10 14338 - Ijpt-21-00043 1
10 14338 - Ijpt-21-00043 1
10 14338 - Ijpt-21-00043 1
Abstract
Purpose: Finite proton range affords improved dose conformality of radiation therapy
when patient regions-of-interest geometries are well characterized. Substantial changes
in patient anatomy necessitate re-planning (RP) to maintain effective, safe treatment.
Regularly planned verification scanning (VS) is performed to ensure consistent treatment
quality. Substantial resources, however, are required to conduct an effective proton plan
verification program, which includes but is not limited to, additional computed
tomography (CT) scanner time and dedicated personnel: radiation therapists, medical
physicists, physicians, and medical dosimetrists.
Materials and Methods: Verification scans (VSs) and re-plans (RPs) of 711 patients
treated with proton therapy between June 2015 and June 2018 were studied. All
treatment RP was performed with the intent to maintain original plan integrity and
coverage. The treatments were classified by anatomic site: brain, craniospinal, bone,
spine, head and neck (H&N), lung or chest, breast, prostate, rectum, anus, pelvis,
Submitted 05 Nov 2021
Accepted 19 May 2022 esophagus, liver, abdomen, and extremity. Within each group, the dates of initial
Published 28 June 2022 simulation scan, number of VSs, number of fractions completed at the time of VS, and
Corresponding Author: Yue- the frequency of RP were collected. Data were analyzed in terms of rate of RP and
Houng Hu, PhD individual likelihood of RP.
Department of Radiation
Oncology Results: A total of 2196 VSs and 201 RPs were performed across all treatment sites.
Division of Medical Physics H&N and lung or chest disease sites represented the largest proportion of plan
and Biophysics modifications in terms of rate of re-plan (RoR: 54% and 58%, respectively) and individual
Brigham and Women’s
Hospital, Dana-Farber Cancer likelihood of RP on a per patient basis (likelihood of RP [RP%]: 46% and 39%,
Institute, and Harvard Medical respectively). These sites required RP beyond 4 weeks of treatment, suggesting
School
75 Francis St continued benefit for frequent, periodic VS. Disease sites in the lower pelvis
Boston, MA 02115, USA demonstrated a low yield for RP per VS (0.01-0.02), suggesting that decreasing VS
Phone: þ1 (617) 732-6310 frequency, particularly late in treatment, may be reasonable.
yuhu@bwh.harvard.edu
Conclusions: A large degree of variation in RoR and individual RP% was observed
Original Article between anatomic treatment sites. The present retrospective analysis provides data to
DOI help develop anatomic site–based VS protocols.
10.14338/IJPT-21-00043.1
Keywords: IMPT; re-planning; verification scanning
*
cc Copyright
Distributed under
Creative Commons CC-BY
OPEN ACCESS
http://theijpt.org
The rate of re-planning in proton therapy
Introduction
While finite proton range potentially allows for superior dose conformality in comparison to other modalities of radiation
therapy, delivery is inherently sensitive to changes in patient geometry. Several approaches to adaptive radiation therapy have
been proposed to address changes in patient anatomy throughout the course of proton treatment. Adaptive therapy solutions
are often categorized as either online [1]—where plans may be adjusted at the time of treatment, based on daily imaging
(typically treatment image guidance) to account for interfraction changes in patient anatomy—or offline, where plans may be
adapted, based on periodic imaging acquired between treatments [2]. Further, changes in patient anatomy may, to some
extent, be addressed in treatment planning and quality assurance, optimizing and evaluating plans for robustness to some
Treatment Planning
All treatment plans were robustly optimized by using the Eclipse treatment planning system (Varian Medical Systems, Palo
Alto, California). Unless otherwise noted, robustness was evaluated with a 3% range uncertainty and 3-mm setup uncertainty.
Typically, a general planning criterion of 95% target coverage by the 95% prescription isodose line is used but is subject to
other planning considerations. Motion mitigation strategies for sites susceptible to breathing motion include layer-by-layer
repainting, passive breath holding, and phase gating. In general, repainting is used when target motion exceeds 5 mm and
breath hold or phase gates (defined by external surrogate motion) are set larger than 1 cm. These are not strict constraints and
motion mitigation may be implemented for cases with smaller target motion, based on dosimetric analysis. For plans using
posterior beams, patients are typically positioned atop a low-density foam pad to improve patient comfort and smooth the
posterior external contour, subsequently improving the reproducibility of patient position.
H&N use 2 anterior oblique beams and 1 posterior beam. Alternative geometries may be used as necessary, for example, to
avoid treating through dental hardware.
Spine
For spine treatment sites, patients are typically immobilized in the same fashion as for H&N treatment sites. The standard
treatment technique recommends single-field optimization (SFO) planning and 2 posterior oblique fields with the possible
addition of a straight posterior field. MFO planning is an acceptable alternative, especially in cases with spinal hardware.
Breast
For smaller targets, including partial breast irradiation (PBI), SFO is used to improve plan robustness. For chest wall and
whole breast plans, MFO is typically used to improve OAR sparing and to accommodate fields not irradiating the entirety of the
volume. Patients are typically set up on a supine breast board with a head rest and open face mask. Planning may be
optionally completed with arms up or down. Additionally, non-PBI treatments are evaluated at 3% range uncertainty and 5-mm
setup uncertainty for robustness.
Esophagus
For esophagus treatment sites, patients are immobilized with arms up in an upper vacuum bag and indexed knee cushion.
This site is commonly treated with SFO but may also be treated with MFO upon physician request or for normal tissue sparing.
The standard beam arrangement is 2 posterior oblique beams. Robustness is assessed for 5-mm shifts and 3% proton range
uncertainty.
Liver
For liver treatment sites, patients are immobilized with arms up in an upper vacuum bag and indexed knee cushion or lower
vacuum bag. SFO or MFO technique may be used depending on skin or normal tissue sparing. Beam selection commonly
includes 2 to 3 fields biased towards patient’s right side. Robustness is evaluated for 5-mm shifts in the lateral and anterior-
posterior directions, 7 mm in the superior-inferior direction, and 3% proton range uncertainty.
Lung or chest
For lung or chest treatment sites, patients are immobilized with either arms up in a vacuum bag or with arms down and a 5-
point thermoplastic mask, depending on target location and patient tolerance. These sites are commonly treated with SFO
technique unless split scanning target volumes [9, 10] are required or improved OAR sparing is necessary. Beam selection
may vary depending on location of target, but typically includes 2 to 3 fields. Robustness is assessed for 5-mm shifts and 3%
proton range uncertainty.
Prostate
For prostate treatment sites, patients are immobilized with an indexed knee cushion and/or lower vacuum bag. This site is
commonly treated with the SFO technique. Beam selection includes 2 lateral beams or 2 lateral and 2 anterior oblique beams.
Analysis
Throughout the course of a given proton treatment, VSs are periodically scheduled (approximately weekly) to monitor potential
gross changes in patient anatomy. VS data are registered to original planning scans and dose from the original plan is
calculated on the current anatomy. Typically, normal structures, OARs, and any other structures with overridden HU values
relevant to planning are rigidly registered to the VS. Target volumes are transferred to VS via rigid and deformable registration
Results
Figure 1 displays bar plots of the observed RoR (Figure 1A) and RP% (Figure 1B) categorized by treatment site and a
comparison of quantities (n) of VS to RPs, as well as the number of RP per VS on a dual-abscissa plot (Figure 1C). H&N and
lung or chest treatment sites represent the greatest encumbrance in terms of RoR and RP%. Although lung or chest plans
exhibit a higher RoR, head and neck patients are individually more likely to require RP. Over all treatment sites, each VS
finding has an approximately 10% likelihood of generating an RP, with the largest probability being observed for craniospinal
as well as bone, spine, chordoma, and sarcoma sites.
Figure 2 plots the total number of patients (blue) and RPs (red) along the left vertical axis, and RoR (green) along the right
vertical axis separated by physician (identified numerically) for each studied disease site. It should be noted that the physician
numbers are assigned in chronological order for each treatment site. Any given physician may participate in treatment of
multiple disease sites and may not necessarily be assigned the same identification number over all groups. Mean and
standard deviations for RoR are tabulated in the Table.
The data were also analyzed for where RP occurred in the course of treatment. Figure 3 displays a series of histograms,
categorized by disease site, of the RoR as a function of treatment week. For all disease sites, RP probabilities were highest
within the first 4 weeks of treatment. Re-plans after week 4 were only found in CSI, lung or chest, and H&N disease sites. One
RP was observed after week 4 for CSI, as well as 11.4% of cases for lung or chest, and 22.2% for H&N treatments.
Similarly, Figures 4 and 5 plot a series of histograms, again categorized by disease site, for VS per patient and for RP per
VS, respectively, as a function of treatment week. In general, as seen in Figure 4, the number of VSs per patient decreased
over the course of treatment. However, in all treatment sites, VSs were acquired at some frequency throughout the duration of
treatment. Figure 5 shows little evidence of a trend in RP per VS as a function of time for any of the disease sites with RP
beyond week 4.
Discussion
Throughout the course of the study period, the H&N disease site represented the largest clinical encumbrance in terms of
gross number of RPs with 91. Normalized by number of patients (Figure 1A and 1B), the lung or chest disease site slightly
exceeded H&N in terms of RoR, whereas RP% was highest for H&N treatments. Thus, while patients will more typically need
RPs with H&N treatments, those who do receive them are less likely to need another when compared to lung or chest patients.
While also considering Figures 3, 4, and 5, the data suggest that continued, periodic VS throughout lung or chest, H&N, and
potentially CSI treatments, would be a reasonable practice.
Figure 2. Dual axis bar plot visualizing comparisons of total number (n) of patients (blue) and RPs (red) along the primary (left vertical) axis, and RoR
(green) plotted on the secondary (right vertical) axis, separated by treating physician (identified numerically) and categorized by treatment site.
Abbreviations: RoR, rate of re-plan; RP, re-plans.
Figure 3. Histograms of RoR as a function of treatment week for each disease site. Abbreviation: RoR, rate of re-plan.
Lower pelvic treatment sites, including prostatic, gynecologic, rectal, and anal, exhibited much lower probabilities of RP as a
function of VS (Figure 1C) as a whole and as a function of time. The low RoR and scarcity of RP after week 2 (Figure 3), in
spite of continued observation via VS (Figure 4) for these disease sites, suggest that a less conservative approach to VS may
be warranted, particularly late in treatment. Although the pelvic treatment site exhibits higher RoR and RP% than the
aforementioned lower pelvic disease sites, all of the RPs were observed before week 4, suggesting the possibility of
discontinuation of VS late in treatment.
While global conclusions may be drawn by collective data by treatment site, some interphysician variation exists (Figure 2,
Table). This variation exists presumably because of physician discretion over RP, though it is unclear from the collected data
whether the discrepancy stems from differences in initial planning approach or tolerance for dosimetric change in the VS. A
particularly dramatic example is observed when considering the breast treatment site. The data for the 3 most prolific breast
physicians—treating 19, 34, and 43 patients—called for RP 5, 10, and 2 times, respectively, resulting in RoR of 26%, 35%, and
4.7%. Such data may be collected for institutions with similar processes and analyzed on a per-physician basis to effectively
review best VS practices.
Finally, RP as a function of time, as reported in Figure 3, suggests continual changes in patient anatomy and geometry
beyond week 4 for lung or chest and H&N disease sites. These late-treatment RPs suggest pragmatism in continued, regular
VS throughout the course of treatment.
Conclusion
A large degree of variation in RoR and individual RP% was observed between treatment sites. The present retrospective
analysis presents data on which to design VS protocols on the basis of this treatment site variability. In general, increases in
gross RP quantities, RoR, and RP% were affected by treatment site (with H&N and lung or chest treatments accounting for the
largest clinical encumbrance of VS and RP) and treating physician.
and editing; Jaden D. Evans: writing – review and editing, investigation; Anita Mahajan: writing – review and editing; Jon J.
Kruse: writing – review and editing; Daniel W. Mundy: conceptualization, project administration, methodology, supervision,
writing – review and editing.
Conflicts of Interest: Anita Mahajan, MD, is an Associate Editor of the International Journal of Particle Therapy. The authors
have no other relevant conflicts of interest to disclose.
Funding: The authors have no funding to disclose.
Ethical Approval: All patient data have been collected under an institutional review board–approved protocol.
References