Parotid gland sparing effect by computed tomography-based modified lower field margin in whole brain radiotherapy

Deep learning-based two-step organs at risk auto-segmentation model for brachytherapy planning in parotid gland carcinoma

Purpose

Delineation of organs at risk (OARs) represents a crucial step for both tailored delivery of radiation doses and prevention of radiation-induced toxicity in brachytherapy. Due to lack of studies on auto-segmentation methods in head and neck cancers, our study proposed a deep learning-based two-step approach for auto-segmentation of organs at risk in parotid carcinoma brachytherapy.

Material and methods

Computed tomography images of 200 patients with parotid gland carcinoma were used to train and evaluate our in-house developed two-step 3D nnU-Net-based model for OARs auto-segmentation. OARs during brachytherapy were defined as the auricula, condyle process, skin, mastoid process, external auditory canal, and mandibular ramus. Auto-segmentation results were compared to those of manual segmentation by expert oncologists. Accuracy was quantitatively evaluated in terms of dice similarity coefficient (DSC), Jaccard index, 95^th-percentile Hausdorff distance (95HD), and precision and recall. Qualitative evaluation of auto-segmentation results was also performed.

Results

The mean DSC values of each OAR were 0.88, 0.91, 0.75, 0.89, 0.74, and 0.93, respectively, indicating close resemblance of auto-segmentation results to those of manual contouring. In addition, auto-segmentation could be completed within a minute, as compared with manual segmentation, which required over 20 minutes. All generated results were deemed clinically acceptable.

Conclusions

Our proposed deep learning-based two-step OARs auto-segmentation model demonstrated high efficiency and good agreement with gold standard manual contours. Thereby, this novel approach carries the potential in expediting the treatment planning process of brachytherapy for parotid gland cancers, while allowing for more accurate radiation delivery to minimize toxicity.

Parotid gland dose reduction in the hippocampus avoidance whole-brain radiotherapy using helical tomotherapy

The present study aimed to reduce the parotid gland dose in the hippocampus avoidance with whole-brain radiotherapy (HA-WBRT) using the helical tomotherapy (HT). Ten patients who had previously undergone WBRT were randomly selected and enrolled in this study. During the treatment planning, two different techniques to the jaw were applied for each patient, namely, 1.0 cm fixed jaw and 2.5 cm dynamic jaw. To efficiently reduce the dose in the bilateral parotid glands, directional block (DB) mode was set. The DB is a function of a treatment planning system for the dose reduction in organs at risk. The standard HA-WBRT plan which did not reduce the parotid gland dose was also designed to compare the plan quality. Compared with the standard HA-WBRT plan, the parotid gland dose could be reduced by approximately 70% without extending the delivery time by adding the parotid gland on the DB mode to the dose constraint. In addition, the differences in dosimetric parameters observed between the plans employing the 1.0 cm fixed jaw and 2.5 cm dynamic jaw were almost negligible. Moreover, delivery time in the 2.5 cm dynamic jaw could be greatly reduced by 60% compared with that in the 1.0 cm fixed jaw.

Examination of the best head tilt angle to reduce the parotid gland dose maintaining a safe level of lens dose in whole-brain radiotherapy using the four-field box technique

The parotid gland is recognized as a major‐risk organ in whole‐brain irradiation; however, the beam delivery from the left and right sides cannot reduce the parotid gland dose. The four‐field box technique using a head‐tilting device has been reported to reduce the parotid gland dose by excluding it from the radiation field. This study aimed to determine the appropriate head tilt angle to reduce the parotid gland dose in the four‐field box technique. The bilateral, anterior, and posterior beams were set for each of ten patients. The orbitomeatal plane angle (OMPA) was introduced as an indicator that expresses the head tilt angle. Next, principal component analysis (PCA) was performed to understand the interrelationship between variables (dosimetric parameters of the lens and parotid gland and OMPA). In PCA, the angle between the OMPA vector and maximum lens dose or mean parotid gland dose vector was approximately opposite or close, indicating a negative or positive correlation [r = −0.627 (p < 0.05) or 0.475 (p < 0.05), respectively]. The OMPA that reduced the maximum lens dose to <10 Gy with a 95% confidence interval was approximately 14°. If the lens dose was not considered, the parotid gland dose could be reduced by decreasing the OMPA.

Assessment of Risk of Xerostomia After Whole-Brain Radiation Therapy and Association With Parotid Dose

Importance

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.

Objective

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.

Results

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.

Prospective Assessment of Patient-Reported Dry Eye Syndrome After Whole Brain Radiation

PURPOSE

Dry eye is not typically considered a toxicity of whole brain radiation therapy (WBRT). We analyzed dry eye syndrome as part of a prospective study of patient-reported outcomes after WBRT.

METHODS AND MATERIALS

Patients receiving WBRT to 25 to 40 Gy were enrolled on a study with dry mouth as the primary endpoint and dry eye syndrome as a secondary endpoint. Patients received 3-dimensional WBRT using opposed lateral fields. Per standard practice, lacrimal glands were not prospectively delineated. Patients completed the Subjective Evaluation of Symptom of Dryness (SESoD, scored 0-4, with higher scores representing worse dry eye symptoms) at baseline, immediately after WBRT (EndRT), and at 1 month (1M), 3 months, and 6 months. Patients with baseline SESoD ≥3 (moderate dry eye) were excluded. The endpoints analyzed were ≥1-point and ≥2-point increase in SESoD score at 1M. Lacrimal glands were retrospectively delineated with fused magnetic resonance imaging scans.

RESULTS

One hundred patients were enrolled, 70 were eligible for analysis, and 54 were evaluable at 1M. Median bilateral lacrimal V20Gy was 79%. At 1M, 17 patients (32%) had a ≥1-point increase in SESoD score, and 13 (24%) a ≥2-point increase. Lacrimal doses appeared to be associated with an increase in SESoD score of both ≥1 point (V10Gy: P = .042, odds ratio [OR] 1.09/%; V20Gy: P = .071, OR 1.03/%) and ≥2 points (V10Gy: P = .038, OR 1.15/%; V20Gy: P = .063, OR 1.04/%). The proportion with increase in dry eye symptoms at 1M for lacrimal V20Gy ≥79% versus <79% was 46% versus 15%, respectively, for ≥1 point SESoD increase (P = .02) and 36% versus 12%, respectively, for ≥2 point SESoD increase (P = .056).

CONCLUSIONS

Dry eye appears to be a relatively common, dose/volume-dependent acute toxicity of WBRT. Minimization of lacrimal gland dose may reduce this toxicity, and patients should be counseled regarding the existence of this potential side effect and treatments for dry eye.

Whole brain radiotherapy using four-field box technique with tilting baseplate for parotid gland sparing

PURPOSE

The aim of this study is to evaluate the efficacy and feasibility of four-field box whole brain radiotherapy (FB-WBRT) with tilting baseplate by comparing bilateral WBRT (B-WBRT).

METHODS AND MATERIALS

Between March 2016 and September 2018, 20 patients with brain metastases underwent WBRT using the four-field box technique. WBRT is performed with a dose of 30 Gy in 10 fractions daily. Two computed tomography simulations per person were performed. One was in the traditional supine position for B-WBRT and the other by applying the tilting acrylic supine baseplate to elevate the head by 40° for FB-WBRT. The B-WBRT used the field-in-field technique, which is the most commonly used method in our institution. The FB-WBRT comprised anterior, posterior, and bilateral beams. A wedge was applied in anterior and posterior fields to compensate for skull convexity.

RESULTS

The average of Dmean of both parotid glands was 10.2 Gy (range, 3.8 to 17.8 Gy) in B-WBRT and 5.4 Gy (range, 2.0 to 11.7 Gy) in FB-WBRT (p < 0.05). Compared to B-WBRT, FB-WBRT reduced the mean dose of the right and left parotid glands from 10.1 Gy to 4.9 Gy and from 10.4 Gy to 5.8 Gy, respectively (p < 0.05). Further, V5, V10, V15, V20, and V25 for the parotid gland decreased significantly in FB-WBRT (p < 0.05). The Dmax and Dmean of lens decreased according to the dose-volume histogram.

CONCLUSION

Compared to B-WBRT, FB-WBRT with a tilting baseplate is a simple and effective method that takes feature of noncoplanar beam to protect the parotid gland.

Non-coplanar whole brain radiotherapy is an effective modality for parotid sparing

Background

The purpose of this study was to evaluate the efficacy and feasibility of non–coplanar whole brain radiotherapy (NC–WBRT) for parotid sparing.

Methods

Fifteen cases, previously treated with WBRT were selected. NC–WBRT plans were generated. The beam arrangement for the non–coplanar plans consisted of superior anterior, right, and left beams. After generation of the non–coplanar plans a field–in–field technique was applied to the bilateral parallel opposed beams in order to reduce maximum dose and increase dose homogeneity. The NC–WBRT plans were subsequently compared with the previously generated bilateral WBRT (B–WBRT) plans. A field–in–field technique was also used with the B–WBRT plans according to our departmental protocol. As per our institutional practice a total dose of 30 Gy in 10 fractions of WBRT was administered 5 days a week.

Results

The mean dose to the parotid gland for the two different plans were 16.2 Gy with B–WBRT and 13.7 Gy with NC–WBRT (p<0.05). In the NC–WBRT plan, the V_5Gy, V_10Gy, V_15Gy, V_20Gy, and V_25Gy of the parotid were significantly lower (p<0.05) than those of the B–WBRT plan. The D_max of the lens was also lower by 10% with NC–WBRT.

Conclusion

The use of NC–WBRT plans could be a simple and effective method to reduce irradiated volumes and improve the dose–volume parameters of the parotid gland.

Volumetric-modulated arc therapy (VMAT) for whole brain radiotherapy: not only for hippocampal sparing, but also for reduction of dose to organs at risk

A prospective clinical trial, Radiation Therapy Oncology Group (RTOG) 0933, has demonstrated that whole brain radiotherapy (WBRT) using conformal radiation delivery technique with hippocampal avoidance is associated with less memory complications. Further sparing of other organs at risk (OARs) including the scalp, ear canals, cochleae, and parotid glands could be associated with reductions in additional toxicities for patients treated with WBRT. We investigated the feasibility of WBRT using volumetric-modulated arc therapy (VMAT) to spare the hippocampi and the aforementioned OARs. Ten patients previously treated with nonconformal WBRT (NC-WBRT) using opposed lateral beams were retrospectively re-planned using VMAT with hippocampal sparing according to the RTOG 0933 protocol. The OARs (scalp, auditory canals, cochleae, and parotid glands) were considered as dose-constrained structures. VMAT plans were generated for a prescription dose of 30 Gy in 10 fractions. Comparison of the dosimetric parameters achieved by VMAT and NC-WBRT plans was performed using paired t-tests using upper bound p-value of < 0.001. Average beam on time and monitor units (MUs) delivered to the patients on VMAT were compared with those obtained with NC-WBRT. All VMAT plans met RTOG 0933 dosimetric criteria including the dose to hippocampi of 100% of the volume (D_100%) of 8.4 ± 0.3 Gy and maximum dose of 15.6 ± 0.4 Gy, respectively. A statistically significant dose reduction (p < 0.001) to all OARs was achieved. The mean and maximum scalp doses were reduced by an average of 9 Gy (32%) and 2 Gy (6%), respectively. The mean and maximum doses to the auditory canals were reduced from 29.5 ± 0.5 Gy and 31.0 ± 0.4 Gy with NC-WBRT, to 21.8 ± 1.6 Gy (26%) and 27.4 ± 1.4 Gy (12%) with VMAT. VMAT also reduced mean and maximum doses to the cochlea by an average of 4 Gy (13%) and 2 Gy (6%), respectively. The parotid glands mean and maximum doses with VMAT were 4.4 ± 1.9 Gy and 15.7 ± 5.0 Gy, compared to 12.8 ± 4.9 Gy and 30.6 ± 0.5 Gy with NC-WBRT, respectively. The average dose reduction of mean and maximum of parotid glands from VMAT were 65% and 50%, respectively. The average beam on time and MUs were 2.3minutes and 719 on VMAT, and 0.7 minutes and 350 on NC-WBRT. This study demonstrated the feasibility of WBRT using VMAT to not only spare the hippocampi, but also significantly reduce dose to OARs. These advantages of VMAT could potentially decrease the toxicities associated with NC-WBRT and improve patients' quality of life, especially for patients with favorable prognosis receiving WBRT or patients receiving prophylactic cranial irradiation (PCI).

Whole-brain Irradiation Field Design: A Comparison of Parotid Dose

Whole-brain radiation therapy (WBRT) plays an important role in patients with diffusely metastatic intracranial disease. Whether the extent of the radiation field design to C1 or C2 affects parotid dose and risk for developing xerostomia is unknown. The goal of this study is to examine the parotid dose based off of the inferior extent of WBRT field to either C1 or C2. Patients treated with WBRT with either 30 Gy or 37.5 Gy from 2011 to 2014 at a single institution were examined. Parotid dose constraints were compared with Radiation Therapy Oncology Group (RTOG) 0615 nasopharyngeal carcinoma for a 33-fraction treatment: mean <26 Gy, volume constraint at 20 Gy (V20) < 20 cc, and dose at 50% of the parotid volume (D50) < 30 Gy. Biologically effective dose (BED) conversions with an α/β of 3 for normal parotid were performed to compare with 10-fraction and 15-fraction treatments of WBRT. The constraints are as follows: mean < BED 32.83 Gy, V15.76 (for 10-fraction WBRT) or V17.35 (for 15-fraction WBRT) < 20 cc, and D50 < BED 39.09 Gy. Nineteen patients treated to C1 and 26 patients treated to C2 were analyzed. Comparing WBRT to C1 with WBRT to C2, the mean left, right, and both parotids' doses were lower when treated to C1. Converting mean dose to BED₃, the parotid doses were lower than BED₃ constraint of 32.83 Gy: left (30.12 Gy), right (30.69 Gy), and both parotids (30.32 Gy). V20 to combined parotids was lower in patients treated to C1. When accounting for fractionation of WBRT received, the mean corrected V20 volume was less than 20 cc when treating to C1. D50 for C1 was lower than C2 for the left parotid, right parotid, and both parotids. BED₃ conversion for the mean D50 of the left, right, and both parotids was less than 39.09 Gy. In conclusion, WBRT to C1 limits parotid dose, and parotid dose constraints are achievable compared with inferior border at C2. A possible mean parotid dose constraint with BED₃ should be less than 32.83 Gy.

Differences in Parotid Dosimetry and Expected Normal Tissue Complication Probabilities in Whole Brain Radiation Plans Covering C1 Versus C2

Objectives

There is no consensus standard regarding the placement of the inferior field border in whole brain radiation therapy (WBRT) plans, with most providers choosing to cover the first versus (vs.) second cervical vertebrae (C1 vs. C2). We hypothesize that extending coverage to C2 may increase predicted rates of xerostomia.

Methods

Fifteen patients underwent computed tomography (CT) simulation; two WBRT plans were then produced, one covering C2 and the other covering C1. The plans were otherwise standard, and patients were prescribed doses of 25, 30 and 37.5 gray (Gy). Dose-volume statistics were obtained and normal tissue complication probabilities (NTCPs) were estimated using the Lyman-Burman-Kutcher model. Mean parotid dose and predicted xerostomia rates were compared for plans covering C2 vs. C1 using a two-sided patient-matched t-test. Plans were also evaluated to determine whether extending the lower field border to cover C2 would result in a violation of commonly accepted dosimetric planning constraints.

Results

The mean dose to both parotid glands was significantly higher in WBRT plans covering C2 compared to plans covering C1 for all dose prescriptions (p<0.01). Normal tissue complication probabilities were also significantly higher when covering C2 vs. C1, for all prescribed doses (p<0.01). Predicted median rates of xerostomia ranged from <0.03%-21% for plans covering C2 vs. <0.001%-12% for patients treated with plans covering C1 (p<0.01), dependent on the treatment dose and NTCP model. Plans covering C2 were unable to constrain at least one parotid to <20 Gy in 31% of plans vs. 9% of plans when C1 was covered. A total parotid dose constraint of <25 Gy was violated in 13% of plans covering C2 vs. 0% of plans covering C1.

Conclusions

Coverage of C2 significantly increases the mean parotid dose and predicted NTCPs and results in more frequent violation of commonly accepted dosimetric planning constraints.

Potential for reduced radiation-induced toxicity using intensity-modulated arc therapy for whole-brain radiotherapy with hippocampal sparing

The purpose of this study was to retrospectively investigate the accuracy, plan quality, and efficiency of using intensity-modulated arc therapy (IMAT) for whole brain radiotherapy (WBRT) patients with sparing not only the hippocampus (following RTOG 0933 compliance criteria) but also other organs at risk (OARs). A total of 10 patients previously treated with nonconformal opposed laterals whole-brain radiotherapy (NC-WBRT) were retrospectively replanned for hippocampal sparing using IMAT treatment planning. The hippocampus was volumetrically contoured on fused diagnostic T1-weighted MRI with planning CT images and hippocampus avoidance zone (HAZ) was generated using a 5 mm uniform margin around the hippocampus. Both hippocampi were defined as one paired organ. Whole brain tissue minus HAZ was defined as the whole-brain planning target volume (WB-PTV). Highly conformal IMAT plans were generated in the Eclipse treatment planning system for Novalis TX linear accelerator consisting of high-definition multileaf collimators (HD-MLCs: 2.5 mm leaf width at isocenter) and 6 MV beam for a prescription dose of 30 Gy in 10 fractions following RTOG 0933 dosimetric criteria. Two full coplanar arcs with orbits avoidance sectors were used. In addition to RTOG criteria, doses to other organs at risk (OARs), such as parotid glands, cochlea, external/middle ear canals, skin, scalp, optic pathways, brainstem, and eyes/lens, were also evaluated. Subsequently, dose delivery efficiency and accuracy of each IMAT plan was assessed by delivering quality assurance (QA) plans with a MapCHECK device, recording actual beam-on time and measuring planed vs. measured dose agreement using a gamma index. On IMAT plans, following RTOG 0933 dosimetric criteria, the maximum dose to WB-PTV, mean WB-PTV D2%, and mean WB-PTV D98% were 34.9 ± 0.3 Gy, 33.2 ± 0.4 Gy, and 26.0± 0.4Gy, respectively. Accordingly, WB-PTV received the prescription dose of 30Gy and mean V30 was 90.5% ± 0.5%. The D100%, and mean and maximum doses to hippocampus were 8.4 ± 0.3 Gy, 11.2 ± 0.3 Gy, and 15.6 ± 0.4 Gy, on average, respectively. The mean values of homogeneity index (HI) and conformity index (CI) were 0.23 ± 0.02 and 0.96 ± 0.02, respectively. The maximum point dose to WB-PTV was 35.3 Gy, well below the optic pathway tolerance of 37.5 Gy. In addition, compared to NC-WBRT, dose reduction of mean and maximum of parotid glands from IMAT were 65% and 50%, respectively. Ear canals mean and maximum doses were reduced by 26% and 12%, and mean and maximum scalp doses were reduced by 9 Gy (32%) and 2 Gy (6%), on average, respectively. The mean dose to skin was 9.7 Gy with IMAT plans compared to 16 Gy with conventional NC-WBRT, demonstrating that absolute reduction of skin dose by a factor of 2. The mean values of the total number of monitor units (MUs) and actual beam on time were 719 ± 44 and 2.34 ± 0.14 min, respectively. The accuracy of IMAT QA plan delivery was (98.1 ± 0.8) %, on average, with a 3%/3 mm gamma index passing rate criteria. All of these plans were considered clinically acceptable per RTOG 0933 criteria. IMAT planning provided highly conformal and homogenous plan with a fast and effective treatment option for WBRT patients, sparing not only hippocampi but also other OARs, which could potentially result in an additional improvement of the quality life (QoL). In the future, we plan to evaluate the clinical potential of IMAT planning and treatment option with hippocampal and other OARs avoidance in our patient's cohort and asses the QoL of the WBRT patients, as well as simultaneous integrated boost (SIB) for the brain metastases diseases.