Concurrent Tissue and Circulating Tumor DNA Molecular Profiling to Detect Guideline-Based Targeted Mutations in a Multicancer Cohort.

Importance: Tissue-based next-generation sequencing (NGS) of solid tumors is the criterion standard for identifying somatic mutations that can be treated with National Comprehensive Cancer Network guideline-recommended targeted therapies. Sequencing of circulating tumor DNA (ctDNA) can also identify tumor-derived mutations, and there is increasing clinical evidence supporting ctDNA testing as a diagnostic tool. The clinical value of concurrent tissue and ctDNA profiling has not been formally assessed in a large, multicancer cohort from heterogeneous clinical settings. Objective: To evaluate whether patients concurrently tested with both tissue and ctDNA NGS testing have a higher rate of detection of guideline-based targeted mutations compared with tissue testing alone. Design, Setting, and Participants: This cohort study comprised 3209 patients who underwent sequencing between May 2020, and December 2022, within the deidentified, Tempus multimodal database, consisting of linked molecular and clinical data. Included patients had stage IV disease (non-small cell lung cancer, breast cancer, prostate cancer, or colorectal cancer) with sufficient tissue and blood sample quantities for analysis. Exposures: Received results from tissue and plasma ctDNA genomic profiling, with biopsies and blood draws occurring within 30 days of one another. Main Outcomes and Measures: Detection rates of guideline-based variants found uniquely by ctDNA and tissue profiling. Results: The cohort of 3209 patients (median age at diagnosis of stage IV disease, 65.3 years [2.5%-97.5% range, 43.3-83.3 years]) who underwent concurrent tissue and ctDNA testing included 1693 women (52.8%). Overall, 1448 patients (45.1%) had a guideline-based variant detected. Of these patients, 9.3% (135 of 1448) had variants uniquely detected by ctDNA profiling, and 24.2% (351 of 1448) had variants uniquely detected by solid-tissue testing. Although largely concordant with one another, differences in the identification of actionable variants by either assay varied according to cancer type, gene, variant, and ctDNA burden. Of 352 patients with breast cancer, 20.2% (71 of 352) with actionable variants had unique findings in ctDNA profiling results. Most of these unique, actionable variants (55.0% [55 of 100]) were found in ESR1, resulting in a 24.7% increase (23 of 93) in the identification of patients harboring an ESR1 mutation relative to tissue testing alone. Conclusions and Relevance: This study suggests that unique actionable biomarkers are detected by both concurrent tissue and ctDNA testing, with higher ctDNA identification among patients with breast cancer. Integration of concurrent NGS testing into the routine management of advanced solid cancers may expand the delivery of molecularly guided therapy and improve patient outcomes.

Learning Using Partially Available Privileged Information and Label Uncertainty: Application in Detection of Acute Respiratory Distress Syndrome.

Acute respiratory distress syndrome (ARDS) is a fulminant inflammatory lung injury that develops in patients with critical illnesses, affecting 200,000 patients in the United States annually. However, a recent study suggests that most patients with ARDS are diagnosed late or missed completely and fail to receive life-saving treatments. This is primarily due to the dependency of current diagnosis criteria on chest x-ray, which is not necessarily available at the time of diagnosis. In machine learning, such an information is known as Privileged Information - information that is available at training but not at testing. However, in diagnosing ARDS, privileged information (chest x-rays) are sometimes only available for a portion of the training data. To address this issue, the Learning Using Partially Available Privileged Information (LUPAPI) paradigm is proposed. As there are multiple ways to incorporate partially available privileged information, three models built on classical SVM are described. Another complexity of diagnosing ARDS is the uncertainty in clinical interpretation of chest x-rays. To address this, the LUPAPI framework is then extended to incorporate label uncertainty, resulting in a novel and comprehensive machine learning paradigm - Learning Using Label Uncertainty and Partially Available Privileged Information (LULUPAPI). The proposed frameworks use Electronic Health Record (EHR) data as regular information, chest x-rays as partially available privileged information, and clinicians' confidence levels in ARDS diagnosis as a measure of label uncertainty. Experiments on an ARDS dataset demonstrate that both the LUPAPI and LULUPAPI models outperform SVM, with LULUPAPI performing better than LUPAPI.

Detection of Acute Respiratory Distress Syndrome by Incorporation of Label Uncertainty and Partially Available Privileged Information.

Acute respiratory distress syndrome (ARDS) is a fulminant inflammatory lung injury that develops in patients with critical illnesses including sepsis, pneumonia, and trauma. However, many patients with ARDS are not recognized when they develop this syndrome nor given outcome-improving treatments. Because ARDS is a clinical syndrome, physicians may not be certain about a patient's diagnosis (label uncertainty). In addition, the diagnosis requires a chest x-ray, which may not be always be available in a clinical setting (privileged information). For this paper, we implemented the Learning Using Label Uncertainty and Partially Available Privileged Information (LULUPAPI) paradigm, built on classical SVM, to detect ARDS using Electronic Health Record (EHR) data and chest radiography. In comparison to SVM, this resulted in a 3.55 percent improvement of test AUC.