Citations

Abstract Changes in the clinical symptoms of acute graft-versus-host disease (GVHD) are currently used to assess treatment responses. The Mount Sinai Acute GVHD International Consortium (MAGIC) consortium has recently revealed that the integration of serum biomarkers with clinical symptoms at the onset of treatment in a MAGIC composite score (MCS) more accurately predicts treatment response and 6-month nonrelapse mortality (NRM) than clinical symptoms alone. In this study, we evaluated whether the integration of serum biomarkers and clinical symptoms on day 28 (D28) would also better predict NRM than clinical response only (CRO). We analyzed data from 1135 patients receiving systemic treatment for acute GVHD and created a fourth MCS category for patients with complete resolution of symptoms and low-risk clinical biomarkers on D28. Using a classification and regression tree model with 6-month NRM as the end point, we identified status of MCS 0 or MCS 1 at D28 as responses, which we termed the MAGIC composite response (MCR). In the validation cohort (n = 309), MCR more accurately predicted 6-month NRM than CRO (area under the curve: 0.77 vs 0.69; P = .014) and demonstrated higher negative and positive predictive values. MCR correctly reclassified both clinical nonresponders and responders: 28 of 213 clinical responders (13%) became nonresponders with fivefold higher NRM (34.3% vs 6.8%, P < .001) and a larger group (29/96, 30%) of clinical nonresponders became responders with sixfold lower NRM (7.6% vs 50.7%, P < .001). These findings support the use of MCR as a superior surrogate end point for long-term GVHD control and survival in future clinical trials.

BACKGROUND: The performance of estimated glomerular filtration rate (eGFR) among patients with heart failure (HF) may be worse than in the general population due to a higher prevalence of confounding factors affecting creatinine and cystatin C. Studies in this area are scarce and not stratified by type of HF. We evaluated the performance of current creatinine and cystatin C equations (eGFRcr, eGFRcys, and eGFRcrcys) compared with measured GFR (mGFR) among patients with HF stratified by ejection fraction. METHODS: We pulled data on Mayo Clinic patients with an mGFR performed for clinical indications from 2011 to 2023, with serum creatinine and cystatin C measured within 7 days and an echocardiogram performed up to 1 year before the mGFR date. HF was identified by the presence of International Classification of Diseases codes within 1 year before the mGFR and subgrouped into ejection fraction (EF) ≥50% (HFEF≥50%, n=182) or <50% (HFEF<50%, n=115) and compared with no-HF controls (n=1871). CKD-EPI (and EKFC) eGFRcr, eGFRcys, and eGFRcrcys equations were calculated and compared for bias (mGFR minus eGFR) and accuracy (1-P30, proportion of people with ≥30% difference between eGFR and mGFR). CIs were generated by bootstrapping. RESULTS: The HF groups were characterized by older age, higher proportion of males, more diabetes, higher creatinine, and higher cystatin C than controls. In terms of bias, eGFRcr overestimated mGFR to a greater extent in both HF groups compared with controls, whereas eGFRcys and eGFRcrcys showed similar bias in both HF groups and controls. In the HF groups, cystatin C-based equations were more accurate than eGFRcr, particularly within HFEF<50% (1-P30 of 28% and 34% for CKD-EPI eGFRcys and eGFRcrcys, respectively, versus 60% for eGFRcr), whereas eGFRcrcys was more accurate in controls. The CKD-EPI and EKFC equations were overall convergent, showing similar results. CONCLUSIONS: Among patients with HF, eGFRcr demonstrates inferior performance (more bias and less accuracy) compared with cystatin C-based eGFRs, with this effect being more pronounced in those with HFEF<50%.

Key Points Automated cyst segmentation in autosomal dominant polycystic kidney disease could transform care by enabling precise monitoring of cyst progression with minimal manual effort. This technology can facilitate more consistent assessments of cyst burden across patients. Translating these segmentation models from magnetic resonance imaging to computed tomography is crucial for providing comprehensive, modality-independent evaluations. Background Total kidney and liver volumes are key image-based biomarkers to predict the severity of kidney and liver phenotype in autosomal dominant polycystic kidney disease (ADPKD). However, magnetic resonance (MR) imaging-based advanced biomarkers like total cyst number and cyst parenchyma surface area have been shown to more accurately assess cyst burden and improve the prediction of disease progression. The main aim of this study was to extend the calculation of advanced biomarkers to other imaging modalities; thus, we propose a fully automated model to segment kidney and liver cysts in computed tomography (CT) images. Methods Abdominal CTs of patients with ADPKD were gathered retrospectively between 2001 and 2018. A three-dimensional deep learning method using the nnU-Net architecture was trained to learn cyst edges-cores and the noncystic kidney/liver parenchyma. Separate segmentation models were trained for kidney cysts in contrast-enhanced CTs and liver cysts in noncontrast CTs using an active learning approach. Two experienced research fellows manually generated the reference standard segmentation, which were reviewed by an expert radiologist for accuracy. Results Two-hundred CT scans from 148 patients (mean age, 51.2±14.1 years; 48% male) were used for model training (80%) and testing (20%). In the test set, both models showed good agreement with the reference standard segmentations, similar to the agreement between two independent human readers (model versus reader: TCN kidney/liver r =0.96/0.97 and CPSA kidney r =0.98, inter-reader: TCN kidney/liver r =0.96/0.98 and CPSA kidney r =0.99). Conclusions Our study demonstrates that automated models can segment kidney and liver cysts accurately in CT scans of patients with ADPKD.

Abstract The aim of this international meta-analysis was to quantify the predictive value of BMI for incident fracture and relationship of this risk with age, sex, follow-up time, and BMD. A total of 1 667 922 men and women from 32 countries (63 cohorts), followed for a total of 16.0 million person-years were studied. 293 325 had FN BMD measured (2.2 million person-years follow-up). An extended Poisson model in each cohort was used to investigate relationships between WHO-defined BMI categories (Underweight: &lt;18.5 kg/m2; Normal: 18.5-24.9 kg/m2; Overweight: 25.0-29.9 kg/m2; Obese I: 30.0-34.9 kg/m2; Obese II: ≥35.0 kg/m2) and risk of incident osteoporotic, major osteoporotic and hip fracture (HF). Inverse-variance weighted β-coefficients were used to merge the cohort-specific results. For the subset with BMD available, in models adjusted for age and follow-up time, the hazard ratio (95% CI) for HF comparing underweight with normal weight was 2.35 (2.10-2.60) in women and for men was 2.45 (1.90-3.17). Hip fracture risk was lower in overweight and obese categories compared to normal weight [obese II vs normal: women 0.66 (0.55-0.80); men 0.91 (0.66-1.26)]. Further adjustment for FN BMD T-score attenuated the increased risk associated with underweight [underweight vs normal: women 1.69 (1.47-1.96); men 1.46 (1.00-2.13)]. In these models, the protective effects of overweight and obesity were attenuated, and in both sexes, the direction of association reversed to higher fracture risk in Obese II category [Obese II vs Normal: women 1.24 (0.97-1.58); men 1.70 (1.06-2.75)]. Results were similar for other fracture outcomes. Underweight is a risk factor for fracture in both men and women regardless of adjustment for BMD. However, while overweight/obesity appeared protective in base models, they became risk factors after additional adjustment for FN BMD, particularly in the Obese II category. This effect in the highest BMI categories was of greater magnitude in men than women. These results will inform the second iteration of FRAX®.

Abstract Sex differences in cancer outcome, including that of glioblastoma (GBM), are shaped by biological, hormonal, and immunological factors and impact disease progression, treatment responses, survival, and tumor microenvironment (TME) interactions. Platelets regulate the immune responses and tumor progression in many cancers, but it is not clear how they contribute to these sex-based differences by affecting the dynamics of the TME. Here, we show that GBM patients exhibit heightened platelet reactivity driven by PAR4 signaling. In murine GBM models, both pharmacological inhibition of PAR4 using BMS986120 and genetic deletion of PAR4 significantly prolong survival in females but not males. This survival advantage is estrogen dependent: it is preserved in chromosomal male-hormonal female mice within the four-core genotype model and is rescued in ovariectomized mice treated with estrogen. The survival benefit is TME specific and is mediated by platelet-driven enhancement of CD8⁺ T cell infiltration into the tumor. Inhibition of platelet PAR4 signaling increases calcium signaling through an estrogen-dependent interaction between PAR4 and estrogen receptor β (ERβ)—a receptor interaction not previously described. PAR4-activated platelets within the TME suppress CD8⁺ T cell function, and depletion of CD8⁺ T cells abolishes both the tumor-induced platelet reactivity and the survival benefit conferred by PAR4 inhibition. These findings establish platelet-mediated PAR4 signaling as a critical driver of tumor progression and identify sex-specific immune responses as key to therapeutic efficacy.

Abstract Objectives Large language models (LLMs) offer potential in natural language processing tasks in healthcare. Due to the need for high accuracy, understanding their limitations is essential. The purpose of this study was to evaluate the performance of LLMs in classifying radiology reports for the presence of pulmonary embolism (PE) under various conditions, including different prompt designs and data perturbations. Materials and Methods In this retrospective, institutional review board approved study, we evaluated 3 Google’s LLMs including Gemini-1.5-Pro, Gemini-1.5-Flash-001, and Gemini-1.5-Flash-002, in classifying 11 999 pulmonary CT angiography radiology reports for PE. Ground truth labels were determined by concordance between a computer vision-based PE detection (CVPED) algorithm and multiple LLM runs under various configurations. Discrepancies between algorithms’ classifications were aggregated and manually reviewed. We evaluated the effects of prompt design, data perturbations, and repeated analyses across geographic cloud regions. Performance metrics were calculated. Results Of 11 999 reports, 1296 (10.8%) were PE-positive. Accuracy across LLMs ranged between 0.953 and 0.996. The highest recall rate for a prompt modified after a review of the misclassified cases (up to 0.997). Few-shot prompting improved recall (up to 0.99), while chain-of-thought generally degraded performance. Gemini-1.5-Flash-002 demonstrated the highest robustness against data perturbations. Geographic cloud region variability was minimal for Gemini-1.5+-Pro, while the Flash models showed stable performance. Discussion and Conclusion LLMs demonstrated high performance in classifying radiology reports, though results varied with prompt design and data quality. These findings underscore the need for systematic evaluation and validation of LLMs for clinical applications, particularly in high-stakes scenarios.

Abstract Migraine and functional neurological disorder (FND) are two of the most common conditions in neurological practice. It is assumed that the two conditions have distinct underlying mechanisms. However, it can be clinically challenging to disentangle their relative contributions to a patient's symptoms. In addition, apart from the relationship between persistent postural perceptual dizziness (PPPD) and migraine, the frequency of co-occurrence has not been characterized in detail. Contemporary conceptualizations of FND have driven a re-evaluation of its relationship to other neurological disorders, including migraine. We carried out a narrative review of the literature examining the co-occurrence of migraine and FND. We also explored their comorbidities, aetiological risk factors and mechanisms, focusing especially on areas of potential overlap. Our review suggests increased frequency of migraine in people with functional seizures compared to epilepsy, but data from people with functional motor symptoms is mixed. Robust epidemiological studies evaluating the frequency of FND in migraine are lacking. Similar to other neurological disorders, migraine is an established trigger of FND. Female gender, adverse childhood experiences and comorbid psychiatric and functional disorders, such as irritable bowel syndrome and fibromyalgia, are more common in both conditions than in controls, but perhaps more so in FND. Mechanistic research in both conditions highlights converging frameworks of dysregulated allostatic/stress responses in the context of predictive processing models of the brain. This has implications for pharmaceutical and rehabilitation treatments. The relationship between migraine and FND is poorly studied. An overview of their overlap offers a model of non-dualistic thinking within a clinical neuroscience framework for future studies.

ABSTRACT Background Blinding mitigates bias in randomized trials and may be especially crucial for surrogate endpoints, such as progression‐free survival (PFS). Here, we characterize utilization of and factors associated with blinding in Phase III oncology trials with PFS primary endpoints. Methods Two‐arm, superiority‐design trials investigating systemic therapy were identified in May 2024 from ClinicalTrials.gov with no date limitation. Trials were required to have a PFS primary endpoint. The study outcomes were the presence of double‐blind designs and blinded independent central review (BICR) for the primary endpoint. Ninety‐five percent credible intervals for binary outcomes were estimated from beta distributions, and multivariable logistic regressions explored associations with trial‐level features. Results After screening, 237 trials were included, enrolling 127,518 patients. A double‐blind design was used in 105 trials (44%, 95% CrI 38%–51%). BICR was used in 111 trials (47%, 95% CrI 41%–53%), including 39 double‐blind trials (16%, 95% CrI 12%–22%). Trials with BICR had higher odds of meeting the primary endpoint (OR 1.84; 95% CI 1.06–3.18; p = 0.03). The PFS assessor identity (central vs. local) or blinding status was not reported in 50 trials (26%, 95% CrI 16%–27%). Trials that met prespecified significance criteria for PFS were more likely to report whether PFS assessors were blinded/unblinded and central/local (OR, 3.05; 95% Cl: 1.60–5.81; p = 0.0007). Conclusions Despite the importance of double blinding in combination with BICR for reducing bias, only a few trials blinded physicians, patients, and primary endpoint assessors. This meta‐epidemiological study illuminates the prevalence of potential assessment biases affecting PFS in Phase III oncology and secondarily emphasizes the need for improved methodology reporting.

Abstract Background Clostridioides difficile infection (CDI) treatment guidelines suggest varied approaches for patients with first (frCDI) or multiply recurrent CDI (mrCDI). Low microbial diversity, elevated primary bile acids (BA), and low secondary BA concentrations favor germination of C. difficile spores into toxin-producing bacteria and are believed to increase rCDI risk. Greater understanding of the gastrointestinal (GI) microbiome in rCDI may inform management of the disease. We describe a post hoc comparison of GI microbiome and bile acid profiles between patients with frCDI and mrCDI in a Phase 3 open-label trial, ECOSPOR IV, of fecal microbiota spores, live-brpk (VOWST®; VOS, formerly SER-109), an orally-administered live microbiome therapeutic. Methods Patients received VOS following symptom resolution after standard-of-care antibiotics. Pretreatment baseline (within 3 days following antibiotic completion) and Week 1 post-dosing stool samples were collected for whole metagenomic sequencing and metabolomics. Diversity was calculated from MetaPhlAn2 species profiles. Concentrations of primary and secondary BAs were measured via targeted LC-MS/MS. Results rCDI rates through Week 8 were similarly low in both frCDI and mrCDI patients (6.5% versus 9.7%, respectively). Baseline microbial diversity was similarly low between frCDI and mrCDI subgroups (P &gt; .05). Diversity and secondary BA concentrations increased in both subgroups, whereas primary BA concentrations declined following VOS dosing, leading to few differences between subgroups at Week 1. Conclusions These data suggest commonalities in microbiome disruption in patients with frCDI and mrCDI that contribute to recurrence and suggest that antibiotics followed by a live microbiome therapy may be an optimal treatment strategy for rCDI, regardless of number of prior CDI recurrences.