Faculty Bibliography

PURPOSE/OBJECTIVE(S): Healthcare data often exist in silos and in unstructured formats that limit interoperability and require tedious manual extraction. Our institution has adopted a flexible and scalable big data platform built on Hadoop that integrates data from Epic/Clarity as well as Aria and allows users to leverage modern data science tools to facilitate access. We hypothesize that a data analytics and visualization dashboard can be built using open-source tools that will (1) allow non-technical users to explore de-identified clinical data within our institutional big data platform and (2) connect with repositories of molecular data to demonstrate potential methods of integrating clinical and basic science data. MATERIALS/METHODS: De-identified patient-level radiation oncology data from the institutional big data platform (Hadoop) were extracted with the python packages pyodbc and pandas. For the purposes of this dashboard, radiation oncology specific clinical data elements were queried including the date of first radiation treatment, treatment location, treatment modality (SBRT, external beam, SRS, TBI, LDR/HDR brachytherapy), ICD10 codes, anatomic treatment site, number of fractions, treatment prescription, and dose per fraction. A python client connection with the publicly accessible instance of cBioPortal for Cancer Genomics was established using the Bravado library. Data transformation and cleaning was performed in python using panda's data frames. A web-based dashboard to facilitate user-defined visualizations was implemented using the Dash python library and interactive visualizations of subsets of extracted data were generated in real-time using the plotly plotting library.
RESULT(S): We developed a web-based dashboard that gives users without extensive programming expertise the ability to explore de-identified clinical data extracted from Hadoop. As proof of principle, the dashboard was used to visualize the clinical impact of the COVID-19 pandemic on radiation oncology patient volumes, revealing a significant decline in new radiation treatments in April and May of 2020 (-54% and -36% compared to 2019) during the initial COVID-19 surge. Furthermore, the dashboard allows users to interact with the cBioPortal for Cancer Genomics repository, which currently houses clinical and molecular data from 301 publicly available studies spanning 869 different cancer types. This interface with cBioPortal illustrates the potential for future integration of clinically meaningful sequencing results with clinical outcomes data.
CONCLUSION(S): We built an interactive web-based dashboard to enable general users' easy access to de-identified clinical data stored within the institutional big data platform. Additional data sources, including external molecular data can be connected to the dashboard allowing for future integration.
Copyright

PURPOSE/OBJECTIVE(S): The supraclavicular (SCV), medial axillary and internal mammary nodes (IMNs) are not typically resected in breast cancer patients (pts). The optimal local therapy of pts with nodal disease in these regions is not well-studied. We aim to evaluate outcomes of breast cancer patients with unresected nodal disease. MATERIALS/METHODS: We identified 79 pts at our institution from 2016- 2021 with unresected nodal disease in the axilla, SCV and/or IMNs defined as grossly enlarged nodes on CT, MRI or PET scan +/- biopsy confirmation. Pts were treated with breast/chest wall and regional nodal irradiation with an additional boost to the unresected nodal region. Distant failure (DF) and local-regional failure (LRF) were assessed. Kaplan-Meier was used to calculate disease-free survival (DFS), overall survival (OS) and local recurrence-free survival (LRFS). Logistic regression was used to identify variables associated with worse DFS. Acute and late toxicity of RT were evaluated.
RESULT(S): 33% of pts were treated with breast-conserving surgery, 65% with mastectomy and all had axillary surgery (81% ALND, 19% SLNB). 47% of pts received IMN boost (IMN), 40% axillary/SCV boost (axSCV) and 15% both IMN and axSCV boost (IMN/axSCV). Most had cT2-3 (72%), hormone receptor positive (75%), and HER-2 negative disease (84%). 57% of axSCV had cN3A disease; 84% of IMN and 83% of IMN/axSCV had cN3b disease. 7% of axSCV and 17% of IMN/axSCV had cN3c disease. Most pts received chemotherapy (97%). Median nodal boost dose was 10 Gy (range 10-20 Gy), with 17% axSCV, 22% IMN, and 17% IMN/axSCV receiving 14-20 Gy. Rates of acute and late grade 3 toxicity did not differ by boost location (acute: IMN: 20%, axSCV: 11% and IMN/axSCV 20%, P=0.559; late: IMN: 40%, axSCV: 25%, IMN/axSCV: 40%, P=0.630) nor by boost dose (10 Gy vs 14-20 Gy). There were no grade 4+ toxicities. With a median follow up of 30 months, the 3-year LRR, DFS, and OS was 94.5%, 86.3% and 93.8% respectively. Crude rates of failure for the entire group were 13.9% (10.1% DF; 3.8% DF+LRF). Rates of failure by boost group were axSCV: 13.3% (10% DF; 3.3% DF+LRF), IMN: 5.4% (2.7% DF, 2.7% DF+LRF), IMN/axSCV 41.7% (33.3% DF, 8.3% DF+LRF). There were no LRFs without DFs. Median time to failure was 23 months (IQR 18-34). On univariate analysis clinical tumor size (cT) and IMN/axSCV vs. IMN or axSCV alone was associated with worse DFS (HR: 9.78 95% CI 2.07-46.2, P=0.004 and HR: 9.49 95% CI 2.67-33.7, P=0.001). On multivariate analysis, cT approached significance (HR 6.15; 95% CI 0.95-39.8, P=0.05). IMN/axSCV vs. IMN or axSCV alone retained significance (HR 4.80; 95% CI 1.27-18.13, P=0.02). The difference between the axSCV vs. IMN group was not significant.
CONCLUSION(S): In this population of pts with unresected nodal disease, boost RT to radiographically positive LN regions can be safely delivered with low rates of grade 3+ toxicity. The majority of failures were distant with no isolated LRFs. Failures were highest in the IMN/axSCV group (~40%). Further treatment escalation is necessary for these pts.
Copyright

PURPOSE/OBJECTIVE(S): Use of local therapy such as stereotactic body radiation therapy (SBRT) to treat oligometastatic malignancy is a well-established paradigm, but whether benefit extends to the oligoprogressive setting remains unclear. We present our institutional series of patients with oligometastatic or oligoprogressive malignancy treated with SBRT. MATERIALS/METHODS: We performed a retrospective study of patients with oligometastatic and oligoprogressive malignancy treated with SBRT between 2014 and 2019. Oligometastatic patients were defined as those with five or fewer metastatic lesions in total. Oligoprogressive patients were defined as those with more than five and up to twenty metastatic lesions in total, of which five or fewer metastases were progressing on current systemic therapy. Patients lacking complete treatment records or follow-up imaging were excluded. The study was approved by the NYU Institutional Review Board.
RESULT(S): A total of 114 patients were treated with 123 courses of SBRT, of which 96 treated oligometastasis and 27 treated oligoprogression. Primary sites of disease included lung (38%), prostate (20%), and GI (12%), as well as gynecologic, abdominal, and cutaneous malignancies. Median follow-up was 21 months. No grade 3 or higher radiation-related adverse events were reported. Patients with oligometastatic malignancy had longer 2-year overall survival (79% vs 59%; P=0.003), local control (73% vs 55%; P=0.01), and progression-free survival (26% vs 8%; P < 0.001), but similar freedom from new systemic therapy (36% vs 31%; P=0.8). This result held true in subgroup analysis regardless of lung vs non-lung primary site, and regardless of the presence or absence of a targetable mutation.
CONCLUSION(S): In this hypothesis-generating retrospective cohort study, patients with oligoprogressive malignancy treated with SBRT have similar freedom from new systemic therapy to patients with oligometastatic malignancy, strengthening the rationale for treating oligoprogressive malignancy with SBRT.
Copyright

PURPOSE/OBJECTIVE(S): TBI is a backbone of many conditioning regimens for hematopoietic stem cell transplants but can lead to both acute and late toxicity including radiation-induced interstitial pneumonitis. The incidence of idiopathic pneumonia syndrome (IPS) after TBI-based myeloablative conditioning regimens ranges from 7% to 35%. The purpose of this study is to implement image guided volumetrically modulated technique (VMAT) for TBI with the goal of lung sparing and improved target coverage. MATERIALS/METHODS: Nine patients have been treated using image-guided VMAT based TBI at our institution as part of a single-arm phase 2 clinical trial for patients undergoing myeloablative conditioning regimens. The trial was approved by our internal review board (IRB) in September 2020 and aims to accrue 15 patients within one year. All patients enrolled in the trial have signed informed consent. The primary endpoints of the study are the following dosimetric constraints: V100% >= 90%, D98% >= 85% of Rx dose for the planning target volume (PTV), and a mean lung dose < 9 Gy. PTV is defined as the body contour cropped 5 mm from the surface and excluding lungs and kidneys but extended 3 mm into these organs. Additional secondary dosimetric endpoints include mean dose to each individual kidney < 11 Gy, and maximum dose to 2cc of the entire body < 130% of Rx dose. Clinical endpoints include the occurrence of IPS in the first 100 days after transplant, occurrence of acute graft versus host disease (GVHD), transplant related mortality or mortality in the first 100 days following transplant.
RESULT(S): Patients were treated to 12 Gy in 8 BID fractions (n=6) or 13.2 Gy in 8 BID fractions (n=3) over four consecutive days. All patients were able to complete treatment to the prescribed dose as planned. All patient plans met dosimetric constraints of the study. The median PTV V100% was 93.2% of Rx dose (Max: 95.6%, Min: 92.1%), the median PTV D98% was 90.2% of Rx dose (Max: 94.3%, Min: 88.3%), and the median lung dose mean was 7.63 Gy (Max: 7.94 Gy, Min: 7.29 Gy). In addition, individual kidney mean doses were < 11 Gy, and body maximum dose (D2cc) was < 130% of Rx dose for all patients. At this time, only one patient (12 Gy treatment) has reached the 100 day post-transplant follow-up with the following findings: no relapse on bone marrow biopsy, no pneumonitis, resolved acute GVHD overall grade 1 (skin: 1, GI: 0, Liver: 0), resolved dermatitis (grade 1), resolved vomiting (grade 2), ongoing diarrhea and nausea (grade 1, previously grade 2).
CONCLUSION(S): Our initial results indicate that primary and secondary dosimetric endpoints were achievable for all protocol patients treated thus far. As the trial progresses, secondary clinical endpoints at 100 day follow-up will be analyzed to evaluate occurrence of IPS, survival, and treatment related toxicities.
Copyright

PURPOSE/OBJECTIVE:Stereotactic radiosurgery (SRS) has become an important modality in the treatment of brain metastases. The purpose of this study is to investigate the potential of radiomic features from planning magnetic resonance (MR) images and dose maps to predict local failure after SRS for brain metastases. MATERIALS/METHODS/METHODS:Twenty-eight patients who received Gamma Knife (GK) radiosurgery for brain metastases were retrospectively reviewed in this IRB-approved study. 179 irradiated tumors included 42 that locally failed within one-year follow-up. Using SRS tumor volumes, radiomic features were calculated on T1-weighted contrast-enhanced MR images acquired for treatment planning and planned dose maps. 125 radiomic features regarding tumor shape, dose distribution, MR intensities and textures were extracted for each tumor. Logistic regression with automatic feature selection was built to predict tumor progression from local control after SRS. Feature selection and model evaluation using receiver operating characteristic (ROC) curves were performed in a nested cross validation (CV) scheme. The associations between selected radiomic features and treatment outcomes were statistically assessed by univariate analysis. RESULTS:The logistic model with feature selection achieved ROC AUC of 0.82 ± 0.09 on 5-fold CV, providing 83% sensitivity and 70% specificity for predicting local failure. A total of 10 radiomic features including 1 shape feature, 6 MR images and 3 dose distribution features were selected. These features were significantly associated with treatment outcomes (p < 0.05). The model was validated on independent holdout data with an AUC of 0.78. CONCLUSIONS:Radiomic features from planning MR images and dose maps provided prognostic information in SRS for brain metastases. A model built on the radiomic features shows promise for early prediction of tumor local failure after treatment, potentially aiding in personalized care for brain metastases.

Purpose: Multi-isocentric treatment delivery for CSI and TBI poses specific challenges for treatment delivery. We have developed a software tool to streamline all aspects of delivery for therapists and physicists at the machine, as well as to inform attending physicians of setup variability and image residuals at different locations.
Method(s): Our institution delivers VMAT-based CSI and TBI with up to 3 and 7 isocenters, respectively. A software tool was developed to assist with treatment delivery including initial patient setup, patient imaging, automatic calculation of the optimal global shift based on each isocenter's ideal shift, and automatic calculation of each isocenter's couch coordinates. Initial treatment couch coordinates are queried via the Eclipse scripting API. The global shift was calculated prioritizing the head isocenter for CSI treatments and the chest isocenter for TBI treatments by first maximizing residual tolerance at any other location prior to accepting any residual deviation at these locations. Maximum residuals tolerance was determined based on target margins, plan uncertainty and as per physician instructions. Delivery parameters are reported to a document uploaded to ARIA via API.
Result(s): The developed tool was employed for 11 cases. The software tool replaced the need for plan shift comments or instructions for therapists. In particular, its use eliminated the need to provide isocenter shifts to therapists by directly providing final couch parameters for treatment, greatly reducing the risk of delivery errors. The software effectively informed the therapists if any expected tolerance was surpassed, triggering a patient setup evaluation.
Conclusion(s): The described software tool is a core component to our multi-isocenter treatment programs and has streamlined delivery of these complex techniques that would otherwise require complicated instructions, including multiple shifts and on-the-fly calculations of optimal image alignment based on multiple imaging locations. This has substantially reduced the possibility of delivery errors

Purpose: In-vivo dosimetry for conventional total body irradiation (TBI) utilize point detectors placed along the patient surface to confirm the delivered dose matches prescription dose. However, in the volumetrically modulated arc therapy (VMAT) approach to TBI, the electronic portal imager device (EPID) can be utilized to acquire a 2-dimensional transmission fluence plane. This work explores the relationship between patient setup accuracy with transit in-vivo dosimetry.
Method(s): A total of 192 fields were investigated. Each VMAT plan consisted of four isocenters: head, chest, abdomen, and pelvis. Prior to treatment, the patient was imaged at the head, pelvis, and chest. Optimal couch shifts were determined for each isocenter under image guidance. The optimal IGRT shifts were determined using an inhouse application that minimized dose deviation using criteria established through plan uncertainty analysis performed in Eclipse. Translational couch residuals were recorded and defined as the difference in the global shift calculated and the optimal couch position with shifts. Transit dosimetry was measured per arc, and analyzed using SNC PerFRACTION with a gamma criteria of 10%/5mm, 5%/5mm, and 5%/7mm.
Result(s): Based on plan uncertainty analysis, clinical threshold for couch residuals were set to 7 mm (5 mm for chest isocenter) as there would be minimal impact on target coverage and organ sparing at those levels. Transit dosimetry showed that the average pass rate across all fields was 99.6%, 97.0%, and 99.2% for 10%/5mm, 5%/5mm, and 5%/7mm gamma criteria, respectively. Pearson correlation tests showed that there was weak correlation between gamma criteria and couch residuals. At stringent 3%/5mm gamma criteria, moderate correlation was found between lateral couch residuals for the head and chest and the head and chest arc analysis.
Conclusion(s): Transit dosimetry showed high pass rates using our couch residual tolerances, which confirmed the plan uncertainty analysis performed during treatment planning

Purpose: Patients receiving myeloablative total body irradiation (TBI) doses >= 12Gy are at risk of developing interstitial pneumonitis. At our institution, TBI transitioned from extended distance opposed Laterals to image-guided VMAT, in an effort to improve coverage while sparing lungs and kidneys. This work presents a dosimetric comparison between 3D Laterals and VMAT.
Method(s): Nine patients were treated with VMAT as part of an ongoing phase II single-arm clinical trial. VMAT patients were CT-simulated supine, with thermoplastic masks for head/neck, chest/abdomen/pelvis and feet. VMAT planning (12Gy (n=6) or 13.2Gy (n=3) in 8-BID fractions) utilizes 6MV multi-isocentric arcs and AP/PA beams to treat the upper and lower body, respectively. Ten 3D Lateral patients were CT-simulated supine with arms positioned/immobilized for lung shielding, with rice compensation between legs/feet. Laterals plan (12Gy in 8-BID fractions) uses 15MV, beam spoiler, head compensation, and subfields to maintain coverage and mean-lungs dose <10.5Gy. Target (Body-5mm, extending 3mm into lungs and kidneys for VMAT; Body-2cm for Laterals) coverage was evaluated at V100%, and D98% (percentage of Rx). Absolute dose to lungs and kidneys were reportedResults: Median Target V100% and D98% for VMAT was 93.2% (Range: 95.6% to 92.1%) and 90.2% (94.3% to 88.3%), whereas for Laterals V100% and D98% was 57.3% (66.5% to 46.3%) and 80.6% (75.5% to 84%). The median Lung mean dose was 7.6Gy (7.3Gy to 7.9Gy) for VMAT. The median mean dose to kidney was 10.4Gy (10.1Gy to 10.7Gy) for VMAT, and 12.5Gy (11.9Gy to 13.5Gy) for Laterals.
Conclusion(s): We have established a VMAT-TBI program for patients requiring myeloablative irradiation. Improvement in target coverage is demonstrated by V100% and D98%, while reducing the mean dose to the lungs significantly from 10.5Gy to 8Gy

Purpose: To investigate the effect of patient positioning in Volumetric Modulated Arc Therapy (VMAT) for Total Body Irradiation (TBI) given the use of multiple isocenters, by simulating offsets in patient positioning and evaluating changes to planned dose distributions.
Method(s): VMAT treatment plans for seven TBI patients treated as part of a prospective stage II clinical trial were evaluated. Plan uncertainties were calculated by introducing 5mm and 10mm translational shifts to the plans' isocenters in the lateral (x), vertical (y), and longitudinal (z) directions. Dose distributions were then re-calculated in the treatment planning system (Eclipse), in order to evaluate dosimetric robustness to one global imaging shift at treatment. Differences in target volume (PTV) coverage and doses to organs at risk were evaluated based on four parameters: lung mean dose, PTV-V100%, PTV-D98%, and kidney mean doses.
Result(s): Lung mean dose increased an average of 8.2cGy, 4.4cGy, and 3.3cGy when shifted 5mm in the x, y, z directions (respectively) across seven patients; 33.2CGy, 18.5cGy, 18.3cGy for 10mm shifts in x, y, z. Target coverage V100% decreased an average of 0.3%, 0.03%, 0.1% for 5mm shifts, and 1.1%, 0.8%, 0.4% for 10mm shifts in x, y, z. D98% decreased 0.9%, 0.3%, 0.3% when shifted 5mm; 3.5%, 2.1%, 1.0% when shifted 10mm in x, y, z. Mean dose to the left kidney increased 6.6cGy, 9.7cGy, 2.8cGy for 5mm, and 28.1cGy, 32.7cGy, 18.0cGy for 10mm shifts in x, y, z. Right kidney mean dose increased 11.9cGy, 8.9cGy, 3.1cGy for 5mm, and 36.5, 30.5, 19.8cGy for 10mm.
Conclusion(s): Though small in relation to total dose, the largest increase in mean lung dose and decrease in coverage was seen with lateral shifts as compared to vertical or longitudinal shifts. These results support the use of an approach with preferential alignment to the chest region (lung-sparing), as long as residual imaging alignment outside the chest is kept below 10mm. Jose Teruel has received honorarium from Varian Medical Systems