Abstract
Background/Aim: To assess intrafractional prostate and patient movement using intra-prostatic fiducials and stereoscopic kilovoltage (kV) X-ray imaging in a 6-dimensional (6D) position correction protocol. To evaluate potential gains of intra-treatment repositioning with respect to treatment margins. Patients and Methods: In intensity-modulated radiotherapy of prostate cancer patients were positioned according to internal fiducials in six dimensions by the use of ExacTrac/Novalis Body™ (ET/NB) System and a robotic couch. Intrafractional displacement of both, prostate and patient were analyzed in 427 treatment fractions of 13 patients. Systematic and random components were specified and used for intra-treatment margin calculation. The potential reduction of treatment margins, and intrafractional repositioning by use of the ET Snap Verification presumed, was simulated. Results: The mean treatment duration was 14.2±2.6 min. Standard deviations (SDs) of the effective intrafractional target displacement in superior-inferior (SI) and anterior-posterior (AP) axes were 2.4 mm and 2.1 mm, respectively. Systematic errors for patient were 1.8 and 1.7 mm, and for prostate movement were 2.1 and 2.0 mm in SI and AP, respectively. The SDs of intrafractional rotation errors of the prostate around SI and left right (LR) were on average 2.2 and 3.6 degrees, respectively. Margins covering intrafractional motion were 4.5 and 4.3 mm in SI and AP without intrafractional correction and were estimated to 2.9 mm and 2.8 mm in SI and AP, respectively for simulated intra-treatment intervention. Conclusion: After positioning according to fiducials, intrafractional motion is significant for treatment margins. Despite correcting rotational deviations by couch angulation, the systematic error for the component of prostate motion was somewhat larger than that of patient displacement. Intrafractional correction could be useful in reducing treatment margins.
Prostate cancer is estimated to account for the highest incidence of cancer in men with 29% (n=241,740) and at 9% (n=28.170), for the second most common cause of cancer death in men in the U.S. (1).
Radiotherapy (RT) is an effective therapeutic option for early prostate cancer. Randomized controlled trials have demonstrated a significant improvement in clinical outcome in terms of better biochemical disease-free survival with dose escalation (2, 3). However, dose escalation may be associated with an increased risk of gastrointestinal morbidity, if delivered with conventional RT techniques (4, 5). The implementation of image-guided RT modalities has enabled the delivery of highly conformal doses in terms of achieving better clinical outcome, while limiting the dose to critical organs, and thus, without increasing injury rates of normal tissue (6, 7).
There is important individual variation among patients with respect to the physiological prostate motion and its consequences for dose distribution and calculation. However, this motion might affect the delivery of an effective high dose to the prostate and thus clinical outcome. The difference in the accuracy of available localization methods currently used to adjust for prostate motion is important in terms of the quality of RT delivery. Organ motion is, as a physiological factor, independent on the trajectory of RT and it is traditionally divided into intrafractional and interfractional motion.
Intrafractional motion occurs after the initial setup and during the process of RT delivery. A large body of data on intrafractional position changes has been gained by tracking methods. Various motion types, mostly translational displacements, have been described and their impact on treatment margins has been evaluated (8-10). Intrafractional rotational errors, despite their relevant dosimetric importance in simulation protocols (11, 12), had to be corrected by means of table translations until recently. Robotic couch positioning devices, initially developed for radiosurgery of the brain, have come into use for repositioning in RT for extracranial lesions (13, 14) allowing for rotational corrections around lateral and longitudinal axes. Promising adaptive corrective strategies have been simulated (12) but are not yet available for routine use and, despite mechanical limitations, rotation correction by couch angulation prevails (15).
We evaluated the intrafractional translational and rotational displacement of patient and prostate and their components with respect to systematic and random errors with focus on the correction of the rotational deviations. In addition, the possible gain of intrafractional position adjustment was quantified.
Patients and Methods
Setup details and therapy. Inter and intrafractional target motion was measured in a non-randomized cohort of 13 men undergoing definitive RT for localized prostate cancer. Each patient had two gold fiducial markers (VisiCoil, Radiomed Corp, Tyngsboro, MA, USA) implanted. Before planning computed tomography, the patients were requested to fill their bladder and were given an enema to empty the rectum. Treatment was performed with empty rectum under monitoring of bladder filling. Standardized head support, knee support (Kneefix™, Sinmed Radiotherapy Products, Reeuwijk, the Netherlands) and a foot restraint preventing rotation of the legs were used for patient immobilization. Radiotherapy was planned and delivered by an intensity-modulated technique as previously described (16). Couch movements with the Varian Exact Couch (Varian Medical Systems, Milpitas, CA, USA) were performed from outside the treatment room with a Robotics Tilt Module™ device. The correction angles were limited to ±3 degrees around SI and ±4 degrees around LR. The ET Snap Verification is a positional verification with fast monoscopic X-ray snapshot imaging allowing interbeam verification X-rays to be performed with deviations easily being corrected by remote couch control. This procedure adds on average one minute to the total treatment time.
Imaging and analysis of displacements. Details have been published elsewhere (16). Initial positioning took place using laser alignment according to skin marks. Prior to each fraction, two non-coplanar oblique isocentric (stereoscopic) kV X-ray images of the bony structures and markers were acquired. The correction vectors resulting from automatic fusion with the digitally reconstructed radiographs (DRRs) according to the pelvic bone were recorded, as well as the correction vectors gained from a (manual) fusion of the fiducials, with their difference vector representing the interfractional error. If displacements exceeded the tolerance levels of 2 mm or one degree, the patient was shifted and rotated according to the correcting transformation set gained from the markers. A second pair of X-ray images was acquired and registered with the corresponding DRRs for verification. The remaining residual errors, to some extent representing displacements resulting from the secondary patient motion (17), were recorded and, if above the action level, again corrected. After treatment delivery, another set of X-rays was performed. The correction vectors after automatic fusion according to the pelvic bone were recorded again as well as the vectors gained from a new manual fusion of the markers. The intrafractional displacement vectors for the pelvic bones were calculated as the sum of the final bone displacement vectors and the interfractional marker displacements, reduced by the residual errors. The total intrafractional target misalignment was calculated as the dislocation of the markers, reduced by the residual error. The intrafractional prostate motion was calculated by subtracting the intrafractional displacement of bones from the total displacement. The calculation of the rotational errors was performed similarly.
Statistics. For each of the 13 patients we calculated the mean of the deviations (in six coordinates) representing the systematic patient error, after summing up their SDs resulting to the systematic group error. The six group random errors were calculated as the SD of the root mean square sum of individual random errors, each gained by subtracting the systematic patient error from each of the recorded deviations out of every fraction. Translation-positive values were left in LR, superior in SI and anterior in AP, rotation-positive values were anterior around X, right around Z and counterclockwise around Y. The 3DV was calculated as the root from the sum of the quadratic mean displacements in the three directions. The treatment margins were calculated according the method described by van Herk (17). For comparison purposes, a correction in the middle of the treatment time was simulated with the assumption that the systematic component could be reset to zero keeping the random error unchanged, thus gaining a kind of time-weighted displacement similar to the method detailed by Mutanga (18) for calculation of the effective time. The statistical analysis was performed using JMP v7.0 (SAS Institute, Cary, NC, USA). Tests for sub-groups were performed using the paired t-test.
Results
We performed the analysis for 13 patients with an average of 33 fractions per patient, totaling in 427 fractions. The mean duration of treatment (pre-therapeutic imaging, therapy and post-therapeutic imaging) was 14.2±2.6 min (median=14.0 min).
For the comparison of the translational errors analyzed refer to Table I. The total intrafractional displacement of the prostate according to isocenter (and beam source) in SI and AP was as high as about two thirds of the interfraction displacement with SDs of 2.4 mm and 2.1 mm. The systematic components were (1SD) as 1.1 mm and 1.2 mm respectively. The intrafractional contribution of the prostate (relative to bone) and patient movement (relative to isocenter) were in about the same range with SD of 1.6 and 1.7 and 2.1 and 2.0, respectively. Residual errors were low.
In Table II we present the intrafractional rotational errors. A total of 345 software-assisted corrections were made. The largest rotational errors occurred around the LR axis with a positive mean towards anterior rotation.
For translational error, the only correlation with mean therapy duration (>14 versus <14 min) was found in SI direction (p=0.04). There were no correlations with intrafractional rotational errors in regard to time or fraction number.
For each patient, treatment margins were computed (Table III). Simulating intrafractional repositioning, the (intrafractional) treatment margin was reduced from 4.5 to 2.9 in SI and from 3.9 to 2.4 in AP, which was statistically significant (p<0.01).
Translational errors for interfractional, residual, total intrafractional, intrafractional patient and prostate error in (mm), including means, (SDs), (3DVs), (Σ) and (σ) for 427 fractions.
Discussion
Main findings. In this study, we compared intrafractional patient and prostate position displacement in a 6D correction setting as assessed by two stereoscopic X-rays taken before and after IMRT. Summarizing our main findings, total intrafractional displacement was about (1SD) 2 mm in SI and AP, with a systematic component of about 1 mm. Both patient and prostate movements contributed to about the half of the intrafractional translational and rotational errors. Midsession correction further reduces the safety margins.
Results in comparison with literature. With respect to overall intrafractional displacement, Skarsgard reported SDs of 1.9 and 2.1 mm in SI and AP (20), comparable to our data of 2.4 and 2.1 mm. Their study also gives a literature overview on published SDs in the three coordinates, data mainly gained by pre- and post-treatment imaging, with our data lying in the upper half of the published values.
With concern regarding the random and systematic components of the total intrafractional displacement, a large patient group of 108 patients was evaluated by Mutanga (19). On the average their earlier (11.1 min vs. 14.2 min) “post-therapeutic” kV images, the systematic errors proved to be larger than ours, in SI and AP of 1.4 and 1.7 mm (versus 1.1 and 1.2 mm in our study). Random errors with SDs of 1.9 and 1.8 (vs. 2.2 and 1.8 in our study) were different than ours. The study of another group of 427 patients evaluated by Kotte (21) showed for both components of total (uncorrected) uncertainty comparably smaller SDs of 0.8 and 0.8 for systematic and 1.4 and 1.6 mm for random deviation in SI and AP, respectively. These values were gained at different time points, within a treatment time of only 5 to 7 min.
Rotational errors for interfractional, residual, total intrafractional, intrafractional patient and prostate error in degree including means, (SDs), (Σ) and (σ) for 427 fractions.
Analyzing intrafractional prostate and patient movements separately, there are data available from Soete for comparison (14). Here, the systematic component for organ motion in a time frame of about 7.5±3 min was 2.4 mm and 1.6 mm in SI and AP (versus 2.1 mm and 2.0 mm in our study); the random one 2.4 mm and 1.6 mm (versus 2.6 mm and 2.2 mm). Boda-Heggemann studying a smaller group, reported SDs of 1.6 mm and 4.3 mm for patient and 3.7 and 4.8 for prostate in the respective directions (22).
Li evaluated tracking results of 775 fractions from 105 patients and reported a mean intrafractional prostate rotation around LR of 1.5° (SD 5.2°), mainly toward the anterior direction, similar to our results of 1.8° and 3.6° (15). Modeling rotation of the prostate yielded a translation displacement of 0.3 mm for every degree of rotational error. With respect to intrafractional rotation errors, the same authors calculated significant margin reduction with adequate control of rotation.
Calculation of intrafractional margins. For the entire group of patients, the standard deviations of systematic errror (Σ) and random error(σ) are given in (LR), (SI) and (AP) directions The margins for (PTV) according the prescription 2.5 Σ + 0.7 σ (12) were calculated in the three directions. Treatment margins were computed after setup according to internal fiducials. Margins with intrafractional repositioning were simulated by supposed correction in the middle of the treatment time with the assumption that the systematic component could be reset to zero, keeping the random error unchanged.
The data reported by Soete confirm these findings (13). Correction of translational and rotational errors in a group of patients proved favorable for systematic and random error setup of the pelvic bone relative to the skin-drawn isocenter in comparison with correction of translational errors alone. From these data, it is concluded that translation and rotation of the prostate cannot be analyzed separately and rotation could lead to translational errors.
Intratreatment adjustment. The applied method of IMRT required comparably long treatment times. Yet from tracking data, it is well-known, that displacement of the prostate is increased in dependence on the treatment time. In a study by Shimizu 35% of patients evaluated required repositioning due to a movement >5 mm at least once in a time frame of 10 min (23). Midsession adjustment has been reported to be feasible with only a minimal elongation of the treatment time (22). For interbeam adjustment in IMRT, Litzenberg et al. estimated the reduction of margins by 0.3 and 0.8 mm in SI and AP (24). Simulating one correction step in the middle of the fraction, in our study we have calculated the possible reduction in treatment margins as being about 1.5 mm.
Limitations. There are limitations to the method of data collection in our study, which mainly concern the lack of continuous monitoring. We analyzed intrafractional motion using images acquired before treatment and images acquired after treatment. Noel reported that the sensitivity of pre- and post-treatment imaging in determining intrafractional prostate motion greater than 5 mm for 30 s was low at about 50% (25). It is true that motions occurring during the time frame between the two image acquisitions, that have already resolved by the end of treatment, could have been missed. However, statistical data can be derived from tracking data in the publication of Li (15), which show that the parameters Σ and σ to calculate treatment margins are similar with and without intrafractional tracking, at least for the vast majority of patients.
There could be also objections to the described method of correcting rotational angles. In the review of Boda-Heggemann (26) it is stated that robotic couches can perform rotational corrections, but the range of rotational motion may be too small for rotations of the prostate, where rotations up to 15° are observed. In our study, the possible correction angles were limited to ±3 degrees around SI and ±4 degrees around LR and from our data, in agreement with prostate rotation histograms published by Li (15), it is clear that a significant proportion of prostate rotations could be corrected. To address the matter of patient fixation when moving the treatment table, Guckenberger argued that correcting rotational deviations of patient and prostate around LR could induce systematic errors by patient drift (27). We noticed that in fact our errors in SI are somewhat larger than in AP, as opposed to most results compiled in the review by Skarsgard (20). The possible role of improved patient fixation has yet to be evaluated. In a review about image-guided RT of the prostate, Boda-Heggemann mentioned the problem of possible counteraction of the patient if correcting rotation of errors by table angulation about the LR (27). Yet in our study this problem was resolved by correcting the secondary patient motions, as is indicated by the routinely taken verification images in a second step after positioning.
Conclusion
Correction of prostate translational and rotational errors is feasible, however intrafractional errors vary in a substantial range. Improved positioning of the patient or intrafractional intervention should be considered if delivering IMRT with treatment times of about 15 min.
Footnotes
-
Conflicts of Interest
All Authors confirm that there is no conflict of interest with regard to any financial and personal relationships with other people or organizations.
- Received June 22, 2013.
- Revision received July 10, 2013.
- Accepted July 12, 2013.
- Copyright© 2013 International Institute of Anticancer Research (Dr. John G. Delinassios), All rights reserved
References
In this issue
Jump to section
Related Articles
Cited By...
- No citing articles found.