Citations

Background: Prostate cancer is the second most common new cancer diagnosis in the United States. It is usually slow-growing, and when it is low-grade and confined to the prostate gland, it can be treated either conservatively (through active surveillance) or with surgery. However, if the cancer has spread beyond the prostate, such as to the lymph nodes, then that indicates a more aggressive cancer, and surgery may not be adequate. Methods: The challenge is that it is often difficult for radiologists reading prostate-specific imaging such as magnetic resonance images (MRIs) to differentiate malignant lymph nodes from non-malignant ones. An emerging field is the development of artificial intelligence (AI) models, including machine learning and deep learning, for medical imaging to assist in diagnostic tasks. Earlier research focused on implementing texture algorithms to extract imaging features used in classification models. More recently, researchers began studying the use of deep learning for both stand-alone feature extraction and end-to-end classification tasks. In order to tackle the challenges inherent in small datasets, this study was designed as a scalable hybrid framework utilizing pre-trained ResNet-18, a deep learning model, to extract features that were subsequently fed into a machine learning classifier to automatically identify malignant lymph nodes in patients with prostate cancer. For comparison, two texture algorithms were implemented, namely the gray-level co-occurrence matrix (GLCM) and Gabor. Results: Using an institutional prostate lymph node dataset (42 positives, 84 negatives), the proposed framework achieved an accuracy of 76.19%, a sensitivity of 79.76%, and a specificity of 69.05%. Using GLCM features, the classification achieved an accuracy of 61.90%, a sensitivity of 74.07%, and a specificity of 42.86%. Using Gabor features, the classification achieved an accuracy of 65.08%, a sensitivity of 73.47%, and a specificity of 52.50%. Conclusions: Our results demonstrate that a hybrid approach, i.e., using a pre-trainined deep learning model for feature extraction, followed by a machine learning classifier, is a viable solution. This hybrid approach is especially useful in medical-imaging-based applications with small datasets.

The mature splenic B‐cell lymphomas are an enigmatic group of lymphoid neoplasms that have long caused significant difficulty for the practicing pathologist due to overlapping diagnostic features among entities and the decreasing availability of splenic tissue for assessment. While some entities have highly characteristic and specific clinicopathological features (e.g. hairy cell leukaemia), others are substantially more difficult to recognise (e.g. splenic diffuse red pulp lymphoma). At the same time, classification systems have been evolving, resulting in multiple changes to the boundaries among these entities and even the existence of some entities in their own right. Moreover, unbiased multi‐omic interrogation (whole genome/transcriptome sequencing, methylome) of the splenic B‐cell lymphomas over the past decade has given us significant insights into the underling biology of these neoplasms. We present a clinicopathological perspective on the historical, current and future state of the diagnosis and classification of splenic B‐cell lymphomas integrating multi‐omic data and highlighting areas of focus for the field in order to continue to strive to improve patient outcomes through accurate diagnosis.

Here we review the ‘what and how’ of molecular techniques used in the context of haematopathological diagnostics of both lymphoid and myeloid neoplasms. Keeping in mind that the required resources for molecular testing are not universally available, we will not only discuss novel and emerging techniques that allow more high‐throughput and sophisticated analyses of lymphoid and myeloid neoplasms, but also the more classical, low‐cost alternatives and even some workarounds for molecular testing approaches. In this review we also address other key aspects around molecular techniques for haematopatholgy diagnostics, including preanalytics, data interpretation, and data management, bioinformatics, and interlaboratory precision and performance evaluation.

ABSTRACTBackgroundPatients with ulcerative colitis (UC) who are hospitalised for acute severe flares represent a high‐risk orphan population.AimTo provide guidance for clinical trial design methodology in these patients.MethodsWe created a multi‐centre consortium to design and conduct a clinical trial for a novel therapeutic intervention (hyperbaric oxygen therapy) in patients with UC hospitalised for moderate–severe flares. During planning, we identified and addressed specific gaps for inclusion/exclusion criteria; disease activity measures; pragmatic trial design considerations within care pathways for hospitalised patients; standardisation of care delivery; primary and secondary outcomes; and sample size and statistical analysis approaches.ResultsThe Truelove‐Witt criteria should not be used in isolation. Endoscopy is critical for defining eligible populations. Patient‐reported outcomes should include rectal bleeding and stool frequency, with secondary measurement of urgency and nocturnal bowel movements. Trial design needs to be tailored to care pathways, with early intervention focused on replacing and/or optimising responsiveness to steroids and later interventions focused on testing novel rescue agents or strategies. The PRECIS‐2 framework offers a means of tailoring to local populations. We provide standardisation of baseline testing, venous thromboprophylaxis, steroid dosing, discharge criteria and post‐discharge follow‐up to avoid confounding by usual care variability. Statistical considerations are provided given the small clinical trial nature of this population.ConclusionWe provide an outline for framework decisions made for the hyperbaric oxygen trial in patients hospitalised for UC flares. Future research should focus on the remaining gaps identified.

Abstract AIMS Methods for disease monitoring in glioma are currently limited. Liquid biopsies are of increasing interest within the field but have been challenging due to the blood-brain barrier. To bypass this challenge, we aimed to assess the technical and clinical feasibility of obtaining intracranial cerebrospinal fluid (CSF) for disease monitoring via cell-free DNA (cfDNA). METHOD Adults with gliomas were recruited for longitudinal intracranial CSF collection via CSF access devices, including Ommaya reservoirs in four patients and an existing VP shunt in one patient. cfDNA was then extracted and analyzed using two sequencing platforms, PredicineCARE for cancer variant profiling and PredicineSCORE for low-pass whole genome sequencing (LP-WGS). This analysis generated genomic aberration profiles and genome wide copy number burden (CNB) score. RESULTS Across five patients, thirty-five CSF samples were obtained (median volume: 3.80 mL; 0.5-5 mL), with 30 samples (85.7%) yielding sufficient cfDNA. Tumor fraction increased by 6.28x and 2.87x in two patients with radiographic disease progression and decreased by 0.08x and 0.04x in another two after chemoradiation. In another patient with pseudoprogression, tumor fraction decreased from 0.26 after resection to unmeasurable levels. Patient specific tumor-associated variant allelic frequencies (VAFs), including TP53, PTEN, and TERT, decreased within individual patients after resection and chemoradiation. In two patients with IDH-mutant gliomas, IDH1 VAF after resection correlated with decreased CSF D-2-hydroxyglutarate (D-2-HG) levels (0.64x and 0.62x, respectively, for the first patient, and 0.01x and 0.07x for the other patient). Both IDH1 VAF and D-2-HG increased in a patient with radiographic progression (2.56x and 9.21x, respectively). CNB decreased below the limit of quantification during treatment and increased above the limit at progression. CONCLUSION Longitudinal CSF cfDNA can feasibly be obtained via CSF access devices in patients with gliomas during their disease course. Further sample collection and analysis are ongoing to determine the thresholds of change indicative of treatment response and progression.

Abstract Pharmacogenomic (PGx) biomarkers integrated using machine learning can be embedded within the electronic health record (EHR) to provide clinicians with individualized predictions of drug treatment outcomes. Currently, however, drug alerts in the EHR are largely generic (not patient‐specific) and contribute to increased clinician stress and burnout. Improving the usability of PGx alerts is an urgent need. Therefore, this work aimed to identify principles for optimal PGx alert design through a health‐system‐wide, mixed‐methods study. Clinicians representing multiple practices and care settings ( N = 1062) in urban, rural, and underserved regions were invited to complete an electronic survey comparing the usability of three drug alerts for citalopram, as a case study. Alert 1 contained a generic warning of pharmacogenomic effects on citalopram metabolism. Alerts 2 and 3 provided patient‐specific predictions of citalopram efficacy with varying depth of information. Primary outcomes included the System's Usability Scale score (0–100 points) of each alert, the perceived impact of each alert on stress and decision‐making, and clinicians' suggestions for alert improvement. Secondary outcomes included the assessment of alert preference by clinician age, practice type, and geographic setting. Qualitative information was captured to provide context to quantitative information. The final cohort comprised 305 geographically and clinically diverse clinicians. A simplified, individualized alert (Alert 2) was perceived as beneficial for decision‐making and stress compared with a more detailed version (Alert 3) and the generic alert (Alert 1) regardless of age, practice type, or geographic setting. Findings emphasize the need for clinician‐guided design of PGx alerts in the era of digital medicine.

Abstract Objective SNOMED CT provides a standardized terminology for clinical concepts, allowing cohort queries over heterogeneous clinical data including Electronic Health Records (EHRs). While it is intuitive that missing and inaccurate subtype (or is-a) relations in SNOMED CT reduce the recall and precision of cohort queries, the extent of these impacts has not been formally assessed. This study fills this gap by developing quantitative metrics to measure these impacts and performing statistical analysis on their significance. Material and Methods We used the Optum de-identified COVID-19 Electronic Health Record dataset. We defined micro-averaged and macro-averaged recall and precision metrics to assess the impact of missing and inaccurate is-a relations on cohort queries. Both practical and simulated analyses were performed. Practical analyses involved 407 missing and 48 inaccurate is-a relations confirmed by domain experts, with statistical testing using Wilcoxon signed-rank tests. Simulated analyses used two random sets of 400 is-a relations to simulate missing and inaccurate is-a relations. Results Wilcoxon signed-rank tests from both practical and simulated analyses (P-values < .001) showed that missing is-a relations significantly reduced the micro- and macro-averaged recall, and inaccurate is-a relations significantly reduced the micro- and macro-averaged precision. Discussion The introduced impact metrics can assist SNOMED CT maintainers in prioritizing critical hierarchical defects for quality enhancement. These metrics are generally applicable for assessing the quality impact of a terminology’s subtype hierarchy on its cohort query applications. Conclusion Our results indicate a significant impact of missing and inaccurate is-a relations in SNOMED CT on the recall and precision of cohort queries. Our work highlights the importance of high-quality terminology hierarchy for cohort queries over EHR data and provides valuable insights for prioritizing quality improvements of SNOMED CT's hierarchy.

Objective To explore associations between general health-related problems and subclinical coronary artery disease (CAD), determined by CT coronary calcium score (CT-CCS), in a general population. Design A cross-sectional design. Setting This study was performed in a prospective population-based cohort, examining the health and health-related behaviour of individuals living in the Northern Netherlands. Participants The initial cohort comprised 6763 participants ≥45 years of age who underwent CT-scanning. Participants were included for the current analysis if they filled in three validated questionnaires (Symptomatic Checklist-90, Research and Development Survey-36 and Reviving the Early Diagnosis of CardioVascular Diseases questionnaire (RED-CVD)) and did not have a history of cardiovascular disease. The final analysis included 6530 participants. Primary outcome measure Backward-stepwise and forward-stepwise logistic regression analyses were performed to determine associations between general health-related problems and subclinical CAD (CCS≥100 and ≥300). Results The median age was 53 years (25th, 75th percentile: 48, 58); 57% were women. CRCS≥100 was found in 1236 (19%) participants, 437 (12%) in women and 799 (29%) men and CCS≥300 in 643 (9.9%) participants of which 180 (4.8%) were women and 463 (16.6%) men. In univariate analysis, in women the expectation of health to worsen (OR=1.13, 95% CI: 1.05 to 1.21), and in men reduced exercise intolerance (OR=1.14, 95% CI: 1.06 to 1.23) were associated with CCS≥100. The total RED-CVD score in women (OR=1.06, (95% CI: 1.05 to 1.08) and men (OR=1.07, 95% CI: 1.06 to 1.09), and in men also reduced exercise intolerance (OR=1.15, 95% CI: 1.06 to 1.25) and headache (OR=0.55, 95% CI: 0.38 to 0.79) were associated with CCS≥300. In multivariate analyses, only general health expectation in women was still significantly associated with subclinical CAD (CCS≥300) (OR=1.92, 95% CI: 1.56 to 2.37). Conclusion Only a few general health-related problems were associated with the presence of subclinical CAD in the general population, however, these problems showed no strong association. Therefore, using health-related symptoms does not seem useful to pre-select for CT-CCS. Trial registration number CCMO Register, NL17981.042.07 and NL58592.042.16.