Individual scores (black open circles) and mean scores (blue filled circles) shown at each assessment period. Mean scores were significantly higher at each period compared with baseline (base) scores. EndRT indicates period immediately after whole-brain radiation therapy. See the Assessments subsection of the Methods for details on the calculation of xerostomia score.
A, Open circles represent individual patient xerostomia scores at 1 month, plotted against parotid V20Gy. B, Mean xerostomia score was significantly higher in patients with parotid V20Gy of 47% or higher (orange dots) at 1 and 3 months, compared with those with parotid V20Gy lower than 47% (blue squares). C, Diamonds represent individual patient maximum bother score at 1 month, plotted against parotid V20Gy. The self-reported bother score is on a 4-point Likert scale (0 indicating bothered not at all; 1, bothered a little bit; 2, bothered quite a bit; 3, bothered very much). D, The proportion of patients bothered quite a bit or very much by xerostomia was significantly higher in those with parotid V20Gy of 47% or greater at 1 and 3 months. Base indicates baseline; EndRT, immediately after whole-brain radiation therapy. See the Assessments subsection of the Methods for definitions and explanation of xerostomia scoring. V20Gy indicates volume of parotid receiving at least 20 Gy.
A, Delivered radiation fields for an enrolled patient who developed persistent xerostomia. The parotids were not prospectively delineated and the parotid V20Gy was 54%. B, Alternative plan with fields adjusted to reduce parotid exposure, resulting in a parotid V20Gy of only 23%. The volume of brain covered by 95% of prescription dose was 99.98% for the original plan, compared with 99.65% for the replan.
eFigure. Dose-Response for Clinically Significant Xerostomia at 1 Month
Customize your JAMA Network experience by selecting one or more topics from the list below.
Wang K, Pearlstein KA, Moon DH, et al. Assessment of Risk of Xerostomia After Whole-Brain Radiation Therapy and Association With Parotid Dose. JAMA Oncol. 2019;5(2):221–228. doi:10.1001/jamaoncol.2018.4951
Is xerostomia a toxic effect of whole-brain radiation therapy?
In this cohort study of 73 patients who received whole-brain radiation therapy, clinically significant xerostomia was common, was bothersome to patients, occurred as early as the end of treatment, and often persisted for months. Furthermore, xerostomia was statistically significantly associated with radiation dose to the parotid glands.
This study appears to justify minimization of parotidradiation doses and associated adverse effects in patients who receive this therapy and who often do not survive long enough for salivary recovery.
Whole-brain radiation therapy (WBRT) delivers a substantial radiation dose to the parotid glands, but the parotid glands are not delineated for avoidance and xerostomia has never been reported as an adverse effect. Minimizing the toxic effects in patients receiving palliative treatments, such as WBRT, is crucial.
To assess whether xerostomia is a toxic effect of WBRT.
Design, Setting, and Participants
This observational cohort study enrolled patients from November 2, 2015, to March 20, 2018, at 1 academic center (University of North Carolina Hospitals) and 2 affiliated community hospitals (High Point Regional Hospital and University of North Carolina Rex Hospital). Adult patients (n = 100) receiving WBRT for the treatment or prophylaxis of brain metastases were enrolled. Patients who had substantial baseline xerostomia or did not complete WBRT or at least 1 postbaseline questionnaire were prospectively excluded from analysis and follow-up. Patients received 3-dimensional WBRT using opposed lateral fields covering the skull and the C1 or C2 vertebra. Per standard practice, the parotid glands were not prospectively delineated.
Main Outcomes and Measures
Patients completed the University of Michigan Xerostomia Questionnaire and a 4-point bother score at baseline, immediately after WBRT, at 1 month, at 3 months, and at 6 months. The primary end point was the 1-month xerostomia score, with a hypothesized worsening score of 10 points from baseline.
Of the 100 patients enrolled, 73 (73%) were eligible for analysis and 55 (55%) were evaluable at 1 month. The 73 patients included 43 women (59%) and 30 men (41%) with a median (range) age of 61 (23-88) years. The median volume of parotid receiving at least 20 Gy (V20Gy) was 47%. The mean xerostomia score was 7 points at baseline and was statistically significantly higher at each assessment period, including 21 points immediately after WBRT (95% CI, 16-26; P < .001), 23 points (95% CI, 16-30; P < .001) at 1 month, 21 points (95% CI, 13-28; P < .001) at 3 months, and 14 points (95% CI, 7-21; P = .03) at 6 months. At 1 month, the xerostomia score increased by 20 points or more in 19 patients (35%). The xerostomia score at 1 month was associated with parotid dose as a continuous variable and was 35 points in patients with parotid V20Gy of 47% or greater, compared with only 9 points in patients with parotid V20Gy less than 47% (P < .001). The proportion of patients who self-reported to be bothered quite a bit or bothered very much by xerostomia at 1 month was 50% in those with parotid V20Gy of 47% or greater, compared with only 4% in those with parotid V20Gy less than 47% (P < .001). At 3 months, this difference was 50% vs 0% (P = .001). Xerostomia was not associated with medication use.
Conclusions and Relevance
Clinically significant xerostomia occurred by the end of WBRT, appeared to be persistent, and appeared to be associated with parotid dose. The findings from this study suggest that the parotid glands should be delineated for avoidance to minimize these toxic effects in patients who undergo WBRT and often do not survive long enough for salivary recovery.
Despite the increasing use of stereotactic radiosurgery (SRS), whole-brain radiation therapy (WBRT) remains a commonly used treatment for the hundreds of thousands of patients who develop brain metastases every year.1-5 The prognosis for most patients who receive WBRT is poor; therefore, minimizing its toxic effects and maximizing quality of life (QOL) are crucial.
Well-known adverse effects of WBRT include fatigue, alopecia, and neurocognitive changes.6-8 To date, xerostomia or dry mouth, however, has not been reported. To our knowledge, no studies exist that describe xerostomia after WBRT. For head and neck radiation therapy, numerous studies over several decades have illustrated the importance of limiting radiation dose to the parotid glands to minimize dry mouth and improve QOL.9-14 However, the parotid glands are not typically delineated nor avoided during WBRT, despite receiving substantial doses.15-20
We hypothesized that WBRT is associated with clinically significant xerostomia and that xerostomia is associated with parotid dose. To investigate this hypothesis, we conducted a prospective observational study of patients receiving routine WBRT.
This study was approved by the University of North Carolina Hospitals Institutional Review Board. Patients provided informed written consent and were enrolled at 1 academic center (University of North Carolina Hospitals in Chapel Hill) and 2 affiliated community hospitals (High Point Regional Hospital in High Point, North Carolina, and University of North Carolina Rex Hospital in Raleigh, North Carolina) from November 2, 2015, to March 20, 2018. Adult patients who were receiving WBRT for any diagnosis at a dose of 25 to 40 Gy (to convert grays to rads, multiply by 100) in 10 to 20 fractions (2 to 3 Gy per fraction) were eligible. Exclusion criteria included previous exposure to parotid radiation and altered mental status. To minimize confounding, we sought to prospectively analyze patients without substantial baseline xerostomia. Thus, only patients with a baseline xerostomia assessment score under 30 points were eligible for analysis. Patients who did not complete WBRT or at least 1 postbaseline questionnaire were also excluded from analysis. Enrolled patients who were ineligible for analysis were not prospectively followed.
Patients were treated with computed tomography–based 3-dimensional (3-D) conformal WBRT, which uses opposed lateral fields designed with digitally reconstructed skull radiographs. All patients were immobilized with a head cast for simulation and treatment. Per current standards, the inferior field border covered the C1 or C2 vertebrae and the parotid glands were not prospectively delineated. Radiation therapy plans were designed by a total of 15 attending radiation oncologists (N.C.S., M.S.K., H.D.P., T.M.Z., L.B.M., B.S.C.).
Patients completed study assessments at baseline; immediately after; and then at 1 month, 3 months, and 6 months after WBRT. The primary study assessment instrument was the validated University of Michigan Xerostomia Questionnaire (xerostomia score). This questionnaire included 8 questions, each of which was scored from 0 to 10 points. The final summation score was linearly converted to a 100-point scale, with higher scores representing worse symptoms.9,21 The secondary study assessment instrument was a 2-question bother score, adapted from the European Organisation for Research and Treatment of Cancer Quality of Life Questionnaire head and neck module,22 that evaluated the degree to which xerostomia bothered patients while eating and while not eating. Each question was answered on a 4-point Likert scale (with 0 as bothered not at all, 1 as bothered a little bit, 2 as bothered quite a bit, and 3 as bothered very much). The higher score of the 2 questions was considered the overall bother score at that period.
Patients were treated without prospective delineation of the parotid glands. The parotid glands were retrospectively delineated by our primary investigator (K.W.) and independently verified by our senior attending head and neck radiation oncologist (B.S.C.). Individual parotid contours (left and right) were combined. The recorded dose metrics included parotid mean dose, as well as volume of parotid receiving at least 5 Gy (V5Gy), 10 Gy (V10Gy), 15 Gy (V15Gy), 20 Gy (V20Gy), and 25 Gy (V25Gy). Because these metrics are collinear and the lowest standard WBRT dose is 20 Gy, we chose parotid V20Gy for most of the statistical analyses. A post hoc exploratory replanning analysis was performed to assess the degree to which parotid doses could be minimized without compromising radiation coverage of the brain. For this analysis, the brain was retrospectively delineated and WBRT fields were redesigned to reduce the overlap with the parotid glands while maintaining a 1-cm geometric margin around the brain contour.
The primary objective was to identify the change in xerostomia score from baseline at 1 month, which is a meaningful period considering the poor survival rate of patients who receive WBRT. Because no previous reports of xerostomia after WBRT existed, to our knowledge, we relied on the head and neck literature for statistical design. In the original study that validated the xerostomia questionnaire, patients who underwent radiation therapy for head and neck cancer reported a mean 20-point increase in the xerostomia score at 1 month.9 Because we expected xerostomia to be less severe after WBRT, we hypothesized a mean 10-point increase in the xerostomia score. Assuming an SD of 24 points,21 48 patients were needed for an 80% power to detect a 10-point difference in the xerostomia score with a 2-sided α = .05, using a paired, 2-tailed t test; 2-sided P < .05 was used to indicate statistical significance. To have 48 evaluable patients at 1 month, we set our accrual goal at 100 patients.
The secondary objective was to investigate the implication of parotid dose. First, we used univariable linear regression to analyze the association of xerostomia score at 1 month with factors including parotid dose, analyzed as a continuous variable. Covariates associated at the P < .10 level on univariable analysis were included in the multivariable linear regression. Collinear covariates (eg, different parotid dose metrics) were not included in the same multivariable analysis. Second, we performed analyses with a median parotid V20Gy cut point at all periods, using unpaired 2-tailed t test for the xerostomia score and Fisher exact test for the bother score. Third, to generate a dose-response plot, we divided parotid V20Gy into 3 equal tertiles and analyzed the probability of clinically significant toxic effects at 1 month as a nominal end point. Other studies have defined minimally important difference as one-third to one-half of the SD,23 which translates to a xerostomia score change of roughly 8 to 12 points, according to the head and neck literature.21 For this analysis, we defined clinically significant toxic effects as at least a 20-point worsening in the xerostomia score, or a bother score of 2 or more points (bothered quite a bit or bothered very much). Statistical analyses were performed using SPSS, version 21 (IBM).
In total, we enrolled 100 patients, 27 (27%) of whom were ineligible for analysis (18 had a baseline xerostomia score of 30 or more points, 4 did not complete WBRT, and 5 did not complete any postbaseline questionnaires). The final analysis included the remaining 73 patients (of which 43 were women [59%] and 30 were men [41%] with a median [range] age of 61 [23-88] years), and 55 were evaluable at 1 month. Patient characteristics are shown in Table 1.
Most patients (60 [82%]) were treated with radiation fields using an inferior field border at the bottom of C1, and 13 (18%) using an inferior field border at the bottom of C2. The most commonly used fractionation was 30 Gy in 10 fractions. Fifty-five patients (75%) completed the 1-month assessment and were included in the primary analysis; 33 patients (45%) completed the 3-month assessment and 28 patients (38%) completed the 6-month assessment. Seven patients (10%) had died by the first month, 19 (26%) by the third month, and 27 (37%) by the sixth month. The median (range) parotid mean dose was 17 (7-29) Gy, and the median (range) parotid V20Gy was 47% (15%-84%). The median (range) follow-up was 7 (0.4-25.6) months, and median overall survival was 8 months.
Figure 1 shows the mean and individual xerostomia scores for patients at each period. Mean xerostomia score was 7 points at baseline and was statistically significantly higher at each period, including 21 points (95% CI, 16-26; P < .001) immediately after WBRT, 23 points (95% CI, 16-30; P < .001) at 1 month, 21 points (95% CI, 13-28; P < .001) at 3 months, and 14 points (95% CI, 7-21; P = .03) at 6 months. By the end of WBRT, 21 patients (30%) already had at least a 20-point worsening in the xerostomia score. Of those who reached the 1-month assessment, 19 patients (35%) had at least a 20-point worsening and 10 patients (18%) at least a 30-point worsening in the xerostomia score compared with the baseline score.
Table 2 shows the results of the analysis of factors associated with xerostomia score at 1 month. Although the baseline xerostomia score was under 30 points in all analyzed patients, it was still associated with xerostomia score at 1 month. No association was found between xerostomia score and higher radiation prescription dose, steroids, narcotics, anticholinergics, or chemotherapy. All parotid dose metrics were associated with the 1-month xerostomia score, including parotid V20Gy. On multivariable analysis, parotid dose remained the strongest factor associated with xerostomia (parotid V20Gy; P < .001), but baseline xerostomia score and Eastern Cooperative Oncology Group performance status were also statistically significant.
Figure 2A shows the 1-month xerostomia score for each patient plotted against parotid V20Gy, and Figure 2B shows the mean xerostomia score at all periods for patients with parotid V20Gy above or below the median (47%). The 1-month xerostomia score was 35 points in patients with parotid V20Gy of 47% or greater, compared with only 9 points in patients with parotid V20Gy less than 47% (P < .001). At 3 months, this difference was 30 points vs 9 points (P = .002).
The proportion of patients who self-reported to be bothered quite a bit or bothered very much by dry mouth was 7% at baseline, 16% immediately after WBRT, 29% at 1 month, 27% at 3 months, and 11% at 6 months and was statistically significantly higher at 1 month (P = .001) and 3 months (P = .04) compared with baseline. Figure 2C shows the maximum bother score at 1 month for each patient plotted against parotid V20Gy, and Figure 2D shows the bother scores at all periods for patients with parotid V20Gy above or below the median (47%). The proportion who reported to be bothered quite a bit or bothered very much at 1 month was 50% in patients with parotid V20Gy of 47% or greater, compared with only 4% in patients with parotid V20Gy less than 47% (P < .001). At 3 months, this difference was 50% vs 0% (P = .001).
The eFigure in the Supplement shows the dose-response plot for clinically significant toxicity at 1 month, dividing patients into 3 equivalent parotid V20Gy tertiles (tertile 1: 15%-39%; tertile 2: 39%-51%; tertile 3: 51%-84%). The proportions of patients at 1 month experiencing toxic effects were 6% for tertile 1, 26% for tertile 2, and 68% for tertile 3 when defining toxic effect as an increase of 20 points or more in the xerostomia score; the proportions were 6% for tertile 1, 21% for tertile 2, and 58% for tertile 3 when defining toxic effects as a bother score of 2 points or higher.
Seventy-three patients underwent a retrospective replanning with radiation fields designed to reduce parotid exposure while maintaining coverage of the brain (eg, Figure 3). For the original plans, the median (range) parotid mean dose was 17 (7-29) Gy and parotid V20Gy was 47% (15%-84%), compared with only 8 (4-16) Gy and 15% (1%-35%) for the parotid-sparing replans. The median (range) proportion of brain receiving 95% of the prescribed dose was 99.99% (97.94%-100%) for the delivered original plans and 99.87% (96.75%-100%) for the parotid-sparing replans.
Xerostomia was common, bothersome, and persistent, and it occurred as early as the last day of WBRT. Furthermore, the degree of xerostomia was associated with parotid dose. To our knowledge, this is the first study to describe this adverse effect of WBRT and supports a change in radiation therapy technique to minimize the toxic effects and preserve QOL for the numerous patients who undergo WBRT. We had 3 sequential questions:
1. Is xerostomia an adverse effect of WBRT?
2. If xerostomia occurs after WBRT, is there an association with parotid dose?
3. If parotid dose is responsible, can the problem be fixed safely?
Regarding question 1, both xerostomia score and bother scores were significantly higher after WBRT, demonstrating the existence of a previously unreported toxic effect. Regarding question 2, xerostomia was strongly associated with all parotid dose metrics on continuous and cut point analyses, suggesting that the parotid gland is an organ at risk. Regarding question 3, parotid doses could have been reduced by almost 70% with essentially no implication for brain coverage, implying that a simple and safe solution exists.
Our findings add to the existing literature that suggests potential QOL advantages of SRS over WBRT. Despite the increasing uptake of SRS, however, WBRT remains a standard treatment that hundreds of thousands of patients receive every year.5 Recent National Cancer Database studies suggest that the use of WBRT remains widespread.1-3 For instance, Rydzewski et al3 reported that of the approximately 18 000 patients who received radiation for brain metastases in the United States between 2010 and 2012, more than 16 000 (87%) underwent WBRT. Not all centers have SRS capabilities, and WBRT is likely to be the most commonly used treatment modality for brain metastases both in the community and worldwide.1 Thus, mitigation of WBRT toxic effects is an important and consequential research question. However, most studies have focused on the neurocognitive adverse effects of WBRT,24-26 and xerostomia has never been described.
Only 1 study, to our knowledge, reports any type of salivary toxic effect after WBRT. In 1980, Cairncross et al27 described acute parotitis with accompanying hyperamylasemia in 4 patients. More recent dosimetric studies in the era of 3-D planning (that makes possible the delineation and avoidance of organs at risk) report parotid doses of 15 to 20 Gy during WBRT, but these studies did not record clinical outcomes.15-20 Therefore, the clinical significance of this parotid exposure is unknown, and the parotid glands are not routinely delineated or avoided during WBRT.25,28 In contrast, radiation therapy for head and neck cancer is well known to be associated with substantial and sometimes permanent xerostomia. Numerous studies have led to guidelines that recommend minimizing parotid doses to less than 20 to 25 Gy during head and neck radiation therapy.13,14
The parotid glands have not been considered to be at risk during WBRT for several reasons. First, clinicians may be unaware of the clinical significance of the parotid dose administered during WBRT. Second, follow-up may not be routine after WBRT, and treating radiation oncologists may not recognize the existence of xerostomia as a toxic effect. Third, patients may underreport xerostomia in the context of the worsening performance status and other neurologic issues that often accompany the diagnosis of brain metastases.
Fourth, one misconception may be that patients who undergo WBRT do not live long enough to experience xerostomia, which is often considered a late effect of radiation therapy. Patients with head and neck cancer are frequently cured, and studies analyzing xerostomia in that population appropriately use late recovery of salivary function as their end point.11,14 However, acute xerostomia is more severe and recovery of salivary function may take years.14,29 In fact, 30% of patients had at least a 20-point worsening in xerostomia score as early as the final day of WBRT. Patients WBRT not only are at risk for substantial acute salivary toxic effects but also may not live long enough to recover salivary function. In this study, the median survival of patients who received WBRT was only 8 months. Thus, patients who develop xerostomia after WBRT may live with this adverse effect for a considerable portion of their remaining life.
Another misconception is that WBRT doses are not high enough to cause xerostomia. The suggested 20- to 25-Gy parotid dose limits for head and neck radiation are based on outcomes of patients who received 6 to 7 weeks of radiation therapy. In contrast, WBRT is typically delivered over only 2 to 3 weeks. Therefore, an equivalent total parotid dose for WBRT (eg, 17 Gy for patients in this study) is both accelerated (delivered more quickly) and hypofractionated (higher dose per fraction) compared with head and neck cancer radiation therapy, and this dose could cause substantial toxic effects despite being lower than the conventional head and neck dose constraints. Patients with parotid V20Gy of 47% or greater who underwent WBRT reported a 1-month xerostomia score similar to that of patients who received unilateral neck radiation therapy for head and neck cancer.9 Because of interpatient or interclinician variation, parotid doses could also be higher than expected.
The results of this study have immediate implications for patients who undergo WBRT. Despite advances in radiation planning, WBRT field design has remained virtually unchanged for decades. The standard approach is to design opposed lateral radiation fields around the skull bones and the first or second cervical vertebrae thereby encompassing the brain with an adequate margin.28 However, this approach is crude and exposes the parotid glands to substantial incidental radiation. In our replanning analysis, we found that parotid doses could be safely minimized, with an apparent negligible change to brain coverage. Even if some underdosing of the brain were to occur, the increasing acceptance of multitarget SRS and hippocampal-sparing techniques challenges the necessity of treating the entire brain in the first place, as long as care is taken to cover posterior fossa metastases (that may be close to the parotid glands). Thus, given the importance of QOL in patients who receive WBRT, we hereby propose a new standard of care in WBRT field design: the routine delineation and avoidance of the parotid glands in all patients while maintaining coverage of the brain contour, without regard for bony anatomy. After all, the target of WBRT is the brain and not the bone.
The results of this study complement other ongoing efforts to improve QOL in patients treated with WBRT. The RTOG 0933 trial reported that hippocampal-sparing intensity-modulated radiation therapy may mitigate the neurocognitive adverse effects of WBRT.25 However, only the lenses, orbits, optic nerves, and optic chiasm (in addition to the hippocampi) were required to be delineated for avoidance in RTOG 0933, and the radiation target was specified as “the whole brain parenchyma to C1 or C2.”30(p18) Because the parotid glands are anatomically located adjacent to the C1 and C2 vertebrae (Figure 3), patients who undergo hippocampal sparing without concurrent parotid sparing may be at risk for parotid toxic effects. Although intensity-modulated radiation therapy may not be necessary to minimize the dose to the parotid glands (Figure 3), it could be used to minimize the dose to both the hippocampi and the parotid glands, providing a dual advantage for patients in whom hippocampal sparing is desired.
This study has several limitations. First, despite enrolling 100 patients, only 73 were eligible for analysis and only 55 remained at the 1-month assessment, with even fewer patients remaining at later periods. The high competing risk of death is a substantial confounder when analyzing adverse effects. However, this is a common issue in similar studies of patients with poor prognosis; to our knowledge, this study is the first to describe this particular toxic effect. Second, other potential confounders were present, such as medication use, that may have affected patient-reported outcomes. However, the association between xerostomia and parotid dose was strong, and we analyzed but did not find any association between xerostomia and the use of medications. Finally, the study assessments have not been validated in patients who receive WBRT, and defining what constitutes clinically significant has been challenging. Nonetheless, the xerostomia questionnaire used in this study has been extensively validated in a different setting, and the results of both study assessments were consistent (eFigure in the Supplement).
This prospective study showed that xerostomia appears to be a common, previously unreported toxic effect of WBRT that is associated with parotid dose. From our findings, it appears that patients who undergo WBRT should be counseled of this risk, and the parotid glands should be delineated and avoided to preserve QOL in this vulnerable population.
Accepted for Publication: August 26, 2018.
Corresponding Author: Kyle Wang, MD, Department of Radiation Oncology, University of North Carolina Hospitals, 101 Manning Dr, CB #7512, Chapel Hill, NC 27599-7512 (firstname.lastname@example.org).
Published Online: November 29, 2018. doi:10.1001/jamaoncol.2018.4951
Author Contributions: Dr K. Wang had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Concept and design: K. Wang, Deal, Zagar, Marks, Chera.
Acquisition, analysis, or interpretation of data: K. Wang, Pearlstein, Moon, Mahbooba, Deal, Y. Wang, Sutton, Motley, Judy, Holmes, Sheets, Kasibhatla, Pacholke, Shen, Zagar, Marks.
Drafting of the manuscript: K. Wang, Deal, Motley.
Critical revision of the manuscript for important intellectual content: K. Wang, Pearlstein, Moon, Mahbooba, Deal, Y. Wang, Sutton, Judy, Holmes, Sheets, Kasibhatla, Pacholke, Shen, Zagar, Marks, Chera.
Statistical analysis: K. Wang, Deal, Y. Wang, Sutton.
Administrative, technical, or material support: K. Wang, Mahbooba, Sutton, Motley, Judy, Kasibhatla, Zagar.
Supervision: K. Wang, Zagar, Marks, Chera.
Other - Patient enrollment: Holmes.
Conflict of Interest Disclosures: None reported.
Funding/Support: This study was funded in part by the Lineberger Comprehensive Cancer Center, Chapel Hill, North Carolina.
Role of the Funder/Sponsor: The funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
Meeting presentation: The results of this study were presented at the American Society for Radiation Oncology Annual Meeting, September 27, 2017, San Diego, California.