Search

Performing measurement-based patient-specific quality assurance (PSQA) is recognized as a resource-intensive and time inefficient task in the radiation therapy treatment workflow. Paired with technological refinements in modern radiation therapy, research toward measurement-free PSQA has seen increased interest during the past 5 years. However, these efforts have not been clinically implemented or prospectively validated in the United States. We propose a virtual QA (VQA) system and workflow to assess the safety and workload reduction of measurement-free PSQA.An XGBoost machine learning model was designed to predict PSQA outcomes of volumetric modulated arc therapy plans, represented as percent differences between the measured ion chamber point dose in a phantom and the corresponding planned dose. The final model was deployed within a web application to predict PSQA outcomes of clinical plans within an existing clinical workflow. The application also displays relevant feature importance and plan-specific distribution analyses relative to database plans for documentation and to aid physicist interpretation and evaluation. VQA predictions were prospectively validated over 3 months of measurements at our clinic to assess safety and efficiency gains.Over 3 months, VQA predictions for 445 volumetric modulated arc therapy plans were prospectively validated at our institution. VQA predictions for these plans had a mean absolute error of 1.08% ± 0.77%, with a maximum absolute error of 2.98%. Using a 1% prediction threshold (ie, plans predicted to have an absolute error1% would not require a measurement) would yield a 69.2% reduction in QA workload, saving 32.5 hours per month on average, with 81.5% sensitivity, 72.4% specificity, and an area under the curve of 0.81 at a 3% clinical threshold and 100% sensitivity, 70% specificity, and an area under the curve of 0.93 at a 4% clinical threshold.This is the first prospective clinical implementation and validation of VQA in the United States, which we observed to be efficient. Using a conservative threshold, VQA can substantially reduce the number of required measurements for PSQA, leading to more effective allocation of clinical resources.

Purpose To suggest PTV margins for liver SBRT with different motion management strategies based on a systematic review and meta-analysis. Methods In accordance with Preferred-Reporting-Items-for-Systematic-Reviews-and-Meta-Analyses (PRISMA), a systematic review in PubMed, Embase and Medline databases was performed for liver tumor position variability. From an initial 533 studies published before October 2020, 36 studies were categorized as 18 free-breathing (FB; npatients=401), 9 abdominal compression (AC; npatients=145) and 9 breath-hold (BH; npatients=126). A meta-analysis was performed on inter- and intra-fraction position variability to report weighted-mean with 95% confidence interval (CI95) in superior-inferior (SI), left-right (LR) and anterior-posterior (AP) directions. Furthermore, weighted-mean ITV margins were computed for FB (nstudies=15, npatients =373) and AC (nstudies=6, npatients =97) and PTV margins were computed for FB (nstudies =6, npatients =95), AC (nstudies =7, npatients =106) and BH (nstudies =8, npatients =133). Results The FB weighted-mean intra-fraction variability, ITV margins and weighted-standard-deviation in mm were SI-9.7, CI95=9.3-10.1, 13.5±4.9; LR-5.4, CI95=5.3-5.6, 7.3±7.9 and AP-4.2, CI95=4.0-4.4, 6.3±7.6. The inter-fraction-based results were SI-4.7, CI95=4.3-5.1, 5.7±1.7; LR-1.4, CI95=1.1-1.6, 3.6±2.7 and AP-2.8, CI95=2.5-3.1, 4.8±2.1. For AC intra-fraction results in mm were SI-1.8, CI95=1.6-2.0, 2.6±1.2; LR-0.7, CI95=0.6-0.8, 1.7±1.5 and AP-0.9, CI95=0.8-1.0, 1.9±1.7. The inter-fraction results were SI-2.6, CI95=2.3-3.0, 5.2±2.9; LR-1.9, CI95=1.7-2.1, 4.0±2.2 and AP-2.9, CI95=2.5-3.2, 5.8±2.7. For BH the inter-fraction variability and the weighted-mean PTV margins and weighted-standard-deviation in mm were SI-2.4, CI95=2.1-2.7, 5.6±2.9; LR-1.8; CI95=1.3-2.2, 5.5±1.7 and AP-1.4; CI95=1.2-1.7, 6.1±2.1. Conclusion Our meta-analysis suggested that weighted-mean PTV margin of 6mm might be appropriate for BH. For AC and FB, asymmetric PTV margins (weighted-mean margin = 4mm (AP), 6mm (SI/LR)) might be appropriate. For FB, if larger (> ITV margin) intra-fraction variability observed, the additional intra- and inter-fraction variability should be accounted in the PTV margin.

BACKGROUND Disparity in mental health care among cancer patients remains understudied. METHODS A large, retrospective, single tertiary-care institution cohort study was conducted based on deidentified electronic health record data of 54,852 adult cancer patients without prior mental health diagnosis (MHD) diagnosed at the University of California, San Francisco between January 2012 and September 2019. The exposure of interest was early-onset MHD with or without psychotropic medication (PM) within 12 months of cancer diagnosis and primary outcome was all-cause mortality. RESULTS There were 8.2% of patients who received a new MHD at a median of 197 days (interquartile range, 61-553) after incident cancer diagnosis; 31.0% received a PM prescription; and 3.7% a mental health-related visit (MHRV). There were 62.6% of patients who were non-Hispanic White (NHW), 10.8% were Asian, 9.8% were Hispanic, and 3.8% were Black. Compared with NHWs, minority cancer patients had reduced adjusted odds of MHDs, PM prescriptions, and MHRVs, particularly for generalized anxiety (Asian odds ratio [OR], 0.66, 95% CI, 0.55-0.78; Black OR, 0.60, 95% CI, 0.45-0.79; Hispanic OR, 0.72, 95% CI, 0.61-0.85) and selective serotonin-reuptake inhibitors (Asian OR, 0.43, 95% CI, 0.37-0.50; Black OR, 0.51, 95% CI, 0.40-0.61; Hispanic OR, 0.79, 95% CI, 0.70-0.89). New early MHD with PM was associated with elevated all-cause mortality (12-24 months: hazard ratio [HR], 1.43, 95% CI, 1.25-1.64) that waned by 24 to 36 months (HR, 1.18, 95% CI, 0.95-1.45). CONCLUSIONS New mental health diagnosis with PM was a marker of early mortality among cancer patients. Minority cancer patients were less likely to receive documentation of MHDs or treatment, which may represent missed opportunities to identify and treat cancer-related mental health conditions.

OBJECTIVE It remains a challenge to predict individual risk for recurrence after primary treatment of ductal carcinoma in situ (DCIS). While DCIS is contained within the duct, prior studies have pointed to the importance of stromal collagen in cancer progression. Using hematoxylin and eosin (H&E) stain-based images, we studied the collagen structure surrounding DCIS in a case-control cohort of women to determine whether stromal collagen could provide additional risk information. METHODS We present a quantitative histology image analysis pipeline (termed “histomics”) that consists of collagen segmentation, quantitative feature extraction and statistical analysis to predict recurrence. A cohort of 73 patients with DCIS (33 cases with recurrence matched with 40 controls) were retrospectively analyzed. H&E images were obtained from pathologic slides and regions-of-interest (ROI) of the collagen in the stroma adjacent to DCIS was automatically segmented using an in-house trained deep learning algorithm. A total of 123 histomics features (intensity, statistical and textural) were then extracted from the segmented ROI for each patient. Kruskal-Wallis test and receiver-operating characteristic methodology was used to assess performance of the features to differentiate patients with and without recurrence. RESULTS We found substantial variations in recorded H&E signal between samples, and the intensity and statistical features did not reach statistical significance. However, twelve percent of the textural features had an AUC between 0.62 to 0.61 with p-value between 0.06 to 0.08. Two textural histomics features in particular were significantly associated with DCIS patient recurrence; inverse variance (AUC: 0.64, p-value: 0.046) feature based on grey-level cooccurrences matrix (GLCM) and complexity (AUC 0.65, p-value: 0.025) feature based on neighborhood grey tone difference matrix (NGTDM). CONCLUSION Textural stromal features extracted from H&E image-based histomics indicate association of some features with tumor progression in DCIS. H&E image based histomics features can be a promising quantitative tool to develop predictive signatures of recurrence in patients with DCIS. We are processing additional patients through our quantitative image analysis pipeline to determine the significance of histomics features compared to conventional clinical features for the prediction of DCIS recurrence. Citation Format: Taman Upadhaya, Ronald Balassanian, Olivier Morin, Catherine Park. Prediction of breast ductal carcinoma in situ recurrence using histomics analysis of stromal features from hematoxylin and eosin stain-based images [abstract]. In: Proceedings of the 2020 San Antonio Breast Cancer Virtual Symposium; 2020 Dec 8-11; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2021;81(4 Suppl):Abstract nr PS4-43.

PurposeThe aim was to develop a novel artificial intelligence (AI)-guided clinical decision support system, to predict radiation doses to subsites of the mandible using diagnostic computed tomography scans acquired before any planning of head and neck radiation therapy (RT).Methods and materialsA dose classifier was trained using RT plans from 86 patients with oropharyngeal cancer; the test set consisted of an additional 20 plans. The classifier was trained to predict whether mandible subsites would receive a mean dose >50 Gy. The AI predictions were prospectively evaluated and compared with those of a specialist head and neck radiation oncologist for 9 patients. Positive predictive value (PPV), negative predictive value (NPV), Pearson correlation coefficient, and Lin concordance correlation coefficient were calculated to compare the AI predictions to those of the physician.ResultsIn the test data set, the AI predictions had a PPV of 0.95 and NPV of 0.88. For 9 patients evaluated prospectively, there was a strong correlation between the predictions of the AI algorithm and physician (P=.72, P < .001). Comparing the AI algorithm versus the physician, the PPVs were 0.82 versus 0.25, and the NPVs were 0.94 versus 1.0, respectively. Concordance between physician estimates and final planned doses was 0.62; this was 0.71 between AI-based estimates and final planned doses.ConclusionAI-guided decision support increased precision and accuracy of pre-RT dental dose estimates.

OBJECTIVE. A clinical challenge of ductal carcinoma in situ (DCIS) is to accurately predict individual risk for recurrence. Prior studies have highlighted the importance of stromal collagen surrounding breast ducts in breast cancer aggression. We studied the collagen structure surrounding DCIS in a matched case-control cohort of women with DCIS using second-harmonic generation (SHG) and hematoxylin and eosin (H&E) images to determine whether stromal collagen could provide additional risk information. METHODS. A cohort of 122 patients with DCIS (61 cases with recurrence matched with 61 controls) were retrospectively analyzed. Whole-side SHG and H&E images were obtained and regions-of-interest (ROI) of stromal collagen adjacent to DCIS was automatically segmented using an in-house trained deep learning algorithm. A quantitative histology image analysis pipeline (termed “histomics”) was exploited to extract a total of 123 histomics features (intensity, statistical and textural) from the segmented stromal region from both SHG and H&E ROI images. Clinical features previously shown to associate with DCIS recurrence including tumor size, margins and treatment were also collected. Conditional logistic regression model with lasso penalty was used to assess the performance of these features to stratify patients with and without recurrence. RESULTS. We found the combination of SHG, H&E and clinical features reached the highest concordance of 0.93 and the combination of only SHG and H&E features reached a concordance of 0.91. In contrast, clinical features alone attained a concordance index of 0.86. Model comparison by the Akaike information criterion (AIC​) shows that the higher concordance of the combination of SHG, H&E and clinical features may indicate better discrimination, whereas the higher concordance of combination of only SHG and H&E features was merely the result of including the larger number of features. Various textural features were found to be statistically significant (p Citation Format: Taman Upadhaya, Mary-Kate Hayward, Mi-Ok Kim, Ronald Balassanian, Valerie Weaver, Olivier Morin, Catherine Park. Prediction of breast ductal carcinoma in situ recurrence using histomics analysis of stromal collagen from second-harmonic generation and hematoxylin and eosin stain-based images [abstract]. In: Proceedings of the 2021 San Antonio Breast Cancer Symposium; 2021 Dec 7-10; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2022;82(4 Suppl):Abstract nr P1-02-16.

Despite widespread adoption of electronic health records (EHRs), most hospitals are not ready to implement data science research in the clinical pipelines. Here, we develop MEDomics, a continuously learning infrastructure through which multimodal health data are systematically organized and data quality is assessed with the goal of applying artificial intelligence for individual prognosis. Using this framework, currently composed of thousands of individuals with cancer and millions of data points over a decade of data recording, we demonstrate prognostic utility of this framework in oncology. As proof of concept, we report an analysis using this infrastructure, which identified the Framingham risk score to be robustly associated with mortality among individuals with early-stage and advanced-stage cancer, a potentially actionable finding from a real-world cohort of individuals with cancer. Finally, we show how natural language processing (NLP) of medical notes could be used to continuously update estimates of prognosis as a given individual’s disease course unfolds. Morin and colleagues develop a data-integration framework capable of performing continuous learning from electronic health records on clinical, social and demographic data collected over a decade to estimate pan-cancer survival prognosis.

Purpose/objective(s) Management of progressive brain tumors following high dose radiotherapy is controversial. In select cases, surgical resection can provide rapid relief of regional mass effect and edema, and highly focal adjuvant re-irradiation can be administered in the form of brachytherapy. Permanent low-dose rate cesium-131 brachytherapy may offer dosimetric and radiobiologic advantages over other isotopes. The purpose of this study was to examine the safety and efficacy of intracranial cesium-131 brachytherapy after surgical resection in a heavily pre-treated patient cohort. Materials/methods A retrospective chart review was conducted of all patients treated with intraoperative intracranial cesium-131 brachytherapy from an institutional database. Treatment consisted of maximal safe resection followed by implantation of suture-stranded cesium-131 seeds, maintaining approximately 10mm spacing between seeds, with a planned target dose of 80 Gy to 5mm depth. Local failure was defined as recurrence in or contacting the resection cavity after salvage resection and brachytherapy. Regional failure was defined as recurrence within 10mm of the resection cavity. Recurrence was defined as a new lesion or growth of ≥25% in any axial dimension on interval MRI. Toxicity was collected based on CTCAE 4.0. Results A total of 35 patients with a median imaging follow up of 17.0 months (interquartile range [IQR] 9.7-25.9) underwent 41 consecutive cesium-131 implants between 2016 and 2020. 20 patients (59%; N = 22 implants) were treated for recurrent brain metastasis, 12 patients (34%; N = 16 implants) were treated for recurrent atypical (N = 8) or anaplastic meningioma (N = 4), and 3 patients (9%) were treated for recurrent glioma (N = 2) or hemangiopericytoma (N = 1). A majority of lesions had received prior surgery (N = 28, 66.7%, range 0-3 prior surgeries) and 34 patients had received prior high dose radiation. 34 of 41 (82.9%) tumors had received prior radiosurgery, and 12 (29.2%) had undergone prior external beam radiation (range 50-60 Gy). 17 of 41 lesions (41.5%) had radiologic evidence of adverse radiation/treatment effect (ARE) prior to salvage therapy. Median tumor size was 3.0cm (IQR 2.3-3.7). Overall actuarial local/loco-regional control at 1-year was 91.6%/84.7% (89.8%/88.9% for metastases and 100%/82.5% for meningiomas). Crude loco-regional control was 86.3% for metastases, 75% for meningiomas, and 66.7% for other lesions. Symptomatic ARE following treatment was observed in 9.8% (N = 4) resection cavities. The surgical complication rate was 14.6% (N = 6 total, N = 3 [7.3%] grade 3-5) including 1 peri-operative death. Conclusion Cesium-131 brachytherapy in this heavily pre-treated cohort resulted in good local control and an acceptable rate of symptomatic adverse radiation effect and toxicity and may be a reasonable treatment option for this select population. Author disclosure W.C. Chen: None. M. Lafreniere: None. C. Phuong: None. M. Lometti: None. O. Morin: None. B.P. Ziemer: None. H. Vasudevan: Research Grant; Children's Tumor Foundation. Patent/License Fees/Copyright; Genentech, Eli Lilly. S. Hervey-Jumper: None. P.V. Theodosopoulos: None. S. Magill: None. S.E. Fogh: Independent Contractor; Accuray. J.L. Nakamura: None. L. Boreta: None. P.K. Sneed: None. M.W. McDermott: Chair; Miami Baptist. D. Raleigh: None. S.E. Braunstein: Advisory Board; Radiation Oncology Questions, LLC.

PURPOSE/OBJECTIVE(S) Previous studies have characterized dose gradient patterns for various hypo-fractionated brain radiosurgery (hSRS) treatment platforms. The Gamma Knife Icon (GKI) has consistently exhibited either superior or non-inferior peripheral dose fall-off and normal brain sparing characteristics when compared to other Linac-based hSRS treatment platforms. Given the prevalence of Linac-based treatments, identifying planning techniques to optimize treatment parameters is desirable. In this study, we investigated a novel Linac-based treatment approach that aimed to create the sharpest dose fall-off for hSRS of large brain lesions. MATERIALS/METHODS A cohort of patient cases (n = 10) with single brain lesions (volume 27.6+/-8.1 mL, range 20.-42.1 mL) treated with GKI at our institution were selected as sample test cases for our study. A non-coplanar unconstrained VMAT (NCU-VMAT) treatment planning approach was developed, and its script was implemented on a commercial treatment planning system for clinical Linac treatment equipped with the latest high-definition multi-leaf collimators (MLCs). Treatment plans produced via the NCU-VMAT approach were then compared against further optimized treatment plans from GKI, as well as conventional coplanar and non-coplanar VMAT treatment planning approaches. The comparison was carried out using DVH-derived parameters including target volume coverage, target dose conformity, modified dose fall-off index defined as the volume of 50% prescribed target dose divided by the target volume (PIV50). RESULTS For each case studied, NCU-VMAT achieved practically identical target coverage (0.98 ± 0.01) and Paddick dose conformity index (0.87 ± 0.02) compared to optimized GKI treatments. The mean PIV50 values were 2.99 ± 0.14, 3.77 ± 0.22, and 3.36 ± 0.19 for optimized GKI, conventional coplanar, and non-coplanar VMAT treatment plans, respectively. These results were in excellent agreement with previously published studies. However, the NCU-VMAT technique yielded a mean PIV50 of 2.41 ± 0.07. This represents an improvement of approximately 40% (P < 0.001, paired two-tailed Student t-test) over all existing approaches including GKI. Further analyses revealed that PIV50 of NCU-VMAT approached the theoretical minimum for all the studied cases. CONCLUSION Our results demonstrate that in treatment of large brain tumors, a treatment planning approach using Bremsstrahlung x-ray based Linacs can achieve superior dose fall-off for intracranial hSRS compared to GKI. We dispel the myth that physical characteristics of Cobalt γ-rays are needed to create the sharpest dose fall-off. Given the prevalence of Linac-based treatments compared to GKI, we present a novel optimized planning technique that achieves a new benchmark of sharpest dose fall-off. AUTHOR DISCLOSURE T. Nano: None. O. Morin: None. B.P. Ziemer: None. D. Raleigh: None. L. Boreta: None. J.L. Nakamura: None. S.E. Fogh: Independent Contractor; Accuray. P.K. Sneed: None. S. Hervey-Jumper: None. P.V. Theodosopoulos: None. S.E. Braunstein: Advisory Board; Radiation Oncology Questions, LLC.L. Ma: Patent/License Fees/Copyright; University of California Regents.

Purpose Our purpose was to assess the use of machine learning methods and Mobius 3D (M3D) dose calculation software to reduce the number of physical ion chamber (IC) dose measurements required for patient-specific quality assurance during corona virus disease 2019. Methods and Materials In this study, 1464 inversely planned treatments using Pinnacle or Raystation treatment planning software (TPS) were delivered using Elekta Versa HD and Varian Truebeam and Truebeam STx linear accelerators between June 2018 and November 2019. For each plan, an independent dose calculation was performed using M3D, and an absolute dose measurement was taken using a Pinpoint IC inside the Mobius phantom. The point dose differences between the TPS and M3D calculation and between TPS and IC measurements were calculated. Agreement between the TPS and IC was used to define the ground truth plan failure. To reduce the on-site personnel during the pandemic, 2 methods of receiver operating characteristic analysis (n = 1464) and machine learning (n = 603) were used to identify patient plans that would require physical dose measurements. Results In the receiver operating characteristic analysis, a predelivery M3D difference threshold of 3% identified plans that failed an IC measurement at a 4% threshold with 100% sensitivity and 76.3% specificity. This indicates that fewer than 25% of plans required a physical dose measurement. A threshold of 1% on a machine learning model was able to identify plans that failed an IC measurement at a 3% threshold with 100% sensitivity and 54.3% specificity, leading to fewer than 50% of plans that required a physical dose measurement. Conclusions It is possible to identify plans that are more likely to fail IC patient-specific quality assurance measurements before delivery. This possibly allows for a reduction of physical measurements taken, freeing up significant clinical resources and reducing the required amount of on-site personnel while maintaining patient safety.

Purpose To investigate the prognostic utility of quantitative 3-dimensional magnetic resonance imaging radiomic analysis for primary pediatric embryonal brain tumors. Methods and Materials Thirty-four pediatric patients with embryonal brain tumor with concurrent preoperative T1-weighted postcontrast (T1PG) and T2-weighted fluid-attenuated inversion recovery (FLAIR) magnetic resonance images were identified from an institutional database. The median follow-up period was 5.2 years. Radiomic features were extracted from axial T1PG and FLAIR contours using MATLAB, and 15 features were selected for analysis based on qualitative radiographic features with prognostic significance for pediatric embryonal brain tumors. Logistic regression, linear regression, receiver operating characteristic curves, the Harrell C index, and the Somer D index were used to test the relationships between radiomic features and demographic variables, as well as clinical outcomes. Results Pediatric embryonal brain tumors in older patients had an increased normalized mean tumor intensity ( P = .05, T1PG), decreased tumor volume ( P = .02, T1PG), and increased markers of heterogeneity ( P ≤ .01, T1PG and FLAIR) relative to those in younger patients. We identified 10 quantitative radiomic features that delineated medulloblastoma, pineoblastoma, and supratentorial primitive neuroectodermal tumor, including size and heterogeneity ( P ≤ .05, T1PG and FLAIR). Decreased markers of tumor heterogeneity were predictive of neuraxis metastases and trended toward significance ( P = .1, FLAIR). Tumors with an increased size (area under the curve = 0.7, FLAIR) and decreased heterogeneity (area under the curve = 0.7, FLAIR) at diagnosis were more likely to recur. Conclusions Quantitative radiomic features are associated with pediatric embryonal brain tumor patient age, histology, neuraxis metastases, and recurrence. These data suggest that quantitative 3-dimensional magnetic resonance imaging radiomic analysis has the potential to identify radiomic risk features for pediatric patients with embryonal brain tumors.

Purpose/Objective(s) The 2019 coronavirus pandemic (COVID-19) had a broad impact on the care of cancer patients, including the rapid adoption of telehealth video visits. On March 15, 2020, our institutional leadership recommended transition to video visits, which posed a risk of altering patients’ access to care. We assessed the impact of this rapid transition on demographic patterns in an urban academic radiation oncology department. Materials/Methods Consultation and follow-up visits from the pre-COVID-19 period (January 1, 2019 to March 14, 2020) and COVID-19 period (March 15, 2020 to August 31, 2020) in a single radiation oncology department were identified. Demographics and appointment data were abstracted from the institutional electronic health record. Time trends and patient and visit characteristics were compared across the pre-COVID-19 and COVID-19 periods. Results During the study period, 9,450 consult and follow-up visits were performed pre-COVID-19 and 3,298 visits in the COVID-19 period. The proportion of video visits increased markedly in the transition period, from 0.6% of all visits in the week of March 2, 2020, to 87% in the week of March 23, 2020. In-person visits decreased from 98% to 3%. Among all visits (in-person and telehealth), those during the COVID-19 period were less likely to be new consultations (43.1% from 60.1%; P < 0.001). There was a small and significant increase in the proportion of visits with patients who: identified as white (61.8% from 58.4%, P = 0.019), spoke English as their primary language (91.3% from 89.4%, P = 0.002), and had commercial insurance (34.1% from 32.0%; P = 0.009). Conclusion The overall COVID-19 clinic population retained demographic features similar to the pre-COVID-19 population despite a very rapid near-complete transition to telehealth. Nonetheless, the telehealth-predominant COVID-19 period had slightly increased visits with patients who were white or primarily English speaking or had commercial insurance. Strategies for ensuring telehealth is accessible to diverse populations should be a priority as telemedicine is integrated into long-term clinical operations.

Leptomeningeal metastasis (LM) is a late stage manifestation of advanced breast cancer frequently managed with whole brain radiotherapy (WBRT) and/or intrathecal chemotherapy. A subset of breast cancer patients who undergo stereotactic radiosurgery (SRS) for limited brain metastases (BM) ultimately develop LM. We hypothesized that this subset of high-risk patients may be identified by patient, disease, and/or treatment parameters. Clinical records from 133 breast cancer patients from a single institution who underwent SRS for BM between February 2010 and March 2018 were retrospectively analyzed. Variables including histopathology, BM features, systemic disease burden, and prior treatments were analyzed. Cumulative incidence rates were estimated with death as a competing risk. Dichotomous variable cutoffs were based on the 75th percentile value. In our cohort, 27 (20.3%) patients ultimately developed LM. With a median follow up of 21.2 months after diagnosis of BM, the actuarial rate of LM at 24 months was 15.2% (95% CI, 8.7%-21.7%). Median OS after diagnosis of LM was 7.0 (95% CI, 3.1–15.4) months. There was significantly increased risk of LM with ≥9 vs < 9 BM at BM diagnosis (28.1% vs 10.8% [24-month actuarial risk], subdistribution HR 2.4, p=0.027), and ≥11 vs < 11 cumulative number of BM treated (25.7% vs 11.7% [24-month actuarial risk], subdistribution HR 2.7, p=0.01). Variables not significantly associated with the risk of LM included tumor receptor status (ER, PR, HER2, triple negative), graded prognostic assessment, KPS, extracranial metastases, total BM volume, prior WBRT, or prior surgical resection. Time intervals between SRS treatments immediately preceding LM diagnosis was not significantly different from other time intervals. In conclusion, patients with a larger number of brain metastases at BM diagnosis (≥9) or cumulatively treated (≥11) appear to be at higher risk of developing LM and may benefit from stronger consideration of WBRT, intrathecal chemotherapy, and/or brain-penetrating systemic therapy.

Artificial intelligence (AI) is emerging as a technology with the power to transform established industries, and with applications from automated manufacturing to advertising and facial recognition to fully auton-omous transportation. Advances in each of these domains have led some to call AI the “fourth” industrial revolution [1]. In healthcare, AI is emerging as both a productive and disruptive force across many disci-plines. This is perhaps most evident in Diagnostic Radiology and Pathology, specialties largely built around the processing and complex interpretation of medical images, where the role of AI is increasingly seen as both a boon and a threat. In Radiation Oncology as well, AI seems poised to reshape the specialty in significant ways, though the impact of AI has been relatively limited at present, and may rightly seem more distant to many, given the predominantly interpersonal and complex interventional nature of the specialty. In this overview, we will explore the current state and anticipated future impact of AI on Radiation Oncology, in detail, focusing on key topics from multiple stakeholder perspectives, as well as the role our specialty may play in helping to shape the future of AI within the larger spectrum of medicine.

OBJECTIVES: Pediatric embryonal brain tumors include medulloblastoma, supratentorial primitive neuroectodermal tumor, and pineoblastoma. The purpose of this study was to investigate the prognostic utility of quantitative three-dimensional (3D) magnetic resonance imaging (MRI) radiomic analysis for primary pediatric embryonal brain tumors. METHODS: Thirty-four pediatric embryonal brain tumor patients with concurrent pre-operative T1-weighted post contrast (T1PG) and T2-weighted fluid-attenuated inversion recovery (FLAIR) MR images were identified from an institutional database. The median imaging follow-up was 5.2 years. Radiomic features were extracted from axial T1PG and FLAIR contours using MATLAB, and 15 features were selected for analysis based on qualitative radiographic features with known prognostic significance for pediatric embryonal brain tumors. Logistic regression, linear regression, receiver operating characteristic curve, Harrells C index and Somers D index were used to test the relationships between radiomic features, demographic variables and clinical outcomes. RESULTS: We found that pediatric embryonal brain tumors in older patients had increased normalized mean tumor intensity (P=0.05, T1PG), decreased tumor volume (P=0.02, T1PG) and increased markers of heterogeneity (P0.01, T1PG and FLAIR) relative to younger patients. We identified 10 quantitative radiomic features that delineated between medulloblastoma, pineoblastoma and supratentorial primitive neuroectodermal tumor, including size and heterogeneity (P0.05, T1PG and FLAIR). Decreased markers of tumor heterogeneity were predictive of neuraxis metastases and trended towards significance (P=0.1, FLAIR). Tumors with increased size (AUC=0.7, FLAIR) and decreased heterogeneity (AUC=0.7, FLAIR) at diagnosis were more likely to recur. CONCLUSIONS: Quantitative radiomic features are associated with pediatric embryonal brain tumor patient age, histology, neuraxis metastases and recurrence, and could be used for risk stratification.

OBJECTIVES: Chordoid meningiomas are a rare histologic variant that follow an aggressive clinical course. Here, we examine histopathologic and radiologic features of chordoid meningiomas to identify risk factors for recurrence. METHODS: Retrospective chart reviews were performed on 11 patients with chordoid meningioma and 15 patients with meningioma with focal chordoid features (

The adoption of enterprise digital imaging, along with the development of quantitative imaging methods and the re-emergence of statistical learning, has opened the opportunity for more personalized cancer treatments through transformative data science research. In the last 5 years, accumulating evidence has indicated that noninvasive advanced imaging analytics (i.e., radiomics) can reveal key components of tumor phenotype for multiple lesions at multiple time points over the course of treatment. Many groups using homegrown software have extracted engineered and deep quantitative features on 3-dimensional medical images for better spatial and longitudinal understanding of tumor biology and for the prediction of diverse outcomes. These developments could augment patient stratification and prognostication, buttressing emerging targeted therapeutic approaches. Unfortunately, the rapid growth in popularity of this immature scientific discipline has resulted in many early publications that miss key information or use underpowered patient data sets, without production of generalizable results. Quantitative imaging research is complex, and key principles should be followed to realize its full potential. The fields of quantitative imaging and radiomics in particular require a renewed focus on optimal study design and reporting practices, standardization, interpretability, data sharing, and clinical trials. Standardization of image acquisition, feature calculation, and statistical analysis (i.e., machine learning) are required for the field to move forward. A new data-sharing paradigm enacted among open and diverse participants (medical institutions, vendors and associations) should be embraced for faster development and comprehensive clinical validation of imaging biomarkers. In this review and critique of the field, we propose working principles and fundamental changes to the current scientific approach, with the goal of high-impact research and development of actionable prediction models that will yield more meaningful applications of precision cancer medicine. (C) 2018 Published by Elsevier Inc.

Author(s): Gennatas, Efstathios D; Wu, Ashley; Braunstein, Steve E; Morin, Olivier; Chen, William C; Magill, Stephen T; Gopinath, Chetna; Villaneueva-Meyer, Javier E; Perry, Arie; McDermott, Michael W; Solberg, Timothy D; Valdes, Gilmer; Raleigh, David R | Abstract: BackgroundMeningiomas are stratified according to tumor grade and extent of resection, often in isolation of other clinical variables. Here, we use machine learning (ML) to integrate demographic, clinical, radiographic and pathologic data to develop predictive models for meningioma outcomes.Methods and findingsWe developed a comprehensive database containing information from 235 patients who underwent surgery for 257 meningiomas at a single institution from 1990 to 2015. The median follow-up was 4.3 years, and resection specimens were re-evaluated according to current diagnostic criteria, revealing 128 WHO grade I, 104 grade II and 25 grade III meningiomas. A series of ML algorithms were trained and tuned by nested resampling to create models based on preoperative features, conventional postoperative features, or both. We compared different algorithms' accuracy as well as the unique insights they offered into the data. Machine learning models restricted to preoperative information, such as patient demographics and radiographic features, had similar accuracy for predicting local failure (AUC = 0.74) or overall survival (AUC = 0.68) as models based on meningioma grade and extent of resection (AUC = 0.73 and AUC = 0.72, respectively). Integrated models incorporating all available demographic, clinical, radiographic and pathologic data provided the most accurate estimates (AUC = 0.78 and AUC = 0.74, respectively). From these models, we developed decision trees and nomograms to estimate the risks of local failure or overall survival for meningioma patients.ConclusionsClinical information has been historically underutilized in the prediction of meningioma outcomes. Predictive models trained on preoperative clinical data perform comparably to conventional models trained on meningioma grade and extent of resection. Combination of all available information can help stratify meningioma patients more accurately.

OBJECTIVES This single institution retrospective study aims to evaluate the association between body mass index (BMI) and survival in glioblastoma (GBM). METHODS Patients were identified using a single institution cancer registry. Baseline height and weight were obtained at first clinical visit after diagnosis. Inclusion criteria were GBM diagnosis after 1999 with available height and weight data. Follow up was through July 2018. RESULTS 779 patients were included who had available BMI data. Median age was 58 (IQR 48–66). 38% of patients were female and 62% were male. Median overall survival (OS) was 1.34 years (IQR 0.78–2.12) and median BMI was 26.01 (IQR 23.3–29.1). There was no strong linear correlation between OS and BMI using Pearson correlation (r=0.045). Patients with higher BMI had significantly improved OS when comparing the top 50% to bottom 50% (2.20 vs 1.89 years, p=0.03) and top 25% to bottom 25% (2.25 vs 1.84 years, p=0.03). When comparing patients classified as overweight or obese (BMI >/= 25) to those within normal range or below (BMI < 25), there was a trend towards increased OS (2.14 vs 1.90 years, p=0.075). There was no significant trend for BMI when comparing patients with higher OS with lower OS. There was also no significance for BMI on Cox proportional hazard multivariate analysis. CONCLUSIONS There was no significant linear correlation between BMI and survival for GBM patients, although there appears to be statistically significant improved survival benefit for patients with higher BMI compared to patients with lower BMI. Further investigation is warranted to explore this finding and if BMI could be used as a potential prognostic marker for GBM.

Purpose To measure the amount of geometric distortions present in the three-dimensional imaging modalities—cone beam computed tomography (CBCT) and magnetic resonance imaging (MRI)—used at University of California, San Francisco, CA, for gynecologic high dose rate brachytherapy. Methods and Materials An MRI- and CT–compatible water phantom with two different sets of support structures was designed and built for this study. The support structures were used to precisely position catheters that were filled with either an MRI contrast agent or a string of radio-opaque markers. The first support structure without anatomy was built to test system-based distortions. A second structure included two types of gynecologic applicators as well as several anatomical structures, including bones and rectum to test object-induced distortions. Images were acquired with CT (for reference), kilovoltage CBCT, and MRI (1.5 T with T1- and T2-weighted images). The difference in catheter positions between the images and the CT images was analyzed. Results For CBCT, the mean of the absolute deviations was below 1 mm in all directions. The inherent uncertainty in the measurement of distortion was less than 0.5 mm. MRI presented mean absolute system-based distortions between 0.6 and 1.1 mm in the central region of the image and between 0.7 and 2.3 mm in the outer region. Images with the applicator and anatomy in place created mean absolute distortions of 0.4, 0.8, and 0.8 mm or less for CBCT, MR-T1, and MR-T2 images, respectively. Conclusions The distortions measured in the presence of applicators are small enough to validate the use of CBCT and 1.5 T MRI for GYN brachytherapy treatment planning and verification.

Purpose: This report describes a new image-guided radiotherapy (IGRT) technique using megavoltage cone-beam computed tomography (MV-CBCT) to treat paraspinous tumors in the presence of orthopedic hardware. Methods and Materials: A patient with a resected paraspinous high-grade sarcoma was treated to 59.4 Gy with an IMRT plan. Daily MV-CBCT imaging was used to ensure accurate positioning. The displacement between MV-CBCT and planning CT images were determined daily and applied remotely to the treatment couch. The dose–volume histograms of the original and a hypothetical IMRT plan (shifted by the average daily setup errors) were compared to estimate the impact on dosimetry. Results: The mean setup corrections in the lateral, longitudinal, and vertical directions were 3.6 mm (95% CI, 2.6–4.6 mm), 4.1 mm (95% CI, 3.2–5.0 mm), and 1.0 mm (95% CI, 0.6–1.3 mm), respectively. Without corrected positioning, the dose to 0.1 cc of the spinal cord increased by 9.4 Gy, and the doses to 95% of clinical target volumes 1 and 2 were reduced by 4 Gy and 4.8 Gy, respectively. Conclusions: Megavoltage-CBCT provides a new alternative image-guided radiotherapy approach for treatment of paraspinous tumors in the presence of orthopedic hardware by providing 3D anatomic information in the treatment position, with clear imaging of metallic objects and without compromising soft-tissue information. © 2006 Elsevier Inc.

Resume Le systeme Megavoltage cone-beam CT (MV CBCT) combine un appareil d'imagerie portale dote d'un detecteur optimise pour les energies du faisceau de traitement, un faisceau de tres faible debit de dose synchronise avec le detecteur et une station d'acquisition, de reconstruction et de fusion avec l'image scanographique de planification. Le systeme d'imagerie MV CBCT genere a faible dose des images tomographiques du patient en position de traitement permettant la verification du positionnement et la detection des distorsions anatomiques, de meme que le calcul de dose. L'objectif de ce travail etait de colliger les recents developpements de la technologie (MV CBCT), et a l'aide de quatre utilisations cliniques specifiques dans les traitements de lesions spinales ou pulmonaires, dans les traitements des cancers de la tete, du cou et de la prostate, d'illustrer les performances et possibilites du systeme, menant a de nouvelles applications cliniques.

In this article, we describe a clinical mega-voltage cone-beam computed tomography (MV CBCT) system, present the image acquisition and patient setup procedure, discuss the positioning accuracy and image quality, and illustrate its potential use for image-guided radiation therapy (IGRT) through selected clinical examples. The MV CBCT system consists of a standard linear accelerator equipped with an amorphous-silicon flat panel electronic portal-imaging device adapted for mega-electron volt (MeV) photons. An integrated computer workspace provides automated acquisition of projection images, image reconstruction, CT to CBCT image registration, and couch shift calculation. The system demonstrates submillimeter localization precision and sufficient soft-tissue resolution to visualize structures such as the prostate. In our clinic, we have used the MV CBCT system to detect nonrigid spinal cord distortions, monitor tumor growth and shrinkage, and locate and position stationary tumors in the lung. MV CBCT has also greatly improved the delineation of structures in CT images that suffer from metal artifacts. MV CBCT has undergone significant development in the last few years. Current image quality has already proven sufficient for many IGRT applications. Moreover, we expect the range of clinical applications for MV CBCT to grow as imaging technology continues to improve.

94 Background: Improving quality of care involves identifying patient health needs and designing systems to meet those needs reliably. Within Radiation Oncology, patient health needs encompass measurable, physical quantities such as a reduction in tumor size, as well as subjective needs, as in freedom from pain. Patient reported outcomes (PRO) have the potential to identify subjective health needs and to guide treatment towards symptoms patients find most bothersome, thereby improving quality of care. Here we report the unmet health needs revealed by systematic electronic PRO data collection in a Radiation Oncology department. Methods: 50 consecutively treated patients for a variety of malignant and nonmalignant, primary and metastatic disease were surveyed within the CNS subdivision of a single Radiation Oncology department using electronic EORTC QLQ C30 or PRO-CTCAE CNS forms displayed on a touch screen tablet. Patients were then followed with PRO surveys at two additional time points, 1-2 weeks post radiotherapy treatment and again at one month. Results: Of 27 planned PRO CTCAE surveys of 9 patients, 22 (81%) were completed and available for evaluation. 6/9 (67%) of patients taking the PRO CTCAE reported “severe” or “very severe” side effects. The most commonly reported “severe” or “very severe” symptoms were fatigue (6/9, 67%), insomnia (3/9, 33%), anxiety (2/9, 22%), and sad/unhappy feelings (2/9, 22%). Of 123 planned EORTC QLQ C30 surveys of 41 patients, 109 (89%) were completed and accessible for evaluation. 38/41 (93%) of patients taking EORTC QLQ C30 reported side effects impacting their lives “quite a bit” or “very much”. The most common symptoms impacting patients “quite a bit” and/or “very much” were fatigue (25/41, 61%), pain (17/41, 41%), insomnia (15/41, 37%), and worry (14/41, 34%). Conclusions: Gaps that may exist between the quality of care we wish to provide and the care patients experience may be recognized and addressed by incorporating PROs into our standard clinical practice. Integrating electronic PROs into our clinical systems will help us to more reliably identify unmet patient health needs, personalize care, and respond to side effects and symptoms that are of the greatest concern to patients.

Purpose: The objective of this work was to demonstrate the feasibility of acquiring low-exposure megavoltage cone-beam CT (MV CBCT) three-dimensional (3D) image data of sufficient quality to register the CBCT images to kilovoltage planning CT images for patient alignment and dose verification purposes. Methods and Materials: A standard clinical 6-MV Primus linear accelerator, operating in arc therapy mode, and an amorphous-silicon (a-Si) flat-panel electronic portal-imaging device (EPID) were employed. The dose-pulse rate of a 6-MV Primus accelerator beam was windowed to expose an a-Si flat panel by using only 0.02 to 0.08 monitor units (MUs) per image. A triggered image-acquisition mode was designed to produce a high signal-tonoise ratio without pulsing artifacts. Several data sets were acquired for an anthropomorphic head phantom and frozen sheep and pig cadaver heads, as well as for a head-and-neck cancer patient on intensity-modulated radiotherapy (IMRT). For each CBCT image, a set of 90 to 180 projection images incremented by 1° to 2° was acquired. The two-dimensional (2D) projection images were then synthesized into a 3D image by use of cone-beam CT reconstruction. The resulting MV CBCT image set was used to visualize the 3D bony anatomy and some soft-tissue details. The 3D image registration with the kV planning CT was performed either automatically by application of a maximization of mutual information (MMI) algorithm or manually by aligning multiple 2D slices. Results: Low-noise 3D MV CBCT images without pulsing artifacts were acquired with a total delivered dose that ranged from 5 to 15 cGy. Acquisition times, including image readout, were on the order of 90 seconds for 180 projection images taken through a continuous gantry rotation of 180°. The processing time of the data required an additional 90 seconds for the reconstruction of a 256 3 cube with 1.0-mm voxel size. Implanted gold markers (1 mm 3 mm) were easily visible for all exposure levels without artifacts. In general, the presence of high Z materials such as tooth fillings or implanted markers did not result in visible streak artifacts. The registration of structures such as the spinal canal and the nasopharynx in the MV CBCT and kV CT data sets was possible with millimeter and degree accuracy as assessed by displacement simulations and subsequent visual evaluation. Conclusions: We believe that the quality of these images, along with the rapid acquisition and reconstruction times, demonstrates that MV CBCT performed by use of a standard linear accelerator equipped with a flat-panel imager can be applied clinically for patient alignment. © 2005 Elsevier Inc. Cone-beam CT, Electronic portal-imaging device, Image-guided radiotherapy.