- Email: [email protected]
Quantification of Prostate and Seminal Vesicle Interfraction Variation During IMRT
Int. J. Radiation Oncology Biol. Phys., Vol. 71, No. 3, pp. 813–820, 2008 Copyright Ó 2008 Elsevier Inc. Printed in the USA. All rights reserved 0360-3016/08/$–see front matter
doi:10.1016/j.ijrobp.2007.10.028
CLINICAL INVESTIGATION
Prostate
QUANTIFICATION OF PROSTATE AND SEMINAL VESICLE INTERFRACTION VARIATION DURING IMRT STEVEN J. FRANK, M.D.,* LEI DONG, PH.D.,y RAJAT J. KUDCHADKER, PH.D.,y RENAUD DE CREVOISIER, M.D.,* ANDREW K. LEE, M.D., M.P.H.,* REX CHEUNG, M.D., PH.D.,* SEUNGTAEK CHOI, M.D.,* JENNIFER O’DANIEL, PH.D.,y SUSAN L. TUCKER, PH.D.,z HE WANG, PH.D.,y AND DEBORAH A. KUBAN, M.D.* Departments of *Radiation Oncology, y Radiation Physics, and z Biostatistics and Applied Mathematics, The University of Texas M.D. Anderson Cancer Center, Houston, TX Purpose: To quantify the interfraction variability in prostate and seminal vesicle (SV) positions during a course of intensity-modulated radiotherapy (IMRT) using an integrated computed tomography (CT)–linear accelerator system and to assess the impact of rectal and bladder volume changes. Methods and Materials: We studied 15 patients who had undergone IMRT for prostate carcinoma. Patients had one pretreatment planning CT scan followed by three in-room CT scans per week using a CT-on-rails system. The prostate, bladder, rectum, and pelvic bony anatomy were contoured in 369 CT scans. Using the planning CT scan as a reference, the volumetric and positional changes were analyzed in the subsequent CT scans. Results: For all 15 patients, the mean systematic internal prostate and SV variation was 0.1 ± 4.1 mm and 1.2 ± 7.3 mm in the anteroposterior axis, 0.5 ± 2.9 mm and 0.7 ± 4.5 mm in the superoinferior axis, and 0.2 ± 0.9 mm and 0.9 ± 1.9 mm in the lateral axis, respectively. The mean magnitude of the three-dimensional displacement vector was 4.6 ± 3.5 mm for the prostate and 7.6 ± 4.7 mm for the SVs. The rectal and bladder volume changes during treatment correlated with the anterior and superior displacement of the prostate and SVs. Conclusion: The dominant prostate and SV variations occurred in the anteroposterior and superoinferior directions. The systematic prostate and SV variation between the treatment planning CT and daily therapy as a result of the rectal and bladder volume changes emphasizes the need for daily directed target localization and/or immobilization techniques. Ó 2008 Elsevier Inc. Prostate cancer, Treatment planning, Intensity-modulated radiotherapy, IMRT, Computed tomography–linear accelerator system, Radiotherapy.
radiation doses are still delivered to surrounding critical organs, which have limited further dose escalation. Moreover, recent studies using implanted markers (6–8) and multiple computed tomography (CT) scans during treatment (5, 9– 14) have shown that the interfraction variability in prostate position might be larger than allowed for by these established margins. Therefore, the target volume could occasionally be missed. We have recently shown, in a retrospective study, that rectal distension on the planning CT scan increased the risk of biochemical and local failure in a series of patients who underwent RT for prostate carcinoma without the use of any repositioning technique based on direct prostate organ localization (15). The hypothesis of that study was that among patients with a distended rectum on the planning CT scan, the prostate will most likely have shifted posteriorly
INTRODUCTION Intensity-modulated radiotherapy (IMRT) and other conformal techniques are increasingly being used for prostate cancer to improve tumor control by increasing the dose to the prostate while minimizing the dose to the surrounding critical organs, namely the bladder and rectum (1–4). Such high-precision RT requires a precise design of the treatment field and, particularly, the tumor margins. These margins must be large enough to avoid a geographic miss and account for patient setup variations and internal organ movement, but small enough to limit the normal-tissue exposure to tolerated levels. Previous studies have established a standard for what the tumor margins should be but used relatively imprecise methods of organ localization (5). However, significant
Therapeutic Radiology and Oncology (ASTRO), Atlanta, GA, October 3–7, 2004. Conflict of interest: none. Received July 31, 2007, and in revised form Oct 16, 2007. Accepted for publication Oct 16, 2007.
Reprint requests to: Steven J. Frank, M.D., Department of Radiation Oncology, Unit 97, The University of Texas M. D. Anderson Cancer Center, 1515 Holcombe Blvd., Houston, TX 77030; Phone: (713) 563-2364; Fax: (713) 563-2366; E-mail: sjfrank@ mdanderson.org Presented at the 46th Annual Meeting of the American Society for 813
I. J. Radiation Oncology d Biology d Physics
814
Volume 71, Number 3, 2008
stage localized prostate adenocarcinoma and scheduled to receive definitive IMRT without concurrent hormonal therapy. Patients underwent a simulation CT scan for treatment planning. To ensure a distended bladder during simulation, patients drank 20 oz. of water 30 min before the CT scan. In addition, a Fleet enema was administered before the simulation CT scan to evacuate the rectum and minimize rectal distension. The patients were immobilized using a Vac-Lok bag (Med-Tec, Orange City, IA) extending from the thigh down to the feet. A pelvic simulation CT scan was performed from the top of the iliac crest to 2 cm below the lesser trochanter and without the use of oral or intravenous contrast material so that the planning scan could be directly compared with the noncontrast CT scans performed during the treatment course. The isocenter was set at the center of the prostate, and marks were tattooed on the skin to indicate the initial daily setup position. The treatment plan was designed to deliver a total dose of 75.6 Gy within 42 fractions to 98% of the planning target volume. All patients were treated using a commercially available integrated CT-LINAC system (EXaCT, Varian Oncology Systems, Palo Alto, CA), which allows for convenient CT imaging at the daily RT session with the patient immobilized in the treatment position (17). Before each CT scan, the patients were aligned using the skin marks tattooed during simulation, and radiopaque fiducial markers were placed on the patient at the laser intersections. CT scans were performed three times weekly just before the daily treatments throughout the RT course for a total of 24 CT scans for each patient. All CT scans (including the planning CT scan) were performed without contrast using 3-mm axial slices throughout. The matrix size was 512 512, and the pixel size was approximately 1 mm. To minimize the daily variations in bladder and rectal distension, the simulated conditions were maintained (i.e., all patients were asked to drink 20 oz. of water 1 h before each fraction and to have the sensation of a full bladder before treatment). Although daily enemas were not required, the diet and medications were monitored to ensure an evacuated rectum before each RT session. The acquired CT scans were imported into the Pinnacle3 treatment planning system (Philips Medical Systems, Andover, MA). The prostate, SV, bladder, and rectum were manually contoured on the axial images for all CT scans, and the contours were verified by two physicians. The volume of gas in the rectum was also manually contoured. The rectum was contoured on the planning CT scan and kept constant within the same patient for the subsequent CT scans performed during RT. For treatment planning, the rectum extended from the rectosigmoid flexure to the anus, and an average
during treatment compared with its position on the planning CT scan, leading to an underdosage of the posterior portion of the prostate tumor and hence a poorer clinical outcome. By combining imaging technology with RT, it is now possible to reduce the effect of patient setup variations and account for variations in the prostate before each treatment fraction is delivered. Recent work at our institution with an ultrasound-based target localization system (BAT, North American Scientific, Chatsworth, CA) has demonstrated that it is feasible to perform daily checks for interfraction variation and that alignment correction is needed for >25% of daily IMRT sessions (16). If the alignment were not corrected for these sessions, patients would have misalignments >5 mm and thus potential underdosing of the planning target volume. However, the accuracy of the BAT system in correcting for patient setup variation and accounting for variations in organ position is still under investigation. Computed tomography images are considered the most reliable source of anatomic information in RT planning, but, until recently, the use of CT scans for daily patient positioning was not possible. Currently, with the use of daily CT scans from an integrated CT-linear accelerator (LINAC) treatment machine (17, 18), we have been able to characterize the variation in the position of the prostate, seminal vesicles (SVs), and critical organs between treatment sessions to determine whether alignment corrections are needed. In this study, the CT images before treatment using the integrated CT-LINAC permit the unique evaluation of volumetric CT data. The purposes of this prospective study were to quantify the interfraction variability in the prostate and SV position during IMRT and to assess the relationship between the prostate and SV interfraction variability and bladder and rectal volume using a combined CT-LINAC treatment machine. METHODS AND MATERIALS Between September 2002 and January 2004, 15 patients were enrolled in an institutional review board-approved prospective study with informed consent to quantify the prostate and SV variability during the course of RT at the University of Texas M.D. Anderson Cancer Center. Eligible patients were newly diagnosed with early-
Table 1. Systematic and random internal variability for prostate and seminal vesicles Systematic internal variation (mm) Organ Prostate Mean SD Range Seminal vesicles Mean SD Range
AP
SI
Random internal variation (mm) Lateral
AP
SI
Lateral
0.1 4.1 10.5 to 6.4
0.5 2.9 5.9 to 6.0
0.2 0.9 1.4 to 1.7
3.0 1.3 1.7 to 6.1
2.1 0.6 1.3 to 3.0
1.2 0.5 0.3 to 2.2
1.2 7.3 20.4 to 20.7
0.7 4.5 11.8 to 11.6
0.9 1.9 5.5 to 7.0
3.3 1.2 1.9 to 6.0
2.4 0.6 1.2 to 3.7
1.2 0.4 0.8 to 2.1
Abbreviations: AP = anteroposterior; SI = superoinferior; SD = standard deviation. Systematic variation refers to magnitude of center of volume mean variability; random variation refers to magnitude of center of volume SD variability.
IMRT interfraction variability during prostate cancer treatment d S. J. FRANK et al.
815
rectal cross-sectional area (CSA) was calculated by dividing the total rectal volume by the rectal length. To quantify the internal organ variability unaffected by the setup variations in daily treatment, a reference bony structure was defined for CT-based three-dimensional image registration. The pubic symphysis was selected because of its proximity to the prostate and lack of rotational variability. The pubic symphysis was aligned using a CT-to-CT rigid-body registration algorithm developed in-house and similar to our previously published algorithm for prostate target alignment (18). For positional analyses, a center of volume (COV) was defined for each structure, including the reference bony structure. The COVs were used to analyze the position of the prostate, SVs, bladder, and rectum relative to the bony reference. After fusion and alignment of the bony-reference COV, the change in the position of the COV of any contoured structure relative to the position of the bony-reference COV represented the daily internal organ displacement. To quantify the positional shifts for each structure (prostate, SVs, bladder, and rectum) during treatment, we calculated the change in COV positions relative to those on the planning CT in each direction. We defined the anterior shifts, inferior shifts, and right displacements as positive values and the posterior shifts, superior shifts, and left displacements as negative values. For the prostate and SVs, both systematic and random COV variations were calculated. For each patient, the systematic COV variation was defined as the mean of the COV variation during the protracted IMRT course, and the random COV variation was defined as the standard deviation (SD) of the COV variation during the protracted IMRT course. Linear regression analyses were used to explore relationships between the prostate and SV displacement and the variations in rectal and bladder volume.
RESULTS Patient characteristics A total of 15 patients were included in the study and had a total of 369 CT scans, corresponding to a median of 24 CT scans (range, 23–26) for each patient. Of the 15 patients, 73% had Stage T1c and 27% had Stage T2 early-stage localized prostate cancer. The median pretreatment prostatespecific antigen value was 5 ng/mL (range, 1.7–11.9 ng/mL). The Gleason score was 6 in 73% of patients and 7 in 27%. The tumors were at low risk in 67% of patients and intermediate risk in 33% for biochemical recurrence. All patients received a total dose of 75.6 Gy delivered in 42 fractions, within a median duration of 58 days (range, 55–62 days). No patient received hormonal therapy or brachytherapy. Interfraction variability in prostate and SV position The systematic and random variations in position for the prostate and SVs for the entire group of patients are shown in Table 1. The mean displacement of the prostate COV in each axis (anteroposterior [AP], superoinferior [SI], and lateral) for each patient is illustrated in Fig. 1. The AP variation was significantly greater than the SI variation statistically, which was significantly greater than the lateral variation. For the entire group of patients, the mean SD systematic internal prostate variation was 0.1 4.1 mm (range, 10.5 to 6.4 mm) in the AP axis, 0.5 2.9 mm (range, 5.9 to 6.0 mm) in the SI axis, and 0.2 0.9 mm (range, 1.4
Fig. 1. Displacement of prostate in (a) anteroposterior (AP) (positive values represent anterior and negative values represent posterior displacement), (b) superoinferior (SI) (positive values represent inferior and negative values represent superior displacement), and (c) lateral (right–left [RL]) (positive values represent right lateral and negative values represent left lateral displacement) directions. Bars represent mean standard deviation (SD).
816
I. J. Radiation Oncology d Biology d Physics
Volume 71, Number 3, 2008
Fig. 2. Ratio of immediate pretreatment rectal volume to rectal volume on simulation (planning) computed tomography (CT) scan for each patient (pt) during intensity-modulated radiotherapy course.
to 1.7 mm) in the lateral axis. The mean SD internal SV variation was 1.2 7.3 mm (range, 20.4 to 20.7 mm) in the AP axis, 0.7 4.5 mm (range, 11.8 to 11.6 mm) in the SI axis, and 0.9 1.9 mm (range, 5.5 to 7.0 mm) in the lateral axis. The magnitude of the mean three-dimensional displacement vector was 4.6 3.5 mm for the prostate and 7.6 4.7 mm for the SVs. During the IMRT course, the variation in the SV COV was greater than the variation in the prostate COV. Interfraction variability in rectal and bladder volume The rectal and bladder volumes for each patient relative to the volumes on the planning CT scan are shown in Figs. 2 and 3.
The mean rectal volume during treatment was 35–140 cm3 and was larger than the rectal volume measured on the planning CT scan in 61% of the daily CT scans. The mean proportion of the rectum filled by gas during treatment was 4–26%; the proportion of rectum filled by gas was 4–10% on the planning CT scan. The mean rectal CSA during treatment was 3.6–12.9 cm2 and on the planning CT scan was 4.0–9.6 cm2. The proportion of the rectum filled by gas correlated with the rectal CSA (R2 = 0.38, p <0.05). The mean bladder volume during treatment was 120–381 cm3, and 90% of the bladder volumes calculated on the daily CT scans were smaller than the bladder volumes measured on the planning CT scan.
Fig. 3. Ratio of immediate pretreatment bladder volume to bladder volume on simulation (planning) computed tomography (CT) scan for each patient (pat) during intensity-modulated radiotherapy course.
IMRT interfraction variability during prostate cancer treatment d S. J. FRANK et al.
817
Prostate, SV, and rectal volume correlations As shown in Fig. 4, the changes in rectal volume correlated with the prostate and SV anterior and superior displacement. For 93% of the patients, the correlation was significant (p <0.05) between the anterior prostate variation and anterior rectal volumetric changes. On multiple regression analysis with the bladder and rectal volume as covariates, the prostate AP and SI displacement during treatment correlated with the rectal volume changes in 73% and 27% of the daily CT scans, respectively. Similarly, the SV AP and SI displacement during treatment correlated with the rectal volume changes in 93% and 53% of the daily scans, respectively. In addition, the rectal CSA calculated from the planning CT scan correlated with the mean prostate AP displacement during treatment (p = 0.03; Fig. 5). Prostate, SV, and bladder volume correlations A decreased bladder volume correlated with the anterior and superior prostate displacement. Multiple regression analysis revealed that the prostate AP and SI variation correlated with the bladder volume changes in 27% and 13% of the daily CT scans, respectively. Similarly, the SV AP and SI variation correlated with the bladder volume changes in 27% and 7% of the daily CT scans, respectively.
DISCUSSION To our knowledge, this is the largest single-institution study to quantify the variability in the prostate and SV position during an IMRT course and to assess the relationships between the prostate and SV position and the bladder and rectal volumes during IMRT. Daily CT scans were obtained using an integrated CT-LINAC treatment machine immediately before treatment for 60% of the treatment fractions. Other studies have reported on CT scan assessment of prostate variability during a RT course but included only a limited number of CT scans per patient (eight or fewer) and, most important, did not obtain the scans at the daily treatment (5, 9–13, 19–24). Ultrasonography (16, 25, 26) and radiopaque markers (6–8) have also been used to evaluate prostate
Fig. 4. Prostate and seminal vesicle (SV) anteroposterior (AP) random variability by change in rectal filling (data from all patients combined).
Fig. 5. Mean prostate and seminal vesicle anteroposterior (AP) displacement by rectal distension on planning computed tomography scan.
motion during treatment, but ultrasonography does not relate exactly anatomically to the planning CT scan, and marker correlation does not allow visualization of the entire prostate and SV. Moreover, neither of these methods can accurately quantify the changes in the rectal and bladder volume during therapy. The definition of the reference point was also more robust in this study, rather than relying on a shift from the isocenter from the therapist’s daily setup. In the present study, guidelines were in place to ensure that the bladder and rectal volumes at each daily treatment were consistent with the volumes on the treatment planning CT. Nevertheless, in 90% of patients, the bladder volume during RT was smaller than the bladder volume on the planning CT for >80% of the treatment fractions (Fig. 3) and correlated with an anterior and superior shift of the prostate and SVs during RT. Similarly, the guidelines to keep the rectal volume at the simulated level were generally unsuccessful (Fig. 2), and the treatment rectal volumes were larger than the rectal volume measured on the planning CT scan in 61% of daily scans. The larger rectal volumes during RT resulted in an anterior and superior shift of the prostate and SVs. The data from this study are consistent with the data from our previous ultrasound-based daily prostate localization study (16), as well as the motion studies of other investigators (5, 6, 9–14). The prostate variation during a protracted RT course was greatest in the AP axis, where it could reach >14 mm (Fig. 1a) and the least in the lateral axis (Fig. 1c). In the present study, for the entire cohort of patients, the random AP prostate variability, representing the variability of the prostate in the AP direction, had a mean value of 3.0 mm and a maximal value of 6.1 mm. In the published studies, the random prostate variation has been 1.5–4.1 mm (27). Although this variability appears relatively small, the systematic AP variability of the prostate (mean variation in prostate position during treatment relative to the position on the planning CT scan) varied widely among the patients in our series and could be significant (range, 10.5 to 6.4 mm),
818
I. J. Radiation Oncology d Biology d Physics
potentially leading to a systematic mistargeting. Thus, a major issue in minimizing the effect of changes in prostate position is to correct for systematic differences between the actual prostate position and its position on the planning CT scan (28). Using the COV of the contoured prostate to measure prostate variability, a 5-mm isotropic margin would successfully cover the internal prostate motion 78%, 88%, and 99% of the time in the AP, SI, and lateral direction, respectively. The frequency of prostate and SV misses as a function of margin size is plotted in Fig. 6. In the era of dose escalation, the posterior treatment margin must be as small as possible to limit irradiation of the rectum and avoid the potential for increased proctitis and a nonhealing anterior rectal wall ulcer. However, the posterior portion of the prostate contains the peripheral zone, where most tumors are located (29, 30), and an insufficient margin might result in undertreatment of the tumor. These two points emphasize the need for direct target localization with image-guided RT to adequately cover the prostate and SVs while minimizing the dose and potential damage to the bladder and rectum. Direct target localization with imageguided RT is no longer optional when the dose to the prostate is escalated to improve biochemical outcomes (31) or the dose is hypofractionated with stereotactic body RT because the therapeutic ratio is decreased. In the present study, as well as in other studies (9, 11, 12, 19, 22), the variability in SV displacement appeared larger than the variability in prostate displacement, suggesting that a greater margin is necessary for the SVs than for the prostate. The AP SV displacement was as large as 20 mm in our series. Therefore, even an AP margin of 1 cm for the SVs might be inadequate in some cases. This margin would adequately compensate for AP SV variation in only 86% of treatments. Moreover, the variation in SV position might
Fig. 6. Frequency of prostate and seminal vesicle (SV) misses as function of margin size (internal organ variation only). Sup-inf = superoinferior.
Volume 71, Number 3, 2008
be dominated by the shape change of this structure. In measuring the variability of the SVs, we used the COV, which does not fully account for the potential effect of the threedimensional shape variation, which might require an even larger margin. For example, a more laterally and symmetrically extended SV might not change the COV value even though significant variation of the individual SV could occur. The effect of the bladder volume on SV variability in the AP and SI axes seems to be greater than the effect of the bladder volume on prostate variability; however, our study has corroborated previous findings that rectal filling remains the most important contributing factor (14). In addition to analyzing the effect of rectal and bladder volume changes on the SV position, we analyzed the effect of rectal and bladder volume changes on the prostate position. Our results concur with those of other investigators in that the AP prostate variability was mainly caused by rectal distension (Fig. 4) (6, 9, 11–13, 19, 23, 24, 32). Rectal distension appeared to correlate with the presence of gas. The absence of a correlation between the rectal volume changes and AP prostate variability in 27% of patients might be explained by asymmetric AP rectal volume variation, with greater increases in rectal volume in the superior than in the inferior part of the rectum, and small variations in organ volume and position. Figure 2 illustrates that for patients with the largest posterior variability (Patients 5, 9, and 14), the rectal volume during RT was less than the rectal volume on the planning CT scan 75%, 86%, and 100% of the time, respectively. As shown in Fig. 6, an empty rectum at the planning CT scan would likely have minimized the posterior shift during treatment. The data in Fig. 6 therefore supports the hypothesis of our earlier study (15), which showed the effect of rectal distension on local control. Thus, if the rectum is distended on the planning CT (as it was for Patients 5, 9, and 14), it will likely be less distended, on average, during the RT course, resulting in a posterior shift of the prostate relative to the simulation position, underdosage of the posterior portion of the prostate tumor, and a poorer clinical outcome. An anterior displacement of the prostate during RT might have only a moderate effect on local control, given the usual large anterior clinical target volume to the planning target volume margin and to the preferential location of the tumor in the posterior part of the gland (29, 30). The effect of the enema given at the planning CT scan on rectal filling and AP prostate motion is clearly shown in Figs. 1a and 2; 67% of the patients had a rectal volume that was larger during treatment than at the planning CT scan >50% of the time. Consequently, 67% of the patients had a positive mean AP prostate variability, indicating a systematic prostate displacement in the anterior direction (Fig. 1a). Other studies have shown that the use of a contrast agent or catheter in the rectum at the planning CT scan can also push the prostate anteriorly as much as 1–2 cm, potentially leading to a systematic posterior positional change during RT relative to the prostate position on the planning CT scan (14, 24, 32, 33). Although we routinely administer an enema to empty the rectum before the simulation CT
IMRT interfraction variability during prostate cancer treatment d S. J. FRANK et al.
scan, the effectiveness of the enema in producing an empty rectum did not appear to be constant. We instructed our patients to maintain regular bowel movements during RT and to refrain from becoming constipated, but stool softeners might have been unnecessary because patients might develop an increased frequency of bowel movements secondary to the effects of RT. Realistically, the degree of rectal filling and gas cannot be controlled consistently during a relatively lengthy IMRT course; thus, for daily localization, we have used the BAT ultrasound-based target localization system before every treatment fraction. In our study, the effect of bladder filling on prostate displacement appeared to be relatively small, and bladder filling did not correlate well with the prostate displacement. Although the bladder volume for each patient varied significantly during the RT course, the prostate movement in the SI direction was generally quite small, and patients with the largest SI prostate displacement showed no difference in bladder filling compared with patients with minimal SI prostate displacement. These findings are similar to those of other investigators (11, 14, 23, 24, 32) and to the findings from a previous study from our institution (5). Additionally, overfilling of the bladder at simulation is not advisable because of the lack of daily reproducibility and the potential effect on the SV dose.
819
CONCLUSION This is one of the largest single-institution studies of variability in prostate and SV position during IMRT, and our study results differ from those of previous such studies in that we performed CT scans frequently during the RT course and performed these scans just before the daily RT sessions. We were able to quantify and analyze the changes in internal organ position and volume of the pelvic organs during a protracted IMRT course. The dominant motion of the prostate and SVs occurred in the AP direction; the next greatest motion was in the SI direction. Our findings indicate that without adequate direct prostate localization or patient immobilization techniques, a 5-mm posterior margin for the prostate could prove inadequate. Rectal distension affected both prostate and SV positions, and the SVs exhibited greater variation in daily position than did the prostate. The systematic difference between the rectal and bladder volume on the planning CT scan and at the daily RT sessions deserves attention. These changes in critical organ volumes raise the question of the relevance of dose–volume histograms calculated solely on the basis of the treatment planning imaging. We are addressing the dosimetric effect of these anatomic variations during the RT course in an effort to make more accurate predictions of normal tissue tolerance.
REFERENCES 1. Pollack A, Zagars GK, Smith LG, et al. Preliminary results of a randomized radiotherapy dose-escalation study comparing 70 Gy with 78 Gy for prostate cancer. J Clin Oncol 2000;18: 3904–3911. 2. Zelefsky MJ, Leibel SA, Gaudin PB, et al. Dose escalation with three-dimensional conformal radiation therapy affects the outcome in prostate cancer. Int J Radiat Oncol Biol Phys 1998; 41:491–500. 3. Hanks GE, Hanlon AL, Epstein B, et al. Dose response in prostate cancer with 8–12 years’ follow-up. Int J Radiat Oncol Biol Phys 2002;54:427–435. 4. Pollack A, Zagars GK, Starkschall G, et al. Prostate cancer radiation dose response: Results of the M.D. Anderson phase III randomized trial. Int J Radiat Oncol Biol Phys 2002;53: 1097–1105. 5. Antolak JA, Rosen II, Childress CH, et al. Prostate target volume variations during a course of radiotherapy. Int J Radiat Oncol Biol Phys 1998;42:661–672. 6. Balter JM, Sandler HM, Lam K, et al. Measurement of prostate movement over the course of routine radiotherapy using implanted markers. Int J Radiat Oncol Biol Phys 1995;31: 113–118. 7. Crook JM, Raymond Y, Salhani D, et al. Prostate motion during standard radiotherapy as assessed by fiducial markers. Radiother Oncol 1995;37:35–42. 8. Millender LE, Aubin M, Pouliot J, et al. Daily electronic portal imaging for morbidly obese men undergoing radiotherapy for localized prostate cancer. Int J Radiat Oncol Biol Phys 2004; 59:6–10. 9. Zelefsky MJ, Crean D, Mageras GS, et al. Quantification and predictors of prostate position variability in 50 patients evaluated with multiple CT scans during conformal radiotherapy. Radiother Oncol 1999;50:225–234.
10. Stroom JC, Koper PCM, Korevaaar GA, et al. Internal organ motion in prostate cancer patients treated in prone and supine treatment position. Radiother Oncol 1999;51:237–248. 11. Beard CJ, Kijewski P, Bussiere M, et al. Analysis of prostate and seminal vesicle motion: Implications for treatment planning. Int J Radiat Oncol Biol Phys 1996;34:451–458. 12. Roeske JC, Forman JD, Mesina CF, et al. Evaluation of changes in the size and location of the prostate, seminal vesicles, bladder, and rectum during a course of external beam radiation therapy. Int J Radiat Oncol Biol Phys 1995;33:1321–1329. 13. Rudat V, Schraube P, Oetzel D, et al. Combined error of patient positioning variability and prostate motion uncertainty in 3D conformal radiotherapy of localized prostate cancer. Int J Radiat Oncol Biol Phys 1996;35:1027–1034. 14. Schild SE, Casale HE, Bellefontaine LP. Movements of the prostate due to rectal and bladder distension: Implications for radiotherapy. Med Dosim 1993;18:13–15. 15. de Crevoisier R, Tucker SL, Dong L, et al. Increased risk of biochemical and local failure in patients with distended rectum on the planning CT for prostate cancer radiotherapy. Int J Radiat Oncol Biol Phys 2005;62:965–973. 16. Chandra A, Dong L, Huang E, et al. Experience of ultrasoundbased daily prostate localization. Int J Radiat Oncol Biol Phys 2003;56:436–447. 17. Court L, Rosen I, Mohan R, et al. Evaluation of mechanical precision and alignment uncertainties for an integrated CT/LINAC system. Med Phys 2003;30:1198–1210. 18. Court LE, Dong L. Automatic registration of the prostate for computed-tomography-guided radiotherapy. Med Phys 2003; 30:2750–2757. 19. Dawson LA, Mah K, Franssen E, et al. Target position variability throughout prostate radiotherapy. Int J Radiat Oncol Biol Phys 1998;42:1155–1161.
820
I. J. Radiation Oncology d Biology d Physics
20. Roach MR, Faillace-Akazawa P, Malfatti C. Prostate volumes and organ movement defined by serial computerized tomographic scans during three-dimensional conformal radiotherapy. Radiat Oncol Invest 1997;5:187–194. 21. Mechalakos JG, Mageras GS, Zelefsky MJ, et al. Time trends in organ position and volume in patients receiving prostate threedimensional conformal radiotherapy. Radiother Oncol 2002;62: 261–265. 22. Tinger A, Michalski JM, Cheng A, et al. A critical evaluation of the planning target volume for 3-D conformal radiotherapy of prostate cancer. Int J Radiat Oncol Biol Phys 1998;42:213–221. 23. Melian E, Mageras GS, Fuks Z, et al. Variation in prostate position quantitation and implications for three-dimensional conformal treatment planning. Int J Radiat Oncol Biol Phys 1997;38:73–81. 24. Lebesque JV, Bruce AM, Guus KAP, et al. Variation in volumes, dose–volume histograms, and estimated normal tissue complication probabilities of rectum and bladder during conformal radiotherapy of T3 prostate cancer. Int J Radiat Oncol Biol Phys 1995;33:1109–1119. 25. Langen KM, Pouliot J, Anezinos C, et al. Evaluation of ultrasound-based prostate localization for image-guided radiotherapy. Int J Radiat Oncol Biol Phys 2003;57:635–644. 26. Lattanzi J, McNeeley S, Hanlon A, et al. Ultrasound-based stereotactic guidance of precision conformal external beam
Volume 71, Number 3, 2008
27. 28.
29. 30. 31.
32.
33.
radiation therapy in clinically localized prostate cancer. Urology 2000;55:73–78. Langen KM, Jones DTL. Organ motion and its management. Int J Radiat Oncol Biol Phys 2001;50:265–278. Yan D, Lockman D, Brabbins D, et al. An off-line strategy for constructing a patient-specific planning target volume in adaptive treatment process for prostate cancer. Int J Radiat Oncol Biol Phys 2000;48:289–302. Chen ME, Johnston DA, Tang K, et al. Detailed mapping of prostate carcinoma foci: Biopsy strategy implications. Cancer 2000;89:1800–1809. Greene DR, Wheeler TM, Egawa S, et al. Relationship between clinical stage and histological zone of origin in early prostate cancer: morphometric analysis. Br J Urol 1991;68:499–509. Zelefsky MJ, Chan H, Hunt M, et al. Long-term outcome of high dose intensity modulated radiation therapy for patients with clinically localized prostate cancer. J Urol 2006;176: 1415–1419. Ten Haken RK, Forman JD, Heimburger DK, et al. Treatment planning issues related to prostate movement in response to differential filling of the rectum and bladder. Int J Radiat Oncol Biol Phys 1991;20:1317–1324. Malone S, Donker R, Dahrouge S, et al. Treatment planning aids in prostate cancer: Friend or foe? Int J Radiat Oncol Biol Phys 2001;51:49–55.
doi:10.1016/j.ijrobp.2007.10.028
CLINICAL INVESTIGATION
Prostate
QUANTIFICATION OF PROSTATE AND SEMINAL VESICLE INTERFRACTION VARIATION DURING IMRT STEVEN J. FRANK, M.D.,* LEI DONG, PH.D.,y RAJAT J. KUDCHADKER, PH.D.,y RENAUD DE CREVOISIER, M.D.,* ANDREW K. LEE, M.D., M.P.H.,* REX CHEUNG, M.D., PH.D.,* SEUNGTAEK CHOI, M.D.,* JENNIFER O’DANIEL, PH.D.,y SUSAN L. TUCKER, PH.D.,z HE WANG, PH.D.,y AND DEBORAH A. KUBAN, M.D.* Departments of *Radiation Oncology, y Radiation Physics, and z Biostatistics and Applied Mathematics, The University of Texas M.D. Anderson Cancer Center, Houston, TX Purpose: To quantify the interfraction variability in prostate and seminal vesicle (SV) positions during a course of intensity-modulated radiotherapy (IMRT) using an integrated computed tomography (CT)–linear accelerator system and to assess the impact of rectal and bladder volume changes. Methods and Materials: We studied 15 patients who had undergone IMRT for prostate carcinoma. Patients had one pretreatment planning CT scan followed by three in-room CT scans per week using a CT-on-rails system. The prostate, bladder, rectum, and pelvic bony anatomy were contoured in 369 CT scans. Using the planning CT scan as a reference, the volumetric and positional changes were analyzed in the subsequent CT scans. Results: For all 15 patients, the mean systematic internal prostate and SV variation was 0.1 ± 4.1 mm and 1.2 ± 7.3 mm in the anteroposterior axis, 0.5 ± 2.9 mm and 0.7 ± 4.5 mm in the superoinferior axis, and 0.2 ± 0.9 mm and 0.9 ± 1.9 mm in the lateral axis, respectively. The mean magnitude of the three-dimensional displacement vector was 4.6 ± 3.5 mm for the prostate and 7.6 ± 4.7 mm for the SVs. The rectal and bladder volume changes during treatment correlated with the anterior and superior displacement of the prostate and SVs. Conclusion: The dominant prostate and SV variations occurred in the anteroposterior and superoinferior directions. The systematic prostate and SV variation between the treatment planning CT and daily therapy as a result of the rectal and bladder volume changes emphasizes the need for daily directed target localization and/or immobilization techniques. Ó 2008 Elsevier Inc. Prostate cancer, Treatment planning, Intensity-modulated radiotherapy, IMRT, Computed tomography–linear accelerator system, Radiotherapy.
radiation doses are still delivered to surrounding critical organs, which have limited further dose escalation. Moreover, recent studies using implanted markers (6–8) and multiple computed tomography (CT) scans during treatment (5, 9– 14) have shown that the interfraction variability in prostate position might be larger than allowed for by these established margins. Therefore, the target volume could occasionally be missed. We have recently shown, in a retrospective study, that rectal distension on the planning CT scan increased the risk of biochemical and local failure in a series of patients who underwent RT for prostate carcinoma without the use of any repositioning technique based on direct prostate organ localization (15). The hypothesis of that study was that among patients with a distended rectum on the planning CT scan, the prostate will most likely have shifted posteriorly
INTRODUCTION Intensity-modulated radiotherapy (IMRT) and other conformal techniques are increasingly being used for prostate cancer to improve tumor control by increasing the dose to the prostate while minimizing the dose to the surrounding critical organs, namely the bladder and rectum (1–4). Such high-precision RT requires a precise design of the treatment field and, particularly, the tumor margins. These margins must be large enough to avoid a geographic miss and account for patient setup variations and internal organ movement, but small enough to limit the normal-tissue exposure to tolerated levels. Previous studies have established a standard for what the tumor margins should be but used relatively imprecise methods of organ localization (5). However, significant
Therapeutic Radiology and Oncology (ASTRO), Atlanta, GA, October 3–7, 2004. Conflict of interest: none. Received July 31, 2007, and in revised form Oct 16, 2007. Accepted for publication Oct 16, 2007.
Reprint requests to: Steven J. Frank, M.D., Department of Radiation Oncology, Unit 97, The University of Texas M. D. Anderson Cancer Center, 1515 Holcombe Blvd., Houston, TX 77030; Phone: (713) 563-2364; Fax: (713) 563-2366; E-mail: sjfrank@ mdanderson.org Presented at the 46th Annual Meeting of the American Society for 813
I. J. Radiation Oncology d Biology d Physics
814
Volume 71, Number 3, 2008
stage localized prostate adenocarcinoma and scheduled to receive definitive IMRT without concurrent hormonal therapy. Patients underwent a simulation CT scan for treatment planning. To ensure a distended bladder during simulation, patients drank 20 oz. of water 30 min before the CT scan. In addition, a Fleet enema was administered before the simulation CT scan to evacuate the rectum and minimize rectal distension. The patients were immobilized using a Vac-Lok bag (Med-Tec, Orange City, IA) extending from the thigh down to the feet. A pelvic simulation CT scan was performed from the top of the iliac crest to 2 cm below the lesser trochanter and without the use of oral or intravenous contrast material so that the planning scan could be directly compared with the noncontrast CT scans performed during the treatment course. The isocenter was set at the center of the prostate, and marks were tattooed on the skin to indicate the initial daily setup position. The treatment plan was designed to deliver a total dose of 75.6 Gy within 42 fractions to 98% of the planning target volume. All patients were treated using a commercially available integrated CT-LINAC system (EXaCT, Varian Oncology Systems, Palo Alto, CA), which allows for convenient CT imaging at the daily RT session with the patient immobilized in the treatment position (17). Before each CT scan, the patients were aligned using the skin marks tattooed during simulation, and radiopaque fiducial markers were placed on the patient at the laser intersections. CT scans were performed three times weekly just before the daily treatments throughout the RT course for a total of 24 CT scans for each patient. All CT scans (including the planning CT scan) were performed without contrast using 3-mm axial slices throughout. The matrix size was 512 512, and the pixel size was approximately 1 mm. To minimize the daily variations in bladder and rectal distension, the simulated conditions were maintained (i.e., all patients were asked to drink 20 oz. of water 1 h before each fraction and to have the sensation of a full bladder before treatment). Although daily enemas were not required, the diet and medications were monitored to ensure an evacuated rectum before each RT session. The acquired CT scans were imported into the Pinnacle3 treatment planning system (Philips Medical Systems, Andover, MA). The prostate, SV, bladder, and rectum were manually contoured on the axial images for all CT scans, and the contours were verified by two physicians. The volume of gas in the rectum was also manually contoured. The rectum was contoured on the planning CT scan and kept constant within the same patient for the subsequent CT scans performed during RT. For treatment planning, the rectum extended from the rectosigmoid flexure to the anus, and an average
during treatment compared with its position on the planning CT scan, leading to an underdosage of the posterior portion of the prostate tumor and hence a poorer clinical outcome. By combining imaging technology with RT, it is now possible to reduce the effect of patient setup variations and account for variations in the prostate before each treatment fraction is delivered. Recent work at our institution with an ultrasound-based target localization system (BAT, North American Scientific, Chatsworth, CA) has demonstrated that it is feasible to perform daily checks for interfraction variation and that alignment correction is needed for >25% of daily IMRT sessions (16). If the alignment were not corrected for these sessions, patients would have misalignments >5 mm and thus potential underdosing of the planning target volume. However, the accuracy of the BAT system in correcting for patient setup variation and accounting for variations in organ position is still under investigation. Computed tomography images are considered the most reliable source of anatomic information in RT planning, but, until recently, the use of CT scans for daily patient positioning was not possible. Currently, with the use of daily CT scans from an integrated CT-linear accelerator (LINAC) treatment machine (17, 18), we have been able to characterize the variation in the position of the prostate, seminal vesicles (SVs), and critical organs between treatment sessions to determine whether alignment corrections are needed. In this study, the CT images before treatment using the integrated CT-LINAC permit the unique evaluation of volumetric CT data. The purposes of this prospective study were to quantify the interfraction variability in the prostate and SV position during IMRT and to assess the relationship between the prostate and SV interfraction variability and bladder and rectal volume using a combined CT-LINAC treatment machine. METHODS AND MATERIALS Between September 2002 and January 2004, 15 patients were enrolled in an institutional review board-approved prospective study with informed consent to quantify the prostate and SV variability during the course of RT at the University of Texas M.D. Anderson Cancer Center. Eligible patients were newly diagnosed with early-
Table 1. Systematic and random internal variability for prostate and seminal vesicles Systematic internal variation (mm) Organ Prostate Mean SD Range Seminal vesicles Mean SD Range
AP
SI
Random internal variation (mm) Lateral
AP
SI
Lateral
0.1 4.1 10.5 to 6.4
0.5 2.9 5.9 to 6.0
0.2 0.9 1.4 to 1.7
3.0 1.3 1.7 to 6.1
2.1 0.6 1.3 to 3.0
1.2 0.5 0.3 to 2.2
1.2 7.3 20.4 to 20.7
0.7 4.5 11.8 to 11.6
0.9 1.9 5.5 to 7.0
3.3 1.2 1.9 to 6.0
2.4 0.6 1.2 to 3.7
1.2 0.4 0.8 to 2.1
Abbreviations: AP = anteroposterior; SI = superoinferior; SD = standard deviation. Systematic variation refers to magnitude of center of volume mean variability; random variation refers to magnitude of center of volume SD variability.
IMRT interfraction variability during prostate cancer treatment d S. J. FRANK et al.
815
rectal cross-sectional area (CSA) was calculated by dividing the total rectal volume by the rectal length. To quantify the internal organ variability unaffected by the setup variations in daily treatment, a reference bony structure was defined for CT-based three-dimensional image registration. The pubic symphysis was selected because of its proximity to the prostate and lack of rotational variability. The pubic symphysis was aligned using a CT-to-CT rigid-body registration algorithm developed in-house and similar to our previously published algorithm for prostate target alignment (18). For positional analyses, a center of volume (COV) was defined for each structure, including the reference bony structure. The COVs were used to analyze the position of the prostate, SVs, bladder, and rectum relative to the bony reference. After fusion and alignment of the bony-reference COV, the change in the position of the COV of any contoured structure relative to the position of the bony-reference COV represented the daily internal organ displacement. To quantify the positional shifts for each structure (prostate, SVs, bladder, and rectum) during treatment, we calculated the change in COV positions relative to those on the planning CT in each direction. We defined the anterior shifts, inferior shifts, and right displacements as positive values and the posterior shifts, superior shifts, and left displacements as negative values. For the prostate and SVs, both systematic and random COV variations were calculated. For each patient, the systematic COV variation was defined as the mean of the COV variation during the protracted IMRT course, and the random COV variation was defined as the standard deviation (SD) of the COV variation during the protracted IMRT course. Linear regression analyses were used to explore relationships between the prostate and SV displacement and the variations in rectal and bladder volume.
RESULTS Patient characteristics A total of 15 patients were included in the study and had a total of 369 CT scans, corresponding to a median of 24 CT scans (range, 23–26) for each patient. Of the 15 patients, 73% had Stage T1c and 27% had Stage T2 early-stage localized prostate cancer. The median pretreatment prostatespecific antigen value was 5 ng/mL (range, 1.7–11.9 ng/mL). The Gleason score was 6 in 73% of patients and 7 in 27%. The tumors were at low risk in 67% of patients and intermediate risk in 33% for biochemical recurrence. All patients received a total dose of 75.6 Gy delivered in 42 fractions, within a median duration of 58 days (range, 55–62 days). No patient received hormonal therapy or brachytherapy. Interfraction variability in prostate and SV position The systematic and random variations in position for the prostate and SVs for the entire group of patients are shown in Table 1. The mean displacement of the prostate COV in each axis (anteroposterior [AP], superoinferior [SI], and lateral) for each patient is illustrated in Fig. 1. The AP variation was significantly greater than the SI variation statistically, which was significantly greater than the lateral variation. For the entire group of patients, the mean SD systematic internal prostate variation was 0.1 4.1 mm (range, 10.5 to 6.4 mm) in the AP axis, 0.5 2.9 mm (range, 5.9 to 6.0 mm) in the SI axis, and 0.2 0.9 mm (range, 1.4
Fig. 1. Displacement of prostate in (a) anteroposterior (AP) (positive values represent anterior and negative values represent posterior displacement), (b) superoinferior (SI) (positive values represent inferior and negative values represent superior displacement), and (c) lateral (right–left [RL]) (positive values represent right lateral and negative values represent left lateral displacement) directions. Bars represent mean standard deviation (SD).
816
I. J. Radiation Oncology d Biology d Physics
Volume 71, Number 3, 2008
Fig. 2. Ratio of immediate pretreatment rectal volume to rectal volume on simulation (planning) computed tomography (CT) scan for each patient (pt) during intensity-modulated radiotherapy course.
to 1.7 mm) in the lateral axis. The mean SD internal SV variation was 1.2 7.3 mm (range, 20.4 to 20.7 mm) in the AP axis, 0.7 4.5 mm (range, 11.8 to 11.6 mm) in the SI axis, and 0.9 1.9 mm (range, 5.5 to 7.0 mm) in the lateral axis. The magnitude of the mean three-dimensional displacement vector was 4.6 3.5 mm for the prostate and 7.6 4.7 mm for the SVs. During the IMRT course, the variation in the SV COV was greater than the variation in the prostate COV. Interfraction variability in rectal and bladder volume The rectal and bladder volumes for each patient relative to the volumes on the planning CT scan are shown in Figs. 2 and 3.
The mean rectal volume during treatment was 35–140 cm3 and was larger than the rectal volume measured on the planning CT scan in 61% of the daily CT scans. The mean proportion of the rectum filled by gas during treatment was 4–26%; the proportion of rectum filled by gas was 4–10% on the planning CT scan. The mean rectal CSA during treatment was 3.6–12.9 cm2 and on the planning CT scan was 4.0–9.6 cm2. The proportion of the rectum filled by gas correlated with the rectal CSA (R2 = 0.38, p <0.05). The mean bladder volume during treatment was 120–381 cm3, and 90% of the bladder volumes calculated on the daily CT scans were smaller than the bladder volumes measured on the planning CT scan.
Fig. 3. Ratio of immediate pretreatment bladder volume to bladder volume on simulation (planning) computed tomography (CT) scan for each patient (pat) during intensity-modulated radiotherapy course.
IMRT interfraction variability during prostate cancer treatment d S. J. FRANK et al.
817
Prostate, SV, and rectal volume correlations As shown in Fig. 4, the changes in rectal volume correlated with the prostate and SV anterior and superior displacement. For 93% of the patients, the correlation was significant (p <0.05) between the anterior prostate variation and anterior rectal volumetric changes. On multiple regression analysis with the bladder and rectal volume as covariates, the prostate AP and SI displacement during treatment correlated with the rectal volume changes in 73% and 27% of the daily CT scans, respectively. Similarly, the SV AP and SI displacement during treatment correlated with the rectal volume changes in 93% and 53% of the daily scans, respectively. In addition, the rectal CSA calculated from the planning CT scan correlated with the mean prostate AP displacement during treatment (p = 0.03; Fig. 5). Prostate, SV, and bladder volume correlations A decreased bladder volume correlated with the anterior and superior prostate displacement. Multiple regression analysis revealed that the prostate AP and SI variation correlated with the bladder volume changes in 27% and 13% of the daily CT scans, respectively. Similarly, the SV AP and SI variation correlated with the bladder volume changes in 27% and 7% of the daily CT scans, respectively.
DISCUSSION To our knowledge, this is the largest single-institution study to quantify the variability in the prostate and SV position during an IMRT course and to assess the relationships between the prostate and SV position and the bladder and rectal volumes during IMRT. Daily CT scans were obtained using an integrated CT-LINAC treatment machine immediately before treatment for 60% of the treatment fractions. Other studies have reported on CT scan assessment of prostate variability during a RT course but included only a limited number of CT scans per patient (eight or fewer) and, most important, did not obtain the scans at the daily treatment (5, 9–13, 19–24). Ultrasonography (16, 25, 26) and radiopaque markers (6–8) have also been used to evaluate prostate
Fig. 4. Prostate and seminal vesicle (SV) anteroposterior (AP) random variability by change in rectal filling (data from all patients combined).
Fig. 5. Mean prostate and seminal vesicle anteroposterior (AP) displacement by rectal distension on planning computed tomography scan.
motion during treatment, but ultrasonography does not relate exactly anatomically to the planning CT scan, and marker correlation does not allow visualization of the entire prostate and SV. Moreover, neither of these methods can accurately quantify the changes in the rectal and bladder volume during therapy. The definition of the reference point was also more robust in this study, rather than relying on a shift from the isocenter from the therapist’s daily setup. In the present study, guidelines were in place to ensure that the bladder and rectal volumes at each daily treatment were consistent with the volumes on the treatment planning CT. Nevertheless, in 90% of patients, the bladder volume during RT was smaller than the bladder volume on the planning CT for >80% of the treatment fractions (Fig. 3) and correlated with an anterior and superior shift of the prostate and SVs during RT. Similarly, the guidelines to keep the rectal volume at the simulated level were generally unsuccessful (Fig. 2), and the treatment rectal volumes were larger than the rectal volume measured on the planning CT scan in 61% of daily scans. The larger rectal volumes during RT resulted in an anterior and superior shift of the prostate and SVs. The data from this study are consistent with the data from our previous ultrasound-based daily prostate localization study (16), as well as the motion studies of other investigators (5, 6, 9–14). The prostate variation during a protracted RT course was greatest in the AP axis, where it could reach >14 mm (Fig. 1a) and the least in the lateral axis (Fig. 1c). In the present study, for the entire cohort of patients, the random AP prostate variability, representing the variability of the prostate in the AP direction, had a mean value of 3.0 mm and a maximal value of 6.1 mm. In the published studies, the random prostate variation has been 1.5–4.1 mm (27). Although this variability appears relatively small, the systematic AP variability of the prostate (mean variation in prostate position during treatment relative to the position on the planning CT scan) varied widely among the patients in our series and could be significant (range, 10.5 to 6.4 mm),
818
I. J. Radiation Oncology d Biology d Physics
potentially leading to a systematic mistargeting. Thus, a major issue in minimizing the effect of changes in prostate position is to correct for systematic differences between the actual prostate position and its position on the planning CT scan (28). Using the COV of the contoured prostate to measure prostate variability, a 5-mm isotropic margin would successfully cover the internal prostate motion 78%, 88%, and 99% of the time in the AP, SI, and lateral direction, respectively. The frequency of prostate and SV misses as a function of margin size is plotted in Fig. 6. In the era of dose escalation, the posterior treatment margin must be as small as possible to limit irradiation of the rectum and avoid the potential for increased proctitis and a nonhealing anterior rectal wall ulcer. However, the posterior portion of the prostate contains the peripheral zone, where most tumors are located (29, 30), and an insufficient margin might result in undertreatment of the tumor. These two points emphasize the need for direct target localization with image-guided RT to adequately cover the prostate and SVs while minimizing the dose and potential damage to the bladder and rectum. Direct target localization with imageguided RT is no longer optional when the dose to the prostate is escalated to improve biochemical outcomes (31) or the dose is hypofractionated with stereotactic body RT because the therapeutic ratio is decreased. In the present study, as well as in other studies (9, 11, 12, 19, 22), the variability in SV displacement appeared larger than the variability in prostate displacement, suggesting that a greater margin is necessary for the SVs than for the prostate. The AP SV displacement was as large as 20 mm in our series. Therefore, even an AP margin of 1 cm for the SVs might be inadequate in some cases. This margin would adequately compensate for AP SV variation in only 86% of treatments. Moreover, the variation in SV position might
Fig. 6. Frequency of prostate and seminal vesicle (SV) misses as function of margin size (internal organ variation only). Sup-inf = superoinferior.
Volume 71, Number 3, 2008
be dominated by the shape change of this structure. In measuring the variability of the SVs, we used the COV, which does not fully account for the potential effect of the threedimensional shape variation, which might require an even larger margin. For example, a more laterally and symmetrically extended SV might not change the COV value even though significant variation of the individual SV could occur. The effect of the bladder volume on SV variability in the AP and SI axes seems to be greater than the effect of the bladder volume on prostate variability; however, our study has corroborated previous findings that rectal filling remains the most important contributing factor (14). In addition to analyzing the effect of rectal and bladder volume changes on the SV position, we analyzed the effect of rectal and bladder volume changes on the prostate position. Our results concur with those of other investigators in that the AP prostate variability was mainly caused by rectal distension (Fig. 4) (6, 9, 11–13, 19, 23, 24, 32). Rectal distension appeared to correlate with the presence of gas. The absence of a correlation between the rectal volume changes and AP prostate variability in 27% of patients might be explained by asymmetric AP rectal volume variation, with greater increases in rectal volume in the superior than in the inferior part of the rectum, and small variations in organ volume and position. Figure 2 illustrates that for patients with the largest posterior variability (Patients 5, 9, and 14), the rectal volume during RT was less than the rectal volume on the planning CT scan 75%, 86%, and 100% of the time, respectively. As shown in Fig. 6, an empty rectum at the planning CT scan would likely have minimized the posterior shift during treatment. The data in Fig. 6 therefore supports the hypothesis of our earlier study (15), which showed the effect of rectal distension on local control. Thus, if the rectum is distended on the planning CT (as it was for Patients 5, 9, and 14), it will likely be less distended, on average, during the RT course, resulting in a posterior shift of the prostate relative to the simulation position, underdosage of the posterior portion of the prostate tumor, and a poorer clinical outcome. An anterior displacement of the prostate during RT might have only a moderate effect on local control, given the usual large anterior clinical target volume to the planning target volume margin and to the preferential location of the tumor in the posterior part of the gland (29, 30). The effect of the enema given at the planning CT scan on rectal filling and AP prostate motion is clearly shown in Figs. 1a and 2; 67% of the patients had a rectal volume that was larger during treatment than at the planning CT scan >50% of the time. Consequently, 67% of the patients had a positive mean AP prostate variability, indicating a systematic prostate displacement in the anterior direction (Fig. 1a). Other studies have shown that the use of a contrast agent or catheter in the rectum at the planning CT scan can also push the prostate anteriorly as much as 1–2 cm, potentially leading to a systematic posterior positional change during RT relative to the prostate position on the planning CT scan (14, 24, 32, 33). Although we routinely administer an enema to empty the rectum before the simulation CT
IMRT interfraction variability during prostate cancer treatment d S. J. FRANK et al.
scan, the effectiveness of the enema in producing an empty rectum did not appear to be constant. We instructed our patients to maintain regular bowel movements during RT and to refrain from becoming constipated, but stool softeners might have been unnecessary because patients might develop an increased frequency of bowel movements secondary to the effects of RT. Realistically, the degree of rectal filling and gas cannot be controlled consistently during a relatively lengthy IMRT course; thus, for daily localization, we have used the BAT ultrasound-based target localization system before every treatment fraction. In our study, the effect of bladder filling on prostate displacement appeared to be relatively small, and bladder filling did not correlate well with the prostate displacement. Although the bladder volume for each patient varied significantly during the RT course, the prostate movement in the SI direction was generally quite small, and patients with the largest SI prostate displacement showed no difference in bladder filling compared with patients with minimal SI prostate displacement. These findings are similar to those of other investigators (11, 14, 23, 24, 32) and to the findings from a previous study from our institution (5). Additionally, overfilling of the bladder at simulation is not advisable because of the lack of daily reproducibility and the potential effect on the SV dose.
819
CONCLUSION This is one of the largest single-institution studies of variability in prostate and SV position during IMRT, and our study results differ from those of previous such studies in that we performed CT scans frequently during the RT course and performed these scans just before the daily RT sessions. We were able to quantify and analyze the changes in internal organ position and volume of the pelvic organs during a protracted IMRT course. The dominant motion of the prostate and SVs occurred in the AP direction; the next greatest motion was in the SI direction. Our findings indicate that without adequate direct prostate localization or patient immobilization techniques, a 5-mm posterior margin for the prostate could prove inadequate. Rectal distension affected both prostate and SV positions, and the SVs exhibited greater variation in daily position than did the prostate. The systematic difference between the rectal and bladder volume on the planning CT scan and at the daily RT sessions deserves attention. These changes in critical organ volumes raise the question of the relevance of dose–volume histograms calculated solely on the basis of the treatment planning imaging. We are addressing the dosimetric effect of these anatomic variations during the RT course in an effort to make more accurate predictions of normal tissue tolerance.
REFERENCES 1. Pollack A, Zagars GK, Smith LG, et al. Preliminary results of a randomized radiotherapy dose-escalation study comparing 70 Gy with 78 Gy for prostate cancer. J Clin Oncol 2000;18: 3904–3911. 2. Zelefsky MJ, Leibel SA, Gaudin PB, et al. Dose escalation with three-dimensional conformal radiation therapy affects the outcome in prostate cancer. Int J Radiat Oncol Biol Phys 1998; 41:491–500. 3. Hanks GE, Hanlon AL, Epstein B, et al. Dose response in prostate cancer with 8–12 years’ follow-up. Int J Radiat Oncol Biol Phys 2002;54:427–435. 4. Pollack A, Zagars GK, Starkschall G, et al. Prostate cancer radiation dose response: Results of the M.D. Anderson phase III randomized trial. Int J Radiat Oncol Biol Phys 2002;53: 1097–1105. 5. Antolak JA, Rosen II, Childress CH, et al. Prostate target volume variations during a course of radiotherapy. Int J Radiat Oncol Biol Phys 1998;42:661–672. 6. Balter JM, Sandler HM, Lam K, et al. Measurement of prostate movement over the course of routine radiotherapy using implanted markers. Int J Radiat Oncol Biol Phys 1995;31: 113–118. 7. Crook JM, Raymond Y, Salhani D, et al. Prostate motion during standard radiotherapy as assessed by fiducial markers. Radiother Oncol 1995;37:35–42. 8. Millender LE, Aubin M, Pouliot J, et al. Daily electronic portal imaging for morbidly obese men undergoing radiotherapy for localized prostate cancer. Int J Radiat Oncol Biol Phys 2004; 59:6–10. 9. Zelefsky MJ, Crean D, Mageras GS, et al. Quantification and predictors of prostate position variability in 50 patients evaluated with multiple CT scans during conformal radiotherapy. Radiother Oncol 1999;50:225–234.
10. Stroom JC, Koper PCM, Korevaaar GA, et al. Internal organ motion in prostate cancer patients treated in prone and supine treatment position. Radiother Oncol 1999;51:237–248. 11. Beard CJ, Kijewski P, Bussiere M, et al. Analysis of prostate and seminal vesicle motion: Implications for treatment planning. Int J Radiat Oncol Biol Phys 1996;34:451–458. 12. Roeske JC, Forman JD, Mesina CF, et al. Evaluation of changes in the size and location of the prostate, seminal vesicles, bladder, and rectum during a course of external beam radiation therapy. Int J Radiat Oncol Biol Phys 1995;33:1321–1329. 13. Rudat V, Schraube P, Oetzel D, et al. Combined error of patient positioning variability and prostate motion uncertainty in 3D conformal radiotherapy of localized prostate cancer. Int J Radiat Oncol Biol Phys 1996;35:1027–1034. 14. Schild SE, Casale HE, Bellefontaine LP. Movements of the prostate due to rectal and bladder distension: Implications for radiotherapy. Med Dosim 1993;18:13–15. 15. de Crevoisier R, Tucker SL, Dong L, et al. Increased risk of biochemical and local failure in patients with distended rectum on the planning CT for prostate cancer radiotherapy. Int J Radiat Oncol Biol Phys 2005;62:965–973. 16. Chandra A, Dong L, Huang E, et al. Experience of ultrasoundbased daily prostate localization. Int J Radiat Oncol Biol Phys 2003;56:436–447. 17. Court L, Rosen I, Mohan R, et al. Evaluation of mechanical precision and alignment uncertainties for an integrated CT/LINAC system. Med Phys 2003;30:1198–1210. 18. Court LE, Dong L. Automatic registration of the prostate for computed-tomography-guided radiotherapy. Med Phys 2003; 30:2750–2757. 19. Dawson LA, Mah K, Franssen E, et al. Target position variability throughout prostate radiotherapy. Int J Radiat Oncol Biol Phys 1998;42:1155–1161.
820
I. J. Radiation Oncology d Biology d Physics
20. Roach MR, Faillace-Akazawa P, Malfatti C. Prostate volumes and organ movement defined by serial computerized tomographic scans during three-dimensional conformal radiotherapy. Radiat Oncol Invest 1997;5:187–194. 21. Mechalakos JG, Mageras GS, Zelefsky MJ, et al. Time trends in organ position and volume in patients receiving prostate threedimensional conformal radiotherapy. Radiother Oncol 2002;62: 261–265. 22. Tinger A, Michalski JM, Cheng A, et al. A critical evaluation of the planning target volume for 3-D conformal radiotherapy of prostate cancer. Int J Radiat Oncol Biol Phys 1998;42:213–221. 23. Melian E, Mageras GS, Fuks Z, et al. Variation in prostate position quantitation and implications for three-dimensional conformal treatment planning. Int J Radiat Oncol Biol Phys 1997;38:73–81. 24. Lebesque JV, Bruce AM, Guus KAP, et al. Variation in volumes, dose–volume histograms, and estimated normal tissue complication probabilities of rectum and bladder during conformal radiotherapy of T3 prostate cancer. Int J Radiat Oncol Biol Phys 1995;33:1109–1119. 25. Langen KM, Pouliot J, Anezinos C, et al. Evaluation of ultrasound-based prostate localization for image-guided radiotherapy. Int J Radiat Oncol Biol Phys 2003;57:635–644. 26. Lattanzi J, McNeeley S, Hanlon A, et al. Ultrasound-based stereotactic guidance of precision conformal external beam
Volume 71, Number 3, 2008
27. 28.
29. 30. 31.
32.
33.
radiation therapy in clinically localized prostate cancer. Urology 2000;55:73–78. Langen KM, Jones DTL. Organ motion and its management. Int J Radiat Oncol Biol Phys 2001;50:265–278. Yan D, Lockman D, Brabbins D, et al. An off-line strategy for constructing a patient-specific planning target volume in adaptive treatment process for prostate cancer. Int J Radiat Oncol Biol Phys 2000;48:289–302. Chen ME, Johnston DA, Tang K, et al. Detailed mapping of prostate carcinoma foci: Biopsy strategy implications. Cancer 2000;89:1800–1809. Greene DR, Wheeler TM, Egawa S, et al. Relationship between clinical stage and histological zone of origin in early prostate cancer: morphometric analysis. Br J Urol 1991;68:499–509. Zelefsky MJ, Chan H, Hunt M, et al. Long-term outcome of high dose intensity modulated radiation therapy for patients with clinically localized prostate cancer. J Urol 2006;176: 1415–1419. Ten Haken RK, Forman JD, Heimburger DK, et al. Treatment planning issues related to prostate movement in response to differential filling of the rectum and bladder. Int J Radiat Oncol Biol Phys 1991;20:1317–1324. Malone S, Donker R, Dahrouge S, et al. Treatment planning aids in prostate cancer: Friend or foe? Int J Radiat Oncol Biol Phys 2001;51:49–55.