Citations

Abstract Objective In acupuncture therapy, the accurate location of acupoints is essential for its effectiveness. The advanced language understanding capabilities of large language models (LLMs) like Generative Pre-trained Transformers (GPTs) and Llama present a significant opportunity for extracting relations related to acupoint locations from textual knowledge sources. This study aims to explore the performance of LLMs in extracting acupoint-related location relations and assess the impact of fine-tuning on GPT’s performance. Materials and Methods We utilized the World Health Organization Standard Acupuncture Point Locations in the Western Pacific Region (WHO Standard) as our corpus, which consists of descriptions of 361 acupoints. Five types of relations (“direction_of”, “distance_of”, “part_of”, “near_acupoint”, and “located_near”) (n = 3174) between acupoints were annotated. Four models were compared: pre-trained GPT-3.5, fine-tuned GPT-3.5, pre-trained GPT-4, as well as pretrained Llama 3. Performance metrics included micro-average exact match precision, recall, and F1 scores. Results Our results demonstrate that fine-tuned GPT-3.5 consistently outperformed other models in F1 scores across all relation types. Overall, it achieved the highest micro-average F1 score of 0.92. Discussion The superior performance of the fine-tuned GPT-3.5 model, as shown by its F1 scores, underscores the importance of domain-specific fine-tuning in enhancing relation extraction capabilities for acupuncture-related tasks. In light of the findings from this study, it offers valuable insights into leveraging LLMs for developing clinical decision support and creating educational modules in acupuncture. Conclusion This study underscores the effectiveness of LLMs like GPT and Llama in extracting relations related to acupoint locations, with implications for accurately modeling acupuncture knowledge and promoting standard implementation in acupuncture training and practice. The findings also contribute to advancing informatics applications in traditional and complementary medicine, showcasing the potential of LLMs in natural language processing.

AbstractIron Dextran is a widely used iron oxide compound to treat iron‐deficiency anemia patients in the clinic. Similar to other iron oxide compounds such as Ferumoxytol, it can also be used off‐label as an intravascular magnetic resonance imaging (MRI) contrast agent due to its strong iron‐induced T2 and T2* shortening effects. In this study, we seek to evaluate the feasibility of using Iron Dextran enhanced multi‐echo susceptibility weighted imaging (SWI) MRI at 7T to image arterial and venous blood vessels in the human brain. Phantom experiments were performed to measure the r2* relaxivity for Iron Dextran in blood, based on which the SWI sequence was optimized. Pre‐ and post‐infusion MR images were acquired in human subjects from which maps of arteries and veins were extracted. The post‐contrast SWI images showed enhanced susceptibility difference between blood and the surrounding tissue in both arteries and veins. Our results showed that the proposed Iron Dextran enhanced multi‐echo SWI approach allowed the visualization of blood vessels with diameters down to ~100 μm, including small blood vessels supplying and draining small brain structures such as the hippocampus. We conclude that Iron Dextran can be an alternative iron‐based MRI contrast agent for blood vessel imaging in the human brain.

INTRODUCTION: Esophageal adenocarcinoma (EAC) is the second-most lethal cancer in the United States, with Barrett esophagus (BE) being the strongest risk factor. Assessing the future risk of neoplastic progression in patients with BE is difficult; however, high-grade dysplasia (HGD) and early EAC are treatable by endoscopic eradication therapy (EET), with survival rates of 90%. Thus, it would be beneficial to develop a molecular assay to identify high-risk patients, who merit more frequent endoscopic surveillance or EET, as well as low-risk patients, who can avoid EET and undergo less frequent surveillance. METHODS: Deidentified endoscopic biopsies were acquired from 240 patients with BE at 6 centers and confirmed as future progressors or nonprogressors. Tissues were analyzed by a set of methylation-specific biomarker assays. Test performance was assessed in an independent validation set using 4 stratification levels: low risks, low-moderate risks, high-moderate risks, and high risks. RESULTS: Relative to patients in the low-risk group, high-risk patients were 15.2 times more likely to progress within 5 years to HGD or EAC. For patients in the high-risk category, the average risk of progressing to HGD or EAC within 5 years was 21.5%, 4-fold the BE population prevalence within 5 years, whereas low-risk patients had a progression risk of only 1.85%. DISCUSSION: This clinical assay, Esopredict, stratifies future neoplastic progression risk to identify higher-risk patients with BE who can benefit from EET or more frequent surveillance and lower-risk patients who can benefit from reduced surveillance.

AbstractSupplemental O2 (hyperoxia) is a critical intervention for premature infants (<34 weeks) but consequently is associated with development of bronchial airway hyperreactivity (AHR) and asthma. Clinical practice shifted toward the use of moderate hyperoxia (<60% O2), but risk for subsequent airway disease remains. In mouse models of moderate hyperoxia, neonatal mice have increased AHR with effects on airway smooth muscle (ASM), a cell type involved in airway tone, bronchodilation, and remodeling. Understanding mechanisms by which moderate O2 during the perinatal period initiates sustained airway changes is critical to drive therapeutic advancements toward treating airway diseases. We propose that cellular clock factor BMAL1 is functionally important in developing mouse airways. In adult mice, cellular clocks target pathways highly relevant to asthma pathophysiology and Bmal1 deletion increases inflammatory response, worsens lung function, and impacts survival outcomes. Our understanding of BMAL1 in the developing lung is limited, but our previous findings show functional relevance of clocks in human fetal ASM exposed to O2. Here, we characterize Bmal1 in our established mouse neonatal hyperoxia model. Our data show that Bmal1 KO deleteriously impacts the developing lung in the context of O2 and these data highlight the importance of neonatal sex in understanding airway disease.

7012 Background: DLBCL with expression of MYC and BCL2 (double expressor lymphoma, DEL), has an adverse prognosis. Venetoclax, an inhibitor of BCL2, was previously studied in combination with R-CHOP. A051701 is a phase II/III randomized trial of chemoimmunotherapy +/- venetoclax in DEL and double hit lymphoma cohorts. Here we report initial phase II results from the DEL cohort. Methods: Patients (pts) age ≥ 18 years (y) with untreated DEL were randomized 1:1 to R-CHOP (Arm 1) or R-CHOP-venetoclax (Arm 2). DEL was defined as MYC expression ≥40% and BCL2 expression ≥50%, without double hit cytogenetics. Enrollment was based on local pathology results that were centrally confirmed. Venetoclax 800 mg was given orally days 4-8 of cycle 1 and days 1-5 of subsequent cycles for up to 6 cycles. All cycles were supported by GCSF or peg-GCSF. The primary endpoint was progression-free survival (PFS) with 84% power to detect a HR of 0.518 corresponding to 2y PFS rate of 65% in Arm 1 and 80% in Arm 2 (1-sided α=0.20). Results: 119 DEL pts were enrolled: 60 Arm 1; 59 Arm 2. Analyses were performed for the modified intent-to-treat population, defined as eligible pts with confirmed DEL histology (n=113; 56 Arm 1, 57 Arm 2). Median age was 64 y (range 22-85); baseline demographic factors were well balanced except for advanced stage (more common in Arm 1; (77% vs 60%, p=0.051). Most pts were non-germinal center B cell-like by immunohistochemistry (Arm 1 72%, Arm 2 58%), high-intermediate/high risk IPI (Arm 1 54%, Arm 2 51%) and ECOG performance status 0-1 (Arm 1 95%, Arm 2 86%). BCL2 rearrangements were detected in 25% and 33% of pts on Arm 1 and Arm 2, respectively. Most pts completed therapy per protocol (84% Arm 1, 75% Arm 2). Median follow up is 27 months (m). No difference in PFS was observed with 12-m PFS estimates of 77% in Arm 1, 76% in Arm 2 [HR = 0.98 (95% CI: 0.48 – 2.01), p=0.95]. The end of treatment (EOT) overall and complete response rates among patients with an EOT assessment were 87.5% and 70% in Arm 1, and 90% and 82% in Arm 2. Overall survival (OS) was not statistically significantly different with 12-mo OS estimates of 94% in Arm 1 and 79% in Arm 2 [HR = 1.27 (95% CI: 0.57 – 2.79), p=0.56]. Grade ≥3 adverse events (AEs) occurred more frequently in Arm 2 (42% vs 76%). The most commonly increased grade ≥3 AEs included neutropenia (20% vs 47%), anemia (4% vs 25.5%), thrombocytopenia (5.5% vs 24%), febrile neutropenia (7% vs 16%) and fatigue (0% vs 11%). No grade 5 AEs on treatment were reported on Arm 1, and 3 were reported on Arm 2 (lung infection and sudden death NOS, both possibly related; respiratory failure, unlikely related). Conclusions: Adding venetoclax to R-CHOP in DEL pts did not improve PFS and resulted in increased toxicity. The study did not meet its primary endpoint in phase II, and so did not proceed to phase III. Clinical trial information: NCT03984448 .

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.

The development and performance of two mass spectrometry (MS) workflows for the intraoperative diagnosis of isocitrate dehydrogenase (IDH) mutations in glioma is implemented by independent teams at Mayo Clinic, Jacksonville, and Huashan Hospital, Shanghai. The infiltrative nature of gliomas makes rapid diagnosis necessary to guide the extent of surgical resection of central nervous system (CNS) tumors. The combination of tissue biopsy and MS analysis used here satisfies this requirement. The key feature of both described methods is the use of tandem MS to measure the oncometabolite 2-hydroxyglutarate (2HG) relative to endogenous glutamate (Glu) to characterize the presence of mutant tumor. The experiments i) provide IDH mutation status for individual patients and ii) demonstrate a strong correlation of 2HG signals with tumor infiltration. The measured ratio of 2HG to Glu correlates with IDH-mutant (IDH-mut) glioma ( P < 0.0001) in the tumor core data of both teams. Despite using different ionization methods and different mass spectrometers, comparable performance in determining IDH mutations from core tumor biopsies was achieved with sensitivities, specificities, and accuracies all at 100%. None of the 31 patients at Mayo Clinic or the 74 patients at Huashan Hospital were misclassified when analyzing tumor core biopsies. Robustness of the methodology was evaluated by postoperative re-examination of samples. Both teams noted the presence of high concentrations of 2HG at surgical margins, supporting future use of intraoperative MS to monitor for clean surgical margins. The power of MS diagnostics is shown in resolving contradictory clinical features, e.g., in distinguishing gliosis from IDH-mut glioma.

AbstractIschemia‒reperfusion (I/R) is a common complication in the clinical treatment of acute myocardial infarction (MI), in which cardiomyocytes play a pivotal role in the recovery of cardiac function after reperfusion injury. The expression of numerous circular ribonucleic acids (circRNAs) is disrupted in I/R‐induced cardiac damage, but the potential role of circRNAs in I/R damage has not been fully elucidated. The purpose of the present study was to clarify the biological action and molecular mechanism of circRNA 002166 (also termed circCL2L13) in postmyocardial I/R. Oxygen–glucose deprivation/reoxygenation (OGD/R) in an in vivo model was performed to simulate I/R damage. real‐time polymerase chain reaction analysis was also conducted to evaluate the relationships of the SOD1, SOD2, NRF2, HO1 and GPX4 indicators with oxidative stress injury. TUNEL immunofluorescence was used to evaluate the degree of cardiomyocyte apoptosis in the different treatment groups. The circBCL2L13 level was markedly upregulated in myocardial tissues from a mouse I/R model. Overexpression of circBCL2L13 markedly attenuated the expression of oxidative stress‐related genes and apoptosis in OGD/R‐induced cardiomyocytes. A mechanistic study revealed that circBCL2L13 functions as a ceRNA for miR‐1246 and modulates paternally expressed gene 3 (PEG3). Eventually, circBCL2L13 was proven to regulate PEG3 by targeting miR‐1246, thereby protecting against OGD/R‐induced cardiomyocyte oxidative damage and apoptosis. In conclusion, our study confirmed that the circBCL2L13/miR‐1246/PEG3 axis suppressed the progression of OGD/R injury in cardiomyocytes, which might lead to new therapeutic strategies for cardiac I/R injury.