Health Equity Achievement in Radiation Therapy (HEART) Score: A Social Prognosis

PURPOSE/OBJECTIVE(S)

The aim of this study was to develop a Health Equity Achievement in Radiation Therapy (HEART) score that can help identify patients at risk of experiencing suboptimal quality-of-care (QoC) early on in the patient-provider encounter and prior to initiation of treatment. Such a score may improve shared decision making to improve QoC.

MATERIALS/METHODS

A retrospective analysis was conducted using the National Cancer Database (NCDB) for prostate cancer cases between 2004-2017. Sociodemographic factors, clinical characteristics, and treatment information were collected. A composite HEART score was built to predict suboptimal QoC, defined as treatment refusal, incomplete treatment, or treatment delay. 70% of the data was allocated to training and 30% to validating a logistic regression model through which a nomogram was constructed.

RESULTS

A total of 1,599,785 patients were included in the analysis, of whom 126,917 (7.9%) had at least one suboptimal QoC. The strongest predictors were Black race, uninsured status, lower educational status, geographic location, and nodal disease (Table). The nomogram demonstrated a fair ability to predict quality metrics, with an area under the receiver operating characteristic curve (AUC) of 0.57 in the test group. The nomogram facilitated graphic interpretation of systemic factors in contributing to suboptimal QoC.

CONCLUSION

With observed potential for predicting suboptimal QoC outcomes in patients with prostate cancer by considering systemic barriers, this NCDB-based nomogram has potential utility as a tool for identifying patients who may benefit from additional social support, including the financial resources associated with these services, to improve access to care. Further validation in diverse datasets is needed to improve performance and generalizability to broader patient populations and different disease sites.

Use of Explainable AI Algorithm Revealing Longitudinal Changes in Practice Patterns and Toxicity Models

PURPOSE/OBJECTIVE(S)

Dosimetric constraints evolve as clinicians implement practice changes, requiring modeling approaches to be dynamic. We applied a semi-automated explainable artificial intelligence (eAI) algorithm and dashboard visualizations to model dysphagia and xerostomia for head and neck cancer patients. We coupled a large, comprehensive, "real-world" database to the eAI for discovery of features with the strongest combined statistical and machine learning based evidence and to identify clinically actionable thresholds.

MATERIALS/METHODS

Cohort included 758 patients treated 2017-2021 for HN cancer with conventional fractionation. Features included age, sex, diagnosis, staging, chemotherapy, smoking and alcohol status, BMI, weight loss, re-simulation, DVH curves, PTV and OAR volumes. Patients were scored for toxicity within 2 yrs of RT for dysphagia grade ≥ 3 and xerostomia grade ≥ 2. Bootstrap resampling of thresholds, ROC-AUC, PR-ROC, SN, SP, F1 and diagnostic odds ratio was used to statistically profile strength of evidence for candidate features. XGBoost models with 10-fold cross validation were repeated (n = 20) to identify mean and CIs for statistical measures of predictions. DVH metrics included standard template values and those with highest statistical evidence and low cross correlation with other features. Backward feature selection was used to identify the most relevant feature subset, where the least informative feature is iteratively removed from the model. This workflow was repeated by year and overall.

RESULTS

Annual incidence of dysphagia averaged 0.13 ± 0.02 overall years. Xerostomia incidence decreased from 0.32 to 0.12 (2017-2021). Box-whisker plots by year showed consistent reductions in standard practice toxicity linked DVH metric values. Median dose to superior constrictors (PCM), contralateral parotid and contralateral submandibular gland (SMG) declined from 2017 to 2021 by 48 to 33 Gy, 17 to 10 Gy, and 28 to 22 Gy respectively. Statistics of XGBoost models of dysphagia for all years were ROC-AUC = 0.72 ± 0.05. Strongest overall years predictors were Oral Cavity (OC) D50%[Gy] < 32, inferior PCM Max [Gy] < 60, contralateral SMG D10%[Gy] < 53 and use of Paclitaxel. Xerostomia models were less predictive with ROC-AUC = 0.65 ± 0.05. Strongest predictors over each year were ipsilateral parotid D30%[Gy] < 35, contralateral SMG D96%[Gy] < 18.4, and overall staging < II. Predictive features varied substantially by year for both, showing the most consistency for SMG doses. For example, OC D50%[Gy] < 27 and contralateral SMG D96%[Gy] < 18 dominated xerostomia model in 2017 but not in 2021 when practice norms shifted to lower doses.

CONCLUSION

As OAR doses were systematically reduced, statistical and AI models evidence highlighted contralateral SMG dose as important to both dysphagia and xerostomia for clinical practice change. The "real-world" database + eAI + visualization dashboards provided a method for continuous learning as clinical practice changes.

Operational Ontology for Radiation Oncology (OORO): A Professional Society-Based, Multi-Stakeholder Consensus Driven Informatics Standard Supporting Clinical and Research Use of Real-World Data

PURPOSE/OBJECTIVE(S)

There is a critical need for large-scale, multi-institutional "real-world" data to evaluate patient, diagnosis and treatment factors affecting oncology patient outcomes. However, lack of data standardization undermines the potential for automated learning from the vast amount of information routinely archived in electronic health records (EHRs), Radiation Oncology Information Systems and other cancer care databases. As next step to promote data standardization beyond the American Association of Physicists in Medicine (AAPM)'s TG-263 guidance for radiotherapy (RT) nomenclature, the AAPM's Big Data Subcommittee (BDSC) has led an international RT professional society collaboration to develop the Operational Ontology for Radiation Oncology (OORO).

MATERIALS/METHODS

Initiated July 2019 to explore issues that typically compromise formation of large inter- and intra- institutional databases from EHRs, the AAPM's BDSC membership includes representatives from the AAPM, American Society of Radiation Oncology (ASTRO), Canadian Organization of Medical Physicists (COMP), Canadian Association of Radiation Oncology (CARO), European Society of Therapeutic Radiation Oncology (ESTRO) and clinical trials experts from NRG Oncology. Multiple external stakeholders were engaged, including government agencies, vendors and RT community members through the iterative and consensus-driven approach to OORO development.

RESULTS

The OORO includes 42 key elements, 359 attributes, 144 value sets, and 155 relationships, ranked for priority of implementation based on clinical significance, likelihood of availability in EHRs, or ability to modify routine clinical processes to permit aggregation. The initial version of OORO includes many disease-site independent concepts common for all cancer patients and a smaller set specific for prostate cancer. The OORO development methodology is currently being applied/adapted to include additional disease site-specific concepts beginning with head and neck cancers.

CONCLUSION

The first of its kind in radiation oncology, the OORO is a professional society-based, multi-stakeholder, consensus driven informatics standard. The iterative and collaborative approach to ontology development and refinement aims to ensure that OORO serves as a « living » guidance document, facilitating incremental expansion of data elements over time, as disease site-specific standards are set and RT concepts evolve. Supporting construction of comprehensive "real-world" datasets and application of advanced analytic techniques, including artificial intelligence (AI), OORO holds the potential to revolutionize patient management and improve outcomes.

Residency, demographics, and movement patterns of North Atlantic right whales Eubalaena glacialis in an offshore wind energy development area in southern New England, USA

Offshore wind energy development is growing quickly around the world. In southern New England, USA, one of the largest commercial offshore wind energy farms in the USA will be established in the waters off Massachusetts and Rhode Island, an area used by the Critically Endangered North Atlantic right whale Eubalaena glacialis. Prior to 2011, little was known about the use of this area by right whales. We examined aerial survey data collected between 2011-2015 and 2017-2019 to quantify right whale distribution, residency, demography, and movements in the region. Right whale occurrence increased during the study period. Since 2017, whales have been sighted in the area nearly every month, with peak sighting rates between late winter and spring. Model outputs suggest that 23% of the species’ population is present from December through May, and the mean residence time has tripled to an average of 13 d during these months. Age and sex ratios of the individuals present in the area are similar to those of the species as a whole, with adult males the most common demographic group. Movement models showed that southern New England is an important destination for right whales, including conceptive and reproductive females, and qualitative observations included animals feeding and socializing. Implementing mitigation procedures in coordination with these findings will be crucial in lessening the potential impacts on right whales from construction noise, increased vessel traffic, and habitat disruption in this region.

Evidence of ‘sickness behaviour’ in bats with white-nose syndrome

Many animals change behaviour in response to pathogenic infections. White-nose syndrome (WNS) is a fungal skin disease causing rapid declines of North American bats. Infection with Pseudogymnoascus destructans causes hibernating bats to arouse from torpor too often, potentially causing starvation. Mechanisms underlying increased arousals are not understood but fungal invasion of the wings could trigger thirst to relieve fluid loss or grooming to relieve skin irritation. Alternatively, bats might exhibit ‘sickness behaviour’, a suite of responses to infection that save energy. We quantified behaviours of healthy and experimentally inoculated little brown bats (Myotis lucifugus) that could reflect active (i.e., drinking, grooming) or inactive (i.e., sickness behaviour) responses to infection. Infected bats groomed less and were less likely to visit their water dish compared to controls. These results are consistent with research suggesting that P. destructans causes sickness behaviour which could help bats compensate for energetic costs associated with infection.

Behaviour of pollinator insects within inflorescences of Scrophularia species from Iberian Peninsula

Several hypotheses have been proposed to explain the pattern of pollinator visits to vertical inflorescences. These range from a response to a pattern of resources to merely instinctive behaviour. In dichogamous plants, such behaviour has been associated with promoting outcrossing and avoiding geitonogamy. We here analyse behaviour of the principal pollinator groups in five protogynous species of Scrophularia with different flower sizes (S. sambucifolia, S. grandiflora, S. lyrata, S. scorodonia and S. canina), and the distribution of sexual phases along the inflorescences. The results in all cases show that pollinators follow a pattern of ascending visits accompanied by movements between flowers of the same whorl (horizontal movements). The relative frequency of these horizontal movements depends on the flower size, with a higher frequency in species with large flowers. In vertical movements of the three more common pollinator groups to several plant species (bumblebees, wasps and small bees), the behaviour was essentially independent of flower size, with bumblebees having the highest ratio of ascents to descents. Behaviour of the pollinators, together with the absence of a definite pattern of distribution of the sexual phases along the inflorescence, implies that geitonogamy is not avoided in any of the Scrophularia species studied.

Brain metastases from lung cancer responding to erlotinib: the importance of EGFR mutation

Median survival of patients with brain metastases from nonsmall cell lung cancer (NSCLC) is poor and more effective treatments are urgently needed. We have evaluated the efficacy of erlotinib in this setting and its association with activating mutations in the epidermal growth factor receptor (EGFR) gene. We retrospectively identified patients with NSCLC and brain metastases treated with erlotinib. EGFR mutations in exons 19 and 21 were analysed by direct sequencing. Efficacy and tolerability were compared according to EGFR mutational status. 69 NSCLC patients with brain metastases were identified, 17 of whom harboured EGFR mutations. Objective response rate in patients with EGFR mutations was 82.4%; no responses were observed in unselected patients (p<0.001). Median (95% CI) time to progression within the brain for patients harbouring EGFR mutations was 11.7 (7.9-15.5) months, compared to 5.8 (5.2-6.4) months for control patients whose EGFR mutational status had not been assessed (p<0.05). Overall survival was 12.9 (6.2-19.7) months and 3.1 (2.5-3.9) months (p<0.001), respectively. The toxicity of erlotinib was as expected and no differences between cohorts were observed. Erlotinib is active in brain metastases from NSCLC; this clinical benefit is related to the presence of activating mutations in exons 19 or 21 of the EGFR gene.