20966 research outputs found
Sort by
PANOVA-3: Phase 3 study of tumor treating fields (TTFields) with gemcitabine and nab-paclitaxel for locally advanced pancreatic ductal adenocarcinoma (LA-PAC)
No full textBackground: To date, no phase 3 clinical trial has demonstrated an overall survival (OS) benefit in patients with locally advanced pancreatic adenocarcinoma (LA-PAC). TTFields are electric fields that disrupt cancer cell division. TTFields therapy is approved for glioblastoma, pleural mesothelioma, and metastatic non-small cell lung cancer. A phase 2 trial in PAC demonstrated the safety and preliminary efficacy of TTFields therapy with gemcitabine with or without nab-paclitaxel. We report final data from PANOVA-3 (NCT03377491), the largest global, phase 3, randomized, open-label trial in LA-PAC to date. Methods: Adult patients with newly diagnosed unresectable LA-PAC were randomized 1:1 to receive TTFields therapy (150 kHz) with gemcitabine/nab-paclitaxel (GnP) or GnP. The primary endpoint was OS. Secondary endpoints included progression-free survival (PFS), local PFS, objective response rate (ORR), and pain-free survival. Distant PFS (metastases beyond the pancreas and regional lymph nodes) was assessed post hoc. Survival data were compared using the Kaplan-Meier method and a log-rank test. Results: 571 patients were randomized. Baseline characteristics were generally well balanced between the study arms. OS was significantly longer with TTFields/GnP than with GnP (median 16.2 [95% CI: 15.0, 18.0] vs 14.2 months [95% CI: 12.8, 15.4]; HR 0.82 [95% CI: 0.68, 0.99], p=0.039). One-year survival rate was also significantly improved with TTFields/GnP vs GnP (68.1% [95% CI: 62.0-73.5] vs 60.2% [95% CI: 54.2-65.7], p=0.029). There was no significant difference in PFS or local PFS between arms. Pain-free survival was significantly longer with TTFields/GnP vs GnP (median 15.2 [95% CI: 10.3, 22.8] vs 9.1 months [95% CI: 7.4, 12.7]; HR 0.74 [95% CI: 0.56, 0.97], p=0.027). Post-hoc analysis showed significant distant PFS benefit (median 13.9 [95% CI: 12.2, 16.8] vs 11.5 months [95% CI: 10.4, 12.9], HR 0.74 [95% CI: 0.57, 0.96], p=0.022) with TTFields/GnP vs GnP. ORR was similar between arms (36.1% [95% CI: 30.0, 42.4] vs 30.0% [95% CI: 24.3, 36.2], p=0.094). 97.8% and 98.9% of patients who received TTFields/GnP and GnP, respectively, had adverse events (AEs) and 88.6% and 84.3% had grade ≥3 AEs. The most frequent grade ≥3 AEs were neutropenia (47.8% and 47.6%) and anemia (21.9% and 22.3). 81% of patients receiving TTFields/GnP had device-related AEs, mostly grade 1/2 skin AEs, e.g., dermatitis (27.7%), rash (17.5%), and pruritus (15.0%); grade 3 and grade 4 device-related AEs occurred in 9.1% and 0.4% of patients, respectively. Conclusions: PANOVA-3 is the largest phase 3 trial exclusively performed in patients with LA-PAC and the first to show a statistically significant OS benefit. With no additive systemic toxicity and a statistically significant pain-free survival benefit, TTFields therapy is a potential new standard treatment for LA-PAC
Benchmarking clinical reasoning and accuracy of large language models on breast oncology multiple-choice questions
No full textBackground: Large language models (LLMs) like GPT-4 (OpenAI) and Claude Opus (Anthropic) showed high accuracy in medical multiple-choice exams, but data on their oncology-specific clinical reasoning and performance is limited. This study evaluates their accuracy and clinical reasoning on breast oncology multiple-choice questions (MCQs) from the American Society of Clinical Oncology (ASCO) question bank. Methods: Using OpenAI and Anthropic Application Programming Interface (APIs), questions were tested without additional prompts under consistent settings (GPT-4 and Claude Opus; Temperature = 0, Tokens = Max). Each question was tested three times to assess precision. Then, Chain-of-thought (COT) prompting was applied to promote LLMs stepwise reasoning to increase their accuracy. Accuracy before and after COT prompting was compared. Incorrect responses were reviewed by board-certified medical on cologists speclized in breast cancer, who scored reasoning clarity, bias, and clinical relevance. Qualitative feedback was descriptively analyzed. Results: A total of 273 breast oncology MCQs were evaluated across the two LLMs. GPT-4 achieved an initial accuracy of 81.3% (222/273; 95% CI: 76.3%-85.5%), compared to Claude Opus, which achieved an accuracy of 79.5% (217/273; 95% CI: 74.3%-83.9%). The Chi-squared test for difference between the models before chain-of-thought (COT) prompting yielded a p-value of 0.59, indicating no statistically significant difference in accuracy between GPT-4 and Claude Opus prior to COT prompting. After COT prompting, the performance of the two models diverged significantly. GPT-4 saw a net decline in accuracy, decreasing by a net of 1 correct answer, resulting in an overall accuracy of 80.95% (221/273). In contrast, Claude Opus experienced a notable improvement, with a net of 19 additional correct answers, leading to an accuracy of 86.4% (236/273). A statistical analysis using a Chi-squared test revealed a difference in accuracy between the two models of 5.5% (p = 0.08), demonstrating that the improvement in accuracy for Claude Opus after COT prompting was borderline statistically significant compared to GPT-4. Thematic analysis of oncologists’ feedback revealed that the most common reasons for incorrect answers were reliance on outdated guidelines, misinterpretation of clinical trial data, and failure to consider multidisciplinary or patient-specific approaches in clinical decision-making. Conclusions: Although AI models can achieve high scores on multiple-choice exams, they still require human supervision. These models rely on potentially outdated training data and lack the ability to individualize patient care or apply data from clinical trials. especially in unique or unconventional/non textbook scenarios
Pre-Implant Renal Function and Optimal Outcomes Among Older LVAD Recipients: An STS-INTERMACS Analysis
No full textPurpose: LVAD recipients are increasingly older and often also have renal dysfunction. For older patients, avoiding adverse events may be as or more important than survival, yet the demographics, comorbidity burden, and impact of renal dysfunction on the risk of sub-optimal outcomes in this population are not well understood. Methods: Using the STS-INTERMACS registry, we analyzed adult patients with INTERMACS profiles 3-7 receiving an isolated first HeartMate 3 LVAD between 6/1/2017 and 6/30/2022. Patients were divided into 3 age groups (\u3c65, 65-70, and \u3e70). To quantify the severities of different adverse events (AEs), we quantified the multivariable-adjusted association between important AEs and 3-year mortality. Regression coefficients were used as the score for each AE, with death assigned a score of 14 (greater than the sum of all individual AE scores, Figure). Patients were classified into outcome score groups at 3 years as follows: 0, alive without AEs; 1-3, alive with low AE burden; 3-\u3c14, alive with high AE burden; 14, dead without AEs; \u3e14, dead with any AEs. Patient characteristics were compared across age and outcome score groups. Results: 4880 patients were included (\u3c65: 3113; 65-70: 767; \u3e70: 1000). Patients \u3c70 were more often white and male, had lower eGFR, and a higher prevalence of frailty compared with patients \u3c65. 3-year survival was 78.3%, 68.6%, and 64.6%, respectively (p\u3c0.001). Patients \u3e70 were less likely to survive with low AE burden compared with patients \u3c65 (62% vs 76%). Survival with low AE burden decreased with increasing age and with decreasing eGFR, with patients \u3e70 and with eGFR\u3c30 having the lowest likelihood of survival with low AE burden (33%, Figure). Conclusion: Older LVAD recipients are less likely to have an optimal outcome at 3 years compared with younger patients, especially those patients with impaired renal function. Understanding the interaction between age and renal function on adverse outcomes will be important to optimizing patient selection for LVAD therapy. [Formula presented
TCT-971 Analysis of Impella and IABP Supported Elective HR-PCI Outcomes and EF Recovery in a Large U.S. Based Electronic Medical Record Database
No full textBackground: PROTECT-II and RESTORE-EF demonstrated LVEF improvements in MCS-supported HRPCI. However, the REVIVED BCIS-2 did not show EF improvement in HF patients undergoing PCI. Methods: We analyzed electronic medical record data (2017-2025) from TRUVETA, a national database aggregating de-identified patient level data across 31 geographically diverse healthcare systems in the U.S. (TRUVETA Inc., Bellevue, WA, USA). We compared patients who received Impella (PLVAD)- with IABP-supported elective high-risk PCI. We excluded patients admitted emergently, STEMI, cardiogenic shock, right heart failure or CABG. A 1:1 propensity score matching (PSM) was performed using co-variates related to the outcomes of interest, including age, sex, history of CHF, reduced LVEF, diabetes, NSTEMI, CKD, cancer, and admissions for ACS, HF, angina, CAD and baseline LVEF. Outcomes and adverse events measured 30 day included all-cause mortality (M-30d), acute kidney injury (AKI), bleeding requiring transfusions (BRT). We evaluated change in ejection fraction (ΔLVEF) defined as difference between pre-PCI LVEF and values reported within 12-months post-PCI. Results: We identified 2794 PLVAD, and 1884 IABP. Prior to PSM adjustment, PLVAD patients were older compared to IABP patients (73±11 vs 68±11 yrs, SMD=0.38), more likely to have a history of CHF (49% vs 24%, SMD=0.54), NSTEMI (34% vs 16%, SMD=0.42) and HFrEF (43% vs 19%, SMD=0.55), anemia (24% vs 15%, SMD=0.23) had lower baseline LVEF (34% vs 40%, SMD=0.22) and more often treated with 2 vessel PCI (50% vs 22% SMD=0.58). The unadjusted rates of M-30d were 12% vs 16% (p\u3c.001) and LVEF gain were 7% vs 3% (p\u3c.001), in PLVAD and IABP, respectively. The PSM identified 1531 patients in each group. After PSM, M-30d were 13% vs 17% (p = 0.003), AKI were 16% vs 20% (p = 0.001), and ΔLVEF were 7% vs 3% (p=. 04), in PLVAD and IABP, respectively. No differences were observed in 30-day or BRT. Conclusion: In this observational analysis of EMR data, patients receiving PLVAD tend to be sicker, have lower baseline EF, and undergo more extensive PCI. After PSM, PLVAD-supported PCI was associated with lower all-cause mortality(30-d), AKI, and greater degree of EF improvement than IABP patients. Categories: CORONARY: Complex and Higher Risk Procedures for Indicated Patients (CHIP
TCT-967 Impella® Protected PCI Outcomes Compared With Intra-Aortic Balloon Pump (IABP) in High-Risk PCI (HRPCI): A Contemporary View
No full textBackground: Traditional trials combine events of varying severity into composites; this may inadequately reflect severity, inherently emphasizes first events, and thus conceals important later signals (particularly in time-to-first-event analytics). Win-Ratio (WR) is a contemporary statistical technique introduced to overcome limitations in reporting composites, and lower sample size needs to demonstrate significance. The Controlled trial of High-Risk Coronary Intervention with Percutaneous Left Ventricular Unloading (CHIP-BCIS3) trial utilizes WR. We analyzed prior trials to compare Impella® Protected PCI to supporting high-risk PCI (HRPCI) with intra-aortic balloon pumps (IABPs), using a WR mimicking the CHIP-BCIS3 definition. Methods: We propensity-matched PROTECT II RCT (P-II) patients with PROTECT III (P-III) Post-Approval Study (PAS) patients who met P-II inclusion criteria to create our Protected PCI cohort. Randomized P-II patients undergoing IABP-supported HRPCI formed the comparator. Major Adverse Cardiac and Cerebrovascular Events (MACCE) in P-II and P-III were independently adjudicated and available fully to 90-days post-PCI. Faithful to CHIP-BCIS3, our WR was defined as follows: Mortality; Stroke; Spontaneous Myocardial Infarction; Re-Hospitalization; and, Peri-Procedural MI. MIs were defined as new Q-waves or CK-MB Elevation ≥8x. All MACCE were analyzed as time-to-event, except Peri-Procedural MIs. Net Benefit (difference between wins and losses, NB) and Win Odds (relative likelihood of winning, WO) are also reported. Results: 719 Protected PCI and 211 IABP HRPCI were included. The WR was 1.691 favoring Impella (95% CI 1.314-2.176, p\u3c0.001*), with NB 0.166 (95% CI 0.084-0.247, p\u3c0.001*) and WO 1.398 (95% CI 1.187-1.645, p\u3c0.001*). Additional reporting can be found in Table 1. [Formula presented] Conclusion: This analysis demonstrates improved Impella Protected PCI outcomes at 90 days, compared to IABP-supported HRPCI. Categories: CORONARY: Complex and Higher Risk Procedures for Indicated Patients (CHIP
TCT-663 Device-Specific Outcomes of Percutaneous Mechanical Aspiration in Right-Sided Infective Endocarditis: Insights from the CLEAR-IE Registry
No full textBackground: Multiple aspiration devices have been used for percutaneous mechanical aspiration (PMA) in right-sided infective endocarditis (RSIE); however, the impact of device-specific characteristics on procedural outcomes remains unclear. Methods: This is an analysis of the CLEAR-IE (Cardiac Lesion Extraction and Aspiration Registry for Infective Endocarditis), a registry of adults with RSIE who underwent PMA from January 2014 to January 2024. Patients were stratified by aspiration technique: continuous-flow (CF) or non-continuous-flow (NCF). Procedural success was defined as ≥70% reduction in vegetation size or residual vegetation ≤1 cm. The primary endpoint was a composite of in-hospital death, new pulmonary embolism, worsening tricuspid regurgitation, or emergency surgery. Secondary endpoints included individual components of the primary outcome and blood loss. Logistic regression and survival analysis were performed. Results: Of 238 patients (median age 43 years, 44.1% female), CF-PMA was used in 202 (84.9%) and NCF-PMA in 36 (15.1%). CF-PMA patients were younger (42 vs. 59.5 years; p=0.004), had higher rates of IV drug use (54% vs. 28.6%; p=0.006), pulmonary embolism (56.4% vs. 30.6%; p=0.004), and persistent sepsis (65.8% vs. 40%; p\u3c0.001). NCF-PMA patients had more atrial fibrillation (36.1% vs. 14.5%; p=0.002), cardiovascular implantable electronic devices (66.7% vs. 19.9%; p\u3c0.001), and lead-related vegetations (54.5% vs. 18.3%; p\u3c0.001). Vegetation size was similar (23 mm vs. 26 mm; p=0.69). Procedural success was similar (87% vs. 97.1%; p=0.091), though, CF-PMA was associated with smaller residual vegetations (0.68 cm vs. 0.95 cm; p=0.005). There was no difference in primary endpoint events (27.2% vs. 30.6%; p=0.68), though CF-PMA had less blood loss (median 50 mL [IQR 50-100] vs. 150 mL [IQR 50-150]; p=0.037). Six-week survival was similar. Conclusion: Both CF and NCF PMA techniques demonstrated high procedural success with comparable rates of adverse events. Differences in baseline patient characteristics may have influenced device selection, and NCF systems were associated with greater blood loss. Prospective studies are needed to confirm these findings and guide optimal device selection. Categories: STRUCTURAL: Valvular Disease and Intervention: Tricuspi
Patterns of transaminase elevation in patients on TPN and association with mortality
No full textBackground There are three primary types of parenteral nutrition- associated liver disease: steatosis- benign, although can progress to cirrhosis in patients receiving long-term PN, cholestasis- serious and can progress to cirrhosis and liver failure, and gallbladder sludge stones. Risk Factors included excess calories or lipids. Our study aims to study different patterns of transaminase elevation, determine If the cholestatic or hepatocellular pattern is more predominant, and finally if a certain kind of elevation has a higher mortality risk. Methods A retrospective chart review was conducted of all adults at our center between 2014-2024, who had a history of elevated liver enzymes while on TPN. Data on indication for TPN, prior history of liver disease, pattern of injury, workup obtained, use of growth factors, and mortality was collected. Results A total of 111 patients with elevated liver enzymes were included. 63 (56.7%) patients had a history of liver disease with the most common indications for TPN being prolonged malnutrition in 41 (36.9%) patients, and short gut syndrome in 39 (35.1%). Mortality occurred in 52 (46.8%) patients. There was no association between the elevation of AST and mortality, p-0.422, the mean time of death from the peak is 38.56 days in the population with transient (\u3c 30 days duration) and 301.69 days in the population with persistent elevation (\u3e 30 days). There was a statistically significant association of ALT with mortality with p- 0.022, with a mean time to mortality of 63.83 days and 328.1 days. There was a statistical association between ALP and mortality with p- 0.034, with a mean time to mortality in 70.8 days and 328.1 days. As for bilirubin, there is no association and the duration to death is 70.8 days and 190.8 days. Further people with TPN change had a higher association with p-0.007, with a mean time to death of 31.79 in the change group versus 36 days in the no change group. Most changes in TPN were associated with ALP and T bili elevation with p- 0.028 and .036 respectively. Conclusions Results indicate ALT elevation is more strongly associated with mortality followed by ALP. The pattern of liver injury more predominant was cholestatic leading to TPN change and further mortality.
TCT-739 Left Atrial Appendage Closure in Patients with Dementia: Data from The National Inpatient Sample Database
No full textBackground: Left atrial appendage closure (LAAC) has emerged as an alternative to oral anticoagulation (OAC) for stroke prevention in patients with atrial fibrillation. Patients with dementia are at increased risk of falls and OAC might be contraindicated. We aimed to evaluate the safety of LAAC in patients with dementia. Methods: The National Inpatient Sample Database was used to identify patients who underwent LAAC between 2018 and 2021. In-hospital complications include mortality, cardiac arrest, myocardial infarction (MI), stroke, pericardial effusion or tamponade, vasopressor use or acute kidney injury (AKI). Multivariate logistic regression analyses were performed. Results: A total of 120,935 patients were included, of whom 4,060 (3%) had dementia. Mean age was 76 (±8) years with 50,165 (42%) females and 103,780 (86%) identified as White. After adjusting for baseline characteristics, dementia was independently associated with significantly higher odds of mortality (OR: 3.46, 95% CI: 1.99-6.02, p\u3c0.0001), myocardial infarction (OR: 4.74, 95% CI: 3.27-6.88, p\u3c0.0001), AKI (OR: 2.15, 95% CI: 1.84-2.51, p\u3c0.0001) stroke (OR: 2.97, 95% CI: 2.35-3.74, p\u3c0.0001) pericardial effusion or tamponade (OR: 2.76, 95% CI: 2.02-3.77, p\u3c0.0001) vasopressor use (OR: 1.68, 95% CI: 1.21-2.33, p=0.001) and cardiac arrest (OR: 2.46, 95% CI: 1.26-4.78, p=0.008) [Formula presented] Conclusion: Patients with dementia undergoing LAAC are at a higher risk of worse in-hospital outcomes. These findings underscore the importance of careful patient selection, risk assessment, and individualized decision-making when considering LAAC in this population. Categories: STRUCTURAL: Left Atrial Appendage Exclusio
What is the Diagnostic Utility of Cardiac Magnetic Resonance Imaging in Unselected Patients with Premature Ventricular Contractions and Non-Sustained Ventricular Tachycardia?
No full textPURPOSE: Premature ventricular contractions (PVCs) and non-sustained ventricular tachycardia (NSVT) are common arrhythmias that may signal underlying structural heart disease (SHD). Cardiac magnetic resonance imaging (CMR) has emerged as a valuable tool for detecting myocardial abnormalities in this population. This study aimed to evaluate the diagnostic utility of CMR in patients with PVCs/NSVT and identify clinical predictors of pathologic late gadolinium enhancement (LGE).
METHODS: We retrospectively reviewed patients who underwent CMR for PVCs or NSVT between 2012 and 2023 at a single health system. The primary outcome was the presence of pathologic LGE. Clinical data were extracted using ICD-10 codes, and cardiac sarcoidosis (CS) was adjudicated by a multidisciplinary team using WASOG criteria.
RESULTS: Among 553 patients (mean age 61.1 ± 14.6 years; 40.7% female), pathologic LGE was identified in 214 (38.6%). Patients with LGE were older and had a greater burden of comorbidities. On multivariable analysis, independent risk factors for LGE included age (aOR 1.04, p=0.001), male sex (aOR 2.37, p\u3c 0.001), heart failure (aOR 2.53, p\u3c 0.001), and polymorphic PVCs (aOR 1.94, p=0.015). Among patients with LGE, 12.6% had highly probable CS. Other diagnoses included non-ischemic cardiomyopathy (53.7%), ischemic cardiomyopathy (11.7%), and idiopathic (34.6%).
CONCLUSION: CMR frequently detects clinically significant myocardial abnormalities in patients with PVCs or NSVT, particularly in those with high-risk features. In this real-world study, nearly 40% of patients had LGE on CMR. An etiology was identified in one-third of these cases. These findings can inform patient selection for CMR in clinical practice to guide diagnosis, risk stratification, and management
