Unlocking the adaptive advantage: correlation and machine learning classification to identify optimal online adaptive stereotactic partial breast candidates.

Objective.Online adaptive radiotherapy (OART) is a promising technique for delivering stereotactic accelerated partial breast irradiation (APBI), as lumpectomy cavities vary in location and size between simulation and treatment. However, OART is resource-intensive, increasing planning and treatment times and decreasing machine throughput compared to the standard of care (SOC). Thus, it is pertinent to identify high-yield OART candidates to best allocate resources.Approach.Reference plans (plans based on simulation anatomy), SOC plans (reference plans recalculated onto daily anatomy), and daily adaptive plans were analyzed for 31 sequential APBI targets, resulting in the analysis of 333 treatment plans. Spearman correlations between 22 reference plan metrics and 10 adaptive benefits, defined as the difference between mean SOC and delivered metrics, were analyzed to select a univariate predictor of OART benefit. A multivariate logistic regression model was then trained to stratify high- and low-benefit candidates.Main results.Adaptively delivered plans showed dosimetric benefit as compared to SOC plans for most plan metrics, although the degree of adaptive benefit varied per patient. The univariate model showed high likelihood for dosimetric adaptive benefit when the reference plan ipsilateral breast V15Gy exceeds 23.5%. Recursive feature elimination identified 5 metrics that predict high-dosimetric-benefit adaptive patients. Using leave-one-out cross validation, the univariate and multivariate models classified targets with 74.2% and 83.9% accuracy, resulting in improvement in per-fraction adaptive benefit between targets identified as high- and low-yield for 7/10 and 8/10 plan metrics, respectively.Significance.This retrospective, exploratory study demonstrated that dosimetric benefit can be predicted using only ipsilateral breast V15Gy on the reference treatment plan, allowing for a simple, interpretable model. Using multivariate logistic regression for adaptive benefit prediction led to increased accuracy at the cost of a more complicated model. This work presents a methodology for clinics wishing to triage OART resource allocation.

Improving Communication of Peer-Review Conference Outcomes: A Practical Experience.

Purpose: The aim of this work was to describe the design and implementation of a more robust workflow for communicating outcomes from a peer-review chart rounds conference. We also provide information regarding cycle times, plan revisions, and other key metrics that we have observed since initial implementation. Methods and Materials: A multidisciplinary team of stakeholders including physicians, physicists, and dosimetrists developed a revised peer-review workflow that addressed key needs to improve the prior process. Consensus terminology was developed to reduce ambiguity regarding the priority of peer-review outcomes and to clarify expectations of the treating physician in response to peer-review outcomes. A custom workflow software tool was developed to facilitate both upstream and downstream processes from the chart rounds conference. The peer-review outcomes of the chart rounds conference and resulting plan changes for the first 18 months of implementation were summarized. Results: In the first 18 months after implementation of the revised processes, 2294 plans were reviewed, and feedback priority levels assigned. Across all cases with feedback, the median time for the treating attending physician to acknowledge conference comments was 1 day and was within 7 calendar days for 89.1% of cases. Conference feedback was acknowledged within 1 day for 74 of 115 (64.3%) cases with level 2 comments and for 18 of 21 (85.7%) cases with level 3 comments (P = .054). Contours were modified in 13 of 116 (11%) cases receiving level 2 feedback and 10 of 21 (48%) cases receiving level 3 feedback (P < .001). The treatment plan was revised in 18 of 116 (16%) cases receiving level 2 feedback and 13 of 21 (61%) cases receiving level 3 feedback (P < .001). Conclusions: We successfully implemented a workflow to improve upstream and downstream processes for a chart rounds conference. Standardizing how peer-review outcomes were communicated and recording physician responses allow for improved ability to monitor conference activities.

Predictors of In-Hospital Death in Patients with Lung Cancer Admitted for Acute Radiation Pneumonitis: A Healthcare Cost and Utilization Project (HCUP) Analysis.

BACKGROUND: Radiation pneumonitis (RP) is a dose-limiting and potentially fatal toxicity of thoracic radiotherapy most often seen in patients treated for primary lung cancer. The purpose of this study was to identify predictors of in-hospital death among lung cancer patients admitted for acute RP in the Healthcare Cost and Utilization Project (HCUP) database. MATERIALS AND METHODS: The HCUP National Inpatient Sample database was queried from 2012 through 2016 to capture adult lung cancer patients admitted to the hospital with a principal diagnosis of acute RP. Multivariate logistic regression modeling and χ2 tests were used to determine predictors of in-hospital death. RESULTS: Of the 882 patients with lung cancer admitted for RP, 67 patients (7.6%) died during the hospitalization and 90 patients (10.2%) required mechanical ventilation. Of those requiring mechanical ventilation, 38 patients (42.2%) died. The average age at hospitalization was 70.4 years (range, 35-90). Of those factors associated with death on univariate analysis, interstitial lung disease (odds ratio [OR] = 6.14; 95% confidence interval [CI], 1.9-19.4; P = .002), pulmonary hypertension (OR = 3.1; 95% CI, 1.6-6.2; P = .001), diabetes mellitus (OR = 2.0; 95% CI, 1.1-3.3; P = .013), and more affluent Zip Code (OR = 1.9; 95% CI, 1.1-3.2; P = .021) remained statistically significant on multivariate logistic regression. CONCLUSION: In the largest reported cohort of patients with lung cancer hospitalized with a principal diagnosis of acute RP, the presence of interstitial lung disease, pulmonary hypertension, diabetes mellitus, and more affluent Zip Code were associated with in-hospital death. Comorbid diagnoses may be useful for risk-stratified management of inpatients with RP.

Longitudinal assessment of anchored transponder migration following lung stereotactic body radiation therapy.

PURPOSE: To assess the long-term stability of the anchored radiofrequency transponders and compare displacement rates with other commercially available lung fiducial markers. We also sought to describe late toxicity attributable to fiducial implantation or migration. MATERIALS AND METHODS: The transponder cohort was comprised of 17 patients at our institution who enrolled in a multisite prospective clinical trial and underwent bronchoscopic implantation of three anchored transponders into small (2-2.5 mm) airways. We generated a comparison cohort of 34 patients by selecting patients from our institutional lung SBRT database and matching 2:1 based on the lobe containing tumor and proximity to the bronchial tree. Assessment of migration was performed by rigidly registering the most recent follow-up CT scan to the simulation scan, and assessing whether the relative geometry of the fiducial markers had changed by more than 5 mm. Toxicity outcomes of interest were hemoptysis and pneumothorax. RESULTS: The median follow-up of patients in the transponder cohort was 25.3 months and the median follow-up in the comparison cohort was 21.7 months. When assessing the most recent CT, all fiducial markers were within 5 mm of their position at CT simulation in 11 (65%) patients in the transponder group as compared to 23 (68%) in the comparison group (P = 0.28). One case of hemoptysis was identified in the transponder cohort, and bronchoscopy confirmed bleeding from recurrent tumor; no cases of hemoptysis were noted in the comparison cohort. No case of pneumothorax was noted in either group. CONCLUSION: No significant difference in the rates of fiducial marker retention and migration were noted when comparing patients who had anchored transponders placed into small airways and a 2:1 matched cohort of patients who had other commercially available lung fiducial markers placed. In both groups, no late or chronic toxicity appeared to be related to the implanted fiducial markers.

Hospice care, cancer-directed therapy, and Medicare expenditures among older patients dying with malignant brain tumors.

Background: End-of-life care for older adults with malignant brain tumors is poorly understood. The purpose of this study is to quantify end-of-life utilization of hospice care, cancer-directed therapy, and associated Medicare expenditures among older adults with malignant brain tumors. Methods: This retrospective cohort study included deceased Medicare beneficiaries age ≥65 with primary malignant brain tumor (PMBT) or secondary MBT (SMBT) receiving care within a southeastern cancer community network including academic and community hospitals from 2012-2015. Utilization of hospice and cancer-directed therapy and total Medicare expenditures in the last 30 days of life were calculated using generalized linear and mixed effect models, respectively. Results: Late (1-3 days prior to death) or no hospice care was received by 24% of PMBT (n = 383) and 32% of SMBT (n = 940) patients. SMBT patients received late hospice care more frequently than PMBT patients (10% vs 5%, P = 0.002). Cancer-directed therapy was administered to 18% of patients with PMBT versus 25% with SMBT (P = 0.003). Nonwhite race, male sex, and receipt of any hospital-based care in the final 30 days of life were associated with increased risk of late or no hospice care. The average decrease in Medicare expenditures associated with hospice utilization for patients with PMBT was $-12,138 (95% CI: $-18,065 to $-6210) and with SMBT was $-1,508 (95% CI: $-3,613 to $598). Conclusions: Receiving late or no hospice care was common among older patients with malignant brain tumors and was significantly associated with increased total Medicare expenditures for patients with PMBT.

What predicts early volumetric edema increase following stereotactic radiosurgery for brain metastases?

A volumetric analysis of pre- and post-radiosurgery (PreSRS and PostSRS) edema in patients with cerebral metastases was performed to determine factors of a predictive model assessing the risk of developing increased edema relatively early after SRS. One-hundred-fourteen metastases in 55 patients were analyzed. Selection for this analysis required an MRI ≤ 30 days before SRS and an MRI ≤ 100 days after SRS. Tumor volumes were calculated on PreSRS, SRS, and PostSRS T1-weighted postgadolinium images while edema volumes were calculating using PreSRS and PostSRS fluid-attenuated inversion recovery MR images. An increase in edema was defined as an increase in measurable edema of at least 5%. We developed and evaluated a model predicting the relative risk (RR) of increased edema after SRS. Peritumoral edema increased in 18% (21/114) of the analyzed lesions. Melanoma/renal histology, recursive partitioning analysis class III, and prior WBRT carried RRs of developing postSRS edema increase of 2.45, 2.48, and 3.16, respectively (all P values <0.05). The PreSRS edema/tumor ratio predicted for a RR of 1.007/ratio unit, and steroid dose at time of SRS predicted for a RR of 0.89/mg (all P values <0.05). A predictive model for assessing the RR of increased edema after SRS was developed based from these data and may be useful in identifying patients who might benefit from prophylactic anti-edema therapies before, during, or after SRS. This model could be used as the basis of inclusion criteria for prospective trials investigating novel anti-edema therapies.

Strategies to prevent brain metastasis in high-risk non-small-cell lung cancer: lessons learned from a randomized study of maintenance temozolomide versus observation.

BACKGROUND: This study investigated whether maintenance temozolomide (TMZ) after definitive therapy for locally advanced non-small-cell lung cancer (NSCLC) could decrease the incidence of brain metastasis (BM). PATIENTS AND METHODS: Eligible patients included those with stage IIIA, IIIB, or IV (for stage IV, only with malignant pleural/pericardial effusion) NSCLC with no BM at diagnosis and stable disease, partial response, or complete response after first-line chemotherapy using at least 2 agents. Patients were randomized to observation or TMZ (75 mg/m(2) for 21 consecutive days followed by a 7-day rest for up to 6 cycles or progression). The primary end point was incidence of radiographically diagnosed BM within 12 months from day 1 of first-line chemotherapy. Secondary end points included overall survival (OS), time to progression, incidence of BM at first progression, and toxicity. RESULTS: The study was closed early on the basis of a futility analysis; 45 of 53 enrolled patients were evaluable from an original target of 100. No difference was noted in the incidence of BM at 1 year in the TMZ and observation groups (18% and 13%, respectively), in median time to progression (11.7 and 10.7 months, respectively), or in median OS (27.1 and 22.5 months, respectively). Common Terminology Criteria for Adverse Events grade 3 or 4 adverse events were 46% in the TMZ group and 19% in the observation group. CONCLUSIONS: TMZ monotherapy does not appear to decrease the incidence of BM in patients with locally advanced NSCLC. These results considered in the context of the existing literature have implications for future clinical trial design.