Implementation of machine learning in the clinic: challenges and lessons in prospective deployment from the System for High Intensity EvaLuation During Radiation Therapy (SHIELD-RT) randomized controlled study.

BACKGROUND: Artificial intelligence (AI) and machine learning (ML) have resulted in significant enthusiasm for their promise in healthcare. Despite this, prospective randomized controlled trials and successful clinical implementation remain limited. One clinical application of ML is mitigation of the increased risk for acute care during outpatient cancer therapy. We previously reported the results of the System for High Intensity EvaLuation During Radiation Therapy (SHIELD-RT) study (NCT04277650), which was a prospective, randomized quality improvement study demonstrating that ML based on electronic health record (EHR) data can direct supplemental clinical evaluations and reduce the rate of acute care during cancer radiotherapy with and without chemotherapy. The objective of this study is to report the workflow and operational challenges encountered during ML implementation on the SHIELD-RT study. RESULTS: Data extraction and manual review steps in the workflow represented significant time commitments for implementation of clinical ML on a prospective, randomized study. Barriers include limited data availability through the standard clinical workflow and commercial products, the need to aggregate data from multiple sources, and logistical challenges from altering the standard clinical workflow to deliver adaptive care. CONCLUSIONS: The SHIELD-RT study was an early randomized controlled study which enabled assessment of barriers to clinical ML implementation, specifically those which leverage the EHR. These challenges build on a growing body of literature and may provide lessons for future healthcare ML adoption. TRIAL REGISTRATION: NCT04277650. Registered 20 February 2020. Retrospectively registered quality improvement study.

Role of stereotactic body radiotherapy in gynecologic radiation oncology.

Stereotactic body radiotherapy (SBRT, also referred to as stereotactic ablative radiotherapy (SABR)) has been used in the treatment of primary and metastatic solid tumors, and increasingly so in gynecologic oncology. This review article aims to summarize the current literature describing the utility of SBRT in the primary, recurrent, and limited metastatic settings for gynecologic malignancies. The use of SBRT in both retrospective and prospective reports has been associated with adequate control of the treated site, particularly in the setting of oligometastatic disease. It is not, however, recommended as an alternative to brachytherapy for intact disease unless all efforts to use brachytherapy are exhausted. While phase I and II trials have established the relative safety and potential toxicities of SBRT, there remains a dearth of phase III randomized evidence, including the use of immunotherapy, in order to better establish the role of this technique as a method of improving more global outcomes for our patients with gynecologic cancers.

Health Care Cost Reductions with Machine Learning-Directed Evaluations during Radiation Therapy - An Economic Analysis of a Randomized Controlled Study.

BACKGROUND: Machine learning (ML) may cost-effectively direct health care by identifying patients most likely to benefit from preventative interventions to avoid negative and expensive outcomes. System for High-Intensity Evaluation During Radiation Therapy (SHIELD-RT; NCT04277650) was a single-institution, randomized controlled study in which electronic health record-based ML accurately identified patients at high risk for acute care (emergency visit or hospitalization) during radiotherapy (RT) and targeted them for supplemental clinical evaluations. This ML-directed intervention resulted in decreased acute care utilization. Given the limited prospective data showing the ability of ML to direct interventions cost-efficiently, an economic analysis was performed. METHODS: A post hoc economic analysis was conducted of SHIELD-RT that included RT courses from January 7, 2019, to June 30, 2019. ML-identified high-risk courses (≥10% risk of acute care during RT) were randomized to receive standard of care weekly clinical evaluations with ad hoc supplemental evaluations per clinician discretion versus mandatory twice-weekly evaluations. The primary outcome was difference in mean total medical costs during and 15 days after RT. Acute care costs were obtained via institutional cost accounting. Physician and intervention costs were estimated via Medicare and Medicaid data. Negative binomial regression was used to estimate cost outcomes after adjustment for patient and disease factors. RESULTS: A total of 311 high-risk RT courses among 305 patients were randomized to the standard (n=157) or the intervention (n=154) group. Unadjusted mean intervention group supplemental visit costs were $155 per course (95% confidence interval, $142 to $168). The intervention group had fewer acute care visits per course (standard, 0.47; intervention, 0.31; P=0.04). Total mean adjusted costs were $3110 per course for the standard group and $1494 for the intervention group (difference in means, $1616 [95% confidence interval, $1450 to $1783]; P=0.03). CONCLUSIONS: In this economic analysis of a randomized controlled, health care ML study, mandatory supplemental evaluations for ML-identified high-risk patients were associated with both reduced total medical costs and improved clinical outcomes. Further study is needed to determine whether economic results are generalizable. (Funded in part by The Duke Endowment, The Conquer Cancer Foundation, the Duke Department of Radiation Oncology, and the National Cancer Institute of the National Institutes of Health [R01CA277782]; ClinicalTrials.gov number, NCT04277650.).

Healthcare provider evaluation of machine learning-directed care: reactions to deployment on a randomised controlled study.

OBJECTIVES: Clinical artificial intelligence and machine learning (ML) face barriers related to implementation and trust. There have been few prospective opportunities to evaluate these concerns. System for High Intensity EvaLuation During Radiotherapy (NCT03775265) was a randomised controlled study demonstrating that ML accurately directed clinical evaluations to reduce acute care during cancer radiotherapy. We characterised subsequent perceptions and barriers to implementation. METHODS: An anonymous 7-question Likert-type scale survey with optional free text was administered to multidisciplinary staff focused on workflow, agreement with ML and patient experience. RESULTS: 59/71 (83%) responded. 81% disagreed/strongly disagreed their workflow was disrupted. 67% agreed/strongly agreed patients undergoing intervention were high risk. 75% agreed/strongly agreed they would implement the ML approach routinely if the study was positive. Free-text feedback focused on patient education and ML predictions. CONCLUSIONS: Randomised data and firsthand experience support positive reception of clinical ML. Providers highlighted future priorities, including patient counselling and workflow optimisation.

Early-stage Breast Cancer: Tailored External Beam Fractionation Approaches for Treatment of the Whole or Partial Breast.

Historically, radiotherapy fractionation for early-stage breast cancer primarily consisted of 1.8-2 Gy per fraction given once daily to a total dose of 45-66 Gy over 5-7 weeks for whole breast treatment. Partial breast treatment employed larger dose per fraction (3.4-3.85 Gy) in 10 fractions given twice daily over 1 week. Radiobiologically, breast cancer is increasingly appreciated as a low alpha-beta ratio malignancy that is best treated with larger dose per fraction. Over the past 10 years, there have been increasing data from multiple large randomized clinical trials that support the use of shorter treatment courses: first hypofractionated regimens consisting of 15-20 treatments, and more recently, ultra-hypofractionated regimens as short as 5 treatments. Simultaneously, data from modern partial breast irradiation (PBI) trials support once daily treatment regimens ranging from 1-5 treatments. Shorter treatment courses represent less treatment burden on patients, reduced financial impact, and potentially improved access to care for patients with transportation and/or socioeconomic barriers. Here we review the evolution of whole and partial breast treatment regimens for early-stage breast cancer.

Evaluating for disparities in place of death for head and neck cancer patients in the United States utilizing the CDC WONDER database.

OBJECTIVE: Evaluate trends in place of death for patients with head and neck cancers (HNC) in the U.S. from 1999 to 2017 based on the CDC WONDER (Wide-ranging Online Data for Epidemiologic Research) database. MATERIALS/METHODS: Using patient-level data from 2015 and aggregate data from 1999 to 2017, multivariable logistic regression analyses (MLR) were performed to evaluate for disparities in place of death. RESULTS: We obtained aggregate data for 101,963 people who died of HNC between 1999 and 2017 (25.9% oral cavity, 24.6% oropharynx/pharynx, 0.4% nasopharynx, and 49.1% larynx/hypopharynx). Most were Caucasian (92.7%) and male (87.0%). Deaths at home or hospice increased over the study period (R2 = 0.96, p < 0.05) from 29.2% in 1999 to 61.2% in 2017. On MLR of patient-level data from 2015, those who were single (ref), ages 85+ (OR 0.78; 95% CI: 0.68, 0.90), African American (OR 0.73; 95% CI: 0.65, 0.82), or Asian/Pacific Islanders (OR 0.66; 95% CI: 0.54, 0.81) were less likely to die at home or hospice. On MLR of the aggregate data (1999-2017), those who were female (OR 0.87; 95% CI: 0.83, 0.91) or ages 75-84 (OR 0.79; 95% CI: 0.76, 0.82) were also less likely to die at home or hospice. In both analyses, those who died from larynx/hypopharynx cancers were less likely to die at home or hospice. CONCLUSIONS: HNC-related deaths at home or hospice increased between 1999 and 2017. Those who were single, female, African American, Asian/Pacific Islander, older (ages 75+), or those with larynx/hypopharynx cancers were less likely to die at home or hospice.

System for High-Intensity Evaluation During Radiation Therapy (SHIELD-RT): A Prospective Randomized Study of Machine Learning-Directed Clinical Evaluations During Radiation and Chemoradiation.

PURPOSE: Patients undergoing outpatient radiotherapy (RT) or chemoradiation (CRT) frequently require acute care (emergency department evaluation or hospitalization). Machine learning (ML) may guide interventions to reduce this risk. There are limited prospective studies investigating the clinical impact of ML in health care. The objective of this study was to determine whether ML can identify high-risk patients and direct mandatory twice-weekly clinical evaluation to reduce acute care visits during treatment. PATIENTS AND METHODS: During this single-institution randomized quality improvement study (ClinicalTrials.gov identifier: NCT04277650), 963 outpatient adult courses of RT and CRT started from January 7 to June 30, 2019, were evaluated by an ML algorithm. Among these, 311 courses identified by ML as high risk (> 10% risk of acute care during treatment) were randomized to standard once-weekly clinical evaluation (n = 157) or mandatory twice-weekly evaluation (n = 154). Both arms allowed additional evaluations on the basis of clinician discretion. The primary end point was the rate of acute care visits during RT. Model performance was evaluated using receiver operating characteristic area under the curve (AUC) and decile calibration plots. RESULTS: Twice-weekly evaluation reduced rates of acute care during treatment from 22.3% to 12.3% (difference, -10.0%; 95% CI, -18.3 to -1.6; relative risk, 0.556; 95% CI, 0.332 to 0.924; P = .02). Low-risk patients had a 2.7% acute care rate. Model discrimination was good in high- and low-risk patients undergoing standard once-weekly evaluation (AUC, 0.851). CONCLUSION: In this prospective randomized study, ML accurately triaged patients undergoing RT and CRT, directing clinical management with reduced acute care rates versus standard of care. This prospective study demonstrates the potential benefit of ML in health care and offers opportunities to enhance care quality and reduce health care costs.

Myeloablative conditioning with total body irradiation for AML: Balancing survival and pulmonary toxicity.

PURPOSE: The purpose of this study was to compare leukemia-free survival (LFS) and other clinical outcomes in patients with acute myelogenous leukemia who underwent a myeloablative allogeneic stem cell transplant with and without total body irradiation (TBI). METHODS AND MATERIALS: Adult patients with acute myelogenous leukemia undergoing myeloablative allogeneic stem cell transplant at Duke University Medical Center between 1995 and 2012 were included. The primary endpoint was LFS. Secondary outcomes included overall survival (OS), nonrelapse mortality, and the risk of pulmonary toxicity. Kaplan-Meier survival estimates and Cox proportional hazards multivariate analyses were performed. RESULTS: A total of 206 patients were evaluated: 90 received TBI-based conditioning regimens and 116 received chemotherapy alone. Median follow-up was 36 months. For all patients, 2-year LFS and OS were 36% (95% confidence interval [CI], 29-43) and 39% (95% CI, 32-46), respectively. After adjusting for known prognostic factors using a multivariate analysis, TBI was associated with improved LFS (hazard ratio: 0.63; 95% CI: 0.44-0.91) and OS (hazard ratio: 0.63; 95% CI, 0.43-0.91). There was no difference in nonrelapse mortality between cohorts, but pulmonary toxicity was significantly more common with TBI (2-year incidence 42% vs 12%, P < .001). High-grade pulmonary toxicity predominated with both conditioning strategies (70% and 93% of cases were grade 3-5 with TBI and chemotherapy alone, respectively). CONCLUSIONS: TBI-based regimens were associated with superior LFS and OS but at the cost of increased pulmonary toxicity.

Human papillomavirus tumor infection in esophageal squamous cell carcinoma.

The association between human papillomavirus (HPV) and esophageal squamous cell carcinoma (ESCC) has been recognized for over three decades. Recently, multiple meta-analyses have drawn upon existing literature to assess the strength of the HPV-ESCC linkage. Here, we review these analyses and attempt to provide a clinically-relevant overview of HPV infection in ESCC. HPV-ESCC detection rates are highly variable across studies. Geographic location likely accounts for a majority of the variation in HPV prevalence, with high-incidence regions including Asia reporting significantly higher HPV-ESCC infection rates compared with low-incidence regions such as Europe, North America, and Oceania. Based on our examination of existing data, the current literature does not support the notion that HPV is a prominent carcinogen in ESCC. We conclude that there is no basis to change the current clinical approach to ESCC patients with respect to tumor HPV status.

Long-Term Outcomes of Lobectomy for Non-Small Cell Lung Cancer After Definitive Radiation Treatment.

BACKGROUND: Salvage surgical resection for non-small cell lung cancer (NSCLC) patients initially treated with definitive chemotherapy and radiotherapy can be performed safely, but the long-term benefits are not well characterized. METHODS: Perioperative complications and long-term survival of all patients with NSCLC who received curative-intent definitive radiotherapy, with or without chemotherapy, followed by lobectomy from 1995 to 2012 were evaluated. RESULTS: During the study period, 31 patients met the inclusion criteria. Clinical stage distribution was stage I in 2 (6%), stage II in 5 (16%), stage IIIA in 15 (48%), stage IIIB in 5 (16%), stage IV in 3 (10%), and unknown in 1 (3%). The reasons surgical resection was initially not considered were: patients deemed medically inoperable (5 [16%]); extent of disease was considered unresectable (21 [68%]); small cell lung cancer misdiagnosis (1 [3%]), and unknown (4 [13%]). Definitive therapy was irradiation alone in 2 (6%), concurrent chemoradiotherapy in 28 (90%), and sequential chemoradiotherapy in 1 (3%). The median radiation dose was 60 Gy. Patients were subsequently referred for resection because of obvious local relapse, medical tolerance of surgical intervention, or posttherapy imaging suggesting residual disease. The median time from radiation to lobectomy was 17.7 weeks. There were no perioperative deaths, and morbidity occurred in 15 patients (48%). None of the 3 patients with residual pathologic nodal disease survived longer than 37 months, but the 5-year survival of pN0 patients was 36%. Patients who underwent lobectomy for obvious relapse (n = 3) also did poorly, with a median overall survival of 9 months. CONCLUSIONS: Lobectomy after definitive radiotherapy can be done safely and is associated with reasonable long-term survival, particularly when patients do not have residual nodal disease.

Multiple p53-independent gene silencing mechanisms define the cellular response to p53 activation.

The cellular response to Nutlin-3, a small-molecule inhibitor of the p53 repressor MDM2, varies widely among human cancer-derived cell types. Whereas HCT116 colorectal carcinoma cells display sustained cell cycle arrest, BV173 leukemia cells undergo rapid apoptosis and other cell lines show an intermediate response. We found that the expression of the p53 target genes p21, 14-3-3sigma and the microRNA miR-34a correlates tightly with the cell fate choice adopted. All three genes were strongly induced in arresting cells, but silenced in cells undergoing Nutlin-3-induced apoptosis. In contrast, key apoptotic p53 target genes were equally expressed in arresting and apoptotic cells. Interestingly, we establish that miR-34a cooperates with p21 and 14-3-3sigma to override the apoptotic signals generated by p53 activation. Strikingly, p53 binding to chromatin and p53-mediated recruitment of certain coactivators to all three target loci does not vary among cell types. Instead, the cell type-specific silencing of these genes is due to enhanced p21 mRNA degradation, 14-3-3sigma promoter DNA methylation and reduced processing of the miR-34a primary transcript. Thus, p53-independent events regulating expression of protein-coding genes and microRNAs within the network can define the cellular outcome of p53 activation.