- Email: [email protected]
Position shifts and volume changes of pelvic and para-aortic nodes during IMRT for patients with cervical cancer
Radiotherapy and Oncology xxx (2014) xxx–xxx
Contents lists available at ScienceDirect
Radiotherapy and Oncology journal homepage: www.thegreenjournal.com
Original article
Position shifts and volume changes of pelvic and para-aortic nodes during IMRT for patients with cervical cancer q Maaike G.A. Schippers a, Gijsbert H. Bol a, Astrid A.C. de Leeuw a, Uulke A. van der Heide a, Bas W. Raaymakers a, Helena M. Verkooijen b,c, Ina M. Jürgenliemk-Schulz a,⇑ a
Department of Radiation Oncology; b Department of Radiology; and c Julius Center for Health Sciences and Primary Care, University Medical Centre Utrecht, The Netherlands
a r t i c l e
i n f o
Article history: Received 21 June 2013 Received in revised form 10 March 2014 Accepted 3 May 2014 Available online xxxx Keywords: Cervix cancer Lymph nodes Image-guided radiotherapy Target volume changes ITV margins
a b s t r a c t Background and purpose: To evaluate volume changes and position shifts and their contribution to treatment margins of pelvic and para-aortic lymph nodes during Intensity Modulated Radiation Therapy (IMRT) for advanced cervical cancer. Materials and methods: Seventeen patients with visible nodes on MR images underwent T2-weighted MR scans before and weekly during the course of IMRT. Thirty-nine pelvic and para-aortic nodes were delineated on all scans. Margins accommodating for volume and position changes were taken from the boundaries of the nodal volumes in the six main directions. Results: Nodal volume regression from the pre-treatment situation to week 4 was 58% on average (range: 11.7% increase to 100% decrease). Nodal volumes partly increased between the pre-treatment scans and the scans in weeks 1–3, but in week 4 all nodes except one had regressed. Around the nodal volumes manually derived ITV margins accounting for volume changes and position shifts of 7.0, 4.0, 7.0, 8.0, 7.0 and 9.0 mm to the medial, lateral, anterior, posterior, superior and inferior directions were needed to cover 95% of all nodes. Conclusions: We used weekly MR scans to derive inhomogeneous margins that accommodate for nodal volume and position changes during treatment. These margins should be taken into consideration when planning external beam radiotherapy (EBRT) boosts, especially for highly conformal boosting techniques. Ó 2014 Elsevier Ireland Ltd. All rights reserved. Radiotherapy and Oncology xxx (2014) xxx–xxx
Since the introduction of image guided radiotherapy for gynecological cancer, effort has been put into optimizing radiation dose delivery. Besides the dosimetric gain of combined conformal adaptive approaches for EBRT and brachytherapy clinical gain, in terms of increased disease control, has been published [1,2]. However, highly conformal treatment can only be delivered safely with adequate treatment margins. These margins have to accommodate for mobility and deformation of the treatment targets during the course of EBRT, as well as for treatment set-up uncertainties. Several publications have given us more knowledge on shifts and volume changes of the primary cervical tumor and the surrounding organs [3–8]. However, affected lymph nodes are also part of the radiation target and have to be taken into account in the process of treatment planning. Nodal disease is not easily controlled and regional recurrences, due to either geometrical misses or dose failure, might contribute to disappointing treatment outcomes [9].
q
Poster presentation at the ESTRO anniversary meeting, London 2011.
⇑ Corresponding author.
E-mail address: [email protected] (I.M. Jürgenliemk-Schulz).
This study was performed in order to evaluate position shifts and volume changes of pelvic nodes during the course of EBRT and their contribution to treatment margins. Materials and methods Patients Seventeen patients with cervical cancer, treated at our department between January 2005 and September 2006, were enrolled in this study. Staging was performed according to the International Federation of Gynecology and Obstetrics (FIGO) classification. Five patients had FIGO stage 1B, 3 stage 2A, 4 stage 2B, 2 stage 3A and 3 stage 3B tumors. All patients had visible lymph nodes on pre-treatment MR images and a total of thirty-nine lymph nodes were delineated in these 17 patients. All nodes had a short axis diameter >5 mm and morphological changes on MRI and were therefore considered potentially pathologic. These nodes were not necessarily diagnosed as containing metastatic disease. We delineated 24 external iliac nodes, 1 common iliac node, 8 internal iliac nodes, 2 obturator
http://dx.doi.org/10.1016/j.radonc.2014.05.013 0167-8140/Ó 2014 Elsevier Ireland Ltd. All rights reserved.
Please cite this article in press as: Schippers MGA et al. Position shifts and volume changes of pelvic and para-aortic nodes during IMRT for patients with cervical cancer. Radiother Oncol (2014), http://dx.doi.org/10.1016/j.radonc.2014.05.013
2
Nodal changes during IMRT
nodes and 4 para-aortal nodes. The location of the nodes was defined using the guidelines published by Taylor et al. [10]. For contouring the individual nodes we did not take potential extra-capsular extension into account. This microscopic component is included in the electively irradiated draining lymphatic regions. Patients were treated according to our clinical protocol with external beam IMRT, delivering 45 Gy in 25 fractions of 1.8 Gy to the primary tumor and elective lymph node planning target volume (PTV), the last mentioned being delineated according to the contouring atlas by Taylor et al. [10]. Patients were treated in supine position with a knee cushion in order to prevent any rotation of the hip. No other immobilization devices were used and no measurements were applied to control bladder or rectum filling. Additionally, patients were treated with two times PDR-brachytherapy using tandem ovoid applicators. The prescribed dose to point A was 17.4 Gy for each application. However, the dose was adjusted depending on the constraints to rectum and bladder. After the second brachytherapy fraction, 7 patients received external nodal boosting due to considered nodal pathology. Currently, based on morphologic MRI information, we would define more nodes as being potentially pathologic and would therefore boost more often. We preferred sequential boosting taking nodal regression and contribution of the brachytherapy dose to the nodes into account. Sixteen patients received concurrent weekly chemotherapy (40 mg/m2 Cisplatin intravenously). Imaging, image registration and delineation Imaging procedures for delineation and treatment planning purposes were performed according to the protocol as described by Van de Bunt et al. [4]. A planning CT scan (CT aura, Philips Medical Systems, Best, The Netherlands) was performed in all patients. Contiguous 3-mm slices were taken from the iliac crest to the ischial tuberosities. To accurately delineate the regions of interest, MR images were generated using a 1.5-T MRI scanner (Gyroscan NT Intera; Philips Medical Systems, Best, The Netherlands). Patients were scanned in the treatment position using a flat tabletop insert. Because of its superior soft-tissue contrast MRI was used to delineate the nodes [8,11–13]. All 17 patients underwent MR imaging before the start of treatment and weekly during IMRT. The first week MRI was taken after about 11 Gy, the second week MRI after about 20 Gy, the third week MRI after about 27 Gy and the fourth week after 36 Gy. In all but 3 patients the pre-treatment MR was taken no longer than 1 month before the MR in the first week. Images were acquired according to the following protocol: axial T2-weighted 6.6-mm-thick slices of the whole abdomen and pelvis, axial and sagittal T2-weighted 4.5-mm-thick slices
using a Synbody coil from the body of L5 to the ischial tuberosities, axial T1-weighted spectral fat saturation inversion recovery (SPIR) 4.5-mm-thick slices, and T1-weighted SPIR slices with gadolinium contrast from the pelvis. A resident radiation oncologist delineated all lymph nodes on the T2 weighted MRI scans before and weekly during radiotherapy. Contouring was performed on the axially and sagittally sliced datasets using our in house developed contouring tool Volumetool [14]. A radiation oncologist specialized in gynecologic oncology and a radiologist, in case of doubt, reviewed the derived contours. The analysis of internal motion and volume changes was performed using the weekly sets of delineations, all registered to the master CT scan. Registration was performed by first extracting the bony anatomy from the CT dataset using HU (hounsfield unit)-based thresholding. The bony anatomy was then registered to the MRI datasets by using a mutual information registration algorithm (VTK CISG Registration Toolkit; Kitware, York). The resulting registrations were not influenced by changes due to internal organ motion and tumor regression, since only the bony anatomy was used for registration. After the registration, structures contoured on the five MR images were transferred to the CT. Volume changes Individual volumes were calculated from the node delineations on the pre-treatment and the weekly MR images using an algorithm present in Volumetool [14]. Nodal volumes of the subsequent weekly MR images were compared with the volumes of the pre-treatment MRI. ITV margins In order to define margins that account for node position shifts, volume changes during treatment and image registration errors (margins to which we will refer as internal target volume (ITV) margins further on), we used a manual approach identical to the one described by Van de Bunt et al. [4] for primary cervical tumors. For each of the four weekly MRI scans inhomogeneous margins were generated around the pre-treatment node delineations that allowed encompassing the boundaries of the weekly delineations in the six main directions (medial, lateral, anterior, posterior, superior and inferior) (Fig. 1). A margin of zero in a given direction does not imply the absence of a shift, only that the boundary of the new volume lies within the pre-treatment delineation volume. Using the data of the 4 intra-treatment MR scans for all our patients resulted in 39 times 4 margin sizes in all 6 directions. For our analyses we excluded the 5% outliers so that the given ITV margins encompass 95% of all nodes.
Fig. 1. Pre-treatment transversal (a) and sagittal (b) T2-weighted MR image of a patient with an enlarged lymph node in the left obturator region. The delineations of the node on the pre-treatment scan (yellow line), scan in week 1 (green line), week 2 (orange line), week 3 (purple line) and week 4 (blue line) are all overlaid on the pretreatment MR image. The red line represents the manually derived margin to encompass the node in all weeks.
Please cite this article in press as: Schippers MGA et al. Position shifts and volume changes of pelvic and para-aortic nodes during IMRT for patients with cervical cancer. Radiother Oncol (2014), http://dx.doi.org/10.1016/j.radonc.2014.05.013
3
M.G.A. Schippers et al. / Radiotherapy and Oncology xxx (2014) xxx–xxx
Statistics
Table 2 Derived ITV margins (mm) in the six main directions for treatment weeks 1–4.
A paired non-parametric Wilcoxon test and a student t-test were used to test for statistical differences in volumes and margins during the course of treatment. P values of less than 5% were considered significant. Bonferroni correction was also applied to correct for multiple testing. All analyses were performed with SPSS predictive analytics software.
39 nodes
Medial Lateral Anterior Posterior Superior Inferior
Results
95% ITV margin (mean, median) MRI week 1
MRI week 2
MRI week 3
MRI week 4
7.0 3.0 7.0 4.0 7.0 7.0
6.0 3.0 5.0 8.0 6.0 9.0
3.0 4.0 4.0 8.0 6.0 7.0
5.0 3.0 3.0 7.0 0.0 6.0
(2.0, (0.8, (1.7, (1.6, (1.7, (1.4,
2.0) 0.0) 1.0) 1.0) 0.0) 0.0)
(1.5, (0.5, (1.1, (2.0, (0.9, (1.7,
1.0) 0.0) 0.0) 0.0) 0.0) 0.0)
(0.9, (0.9, (0.7, (1.8, (1.0, (1.2,
0.0) 0.0) 0.0) 1.0) 0.0) 0.0)
(0.9, (0.5, (0.4, (1.8, (0.1, (0.8,
0.0) 0.0) 0.0) 0.0) 0.0) 0.0)
The 95% ITV margin in a given direction encompasses 95% of all nodes in that direction from the pre-treatment MRI scan.
Nodal volume changes Forty nodes were visible and could be delineated on the pre-treatment, and intra-treatment MR images in weeks 1, 2, 3 and 4 respectively. Due to its exceptional volume (30.4 cm3) we considered one node to be an outlier and excluded it from the analysis. Thirty-nine nodes were taken into analyses. Compared to the pre-treatment MRI 16 nodes had larger volumes in week 1, 10 nodes in week 2, 5 nodes in week 3 and 1 node even in week 4. This last node progressed quickly and the patient was diagnosed with distant metastases only 1 month after having finished radiation therapy. On average, in week 4 there was a 58% reduction (range: 11.7% increase to 100% decrease) compared to the pre-treatment nodal volume. The calculated volumes showed a wide variety, with average volumes on the pre-treatment MRI of 2.1 and 1.9, 1.7, 1.4 and 1.0 cm3 for weeks 1, 2, 3 and 4, respectively (Table 1). Node regression was significant when comparing nodal volumes of weeks 2, 3 and 4 with the ones on pre-treatment MRIs (p = 0.003, p < 0.001 and p < 0.001 respectively, taking Bonferroni correction into account (p-values <0.0125 were considered significant)). The difference was not significant between the nodal volumes in week 1 and before treatment (p = 0.157). Manually derived ITV margins When comparing the ITV margin sizes measured on the first week MR scans with the ones in week 4, we found a significant reduction in average size in the medial (from 2.0 to 0.9 mm, p = 0.005), anterior (from 1.7 to 0.4 mm, p = 0.001) and superior (from 1.7 to 0.1 mm, p = 0.001) directions. There was no significant difference in average margins in the lateral (from 0.8 to 0.5 mm), posterior (from 1.6 to 1.8 mm) and inferior (1.4 to 0.8 mm) directions. See Table 2 and Fig. 2 for weekly variations. Generic margin sizes were also calculated from the numbers derived in all 4 weeks, allowing inclusion of at least 95% of the delineated nodes at all times. Margins of 7.0, 4.0, 7.0, 8.0, 7.0 and 9.0 mm to the medial, lateral, anterior, posterior, superior and inferior directions around the pre-treatment delineations met this requirement. In order to cover 100% of all nodal volumes, 7 mm was sufficient in the lateral direction and 9 mm in all of the others. Discussion We investigated volume changes and possible shifts of pelvic and para-aortic nodes, during the course of (chemo-) radiation in
Fig. 2. Box plots showing ITV margins in the six main directions with manually derived margins that account for position shifts and volume changes; box: 5–95%, line: maximum value.
cervical cancer patients. Control of primary tumors is increasing due to 3D image guided adaptive radiotherapy and especially brachytherapy [1,2]. However, regional control of nodal disease is still a matter of concern. Pelvic and para-aortic recurrences are described in an essential proportion of the patients (Beadle et al. [9] found up to 17%), having been treated with conventional radiotherapy techniques without surgery on suspicious nodes or lymphatic spaces. In 66% of the cases with regional recurrence a component of marginal failure was observed, usually just superior to the radiation field, suggesting a deficiency in treated volume. Recurrences also occurred inside the irradiated volumes, suggesting a deficiency in delivered dose. The problem may partly be due to suboptimal diagnostic procedures and partly because of suboptimal treatment. Image guided conformal adaptive treatment optimization may help improve this, but before introducing higher conformity one needs to know how treatment targets behave during the course of treatment. We found that on average lymph nodes shrink and change their position. However, the order of magnitude of volume regression and mobility is different from what we know about the primary tumor. Several studies have been published describing the regression rate of primary cervical cancers. Repeated CT or MRI images have been performed during the course of EBRT in these studies. In the study of Van de Bunt et al. [3] regression of the primary tumor was 46% on average, when comparing pre-treatment MR images
Table 1 Absolute nodal volumes (cm3) and nodal changes during IMRT.
*
39 nodes
Mean (cm3) [min, max, median]
Percentage (%)
Mean decrease (%) pre-treatment MRI (range)
Pre-treatment MRI MRI week 1 MRI week 2 MRI week 3 MRI week 4
2.1 1.9 1.7 1.4 1.0
100 90.4 75.6 59.2 42.0
– 9.6 ( 69.8*/68.5) 24.4 ( 48.3*/69.8) 40.8 ( 46.6*/89.5) 58.0 ( 11.7*/100)
(0.3, 8.1, 1.0) (0.2, 8.2, 0.9) (0.2, 9.7, 0.7) (0.1, 10.7, 0.5) (0, 5.5, 0.3)
A negative value reflecting an increase in size compared to the pre-treatment volume.
Please cite this article in press as: Schippers MGA et al. Position shifts and volume changes of pelvic and para-aortic nodes during IMRT for patients with cervical cancer. Radiother Oncol (2014), http://dx.doi.org/10.1016/j.radonc.2014.05.013
4
Nodal changes during IMRT
with the images taken in week four of EBRT. These data are in the same range as the ones published by Lee et al. [15]. As far as we know, no other studies published yet describe the regression rate of lymph nodes during the course of EBRT. As expected, we found the nodal volumes decreasing in time (on average 58% in week 4), but additionally we observed that nodal volumes in the first weeks of treatment often increase. This temporary volume expansion may be due to edema, inflammatory reactions, and probably delineation uncertainties, due to the slice thickness of the MR images in comparison to the relatively small volumes of the nodes. Also the time interval between the pre-treatment MRI and MRI in week 1 has to be taken into consideration. Without treatment, affected nodes will grow and the effect of therapy is often not yet visible after 1 week. However, at the end, all nodes except one were smaller than at the time of diagnosis. The second part of this study deals with position changes of regional nodes during EBRT. From several studies we know how primary cervical tumors and the uterus move during EBRT and how this is influenced by different filling situations of the surrounding organs in the pelvis [4]. Especially the influence of changes in bladder filling has been investigated in the last years [16]. These studies have demonstrated that position shifts of up to 3 cm or more can occur in cases of cervical tumors, especially in the anterior–posterior direction. Lateral or cranio-caudal shifts are somewhat smaller, but still in the order of 1–2 cm. In our study, we found that lymph nodes can also change their position, but not in centimeters but in millimeters. Margins of up to 9 mm were necessary in the different directions in order to cover at least 95% of all nodal volumes, with the smallest distance being in the lateral direction, where the bony anatomy restricts position shifts. Most of the more extended margins are due to nodal volume and position changes in the first 2 weeks of treatment. After 4 weeks of IMRT, derived margins are smaller due to the observed nodal volume regressions. Many of the median margins in a given direction are zero, but this does not imply an absence of shifts, but means that the surfaces of the new volumes were located inside the pre-treatment ones. The average and median margin sizes are smaller than the maximal margins needed to encompass 95% of all nodes. This is because several nodes need more than the average margin individually, due to volume expansion and more elevated position shifts during the course of treatment. The distribution is skewed to the right and due to several outliers the 95% margins are larger than the averages. Both, the pattern of volume changes and the position shifts have to be kept in mind when external node boosting is part of an adaptive treatment approach. If nodal target volumes (GTVs) change during time, clinical target volumes (CTVs) will also change and internal target volumes (ITVs) have to accommodate accordingly. In the end, these findings will affect PTV boost volumes and the amount of surrounding tissue irradiated. Volume and position changes are part of the uncertainty in radiotherapy and contribute to the margin concept, which can vary for different nodal boost approaches. In case of simultaneous integration, the boosts are delivered during the course of elective EBRT. One has to keep in mind that essential regression, or position shifts of nodal volumes, may increase radiation of the surrounding tissues in higher doses than initially planned. Monitoring during treatment with repeated MR imaging and re-planning if necessary might help to reduce this risk. For sequential boosting the situation is different. Nodal boosts are planned after EBRT and brachytherapy and adaptation of boosted volumes should be considered following the pattern of nodal volume regression. Note that part of the ITV margins as described above is also caused by the image registration errors between MRI and CT (on average 61.4 mm in our clinical practice [4]). In order to generate adequate PTVs another margin component should be included in
the CTV–PTV margin. This has to account for external uncertainties and is for example influenced by the chosen treatment strategies (serial versus integrated boosting), position verification procedures (on-line or off-line and bony anatomy versus soft tissue contrast) and external treatment set up uncertainties (treatment machine and beam). The definite PTV margins also depend on registration errors and set-up uncertainties, and can be center or even machine specific. The results of this study are not valid to estimate the margins needed to treat draining lymphatic regions electively, as the visible nodes with their volume changes and position shifts are only part of the total system. Conclusion During IMRT volume changes and position shifts of pelvic and para-aortal nodes occur, although to a lesser extent than the primary cervical tumors. These changes have to be taken into account when deciding on appropriate ITV margins, especially when highly conformal and simultaneous integrated treatment planning strategies are used for nodal boosting. Conflict of interest No actual or potential conflicts of interest exist. References [1] Potter R, Georg P, Dimopoulos JC, et al. Clinical outcome of protocol based image (MRI) guided adaptive brachytherapy combined with 3D conformal radiotherapy with or without chemotherapy in patients with locally advanced cervical cancer. Radiother Oncol 2011;100:116–23. [2] Potter R, Dimopoulos J, Georg P, et al. Clinical impact of MRI assisted dose volume adaptation and dose escalation in brachytherapy of locally advanced cervix cancer. Radiother Oncol 2007;83:148–55. [3] Van de Bunt L, van der Heide UA, Ketelaars M, et al. Conventional, conformal, and intensity-modulated radiation therapy treatment planning of external beam radiotherapy for cervical cancer: The impact of tumor regression. Int J Radiat Oncol Biol Phys 2006;64:189–96. [4] Van de Bunt L, Jurgenliemk-Schulz IM, de Kort GA, et al. Motion and deformation of the target volumes during IMRT for cervical cancer: what margins do we need? Radiother Oncol 2008;88:233–40. [5] Chan P, Dinniwell R, Haider MA, et al. Inter- and intrafractional tumor and organ movement in patients with cervical cancer undergoing radiotherapy: a cinematic-MRI point-of-interest study. Int J Radiat Oncol Biol Phys 2008;70:1507–15. [6] Huh SJ, Park W, Han Y. Interfractional variation in position of the uterus during radical radiotherapy for cervical cancer. Radiother Oncol 2004;71:73–9. [7] Taylor A, Powell ME. An assessment of interfractional uterine and cervical motion: implications for radiotherapy target volume definition in gynaecological cancer. Radiother Oncol 2008;88:250–7. [8] Kerkhof EM, van der Put RW, Raaymakers BW, et al. Intrafraction motion in patients with cervical cancer: The benefit of soft tissue registration using MRI. Radiother Oncol 2009;93:115–21. [9] Beadle BM, Jhingran A, Yom SS, et al. Patterns of regional recurrence after definitive radiotherapy for cervical cancer. Int J Radiat Oncol Biol Phys 2010;76:1396–403. [10] Taylor A, Rockall AG, Powell ME. An atlas of the pelvic lymph node regions to aid radiotherapy target volume definition. Clin Oncol (R Coll Radiol) 2007;19:542–50. [11] Mayr NA, Tali ET, Yuh WT, et al. Cervical cancer: application of MR imaging in radiation therapy. Radiology 1993;189:601–8. [12] Barillot I, Reynaud-Bougnoux A. The use of MRI in planning radiotherapy for gynaecological tumours. Cancer Imaging 2006;6:100–6. [13] Thomas L, Chacon B, Kind M, et al. Magnetic resonance imaging in the treatment planning of radiation therapy in carcinoma of the cervix treated with the four-field pelvic technique. Int J Radiat Oncol Biol Phys 1997;37:827–32. [14] Bol GH, Kotte AN, van der Heide UA, et al. Simultaneous multi-modality ROI delineation in clinical practice. Comput Methods Programs Biomed 2009;96:133–40. [15] Lee CM, Shrieve DC, Gaffney DK. Rapid involution and mobility of carcinoma of the cervix. Int J Radiat Oncol Biol Phys 2004;58:625–30. [16] Ahamad A, D’Souza W, Salehpour M, et al. Intensity-modulated radiation therapy after hysterectomy: comparison with conventional treatment and sensitivity of the normal-tissue-sparing effect to margin size. Int J Radiat Oncol Biol Phys 2005;62:1117–24.
Please cite this article in press as: Schippers MGA et al. Position shifts and volume changes of pelvic and para-aortic nodes during IMRT for patients with cervical cancer. Radiother Oncol (2014), http://dx.doi.org/10.1016/j.radonc.2014.05.013
Contents lists available at ScienceDirect
Radiotherapy and Oncology journal homepage: www.thegreenjournal.com
Original article
Position shifts and volume changes of pelvic and para-aortic nodes during IMRT for patients with cervical cancer q Maaike G.A. Schippers a, Gijsbert H. Bol a, Astrid A.C. de Leeuw a, Uulke A. van der Heide a, Bas W. Raaymakers a, Helena M. Verkooijen b,c, Ina M. Jürgenliemk-Schulz a,⇑ a
Department of Radiation Oncology; b Department of Radiology; and c Julius Center for Health Sciences and Primary Care, University Medical Centre Utrecht, The Netherlands
a r t i c l e
i n f o
Article history: Received 21 June 2013 Received in revised form 10 March 2014 Accepted 3 May 2014 Available online xxxx Keywords: Cervix cancer Lymph nodes Image-guided radiotherapy Target volume changes ITV margins
a b s t r a c t Background and purpose: To evaluate volume changes and position shifts and their contribution to treatment margins of pelvic and para-aortic lymph nodes during Intensity Modulated Radiation Therapy (IMRT) for advanced cervical cancer. Materials and methods: Seventeen patients with visible nodes on MR images underwent T2-weighted MR scans before and weekly during the course of IMRT. Thirty-nine pelvic and para-aortic nodes were delineated on all scans. Margins accommodating for volume and position changes were taken from the boundaries of the nodal volumes in the six main directions. Results: Nodal volume regression from the pre-treatment situation to week 4 was 58% on average (range: 11.7% increase to 100% decrease). Nodal volumes partly increased between the pre-treatment scans and the scans in weeks 1–3, but in week 4 all nodes except one had regressed. Around the nodal volumes manually derived ITV margins accounting for volume changes and position shifts of 7.0, 4.0, 7.0, 8.0, 7.0 and 9.0 mm to the medial, lateral, anterior, posterior, superior and inferior directions were needed to cover 95% of all nodes. Conclusions: We used weekly MR scans to derive inhomogeneous margins that accommodate for nodal volume and position changes during treatment. These margins should be taken into consideration when planning external beam radiotherapy (EBRT) boosts, especially for highly conformal boosting techniques. Ó 2014 Elsevier Ireland Ltd. All rights reserved. Radiotherapy and Oncology xxx (2014) xxx–xxx
Since the introduction of image guided radiotherapy for gynecological cancer, effort has been put into optimizing radiation dose delivery. Besides the dosimetric gain of combined conformal adaptive approaches for EBRT and brachytherapy clinical gain, in terms of increased disease control, has been published [1,2]. However, highly conformal treatment can only be delivered safely with adequate treatment margins. These margins have to accommodate for mobility and deformation of the treatment targets during the course of EBRT, as well as for treatment set-up uncertainties. Several publications have given us more knowledge on shifts and volume changes of the primary cervical tumor and the surrounding organs [3–8]. However, affected lymph nodes are also part of the radiation target and have to be taken into account in the process of treatment planning. Nodal disease is not easily controlled and regional recurrences, due to either geometrical misses or dose failure, might contribute to disappointing treatment outcomes [9].
q
Poster presentation at the ESTRO anniversary meeting, London 2011.
⇑ Corresponding author.
E-mail address: [email protected] (I.M. Jürgenliemk-Schulz).
This study was performed in order to evaluate position shifts and volume changes of pelvic nodes during the course of EBRT and their contribution to treatment margins. Materials and methods Patients Seventeen patients with cervical cancer, treated at our department between January 2005 and September 2006, were enrolled in this study. Staging was performed according to the International Federation of Gynecology and Obstetrics (FIGO) classification. Five patients had FIGO stage 1B, 3 stage 2A, 4 stage 2B, 2 stage 3A and 3 stage 3B tumors. All patients had visible lymph nodes on pre-treatment MR images and a total of thirty-nine lymph nodes were delineated in these 17 patients. All nodes had a short axis diameter >5 mm and morphological changes on MRI and were therefore considered potentially pathologic. These nodes were not necessarily diagnosed as containing metastatic disease. We delineated 24 external iliac nodes, 1 common iliac node, 8 internal iliac nodes, 2 obturator
http://dx.doi.org/10.1016/j.radonc.2014.05.013 0167-8140/Ó 2014 Elsevier Ireland Ltd. All rights reserved.
Please cite this article in press as: Schippers MGA et al. Position shifts and volume changes of pelvic and para-aortic nodes during IMRT for patients with cervical cancer. Radiother Oncol (2014), http://dx.doi.org/10.1016/j.radonc.2014.05.013
2
Nodal changes during IMRT
nodes and 4 para-aortal nodes. The location of the nodes was defined using the guidelines published by Taylor et al. [10]. For contouring the individual nodes we did not take potential extra-capsular extension into account. This microscopic component is included in the electively irradiated draining lymphatic regions. Patients were treated according to our clinical protocol with external beam IMRT, delivering 45 Gy in 25 fractions of 1.8 Gy to the primary tumor and elective lymph node planning target volume (PTV), the last mentioned being delineated according to the contouring atlas by Taylor et al. [10]. Patients were treated in supine position with a knee cushion in order to prevent any rotation of the hip. No other immobilization devices were used and no measurements were applied to control bladder or rectum filling. Additionally, patients were treated with two times PDR-brachytherapy using tandem ovoid applicators. The prescribed dose to point A was 17.4 Gy for each application. However, the dose was adjusted depending on the constraints to rectum and bladder. After the second brachytherapy fraction, 7 patients received external nodal boosting due to considered nodal pathology. Currently, based on morphologic MRI information, we would define more nodes as being potentially pathologic and would therefore boost more often. We preferred sequential boosting taking nodal regression and contribution of the brachytherapy dose to the nodes into account. Sixteen patients received concurrent weekly chemotherapy (40 mg/m2 Cisplatin intravenously). Imaging, image registration and delineation Imaging procedures for delineation and treatment planning purposes were performed according to the protocol as described by Van de Bunt et al. [4]. A planning CT scan (CT aura, Philips Medical Systems, Best, The Netherlands) was performed in all patients. Contiguous 3-mm slices were taken from the iliac crest to the ischial tuberosities. To accurately delineate the regions of interest, MR images were generated using a 1.5-T MRI scanner (Gyroscan NT Intera; Philips Medical Systems, Best, The Netherlands). Patients were scanned in the treatment position using a flat tabletop insert. Because of its superior soft-tissue contrast MRI was used to delineate the nodes [8,11–13]. All 17 patients underwent MR imaging before the start of treatment and weekly during IMRT. The first week MRI was taken after about 11 Gy, the second week MRI after about 20 Gy, the third week MRI after about 27 Gy and the fourth week after 36 Gy. In all but 3 patients the pre-treatment MR was taken no longer than 1 month before the MR in the first week. Images were acquired according to the following protocol: axial T2-weighted 6.6-mm-thick slices of the whole abdomen and pelvis, axial and sagittal T2-weighted 4.5-mm-thick slices
using a Synbody coil from the body of L5 to the ischial tuberosities, axial T1-weighted spectral fat saturation inversion recovery (SPIR) 4.5-mm-thick slices, and T1-weighted SPIR slices with gadolinium contrast from the pelvis. A resident radiation oncologist delineated all lymph nodes on the T2 weighted MRI scans before and weekly during radiotherapy. Contouring was performed on the axially and sagittally sliced datasets using our in house developed contouring tool Volumetool [14]. A radiation oncologist specialized in gynecologic oncology and a radiologist, in case of doubt, reviewed the derived contours. The analysis of internal motion and volume changes was performed using the weekly sets of delineations, all registered to the master CT scan. Registration was performed by first extracting the bony anatomy from the CT dataset using HU (hounsfield unit)-based thresholding. The bony anatomy was then registered to the MRI datasets by using a mutual information registration algorithm (VTK CISG Registration Toolkit; Kitware, York). The resulting registrations were not influenced by changes due to internal organ motion and tumor regression, since only the bony anatomy was used for registration. After the registration, structures contoured on the five MR images were transferred to the CT. Volume changes Individual volumes were calculated from the node delineations on the pre-treatment and the weekly MR images using an algorithm present in Volumetool [14]. Nodal volumes of the subsequent weekly MR images were compared with the volumes of the pre-treatment MRI. ITV margins In order to define margins that account for node position shifts, volume changes during treatment and image registration errors (margins to which we will refer as internal target volume (ITV) margins further on), we used a manual approach identical to the one described by Van de Bunt et al. [4] for primary cervical tumors. For each of the four weekly MRI scans inhomogeneous margins were generated around the pre-treatment node delineations that allowed encompassing the boundaries of the weekly delineations in the six main directions (medial, lateral, anterior, posterior, superior and inferior) (Fig. 1). A margin of zero in a given direction does not imply the absence of a shift, only that the boundary of the new volume lies within the pre-treatment delineation volume. Using the data of the 4 intra-treatment MR scans for all our patients resulted in 39 times 4 margin sizes in all 6 directions. For our analyses we excluded the 5% outliers so that the given ITV margins encompass 95% of all nodes.
Fig. 1. Pre-treatment transversal (a) and sagittal (b) T2-weighted MR image of a patient with an enlarged lymph node in the left obturator region. The delineations of the node on the pre-treatment scan (yellow line), scan in week 1 (green line), week 2 (orange line), week 3 (purple line) and week 4 (blue line) are all overlaid on the pretreatment MR image. The red line represents the manually derived margin to encompass the node in all weeks.
Please cite this article in press as: Schippers MGA et al. Position shifts and volume changes of pelvic and para-aortic nodes during IMRT for patients with cervical cancer. Radiother Oncol (2014), http://dx.doi.org/10.1016/j.radonc.2014.05.013
3
M.G.A. Schippers et al. / Radiotherapy and Oncology xxx (2014) xxx–xxx
Statistics
Table 2 Derived ITV margins (mm) in the six main directions for treatment weeks 1–4.
A paired non-parametric Wilcoxon test and a student t-test were used to test for statistical differences in volumes and margins during the course of treatment. P values of less than 5% were considered significant. Bonferroni correction was also applied to correct for multiple testing. All analyses were performed with SPSS predictive analytics software.
39 nodes
Medial Lateral Anterior Posterior Superior Inferior
Results
95% ITV margin (mean, median) MRI week 1
MRI week 2
MRI week 3
MRI week 4
7.0 3.0 7.0 4.0 7.0 7.0
6.0 3.0 5.0 8.0 6.0 9.0
3.0 4.0 4.0 8.0 6.0 7.0
5.0 3.0 3.0 7.0 0.0 6.0
(2.0, (0.8, (1.7, (1.6, (1.7, (1.4,
2.0) 0.0) 1.0) 1.0) 0.0) 0.0)
(1.5, (0.5, (1.1, (2.0, (0.9, (1.7,
1.0) 0.0) 0.0) 0.0) 0.0) 0.0)
(0.9, (0.9, (0.7, (1.8, (1.0, (1.2,
0.0) 0.0) 0.0) 1.0) 0.0) 0.0)
(0.9, (0.5, (0.4, (1.8, (0.1, (0.8,
0.0) 0.0) 0.0) 0.0) 0.0) 0.0)
The 95% ITV margin in a given direction encompasses 95% of all nodes in that direction from the pre-treatment MRI scan.
Nodal volume changes Forty nodes were visible and could be delineated on the pre-treatment, and intra-treatment MR images in weeks 1, 2, 3 and 4 respectively. Due to its exceptional volume (30.4 cm3) we considered one node to be an outlier and excluded it from the analysis. Thirty-nine nodes were taken into analyses. Compared to the pre-treatment MRI 16 nodes had larger volumes in week 1, 10 nodes in week 2, 5 nodes in week 3 and 1 node even in week 4. This last node progressed quickly and the patient was diagnosed with distant metastases only 1 month after having finished radiation therapy. On average, in week 4 there was a 58% reduction (range: 11.7% increase to 100% decrease) compared to the pre-treatment nodal volume. The calculated volumes showed a wide variety, with average volumes on the pre-treatment MRI of 2.1 and 1.9, 1.7, 1.4 and 1.0 cm3 for weeks 1, 2, 3 and 4, respectively (Table 1). Node regression was significant when comparing nodal volumes of weeks 2, 3 and 4 with the ones on pre-treatment MRIs (p = 0.003, p < 0.001 and p < 0.001 respectively, taking Bonferroni correction into account (p-values <0.0125 were considered significant)). The difference was not significant between the nodal volumes in week 1 and before treatment (p = 0.157). Manually derived ITV margins When comparing the ITV margin sizes measured on the first week MR scans with the ones in week 4, we found a significant reduction in average size in the medial (from 2.0 to 0.9 mm, p = 0.005), anterior (from 1.7 to 0.4 mm, p = 0.001) and superior (from 1.7 to 0.1 mm, p = 0.001) directions. There was no significant difference in average margins in the lateral (from 0.8 to 0.5 mm), posterior (from 1.6 to 1.8 mm) and inferior (1.4 to 0.8 mm) directions. See Table 2 and Fig. 2 for weekly variations. Generic margin sizes were also calculated from the numbers derived in all 4 weeks, allowing inclusion of at least 95% of the delineated nodes at all times. Margins of 7.0, 4.0, 7.0, 8.0, 7.0 and 9.0 mm to the medial, lateral, anterior, posterior, superior and inferior directions around the pre-treatment delineations met this requirement. In order to cover 100% of all nodal volumes, 7 mm was sufficient in the lateral direction and 9 mm in all of the others. Discussion We investigated volume changes and possible shifts of pelvic and para-aortic nodes, during the course of (chemo-) radiation in
Fig. 2. Box plots showing ITV margins in the six main directions with manually derived margins that account for position shifts and volume changes; box: 5–95%, line: maximum value.
cervical cancer patients. Control of primary tumors is increasing due to 3D image guided adaptive radiotherapy and especially brachytherapy [1,2]. However, regional control of nodal disease is still a matter of concern. Pelvic and para-aortic recurrences are described in an essential proportion of the patients (Beadle et al. [9] found up to 17%), having been treated with conventional radiotherapy techniques without surgery on suspicious nodes or lymphatic spaces. In 66% of the cases with regional recurrence a component of marginal failure was observed, usually just superior to the radiation field, suggesting a deficiency in treated volume. Recurrences also occurred inside the irradiated volumes, suggesting a deficiency in delivered dose. The problem may partly be due to suboptimal diagnostic procedures and partly because of suboptimal treatment. Image guided conformal adaptive treatment optimization may help improve this, but before introducing higher conformity one needs to know how treatment targets behave during the course of treatment. We found that on average lymph nodes shrink and change their position. However, the order of magnitude of volume regression and mobility is different from what we know about the primary tumor. Several studies have been published describing the regression rate of primary cervical cancers. Repeated CT or MRI images have been performed during the course of EBRT in these studies. In the study of Van de Bunt et al. [3] regression of the primary tumor was 46% on average, when comparing pre-treatment MR images
Table 1 Absolute nodal volumes (cm3) and nodal changes during IMRT.
*
39 nodes
Mean (cm3) [min, max, median]
Percentage (%)
Mean decrease (%) pre-treatment MRI (range)
Pre-treatment MRI MRI week 1 MRI week 2 MRI week 3 MRI week 4
2.1 1.9 1.7 1.4 1.0
100 90.4 75.6 59.2 42.0
– 9.6 ( 69.8*/68.5) 24.4 ( 48.3*/69.8) 40.8 ( 46.6*/89.5) 58.0 ( 11.7*/100)
(0.3, 8.1, 1.0) (0.2, 8.2, 0.9) (0.2, 9.7, 0.7) (0.1, 10.7, 0.5) (0, 5.5, 0.3)
A negative value reflecting an increase in size compared to the pre-treatment volume.
Please cite this article in press as: Schippers MGA et al. Position shifts and volume changes of pelvic and para-aortic nodes during IMRT for patients with cervical cancer. Radiother Oncol (2014), http://dx.doi.org/10.1016/j.radonc.2014.05.013
4
Nodal changes during IMRT
with the images taken in week four of EBRT. These data are in the same range as the ones published by Lee et al. [15]. As far as we know, no other studies published yet describe the regression rate of lymph nodes during the course of EBRT. As expected, we found the nodal volumes decreasing in time (on average 58% in week 4), but additionally we observed that nodal volumes in the first weeks of treatment often increase. This temporary volume expansion may be due to edema, inflammatory reactions, and probably delineation uncertainties, due to the slice thickness of the MR images in comparison to the relatively small volumes of the nodes. Also the time interval between the pre-treatment MRI and MRI in week 1 has to be taken into consideration. Without treatment, affected nodes will grow and the effect of therapy is often not yet visible after 1 week. However, at the end, all nodes except one were smaller than at the time of diagnosis. The second part of this study deals with position changes of regional nodes during EBRT. From several studies we know how primary cervical tumors and the uterus move during EBRT and how this is influenced by different filling situations of the surrounding organs in the pelvis [4]. Especially the influence of changes in bladder filling has been investigated in the last years [16]. These studies have demonstrated that position shifts of up to 3 cm or more can occur in cases of cervical tumors, especially in the anterior–posterior direction. Lateral or cranio-caudal shifts are somewhat smaller, but still in the order of 1–2 cm. In our study, we found that lymph nodes can also change their position, but not in centimeters but in millimeters. Margins of up to 9 mm were necessary in the different directions in order to cover at least 95% of all nodal volumes, with the smallest distance being in the lateral direction, where the bony anatomy restricts position shifts. Most of the more extended margins are due to nodal volume and position changes in the first 2 weeks of treatment. After 4 weeks of IMRT, derived margins are smaller due to the observed nodal volume regressions. Many of the median margins in a given direction are zero, but this does not imply an absence of shifts, but means that the surfaces of the new volumes were located inside the pre-treatment ones. The average and median margin sizes are smaller than the maximal margins needed to encompass 95% of all nodes. This is because several nodes need more than the average margin individually, due to volume expansion and more elevated position shifts during the course of treatment. The distribution is skewed to the right and due to several outliers the 95% margins are larger than the averages. Both, the pattern of volume changes and the position shifts have to be kept in mind when external node boosting is part of an adaptive treatment approach. If nodal target volumes (GTVs) change during time, clinical target volumes (CTVs) will also change and internal target volumes (ITVs) have to accommodate accordingly. In the end, these findings will affect PTV boost volumes and the amount of surrounding tissue irradiated. Volume and position changes are part of the uncertainty in radiotherapy and contribute to the margin concept, which can vary for different nodal boost approaches. In case of simultaneous integration, the boosts are delivered during the course of elective EBRT. One has to keep in mind that essential regression, or position shifts of nodal volumes, may increase radiation of the surrounding tissues in higher doses than initially planned. Monitoring during treatment with repeated MR imaging and re-planning if necessary might help to reduce this risk. For sequential boosting the situation is different. Nodal boosts are planned after EBRT and brachytherapy and adaptation of boosted volumes should be considered following the pattern of nodal volume regression. Note that part of the ITV margins as described above is also caused by the image registration errors between MRI and CT (on average 61.4 mm in our clinical practice [4]). In order to generate adequate PTVs another margin component should be included in
the CTV–PTV margin. This has to account for external uncertainties and is for example influenced by the chosen treatment strategies (serial versus integrated boosting), position verification procedures (on-line or off-line and bony anatomy versus soft tissue contrast) and external treatment set up uncertainties (treatment machine and beam). The definite PTV margins also depend on registration errors and set-up uncertainties, and can be center or even machine specific. The results of this study are not valid to estimate the margins needed to treat draining lymphatic regions electively, as the visible nodes with their volume changes and position shifts are only part of the total system. Conclusion During IMRT volume changes and position shifts of pelvic and para-aortal nodes occur, although to a lesser extent than the primary cervical tumors. These changes have to be taken into account when deciding on appropriate ITV margins, especially when highly conformal and simultaneous integrated treatment planning strategies are used for nodal boosting. Conflict of interest No actual or potential conflicts of interest exist. References [1] Potter R, Georg P, Dimopoulos JC, et al. Clinical outcome of protocol based image (MRI) guided adaptive brachytherapy combined with 3D conformal radiotherapy with or without chemotherapy in patients with locally advanced cervical cancer. Radiother Oncol 2011;100:116–23. [2] Potter R, Dimopoulos J, Georg P, et al. Clinical impact of MRI assisted dose volume adaptation and dose escalation in brachytherapy of locally advanced cervix cancer. Radiother Oncol 2007;83:148–55. [3] Van de Bunt L, van der Heide UA, Ketelaars M, et al. Conventional, conformal, and intensity-modulated radiation therapy treatment planning of external beam radiotherapy for cervical cancer: The impact of tumor regression. Int J Radiat Oncol Biol Phys 2006;64:189–96. [4] Van de Bunt L, Jurgenliemk-Schulz IM, de Kort GA, et al. Motion and deformation of the target volumes during IMRT for cervical cancer: what margins do we need? Radiother Oncol 2008;88:233–40. [5] Chan P, Dinniwell R, Haider MA, et al. Inter- and intrafractional tumor and organ movement in patients with cervical cancer undergoing radiotherapy: a cinematic-MRI point-of-interest study. Int J Radiat Oncol Biol Phys 2008;70:1507–15. [6] Huh SJ, Park W, Han Y. Interfractional variation in position of the uterus during radical radiotherapy for cervical cancer. Radiother Oncol 2004;71:73–9. [7] Taylor A, Powell ME. An assessment of interfractional uterine and cervical motion: implications for radiotherapy target volume definition in gynaecological cancer. Radiother Oncol 2008;88:250–7. [8] Kerkhof EM, van der Put RW, Raaymakers BW, et al. Intrafraction motion in patients with cervical cancer: The benefit of soft tissue registration using MRI. Radiother Oncol 2009;93:115–21. [9] Beadle BM, Jhingran A, Yom SS, et al. Patterns of regional recurrence after definitive radiotherapy for cervical cancer. Int J Radiat Oncol Biol Phys 2010;76:1396–403. [10] Taylor A, Rockall AG, Powell ME. An atlas of the pelvic lymph node regions to aid radiotherapy target volume definition. Clin Oncol (R Coll Radiol) 2007;19:542–50. [11] Mayr NA, Tali ET, Yuh WT, et al. Cervical cancer: application of MR imaging in radiation therapy. Radiology 1993;189:601–8. [12] Barillot I, Reynaud-Bougnoux A. The use of MRI in planning radiotherapy for gynaecological tumours. Cancer Imaging 2006;6:100–6. [13] Thomas L, Chacon B, Kind M, et al. Magnetic resonance imaging in the treatment planning of radiation therapy in carcinoma of the cervix treated with the four-field pelvic technique. Int J Radiat Oncol Biol Phys 1997;37:827–32. [14] Bol GH, Kotte AN, van der Heide UA, et al. Simultaneous multi-modality ROI delineation in clinical practice. Comput Methods Programs Biomed 2009;96:133–40. [15] Lee CM, Shrieve DC, Gaffney DK. Rapid involution and mobility of carcinoma of the cervix. Int J Radiat Oncol Biol Phys 2004;58:625–30. [16] Ahamad A, D’Souza W, Salehpour M, et al. Intensity-modulated radiation therapy after hysterectomy: comparison with conventional treatment and sensitivity of the normal-tissue-sparing effect to margin size. Int J Radiat Oncol Biol Phys 2005;62:1117–24.
Please cite this article in press as: Schippers MGA et al. Position shifts and volume changes of pelvic and para-aortic nodes during IMRT for patients with cervical cancer. Radiother Oncol (2014), http://dx.doi.org/10.1016/j.radonc.2014.05.013