Predictors of Distant Metastases in Triple-Negative Breast Cancer Without Pathologic Complete Response After Neoadjuvant Chemotherapy.

BACKGROUND: Pathologic complete response (pCR) after neoadjuvant chemotherapy (NAC) for triple-negative breast cancer (TNBC) predicts decreased distant metastasis. However, most patients do not experience pCR, and other risk factors for distant metastasis after NAC are poorly characterized. This study investigated factors predictive of distant metastasis in TNBC without pCR after NAC. METHODS: Women with TNBC treated with NAC, surgery, and radiation therapy in 2000 through 2013 were reviewed. Freedom from distant metastasis (FFDM) was compared between patients with and without pCR using the Kaplan-Meier method. In patients without pCR, univariate and multivariable Cox analyses were used to determine factors predictive of distant metastasis. RESULTS: We identified 153 patients with median follow-up of 4.0 years (range, 0.5-14.0 years). After NAC, 108 had residual disease (pCR, 29%). Five-year FFDM was 98% and 55% in patients with and without pCR, respectively (P<.001). Factors independently predicting FFDM in patients without pCR were pathologic nodal positivity (hazard ratio, 3.08; 95% CI, 1.54-6.14; P=.001) and lymphovascular space invasion (hazard ratio, 1.91; 95% CI, 1.07-3.43; P=.030). Patients with a greater number of factors had worse FFDM; 5-year FFDM was 76.5% for patients with no factors (n=38) versus 54.9% and 27.5% for patients with 1 (n=44) and 2 factors (n=26), respectively (P<.001). CONCLUSIONS: Lack of pCR after NAC resulted in worse overall survival and FFDM, despite trimodality therapy. In patients with residual disease after NAC, pathologic lymph node positivity and lymphovascular space invasion predicted worse FFDM.

Predictors of Locoregional Recurrence After Failure to Achieve Pathologic Complete Response to Neoadjuvant Chemotherapy in Triple-Negative Breast Cancer.

BACKGROUND: This study evaluated factors predictive of locoregional recurrence (LRR) in women with triple-negative breast cancer (TNBC) treated with neoadjuvant chemotherapy who do not experience pathologic complete response (pCR). METHODS: This is a single-institution retrospective review of women with TNBC treated with neoadjuvant chemotherapy, surgery, and radiation therapy in 2000 through 2013. LRR was estimated between patients with and without pCR using the Kaplan-Meier method. Patient-, tumor-, and treatment-specific factors in patients without pCR were analyzed using the Cox proportional hazards method to evaluate factors predictive of LRR. Log-rank statistics were then used to compare LRR among these risk factors. RESULTS: A total of 153 patients with a median follow-up of 48.6 months were included. The 4-year overall survival and LRR were 70% and 15%, respectively, and the 4-year LRR in patients with pCR was 0% versus 22.0% in those without (P<.001). In patients without pCR, lymphovascular space invasion (LVSI; hazard ratio, 3.92; 95% CI, 1.64-9.38; P=.002) and extranodal extension (ENE; hazard ratio, 3.32; 95% CI, 1.35-8.15; P=.009) were significant predictors of LRR in multivariable analysis. In these patients, the 4-year LRR with LVSI was 39.8% versus 15.0% without (P<.001). Similarly, the 4-year LRR was 48.1% with ENE versus 16.1% without (P=.002). In patients without pCR, the presence of both LVSI and ENE were associated with an even further increased risk of LRR compared with patients with either LVSI or ENE alone and those with neither LVSI nor ENE in the residual tumor (P<.001). CONCLUSIONS: In patients without pCR, the presence of LVSI and ENE increases the risk of LRR in TNBC. The risk of LRR is compounded when both LVSI and ENE are present in the same patient. Future clinical trials are warranted to lower the risk of LRR in these high-risk patients.

Treatment response as predictor for brain metastasis in triple negative breast cancer: A score-based model.

BACKGROUND: Triple negative breast cancer (TNBC) has worse prognosis than other subtypes of breast cancer, and many patients develop brain metastasis (BM). We developed a simple predictive model to stratify the risk of BM in TNBC patients receiving neo-adjuvant chemotherapy (NAC), surgery, and radiation therapy (RT). METHODS: Patients with TNBC who received NAC, surgery, and RT were included. Cox proportional hazards method was used to evaluate factors associated with BM. Significant factors predictive for BM on multivariate analysis (MVA) were used to develop a risk score. Patients were divided into three risk groups: low, intermediate, and high. A receiver operating characteristic (ROC) curve was drawn to evaluate the value of the risk group in predicting BM. This predictive model was externally validated. RESULTS: A total of 160 patients were included. The median follow-up was 47.4 months. The median age at diagnosis was 49.9 years. The 2-year freedom from BM was 90.5%. Persistent lymph node positivity, HR 8.75 (1.76-43.52, P = 0.01), and lack of downstaging, HR 3.46 (1.03-11.62, P = 0.04), were significant predictors for BM. The 2-year rate of BM was 0%, 10.7%, and 30.3% (P < 0.001) in patients belonging to low-, intermediate-, and high-risk groups, respectively. Area under the ROC curve was 0.81 (P < 0.001). This model was externally validated (C-index = 0.79). CONCLUSIONS: Lack of downstaging and persistent lymph node positivity after NAC are associated with development of BM in TNBC. This model can be used by the clinicians to stratify patients into the three risk groups to identify those at increased risk of developing BM and potentially impact surveillance strategies.

Accelerated Partial Breast Irradiation: A Safe, Effective, and Convenient Early Breast Cancer Treatment Option.

Accelerated partial breast irradiation (APBI) is an excellent treatment option for many women with early stage breast cancer. Patient selection criteria include age over 40, status post lumpectomy, breast cancer (invasive or in situ disease) measuring <3 cm, negative margins (at least 2 mm), negative lymph nodes, and no lymphovascular space invasion. APBI is effective, well tolerated, and convenient. Women with early stage breast cancer and theii caregivers should be aware of this potential treatment option.

A quality assurance framework for routine monitoring of deep learning cardiac substructure computed tomography segmentation models in radiotherapy.

BACKGROUND: For autosegmentation models, the data used to train the model (e.g., public datasets and/or vendor-collected data) and the data on which the model is deployed in the clinic are typically not the same, potentially impacting the performance of these models by a process called domain shift. Tools to routinely monitor and predict segmentation performance are needed for quality assurance. Here, we develop an approach to perform such monitoring and performance prediction for cardiac substructure segmentation. PURPOSE: To develop a quality assurance (QA) framework for routine or continuous monitoring of domain shift and the performance of cardiac substructure autosegmentation algorithms. METHODS: A benchmark dataset consisting of computed tomography (CT) images along with manual cardiac substructure delineations of 241 breast cancer radiotherapy patients were collected, including one "normal" image domain of clean images and five "abnormal" domains containing images with artifact (metal, contrast), pathology, or quality variations due to scanner protocol differences (field of view, noise, reconstruction kernel, and slice thickness). The QA framework consisted of an image domain shift detector which operated on the input CT images and a shape quality detector on the output of an autosegmentation model, and a regression model for predicting autosegmentation model performance. The image domain shift detector was composed of a trained denoising autoencoder (DAE) and two hand-engineered image quality features to detect normal versus abnormal domains in the input CT images. The shape quality detector was a variational autoencoder (VAE) trained to estimate the shape quality of the auto-segmentation results. The output from the image domain shift and shape quality detectors was used to train a regression model to predict the per-patient segmentation accuracy, measured by Dice coefficient similarity (DSC) to physician contours. Different regression techniques were investigated including linear regression, Bagging, Gaussian process regression, random forest, and gradient boost regression. Of the 241 patients, 60 were used to train the autosegmentation models, 120 for training the QA framework, and the remaining 61 for testing the QA framework. A total of 19 autosegmentation models were used to evaluate QA framework performance, including 18 convolutional neural network (CNN)-based and one transformer-based model. RESULTS: When tested on the benchmark dataset, all abnormal domains resulted in a significant DSC decrease relative to the normal domain for CNN models ( p < 0.001 $p < 0.001$ ), but only for some domains for the transformer model. No significant relationship was found between the performance of an autosegmentation model and scanner protocol parameters ( p = 0.42 $p = 0.42$ ) except noise ( p = 0.01 $p = 0.01$ ). CNN-based autosegmentation models demonstrated a decreased DSC ranging from 0.07 to 0.41 with added noise, while the transformer-based model was not significantly affected (ANOVA, p = 0.99 $p=0.99$ ). For the QA framework, linear regression models with bootstrap aggregation resulted in the highest mean absolute error (MAE) of 0.041 ± 0.002 $0.041 \pm 0.002$ , in predicted DSC (relative to true DSC between autosegmentation and physician). MAE was lowest when combining both input (image) detectors and output (shape) detectors compared to output detectors alone. CONCLUSIONS: A QA framework was able to predict cardiac substructure autosegmentation model performance for clinically anticipated "abnormal" domain shifts.

Temporal analysis of intratumoral metabolic heterogeneity characterized by textural features in cervical cancer.

PURPOSE: The aim of this pilot study was to explore heterogeneity in the temporal behavior of intratumoral [(18)F]fluorodeoxyglucose (FDG) accumulation at a regional scale in patients with cervical cancer undergoing chemoradiotherapy. METHODS: Included in the study were 20 patients with FIGO stages IB1 to IVA cervical cancer treated with combined chemoradiotherapy. Patients underwent FDG PET/CT before treatment, during weeks 2 and 4 of treatment, and 12 weeks after completion of therapy. Patients were classified based on response to therapy as showing a complete metabolic response (CMR), a partial metabolic response (PMR), or residual disease and the development of new disease (NEW). Based on the presence of residual primary tumor following therapy, patients were divided into two groups, CMR and PMR/NEW. Temporal profiles of intratumoral FDG heterogeneity as characterized by textural features at a regional scale were assessed and compared with those of the standardized uptake value (SUV) indices (SUVmax and SUVmean) within the context of differentiating response groups. RESULTS: Textural features at a regional scale with emphasis on characterizing contiguous regions of high uptake in tumors decreased significantly with time (P < 0.001) in the CMR group, while features describing contiguous regions of low uptake along with those measuring the nonuniformity of contiguous isointense regions in tumors exhibited significant temporal changes in the PMR/NEW group (P < 0.03) but showed no persistent trends with time. Both SUV indices showed significant changes during the course of the disease in both patient groups (P < 0.001 for SUVmax and SUVmean in the CMR group; P = 0.0109 and 0.0136, respectively, for SUVmax and SUVmean in the PMR/NEW group), and also decreased at a constant rate in the CMR group and decreased up to the 4th week of treatment and then increased in the PMR/NEW group. CONCLUSION: The temporal changes in the heterogeneity of intratumoral FDG distribution characterized at a regional scale using image-based textural features may provide an adjunctive or alternative option for understanding tumor response to chemoradiotherapy and interpreting FDG accumulation dynamics in patients with malignant cervical tumors during the course of the disease.

Robustness of deep learning segmentation of cardiac substructures in noncontrast computed tomography for breast cancer radiotherapy.

PURPOSE: To develop and evaluate deep learning-based autosegmentation of cardiac substructures from noncontrast planning computed tomography (CT) images in patients undergoing breast cancer radiotherapy and to investigate the algorithm sensitivity to out-of-distribution data such as CT image artifacts. METHODS: Nine substructures including aortic valve (AV), left anterior descending (LAD), tricuspid valve (TV), mitral valve (MV), pulmonic valve (PV), right atrium (RA), right ventricle (RV), left atrium (LA), and left ventricle (LV) were manually delineated by a radiation oncologist on noncontrast CT images of 129 patients with breast cancer; among them 90 were considered in-distribution data, also named as "clean" data. The image/label pairs of 60 subjects were used to train a 3D deep neural network while the remaining 30 were used for testing. The rest of the 39 patients were considered out-of-distribution ("outlier") data, which were used to test robustness. Random rigid transformations were used to augment the dataset during training. We investigated multiple loss functions, including Dice similarity coefficient (DSC), cross-entropy (CE), Euclidean loss as well as the variation and combinations of these, data augmentation, and network size on overall performance and sensitivity to image artifacts due to infrequent events such as the presence of implanted devices. The predicted label maps were compared to the ground-truth labels via DSC and mean and 90th percentile symmetric surface distance (90th-SSD). RESULTS: When modified Dice combined with cross-entropy (MD-CE) was used as the loss function, the algorithm achieved a mean DSC = 0.79 ± 0.07 for chambers and  0.39 ± 0.10 for smaller substructures (valves and LAD). The mean and 90th-SSD were 2.7 ± 1.4 and 6.5 ± 2.8 mm for chambers and 4.1 ± 1.7 and 8.6 ± 3.2 mm for smaller substructures. Models with MD-CE, Dice-CE, MD, and weighted CE loss had highest performance, and were statistically similar. Data augmentation did not affect model performances on both clean and outlier data and model robustness was susceptible to network size. For a certain type of outlier data, robustness can be improved via incorporating them into the training process. The execution time for segmenting each patient was on an average 2.1 s. CONCLUSIONS: A deep neural network provides a fast and accurate segmentation of large cardiac substructures in noncontrast CT images. Model robustness of two types of clinically common outlier data were investigated and potential approaches to improve them were explored. Evaluation of clinical acceptability and integration into clinical workflow are pending.

Implementing stereotactic accelerated partial breast irradiation using magnetic resonance guided radiation therapy.

INTRODUCTION: Accelerated partial breast irradiation (APBI) seeks to reduce irradiated volumes and radiation exposure for patients while maintaining acceptable clinical outcomes. Magnetic resonance image-guided radiotherapy (MRgRT) provides excellent soft-tissue contrast for treatment localization, which can reduce setup uncertainty, thus reducing margins in the external beam setting. Additionally, stereotactic body radiotherapy (SBRT)-style regimens with high gradients can also be executed. This MR-guided stereotactic APBI (MRgS-APBI) approach can be utilized for a lower number of fractions and spare a greater volume of healthy tissues compared to conventional 3D external beam APBI. METHODS: Our MRgS-APBI program was developed for two prospective non-randomized phase I/II clinical trials (20Gyx1 and 8.5Gyx3). Both breast SBRT treatment planning and MRgRT delivery techniques were described in this study. Simulation included both CT and MRI with specialized immobilization to accommodate MR-guided setup and cine-MRI treatment gating. Dosimetry data from 48 single-fraction and 19 three-fraction patients were collected and evaluated. This included planning objectives and SBRT-specific indices. During treatment, setup errors were calculated to evaluate setup reproducibility and duty cycle was calculated using cine-MRI data during gated delivery. RESULTS: In both the single- and three- fraction trials combined, 88.5% of the possible dosimetric objectives across all patients were met during planning. The majority of the planning objectives were easily achievable indicating the potential for stricter objectives for subsequent S-APBI treatments. The average magnitude of setup uncertainties was 1.0 cm ±â€¯0.6 cm across all treatments. In the three-fraction trial, the average beam-on duty-cycle for the MRI-gated delivery was 83.0 ±â€¯13.0%. There were no technical MRgS-APBI related issues that resulted in discontinuation of treatment across all patients. CONCLUSION: SBRT-style dosimetry and delivery for APBI is feasible using MR-guidance. The program development and dosimetric outcomes reported here can serve as a guide for other institutions considering the clinical implementation of MR-guided stereotactic APBI.

Characterization of Dosimetric Differences in Strut-Adjusted Volume Implant Treatment Plans Calculated With TG-43 Formalism and a Model-Based Dose Calculation Algorithm.

PURPOSE: To comprehensively characterize dosimetric differences between calculations with a commercial model-based dose calculation algorithm (MBDCA) and the TG-43 formalism in application to accelerated partial breast irradiation (APBI) with the strut-adjusted volume implant (SAVI) applicator. METHODS: Dose for 100 patients treated with the SAVI applicator was recalculated with an MBDCA for comparison to dose calculated via TG-43. For every pair of dose calculations, dose-volume histogram (DVH) metrics including V90%, V95%, V100%, V150%, and V200% for the PTV_EVAL were compared. Features were defined for each case including (1) applicator size, (2) ratio between PTV_EVAL contour and 1-cm rind surrounding SAVI applicator, (3) ratio between dwell time in central catheter and total dwell time, and (4) mean computed tomography (CT) number within the lumpectomy cavity. Wilcoxon rank sum tests were performed to test whether treatment plans could be stratified according to feature values into groups with statistically significant dosimetry differences between MBDCA and TG-43. RESULTS: For all DVH metrics, differences between TG-43 and MBDCA calculations were statistically significant (P < .05). Minimum (maximum) relative percent differences between the MBDCA and TG-43 for V90%, V95%, and V100% were -2.1% (0.1%), -3.1% (-0.1%), and -5.0% (-0.5%), respectively. The median relative percent difference in mean PTV_EVAL dose between the MBDCA and TG-43 was -3.9%, with minimum (maximum) difference of -6.5% (-1.8%). For V90%, V95%, and V100%, plan quality worsened beyond defined thresholds in 26, 23, and 31 cases with no instances of coverage improvement. Features 1, 2, and 4 were shown to be able to stratify treatment plans into groups with statistically significant differences in dosimetry metrics between MBDCA and TG-43. CONCLUSIONS: Investigated dose metrics for SAVI treatments were found to be systematically lower with MBDCA calculation in comparison to TG-43. Plans could be stratified according to several features by the magnitude of dosimetric differences between these calculations.

Radiation-Induced Brachial Plexopathy in Patients With Breast Cancer Treated With Comprehensive Adjuvant Radiation Therapy.

PURPOSE: Our purpose was to describe the risk of radiation-induced brachial plexopathy (RIBP) in patients with breast cancer who received comprehensive adjuvant radiation therapy (RT). METHODS AND MATERIALS: Records for 498 patients who received comprehensive adjuvant RT (treatment of any residual breast tissue, the underlying chest wall, and regional nodes) between 2004 and 2012 were retrospectively reviewed. All patients were treated with conventional 3 to 5 field technique (CRT) until 2008, after which intensity modulated RT (IMRT) was introduced. RIBP events were determined by reviewing follow-up documentation from oncologic care providers. Patients with RIBP were matched (1:2) with a control group of patients who received CRT and a group of patients who received IMRT. Dosimetric analyses were performed in these patients to determine whether there were differences in ipsilateral brachial plexus dose distribution between RIBP and control groups. RESULTS: Median study follow-up was 88 months for the overall cohort and 92 months for the IMRT cohort. RIBP occurred in 4 CRT patients (1.6%) and 1 IMRT patient (0.4%) (P = .20). All patients with RIBP in the CRT cohort received a posterior axillary boost. Maximum dose to the brachial plexus in RIBP, CRT control, and IMRT control patients had median values of 56.0 Gy (range, 49.7-65.1), 54.8 Gy (47.4-60.5), and 54.8 Gy (54.2-57.3), respectively. CONCLUSIONS: RIBP remains a rare complication of comprehensive adjuvant breast radiation and no clear dosimetric predictors for RIBP were identified in this study. The IMRT technique does not appear to adversely affect the development of this late toxicity.

Long-Term Outcomes with 3-Dimensional Conformal External Beam Accelerated Partial Breast Irradiation.

PURPOSE: Long-term tumor control and cosmetic outcomes for accelerated partial breast radiation (APBI) delivered with 3-dimensional conformal external beam radiation (3D-CRT) remain limited. We seek to address these concerns by reporting our experience of 3D-CRT APBI with extended follow-up. METHODS AND MATERIALS: All patients treated with APBI delivered with 3D-CRT from January 2006 through December 2012 at a single institution were identified. Those with more than a year of follow-up were analyzed for ipsilateral breast tumor recurrence (IBTR), progression-free survival (PFS), cosmesis, and pain. Disease outcomes were analyzed by margin status (<2 mm, ≥2 mm), total radiation dose prescribed, presence of invasive disease, and American Society for Radiation Oncology (ASTRO) 2016 updated consensus groupings (suitable, cautionary, and unsuitable). RESULTS: Two hundred ninety-three patients were identified, of whom 266 had >1 year of follow-up. Median follow-up was 87 months (range, 13-156). Of the 266, 162 (60.9%) were ASTRO "suitable," 87 (32.7%) were "cautionary," and 17 (6.4%) were "unsuitable." Seven-year rates of IBTR and PFS were 1.8% and 95.2%, respectively. Margin status, invasive versus in situ disease, prescribed dose, and ASTRO grouping were not prognostic for either IBTR or PFS on univariate analysis. Cosmesis was good to excellent in 75.2%. Two patients (0.8%) had subsequent plastic surgery owing to poor cosmesis. Narcotic medication for treatment site pain was needed by 6 (2.3%). CONCLUSIONS: External beam APBI results in excellent long-term disease control. Good to excellent cosmetic outcomes are achieved in most patients, although increasing dose per fraction and greater percentage of irradiated breast were predictive of adverse posttreatment cosmetic outcomes. Select patients in "cautionary" and "unsuitable" consensus groupings do not appear to have inferior outcomes.

Single-Institution Phase 1/2 Prospective Clinical Trial of Single-Fraction, High-Gradient Adjuvant Partial-Breast Irradiation for Hormone Sensitive Stage 0-I Breast Cancer.

PURPOSE: We sought to evaluate the feasibility and tolerability of a novel accelerated partial breast irradiation regimen delivered in a single fraction postoperatively. METHODS AND MATERIALS: We enrolled 50 patients with low-risk, hormone-sensitive breast cancer from 2015 to 2018 on a prospective phase 1/2 trial to receive single-fraction, high-gradient partial-breast irradiation (SFHGPBI) 2 to 8 weeks after lumpectomy for node-negative, invasive, or in situ breast cancer. The high gradient was achieved by prescribing 20 Gy to the surgical bed and 5 Gy to the breast tissue within 1 cm of the surgical bed simultaneously in 1 fraction using external beam. RESULTS: The median age was 65 (range, 52-84). Ten patients (20%) had small-volume ductal carcinoma in situ while the remainder had stage I disease. At a median follow-up of 25 months, we evaluated toxicity, patient- and physician-reported cosmesis, patient-reported quality of life (QOL), and initial tumor control. There was no Common Terminology Criteria for Adverse Events v4.0 grade 3+ toxicity. Only 34% of patients experienced grade 1 erythema. Good-to-excellent pretreatment cosmesis was present in 100% and 98% per physicians and patients, respectively, and did not change post-SFHGPBI. Quantitative cosmesis by percentage of breast retraction assessment significantly improved over time during the post-SFHGPBI period per mixed repeated measures modeling (P = .0026). QOL per European Organization for Research and Treatment of Cancer QOL Questionnaires C30 and BR-23 did not decline other than temporarily in the systemic therapy effects and hair loss domains, both of which returned to pretreatment values. There was 1 noninvasive in-breast recurrence in a separate untreated quadrant 18 months post-SFHGPBI and 1 isolated axillary recurrence 30 months post-SFHGPBI, both salvaged successfully. There were no distant recurrences or cancer-related deaths observed. CONCLUSIONS: Accelerated partial-breast irradiation delivered in a single fraction postoperatively using external beam techniques is a novel, feasible, well-tolerated regimen. SFHGPBI does not adversely affect cosmesis or QOL as reported by both physicians and patients. Initial tumor control rates are excellent, with longer follow-up required to confirm efficacy.

Immunosuppression enhances oncolytic adenovirus replication and antitumor efficacy in the Syrian hamster model.

We recently described an immunocompetent Syrian hamster model for oncolytic adenoviruses (Ads) that permits virus replication in tumor cells as well as some normal tissues. This model allows exploration of interactions between the virus, tumor, normal organs, and host immune system that could not be examined in the immunodeficient or nonpermissive animal models previously used in the oncolytic Ad field. Here we asked whether the immune response to oncolytic Ad enhances or limits antitumor efficacy. We first determined that cyclophosphamide (CP) is a potent immunosuppressive agent in the Syrian hamster and that CP alone had no effect on tumor growth. Importantly, we found that the antitumor efficacy of oncolytic Ads was significantly enhanced in immunosuppressed animals. In animals that received virus therapy plus immunosuppression, significant differences were observed in tumor histology, and in many cases little viable tumor remained. Notably, we also determined that immunosuppression allowed intratumoral virus levels to remain elevated for prolonged periods. Although favorable tumor responses can be achieved in immunocompetent animals, the rate of virus clearance from the tumor may lead to varied antitumor efficacy. Immunosuppression, therefore, allows sustained Ad replication and oncolysis, which leads to substantially improved suppression of tumor growth.

Clinical outcomes and patterns of care in the treatment of carcinosarcoma of the breast.

PURPOSE: Carcinosarcoma of the breast is a rare yet highly aggressive tumor accounting for <1% of all breast cancers, for which guidance on optimal management and prognosis are sparse. The purpose of this study was to investigate population-based treatment patterns and overall survival (OS) outcomes in patients with this diagnosis. MATERIALS AND METHODS: We queried the National Cancer Database for patients diagnosed with carcinosarcoma of the breast. All patients included were treated with surgery in the form of mastectomy or lumpectomy, with or without chemotherapy and/or radiation therapy. Patients with metastatic disease were excluded. Kaplan-Meier analysis was used to estimate OS. Univariate and multivariable Cox analyses were used to determine predictive factors of OS. RESULTS: A total of 329 patients from 2004 to 2012 were identified. Median age at diagnosis was 58 years (range, 24-90). Patients had T1 (21%), T2 (44%), T3 (25%), or T4 disease (10%). Most patients were node-negative at diagnosis (77%). Breast conservation surgery was utilized in 33% of patients. Chemotherapy was used in 66% of patients. Less than half (44%) of patients received radiation therapy to a median dose of 50.4 Gy (range 35-56 Gy), with a median 10 Gy boost used in 76%. With a median follow-up of 40.0 months, 3- and 5-year OS for all patients was 74% and 60%, respectively. Kaplan-Meier estimates revealed the 3-yr OS was 80% in patients receiving chemotherapy vs 59% without chemotherapy (P < 0.001). The 3-yr OS was 82% in patients receiving RT vs 66% without RT (P = 0.001). On multivariable analysis, OS was significantly influenced by Charlson-Deyo comorbidity index, insurance status, clinical T stage, surgical margin status, and treatment group, with trimodality therapy (HR: 0.45, 95% CI: 0.27-0.78; P = 0.004) and surgery plus CT (HR: 0.54, 95% CI: 0.33-0.90; P = 0.02) being associated with the greatest OS. Logistic regression revealed only younger patients were more likely to receive trimodality therapy. CONCLUSIONS: Carcinosarcoma of the breast is associated with relatively poor rates of OS. The addition of CT and RT to surgery improves OS. Trimodality therapy and surgery plus CT were associated with the greatest OS compared to surgery alone.

Clinical outcomes and toxicity of proton beam radiation therapy for re-irradiation of locally recurrent breast cancer.

PURPOSE: Repeat radiation therapy (RT) using photons/X-rays for locally recurrent breast cancer results in increased short and long-term toxicity. Proton beam RT (PBRT) can minimize dose to surrounding organs, thereby potentially reducing toxicity. Here, we report the toxicity and clinical outcomes for women who underwent re-irradiation to the chest wall for locally recurrent breast cancer using PBRT. MATERIALS AND METHODS: This was a retrospective study analyzing 16 consecutive patients between 2013 and 2018 with locally recurrent breast cancer who underwent re-irradiation to the chest wall with PBRT. For the recurrent disease, patients underwent maximal safe resection, including salvage mastectomy, wide local excision, or biopsy only per surgeons recommendations. Systemic therapy was used per the recommendation of the medical oncologist. Patients were treated with median dose of 50.4 Cobalt Gray Equivalent (CGyE) in 28 fractions at the time of re-irradiation. Follow-up was calculated from the start of second RT course. Acute toxicities were defined as those occurring during treatment or up to 8 weeks after treatment. Late toxicities were defined as those occurring more than 8 weeks after the completion of therapy. Toxicities were based on CTCAE 4.0. RESULTS: The median age at original diagnosis and at recurrence was 49.8 years and 60.2 years, respectively. The median time between the two RT courses was 10.2 (0.7-20.2) years. The median follow-up time was 18.7 (2.5-35.2) months. No local failures were observed after re-irradiation. One patient developed distant metastasis and ultimately died. Grade 3-4 acute skin toxicity was observed in 5 (31.2%) patients. Four (25%) patients developed chest wall infections during or shortly (2 weeks) after re-irradiation. Late grade 3-4 fibrosis was observed in only 3 (18.8%) patients. Grade 5 toxicities were not observed. Hyperpigmentation was seen in 12 (75%) patients. Pneumonitis, telangiectasia, rib fracture, and lymphedema occurred in 2 (12.5%), 4 (25%), 1 (6.3%), and 1 (6.3%) patients, respectively. CONCLUSIONS: Re-irradiation with PBRT for recurrent breast cancer has acceptable toxicities. There was a high incidence of acute grade 3-4 skin toxicity and infections, which resolved, however, with skin care and antibiotics. Longer follow-up is needed to determine long-term clinical outcomes.

Standardization and automation of quality assurance for high-dose-rate brachytherapy planning with application programming interface.

PURPOSE: To standardize and automate the high-dose-rate (HDR) brachytherapy planning quality assurance (QA) process utilizing scripting with application programming interface (API) in a commercially available treatment planning system (TPS). METHODS AND MATERIALS: Site- and applicator-dependent plan quality (PQ) evaluation criteria and plan integrity (PI) checklists were established based on published guidelines, clinical protocols, and institutional experience. User designed C# programs ("scripts") were created and executed through the API to access planning information in TPS. A set of standardized quality control reports, focusing on PQ evaluations and PI checks, were automatically generated. Information derived from the TPS was compared against predetermined QA metrics with color-coded pass/fail indicators to aid and enhance the efficiency of plan evaluation. Five independent, blinded observers reviewed mock plans with simulated errors to validate the scripts and to quantify the improvement of plan review efficiency. RESULTS: Scripts were developed for HDR prostate and breast. Forty-one parameters were reported/checked in the PI report; the PQ report returned dose-volume indices and an independent check of dwell time. All simulated errors were detected by the PI scripts with appropriate warning messages displayed, and any values failing to meet the planning constraints were red-flagged successfully in the PQ report. An average time reduction of 16 min for plan review was observed when using the scripts. CONCLUSIONS: API scripting-based automated planning QA for HDR brachytherapy including PI checks and PQ evaluations was designed and implemented. The simulated error study showed promising results in terms of error catching and efficiency improvement.

Synthetic CT reconstruction using a deep spatial pyramid convolutional framework for MR-only breast radiotherapy.

PURPOSE: The superior soft-tissue contrast achieved using magnetic resonance imaging (MRI) compared to x-ray computed tomography (CT) has led to the popularization of MRI-guided radiation therapy (MR-IGRT), especially in recent years with the advent of first and second generation MRI-based therapy delivery systems for MR-IGRT. The expanding use of these systems is driving interest in MRI-only RT workflows in which MRI is the sole imaging modality used for treatment planning and dose calculations. To enable such a workflow, synthetic CT (sCT) data must be generated based on a patient's MRI data so that dose calculations may be performed using the electron density information derived from CT images. In this study, we propose a novel deep spatial pyramid convolutional framework for the MRI-to-CT image-to-image translation task and compare its performance to the well established U-Net architecture in a generative adversarial network (GAN) framework. METHODS: Our proposed framework utilizes atrous convolution in a method named atrous spatial pyramid pooling (ASPP) to significantly reduce the total number of parameters required to describe the model while effectively capturing rich, multi-scale structural information in a manner that is not possible in the conventional framework. The proposed framework consists of a generative model composed of stacked encoders and decoders separated by the ASPP module, where atrous convolution is applied at increasing rates in parallel to encode large-scale features. The performance of the proposed method is compared to that of the conventional GAN framework in terms of the time required to train the model and the image quality of the generated sCT as measured by the root mean square error (RMSE), structural similarity index (SSIM), and peak signal-to-noise ratio (PSNR) depending on the size of the training data set. Dose calculations based on sCT data generated using the proposed architecture are also compared to clinical plans to evaluate the dosimetric accuracy of the method. RESULTS: Significant reductions in training time and improvements in image quality are observed at every training data set size when the proposed framework is adopted instead of the conventional framework. Over 1042 test images, values of 17.7 ± 4.3 HU, 0.9995 ± 0.0003, and 71.7 ± 2.3 are observed for the RMSE, SSIM, and PSNR metrics, respectively. Dose distributions calculated based on sCT data generated using the proposed framework demonstrate passing rates equal to or greater than 98% using the 3D gamma index with a 2%/2 mm criterion. CONCLUSIONS: The deep spatial pyramid convolutional framework proposed here demonstrates improved performance compared to the conventional GAN framework that has been applied to the image-to-image translation task of sCT generation. Adopting the method is a first step toward an MRI-only RT workflow that enables widespread clinical applications for MR-IGRT including online adaptive therapy.

Delineation of a Cardiac Planning Organ-At-Risk Volume Using Real-Time Magnetic Resonance Imaging for Cardiac Protection in Thoracic and Breast Radiation Therapy.

PURPOSE: Cardiac radiation is associated with cardiotoxicity in patients with thoracic and breast malignancies. We conducted a prospective study using cine magnetic resonance imaging (MRI) scans to evaluate heart motion. We hypothesized that cine MRI could be used to define population-based cardiac planning organ-at-risk volumes (PRV). METHODS AND MATERIALS: A total of 16 real-time acquisitions were obtained per subject on a 1.5 Tesla MRI (Philips Ingenia). Planar cine MRI was performed in 4 sequential sagittal and coronal planes at free-breathing (FB) and deep-inspiratory breath hold (DIBH). In-plane cardiac motion was assessed using a scale-invariant feature transformation-based algorithm. Subject-specific pixel motion ranges were defined in anteroposterior (AP), left-right (LR), and superoinferior (SI) planes. Averages of the 98% and 67% of the maximum ranges of pixel displacement were defined by subject, then averaged across the cohort to calculate PRV expansions at FB and DIBH. RESULTS: Data from 20 subjects with a total of 3120 image frames collected per subject in coronal and sagittal planes at DIBH and FB, and 62,400 total frames were analyzed. Cohort averages of 98% of the maximum cardiac motion ranges comprised margin expansions of 12.5 ± 1.1 mm SI, 5.8 ± 1.2 mm AP, and 6.6 ± 1.0 mm LR at FB and 6.7 ± 1.5 mm SI, 4.7 ± 1.3 mm AP, and 5.3 ± 1.3 mm LR at DIBH. Margins for 67% of the maximum range comprised 7.7 ± 0.7 mm SI, 3.2 ± 0.6 mm AP, and 3.7 ± 0.6 mm LR at FB and 4.1 ± 0.9 mm SI, 2.7 ± 0.8 mm AP, and 3.2 ± 0.8 mm LR at DIBH. Subsequently, these margins were simplified to form PRVs for treatment planning. CONCLUSIONS: We implemented scale-invariant feature transformation-based motion tracking for analysis of the cardiac cine MRI scans to quantify motion and create cohort-based cardiac PRVs to improve cardioprotection in breast and thoracic radiation.

Syrian hamster as a permissive immunocompetent animal model for the study of oncolytic adenovirus vectors.

Oncolytic adenoviruses represent an innovative approach to cancer therapy. These vectors are typically evaluated in immunodeficient mice with human xenograft tumors. However, in addition to being immunodeficient, this model is limited because normal and cancerous mouse tissues are poorly permissive for human adenovirus replication. This prompted us to search for a model that more accurately reflects the use of oncolytic adenoviruses in cancer patients. To this end, we developed a novel Syrian hamster model that is both immunocompetent and replication-permissive. We found that human adenovirus replicates well in Syrian hamster cell lines and confirmed replication in the lungs. Oncolytic adenovirus injection showed efficacy in three different hamster tumor models. Furthermore, i.t. oncolytic adenovirus injection resulted in suppression of primary and metastatic lesions, i.t. replication and necrosis, vector entrance into the bloodstream, replication in the liver and lungs, and anti-adenovirus antibody induction. Our findings show that the Syrian hamster is a promising immunocompetent model that is permissive to human adenovirus replication in tumors as well as normal tissues. Therefore, the Syrian hamster model may become a valuable tool for the field of oncolytic adenovirus vectors in which vector safety and efficacy can be evaluated.

Efficiency of using the day-of-implant CT for planning of SAVI APBI.

PURPOSE: The purpose of the study was to develop an optimized, efficient workflow for using the day-of-implant (DOI) CT for treatment planning of accelerated partial breast irradiation brachytherapy using the strut-adjusted volume implant (SAVI) device. METHODS AND MATERIALS: For 62 consecutive SAVI patients, a DOI CT was acquired and used for treatment planning. A "verification" CT was acquired 24-72 h after implant and immediately before the first fraction, then registered to the DOI CT. If the DOI CT-based plan was no longer optimal, a replan was performed. An array of metrics describing the geometry of the device and its relative position in the patient from the DOI CTs for these patients was collected. These metrics from the DOI CT were evaluated to determine what features could predict for the need to replan before the first treatment fraction. Logistical regression analysis including χ2 tests was used to determine if different factors correlated with replanning. RESULTS: Twenty-two of 62 patients (35%) required replanning. Only the presence of splayed struts, where splay was toward the skin, and the use of a nine strut ("8-1") SAVI were significantly correlated (p < 0.05) with replanning. Within these individual populations, no additional factors showed a significant statistical correlation for requiring replanning. CONCLUSIONS: For strut-based accelerated partial breast irradiation brachytherapy, it was feasible to treat with a plan based on the DOI CT for a majority (65%) of patients. Some factors correlate to needing replanning; recognizing these could be used to optimize treatment workflow for certain patients, increasing clinical efficiency while enhancing the quality of patient care.