Warner, Jeremy L

Introduction Registrational trials are clinical trials which are used to justify regulatory approval of a new medication by the FDA. However, clinical trials may not be designed around patient-centred outcomes, like overall survival (OS) or quality of life. This disconnect raises concerns about whether current trial designs are truly aligned with the long-term needs and priorities of patients. A comprehensive analysis of trial designs and endpoints is necessary to ensure that future oncological therapies deliver meaningful clinical improvements. Our objective was to describe types of clinical trial design and primary endpoints used for registrational trials, and their trends over time. Research design and methods In this cohort study, the HemOnc database was used to identify registrational trials supporting cancer drug approval from 2004 through 2023. Clinical trial design (escalation, de-escalation, in-class switch, out-of-class switch) and primary endpoint (OS; time-to-event including progression-free survival; response-based endpoint) were characterized. Results Escalation trial designs were more common than in-class switches or out-of-class switches. De-escalation and mixed designs were exceedingly rare. The most common primary endpoint was a composite time-to-event endpoint. Conclusion Most clinical trials leading to FDA approval for cancer-directed therapies employ escalation designs with composite time-to-event endpoints. Designs reflecting paradigm shifts in therapy (in-class switches and out-of-class switches) are sparsely used. This approach raises the concern that trials may increase therapeutic toxicity without delivering meaningful improvements in OS or quality of life.

PURPOSE The RECIST guidelines provide a standardized approach for evaluating the response of cancer to treatment, allowing for consistent comparison of treatment efficacy across different therapies and patients. However, collecting such information from electronic health records manually can be extremely labor-intensive and time-consuming because of the complexity and volume of clinical notes. The aim of this study is to apply natural language processing (NLP) techniques to automate this process, minimizing manual data collection efforts, and improving the consistency and reliability of the results. METHODS We proposed a complex, hybrid NLP system that automates the process of extracting, linking, and summarizing anticancer therapy and associated RECIST-like responses from narrative clinical text. The system consists of multiple machine learning–/deep learning–based and rule-based modules for diverse NLP tasks such as named entity recognition, assertion classification, relation extraction, and text normalization, to address different challenges associated with anticancer therapy and response information extraction. We then evaluated the system performances on two independent test sets from different institutions to demonstrate its effectiveness and generalizability. RESULTS The system used domain-specific language models, BioBERT and BioClinicalBERT, for high-performance therapy mentions identification and RECIST responses extraction and categorization. The best-performing model achieved a 0.66 score in linking therapy and RECIST response mentions, with end-to-end performance peaking at 0.74 after relation normalization, indicating substantial efficacy with room for improvement. CONCLUSION We developed, implemented, and tested an information extraction system from clinical notes for cancer treatment and efficacy assessment information. We expect this system will support future cancer research, particularly oncologic studies that focus on efficiently assessing the effectiveness and reliability of cancer therapeutics.

PURPOSEThe advancement of natural language processing (NLP) has promoted the use of detailed textual data in electronic health records (EHRs) to support cancer research and to facilitate patient care. In this review, we aim to assess EHR for cancer research and patient care by using the Minimal Common Oncology Data Elements (mCODE), which is a community-driven effort to define a minimal set of data elements for cancer research and practice. Specifically, we aim to assess the alignment of NLP-extracted data elements with mCODE and review existing NLP methodologies for extracting said data elements.METHODSPublished literature studies were searched to retrieve cancer-related NLP articles that were written in English and published between January 2010 and September 2020 from main literature databases. After the retrieval, articles with EHRs as the data source were manually identified. A charting form was developed for relevant study analysis and used to categorize data including four main topics: metadata, EHR data and targeted cancer types, NLP methodology, and oncology data elements and standards.RESULTSA total of 123 publications were selected finally and included in our analysis. We found that cancer research and patient care require some data elements beyond mCODE as expected. Transparency and reproductivity are not sufficient in NLP methods, and inconsistency in NLP evaluation exists.CONCLUSIONWe conducted a comprehensive review of cancer NLP for research and patient care using EHRs data. Issues and barriers for wide adoption of cancer NLP were identified and discussed.

PURPOSE Multiple large clinical trials have investigated adjuvant tyrosine kinase inhibitors (TKIs) to reduce the risk of cancer recurrence and progression to metastasis in high-risk renal cell carcinoma. We sought to maintain living and interactive evidence on this topic, until a high level of certainty is reached for key clinical outcomes such that further updates become unnecessary and unlikely to change clinical practice. METHODS We created a living interactive evidence synthesis platform to maintain a continuously updated meta-analysis on TKI monotherapy in adjuvant renal cell carcinoma. We implemented an automated search strategy with weekly updates to identify randomized phase 2 and 3 clinical trials. Study selection, appraisal, and data extraction were done in duplicate. Cumulative meta-analysis was performed using Analyzer Module in Living Interactive Evidence platform. For each outcome (overall survival [OS], disease-free survival [DFS], and all-cause and treatment-related adverse events), we assessed certainty of evidence using GRADE approach and conducted trial sequential analysis. RESULTS This final update includes five randomized trials including recently updated data from PROTECT trial. Meta-analysis shows that adjuvant TKI monotherapy offers no benefit in OS (hazard ratio, 1.01; 95% CI, 0.91 to 1.12, high certainty) or DFS (hazard ratio, 0.92; 95% CI, 0.86 to 1.00, high certainty) and significantly increases adverse event risk. Lack of benefit was consistent across subgroups including highest-risk patients (test for subgroup differences: P = .32). Optimal information size criteria were met, and there was high certainty of evidence for lack of DFS and OS benefit for adjuvant TKIs. CONCLUSION There is no guidance on when to stop maintaining a living review. In this example, we used trial sequential analysis and high certainty of evidence (future clinical trials unlikely to change current conclusions) as a benchmark to conclude a living review in view of convincing evidence.

Abstract Defining patient-to-patient similarity is essential for the development of precision medicine in clinical care and research. Conceptually, the identification of similar patient cohorts appears straightforward; however, universally accepted definitions remain elusive. Simultaneously, an explosion of vendors and published algorithms have emerged and all provide varied levels of functionality in identifying patient similarity categories. To provide clarity and a common framework for patient similarity, a workshop at the American Medical Informatics Association 2019 Annual Meeting was convened. This workshop included invited discussants from academics, the biotechnology industry, the FDA, and private practice oncology groups. Drawing from a broad range of backgrounds, workshop participants were able to coalesce around 4 major patient similarity classes: (1) feature, (2) outcome, (3) exposure, and (4) mixed-class. This perspective expands into these 4 subtypes more critically and offers the medical informatics community a means of communicating their work on this important topic.

Abstract Among 2,186 U.S. adults with invasive cancer and laboratory-confirmed SARS-CoV-2 infection, we examined the association of COVID-19 treatments with 30-day all-cause mortality and factors associated with treatment. Logistic regression with multiple adjustments (e.g., comorbidities, cancer status, baseline COVID-19 severity) was performed. Hydroxychloroquine with any other drug was associated with increased mortality versus treatment with any COVID-19 treatment other than hydroxychloroquine or untreated controls; this association was not present with hydroxychloroquine alone. Remdesivir had numerically reduced mortality versus untreated controls that did not reach statistical significance. Baseline COVID-19 severity was strongly associated with receipt of any treatment. Black patients were approximately half as likely to receive remdesivir as white patients. Although observational studies can be limited by potential unmeasured confounding, our findings add to the emerging understanding of patterns of care for patients with cancer and COVID-19 and support evaluation of emerging treatments through inclusive prospective controlled trials. Significance: Evaluating the potential role of COVID-19 treatments in patients with cancer in a large observational study, there was no statistically significant 30-day all-cause mortality benefit with hydroxychloroquine or high-dose corticosteroids alone or in combination; remdesivir showed potential benefit. Treatment receipt reflects clinical decision-making and suggests disparities in medication access. This article is highlighted in the In This Issue feature, p. 1426