patients are instructed to maintain a full bladder and are initially set up based on their tattoos in the supine position. The system is attached to the accelerator collimator and imports the coordinates of the isocenter as well as the target contours from the planning CT. Computer software facilitates 3D matching of the target contours with the prostate visualized on ultrasonography and determination of the necessary patient position modification. This system may not be reliable for patients who cannot maintain a full bladder and those with a large body habitus or other anatomic constraints because of anticipated poor image quality. It has also been noted that BAT measurements can be associated with a 2- to 3-mm error, which is sometimes in the range of the required shifts. Nevertheless, this system has been used by several investigators to verify the prostate position and results are consistent with improved accuracy of the daily treatment delivery (70,123,129,130,186,217,225,228,229,240,282,363). One such study, by Morr et al. (282), evaluated the BAT system on 23 patients undergoing IMRT treatment to the prostate. Once users of the system were sufficiently experienced, pretreatment ultrasound procedures could be performed in an average of about 5 minutes. Positional corrections averaged 2.6 ± 2.1 mm, 4.7 ± 2.7 mm, and 4.2 ± 2.8 mm in the lateral, anterior-posterior, and superior-inferior directions, respectively. The authors concluded that daily system use was feasible and resulted in clinically significant adjustments that would not have been possible using other conventional verification methods.
Recent technological advances have opened up the possibility of acquiring pre- or posttreatment megavoltage or kilovoltage CT images directly on the linear accelerator with the patient in the treatment position. One example of such a device, known as a tomotherapy unit (TomoTherapy Hi-ART, Madison, WI), consists of a 6-MV accelerator mounted within a CT-type gantry. Megavoltage CT images can be obtained prior to treatment, registered with the planning CT study, and the resulting positional corrections can then be applied prior to treatment. IMRT treatment is delivered through synchronized circular motion of the accelerator, couch translational motion, and multileaf collimation. Langen et al. (226) compared three methods of registering megavoltage CT images from a tomotherapy unit with the kilovoltage planning CT images and found that manual registration performed using implanted fiducial markers exhibited the least interobserver variability and agreed best with automatic registration computed from the center of mass of the three implanted fiducial markers.
Linac-based kilovoltage image guidance systems have recently become commercially available. These systems are comprised of a kilovoltage x-ray tube mounted 90 degrees from the accelerator head and a kilovotage imaging plate mounted 90 degrees from the standard megavoltage imaging device. They possess capabilities for kilovoltage two-dimensional projection imaging (radiographs), fluoroscopy, and 3D cone-beam CT, and are thus ideally suited for monitoring of inter- and intrafractional motion. Although little clinical experience using these kilovoltage gantry-mounted systems has been obtained to date, several groups have extensive experience monitoring and correcting for changes due to patient setup and organ motion with either two-dimensional electronic megavoltage portal imaging (EPID) or conventional CT systems (185,238,250,373,374,432,434,435). As a result, several methods for fast prostate localization on CT images, appropriate for either off-line or on-line, image-guided radiotherapy have already been described. Smitsmans et al. (373,374) have developed an automated 3D gray scale registration method that they have applied to both conventional and 3D cone-beam CT images. Collimating the field of view during the cone-beam CT acquisition significantly improves the cone-beam CT image quality and hence the registration success rate. Cone-beam CT artifacts caused by gas pockets moving during the CT acquisition are the main cause of unsuccessful registration.
P.1463
Hua et al. (185) have developed a semiautomatic method for localizing the prostate on pretreatment CT images based on manual identification of the posterior, anterior, left, and right extents of the prostate on the CT slices. The prostate displacement relative to the planning scan is then estimated through a simultaneous fitting of these “extents” to a finely spaced contour template from the planning scan. Identification of the prostatic extents on five pretreatment CT slices was found to be sufficient for reliable determination of the prostatic displacement. The approach of Yan et al. (435) has been to acquire an initial sequence of daily CT scans (typically 5 to 10 serial scans) from which the organ motion and patient setup inaccuracy can be reliably estimated. Based on these data, a confidence-limited PTV can be constructed that ensures, to within a defined statistical limit, that the CTV will receive a dose within a predefined tolerance. For example, the authors determined that a confidence-limited PTV constructed from daily CT scans obtained during the first week of 3DCRT treatment was sufficient to achieve a maximum dose reduction of ≤2% in the CTV for at least 80% of the patients or a 4.5% reduction for 95% of the patients. IMRT treatment required 2 weeks of CT data to achieve the same level of dosimetric coverage. Referred to as adaptive radiotherapy, this off-line correction strategy is capable of excluding systematic error and compensating for random uncertainties. It requires serial scans, as previously outlined, and a single plan modification after the first or second week of therapy.
posted by stefanoxx at 12:38 PM
0 Comments:
Post a Comment
<< Home