Geometrical variability of esophageal tumors and its implications for accurate radiation therapy - Summary

(1)

UvA-DARE is a service provided by the library of the University of Amsterdam (https://dare.uva.nl)

Geometrical variability of esophageal tumors and its implications for accurate

radiation therapy

Jin, P.

Publication date

2019

Document Version

Other version

License

Other

Link to publication

Citation for published version (APA):

Jin, P. (2019). Geometrical variability of esophageal tumors and its implications for accurate

radiation therapy.

General rights

It is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s) and/or copyright holder(s), other than for strictly personal, individual use, unless the work is under an open content license (like Creative Commons).

Disclaimer/Complaints regulations

If you believe that digital publication of certain material infringes any of your rights or (privacy) interests, please let the Library know, stating your reasons. In case of a legitimate complaint, the Library will make the material inaccessible and/or remove it from the website. Please Ask the Library: https://uba.uva.nl/en/contact, or a letter to: Library of the University of Amsterdam, Secretariat, Singel 425, 1012 WP Amsterdam, The Netherlands. You will be contacted as soon as possible.

(2)

527155-L-bw-Jin 527155-L-bw-Jin 527155-L-bw-Jin 527155-L-bw-Jin Processed on: 18-12-2018 Processed on: 18-12-2018 Processed on: 18-12-2018

Processed on: 18-12-2018 PDF page: 195PDF page: 195PDF page: 195PDF page: 195

Esophageal cancer has been estimated worldwide as the eighth most common cancer. The inci-dence of esophageal cancer is also rapidly rising especially for esophageal adenocarcinoma. Al-though surgery is the primary treatment modality, an improved survival rate is associated with preoperative chemoradiation therapy (CRT) compared with surgery alone. For inoperable or unresectable esophageal cancer, definitive CRT is also applied with a curative intent. One major concern in esophageal cancer radiation therapy (RT) is the geometrical uncertainty of the clinical target volume (CTV). Therefore, it is essential to apply a safety margin to the CTV to compose the planning target volume (PTV). The advances in imaging and dose delivery techniques allow better conformal dose to the target and more sparing of organs at risk (OARs). However, the soft-tissue contrast in the planning computed tomography (pCT) and in-room cone-beam computed tomography (CBCT) is limited. Consequently, the inter- and intra-observer variation of the tar-get delineations remain large and the interfractional tumor position variation during the four- or five-week treatment cannot easily be determined, resulting in a large CTV-to-PTV margin. The intrafractional target mobility induced by respiratory motion also hinders improving the accuracy of dose delivery. Moreover, the presence and variation of gastrointestinal gas volume could cause undesired degradation of the dose distribution in RT of esophageal cancer.

At the department of Radiation Oncology in our clinic, the implantation of radio-opaque fidu-cial markers has become a standard procedure for the esophageal cancer patient who has poorly visible tumor borders in CT or an inconsistent extent in CT and endoscopy since 2014. The fidu-cial markers are placed at the tumor borders under the guidance of endoscopy or endoscopic ul-trasound. The use of these markers is expected to reduce the inter- and intra-observer variation in delineation of the gross tumor volume (GTV) in the pCT. Therefore, the inter- and intra-observer variation of the GTV delineation with and without markers were compared in Chapter 2. The overall inter- and intra-observer variation of the GTV delineation with markers were significantly smaller than those without markers. The reductions of inter- and intra-observer variation were more evident in the longitudinal direction than in the radial direction. Since the delineation vari-ation is usually considered as a systematic error in the margin recipe, implanting markers at the cranial and caudal borders of the primary tumor can evidently reduce the CTV-to-PTV margin.

Apart from the delineation uncertainty, the interfractional tumor position uncertainty is

(3)

other main source of the systematic and random errors. Because bony anatomy is commonly used as the surrogate for the target in setup verification, there is still a residual variation of the tumor po-sition relative to bony anatomy. With the help of the markers and the 3D-CBCT acquired over the treatment course, the interfractional tumor position variation relative to bony anatomy was quan-tified in Chapter 3. After the rigid bony anatomy-based registration between the pCT and the CBCT, differences in position of the markers in the CBCT relative to those in the pCT were mea-sured in the left–right (LR), cranial–caudal (CC), and anterior–posterior (AP) directions. These were compared between the three directions and between four regions: the proximal, middle, dis-tal esophagus, and the cardia. The systematic and random errors derived from the interfractional tumor position variation were most pronounced in the CC direction and in the cardia. These sig-nify that when bony anatomy-based registration is used for setup verification of esophageal cancer patients, the systematic and random errors incorporated in the margin should be anisotropic and region-specific. Because the fiducial markers were expected to be a better surrogate for the target, we also performed rigid marker-based registration to minimize the residual setup errors. However, we encountered many difficulties in both automatic and manual marker-based registrations due to the large deformation of the elongated tumor. Based on our experience, bony anatomy-based reg-istration is the most reliable approach for setup verification in esophageal cancer RT. Further, the good visibility of markers in the 3D-CBCT can allow the visual inspection of whether the target is located within the PTV during the treatment course.

In addition to bony anatomy, the carina has also been used as the surrogate for thoracic tumors in setup verification. Because of its proximity to the esophagus, a smaller interfractional tumor position variation relative to the carina was expected compared to the bony anatomy, especially for tumors located in the middle esophagus. In Chapter 4, the carina-based setup verification was compared with the bony anatomy-based setup verification in terms of systematic and random errors and resulting margins. Using an approach similar to the one presented in Chapter 3, the interfractional tumor position variation relative to the carina was quantified. Only for the distally located markers we found significantly larger systematic and random errors in the AP direction using the carina-based registration than using the bony anatomy-based registration. The observed slight reduction in the systematic errors in the middle esophagus implied that the carina-based registration might be favorable for patients with middle esophageal tumors. However, esophageal tumors are rarely confined only to the middle esophagus. Moreover, most esophageal primary tumors are located in the distal esophagus and cardia. Therefore, there might be no clinical benefit of using the carina-based setup verification. Currently, the bony anatomy-based registration is still endorsed for setup verification for the majority of patients.

The variation of gastrointestinal gas volumes is another source of interfractional geometrical uncertainty. Because gas has a different mass density than soft tissue, it can induce dose attenu-ation if it appears in the irradiated region. Accordingly, the dosimetric effect of gastrointestinal gas pockets was investigated in Chapter 5. By calculating the fractional dose using retrospectively

(4)

synthesized fractional CT scans, a linear model was fitted between the gas volume changes relative to the initial volume measured in the pCT and the dose difference relative to the planned dose in the CTV. This linear model can be exploited for estimating the underdose or overdose in the case of a decreased or increased gas volume, respectively. To mitigate the target underdose induced by a decreased gas volume, we proposed the use of density override (DO) in the gas pockets, i.e., overriding the relative mass density (RMD) of the gas to water in the pCT, at treatment plan-ning. To avoid the overdose induced by an increased gas volume, re-planning could serve as the solution. A plan-selection strategy and a single-plan strategy were proposed and compared with respect to the accumulated dose. The plan-selection strategy required a plan library consisting of plans with different RMDs in the gas for the DO and monitoring of the gas volume changes; the single-plan strategy, on the contrary, relied on a single plan with a certain RMD for the DO. In our patient group, the plan-selection strategy showed good target coverage and minimum over-doses compared with the single-plan strategy. The overall performance of the single-plan strategy using the DO with RMD value of 0.5 and 1 effectively prevented the gas-induced underdose but resulted in undesired overdose. On selecting the plan-selection or the single-plan strategy, the tradeoff between the workload and the clinical relevance should be assessed in the clinic.

Apart from interfractional tumor position variation, intrafractional tumor motion, primarily induced by respiration, plays an important role in the geometrical uncertainties for esophageal cancer RT. We therefore quantified the respiration-induced tumor motion using 4D-CT and im-planted fiducial markers in Chapter 6. Similar to the interfractional esophageal tumor position variation relative to bony anatomy, the respiration-induced tumor motion amplitude was also found most pronounced in the CC direction compared with those in the LR and AP directions. It was also larger in the distal esophagus and the cardia than in the proximal and middle esophagus. Moreover, the respiration-induced tumor motion amplitude varied among patients. These find-ings suggest that when the passive motion management approach is used, anisotropic and region-specific margins are required to compensate for the respiration-induced tumor motion and that these margins should be individualized for the patients.

Nonetheless, the margin derived from a single 4D-CT to account for respiration-induced tu-mor motion would not be sufficient in case of a large interfractional variability of the respiration-induced tumor motion. To investigate its interfractional variability, fiducial markers and CBCT can be used to quantify the daily respiration-induced tumor motion. However, the 4D-CBCT has not yet been implemented in our clinical practice for esophageal cancer RT. Alterna-tively, a 4D-CBCT can be retrospectively reconstructed based on the available projection images that constructed the 3D-CBCT. As a result, in Chapter 7, different CBCT acquisition settings with a 4D reconstruction were tested for marker visibility and accuracy of motion quantification using a dynamic phantom with fiducial markers. Although the visibility of markers quantified by the contrast-to-noise ratio significantly varied among these 4D-CBCT scans acquired with differ-ent settings, the mean error in quantification of the marker motion was smaller than 1.4 mm. This

(5)

error was also comparable with the error in motion quantification using 4D-CT. Therefore, we be-lieve that it is feasible and highly accurate to quantify the respiration-induced tumor motion using the markers present in the retrospectively reconstructed 4D-CBCT.

To investigate the interfractional variability of respiration-induced tumor motion, we applied retrospectively reconstructed 4D-CBCT to quantify the marker motion over the treatment course in the patients in Chapter 8. These quantifications were also done in the three directions and four regions of the esophagus. The quantified respiration-induced tumor motion was in line with the results found in Chapter 6. However, the interfractional variation of the respiration-induced tumor motion in terms of amplitude and trajectory shape was found to be limited with a mean variation up to 1.4 mm. Moreover, the differences in the amplitude of respiration-induced tu-mor motion between the 4D-CT and 4D-CBCT measurements were also small, with respect to a mean absolute difference up to 1.0 mm. These findings suggest that quantification using a single 4D-CT could provide a sufficient prediction of the respiration-induced tumor motion during the treatment course.

Although the 4D-CT is preferred to be used for taking the respiration-induced tumor motion into account in treatment planning, it is under discussion how to incorporate this motion uncer-tainty into the CTV-to-PTV margin. For esophageal cancer RT, the internal target volume (ITV) concept, i.e., completely enveloping the CTVs in all breathing phases, is mostly used. However, the mid-position (MidP) strategy, i.e., applying a probability-based margin on the mid-position of all 10 breathing phases to account for the respiration-induced tumor motion, can potentially reduce the PTV volume and thereby reduce the dose to OARs. Accordingly, for patients with esophageal cancer, the MidP strategy was compared with the ITV strategy in terms of dosimet-ric benefits in Chapter 9. The MidP-CT was reconstructed using deformable image registration on the 10 breathing phases of the 4D-CT. Based on the quantified delineation uncertainty, inter-fractional tumor position variation, and patient-specific respiration-induced tumor motion, the anisotropic and region-specific CTV-to-PTV and ITV-to-PTV margins were designed. Subse-quently, the accumulated 4D-dose associated with the MidP and ITV strategies were compared with respect to the target coverage and dose to OARs. Without compromising the target cover-age, the MidP strategy showed a reduction of approximately 10% in the dose to OARs, compared with the ITV strategy. Based on our patient group, we recommend the use of the MidP strategy with anisotropic and region-specific margins for the sake of reduced dose to OARs in esophageal cancer RT.

In Chapter 10, we discussed other clinically available solutions to mitigate the geometrical un-certainties in esophageal cancer RT. These solutions are highly subject to the availability of some techniques and facilities in each clinic. Although we proposed the CTV-to-PTV margins in the previous chapters, these margins were made on the basis of our clinical settings. In clinical prac-tice, we recommend using fiducial markers for the fractionated RT of esophageal cancer to facil-itate the target delineation and daily inspection of the target position. When the bony

(6)

based setup verification and the MidP strategy are applied, anisotropic and region-specific CTV-to-PTV margins are preferred. For patients with considerable gastrointestinal gas pockets, the use of DO in treatment planning is a simple and sound approach to serve the purpose of mitigating the gas-induced dosimetric effect. Compared to the CBCT-guided photon therapy, the magnetic resonance (MR) guided RT is advanced in image quality and proton therapy is superior in dose shaping. Currently, MR-guided RT and proton therapy are therefore becoming the most promis-ing modalities of RT.