Citations

PURPOSE Patient-centered outcomes (PCOs) are pivotal in cancer treatment, as they directly reflect patients' quality of life. Although multiple studies suggest that factors affecting breast cancer–related morbidity and survival are influenced by treatment side effects and adherence to long-term treatment, such data are generally only available on a smaller scale or from a single center. The primary challenge with collecting these data is that the outcomes are captured as free text in clinical narratives written by clinicians. MATERIALS AND METHODS Given the complexity of PCO documentation in these narratives, computerized methods are necessary to unlock the wealth of information buried in unstructured text notes that often document PCOs. Inspired by the success of large language models (LLMs), we examined the adaptability of three LLMs, GPT-2, BioGPT, and PMC-LLaMA, on PCO tasks across three institutions, Mayo Clinic, Emory University Hospital, and Stanford University. We developed an open-source framework for fine-tuning LLM that can directly extract the five different categories of PCO from the clinic notes. RESULTS We found that these LLMs without fine-tuning (zero-shot) struggle with challenging PCO extraction tasks, displaying almost random performance, even with some task-specific examples (few-shot learning). The performance of our fine-tuned, task-specific models is notably superior compared with their non–fine-tuned LLM models. Moreover, the fine-tuned GPT-2 model has demonstrated a significantly better performance than the other two larger LLMs. CONCLUSION Our discovery indicates that although LLMs serve as effective general-purpose models for tasks across various domains, they require fine-tuning when applied to the clinician domain. Our proposed approach has the potential to lead more efficient, adaptable models for PCO information extraction, reducing reliance on extensive computational resources while still delivering superior performance for specific tasks.

Abstract Background African-American (AA) women are less likely to achieve ideal cardiovascular (CV) health compared with women of other racial/ethnic subgroups, primarily due to structural and psychosocial barriers. A potential psychosocial construct relevant to ideal CV health is the superwoman schema (SWS). Purpose We explored whether the SWS was associated with perceived stress, CV risk factors, and overall CV health among AA women. Methods This cross-sectional analysis of the FAITH! Heart Health+ Study was conducted among AA women with high cardiometabolic risk. Pearson correlation evaluated associations between SWS and CV risk factors (e.g., stress, hypertension, diabetes, etc.). The 35-item SWS questionnaire includes five domains. Stress was measured by the 8-item Global Perceived Stress Scale (GPSS). CV health was assessed using the American Heart Association Life’s Simple 7 (LS7) rubric of health behaviors/biometrics. Data acquisition spanned from February to August 2022. Results The 38 women included in the analysis (mean age 54.3 [SD 11.5] years) had a high CV risk factor burden (71.1% hypertension, 76.3% overweight/obesity, 28.9% diabetes, 39.5% hyperlipidemia). Mean GPSS level was 7.7 (SD 5.2), CV health score 6.7 (SD 1.8), and SWS score 60.3 (SD 18.0). Feeling an “obligation to help others” and “obligation to present an image of strength” had strongest correlations with GPSS score among all SWS domains (r = 0.51; p = .002 and r = 0.39; p = .02, respectively). Correlation among the SWS domains and traditional CV risk factors was not statistically significant. Conclusion Our findings suggest that an obligation to help others and to project an image of strength could be contributing to stress among AA women.

Abstract Context Multiple common genetic variants have been associated with type 2 diabetes, but the mechanism by which they predispose to diabetes is incompletely understood. One such example is variation in MTNR1B, which implicates melatonin and its receptor in the pathogenesis of type 2 diabetes. Objective To characterize the effect of diabetes-associated genetic variation at rs10830963 in the MTNR1B locus on islet function in people without type 2 diabetes. Design The association of genetic variation at rs10830963 with glucose, insulin, C-peptide, glucagon, and indices of insulin secretion and action were tested in a cohort of 294 individuals who had previously undergone an oral glucose tolerance test (OGTT). Insulin sensitivity, β-cell responsivity to glucose, and Disposition Indices were measured using the oral minimal model. Setting The Clinical Research and Translation Unit at Mayo Clinic, Rochester, MN. Participants Two cohorts were utilized for this analysis: 1 cohort was recruited on the basis of prior participation in a population-based study in Olmsted County. The other cohort was recruited on the basis of TCF7L2 genotype at rs7903146 from the Mayo Biobank. Intervention Two-hour, 7-sample OGTT. Main Outcome Measures Fasting, nadir, and integrated glucagon concentrations. Results One or 2 copies of the G-allele at rs10830963 were associated with increased postchallenge glucose and glucagon concentrations compared to subjects with the CC genotype. Conclusion The effects of rs10830963 on glucose homeostasis and predisposition to type 2 diabetes are likely to be partially mediated through changes in α-cell function.

AbstractA 44‐year‐old male presented with a 2‐month history of erythematous ulcerative papules and plaques on the scalp, face, and bilateral lower legs. He had a 5‐year history of well‐controlled HIV on antiretroviral therapy and recurrent syphilis infections. His face had violaceous plaques, while bilateral ankles and calves had ulcerative lesions with necrotic centers and purple borders. The morphologies clinically mimicked pyoderma gangrenosum on the lower extremities and cutaneous lymphoma on the face. Biopsy and reactive rapid plasma reagin confirmed a diagnosis of lues maligna, and the patient was successfully treated with penicillin G benzathine.

PURPOSE Cancer center clinical trial offices (CCTOs) support trial development, activation, conduct, regulatory adherence, data integrity, and compliance. In 2018, the Association of American Cancer Institutes (AACI) Clinical Research Innovation (CRI) Steering Committee conducted and published survey results to benchmark North American CCTOs, including trial volume, accrual, full time equivalents (FTEs), and budget. The survey was readministered in 2023 to assess contemporary CCTO performance and capacity with results presented here. METHODS The 28 question 2023 survey was sent to directors of AACI's clinical member cancer centers. Survey participation was voluntary, no compensation was provided, and data requested covered operations during 2022. Definitions were consistent with National Cancer Institute (NCI) CCTO reporting requirements and AACI staff anonymously compiled results for descriptive statistical reporting. RESULTS The survey response rate was 61% (60/99). The median annual CCTO budget was $11.5 million (M) US dollars (USD) versus $8.2M USD in 2018. These budgets support a median of 150 FTEs versus 104 previously, and a median total of 384 versus 280 interventional treatment trials and a median of 479 versus 531 interventional treatment accruals. Sources of support for CCTO annual budgets were primarily from industry revenue (45.3%) or institutional support (31.7%). Nearly 60% of centers reported activating NCI-sponsored studies within 90 days but only 9% reported meeting a 90-day activation timeline for industry sponsored studies. CONCLUSION Contemporary benchmarks for CCTO operations through this survey demonstrate larger staff sizes, larger budgets, more trials supported, but fewer patients enrolled to interventional treatment trials in comparison with 2018. These data shine a critical light on the increasing complexity of cancer clinical trials, the importance of external funding sources, and necessary operational efficiency upgrades to provide cutting-edge cancer research and care.

Abstract Background Despite advances in cardiovascular (CV) care, African American adults experience worse CV disease (CVD) outcomes than White adults, warranting the investigation of CVD prediction biomarkers. Plasma ceramides are bioactive lipids that are biomarkers of adverse cardiovascular risk for both primary and secondary prevention. However, racial/ethnic differences in ceramide levels and their link to CV health (CVH) in African American adults remain underexplored. Purpose This study aimed to: 1) assess the association between ceramide scores and CVH and atherosclerotic CVD (ASCVD) risk among African American adults and 2) compare ceramide scores between African American and White adults. Methods Health surveys and ceramides measurements were conducted for African American and White adults, both recruited from population-based observational study cohorts. Ten-year pooled cohort risk ASCVD scores were calculated based on American College of Cardiology guidelines. CVH was assessed using Life’s Essential 8 (LE8) scores, as per American Heart Association rubrics. CV risk factors in the two cohorts were compared using Chi-squared tests. Linear regression and Pearson correlation assessed the association between ceramide, LE8, and ASCVD risk scores. Linear regression, adjusted for age and sex, compared ASCVD risk and ceramide scores between the two cohorts. Results In total, 58 African American (mean age [SD] 54.5 [11.9] years, 67.9% women) and 1103 White (mean age [SD] 64.5 [11.9] years, 52.1% women) adults were included (Table 1). Compared to White adults, African American adults had a higher prevalence of overweight/obesity (92.9% vs 33.6%; p<0.001), hypertension (69.6% vs 42.5%; p<0.001), diabetes (30.4% vs 12.2%; p<0.001), and hyperlipidaemia (42.9% vs 30.0%; p=0.03). No difference in mean ASCVD risk was observed between the African American (12.8% [SD 10.4]) and White (12.6% [SD 11.1]) cohorts (p= 0.91). African American adults had lower 18:0 (0.08 [0.03] vs 0.10 [0.04] μmol/L) and 24:1 (0.91 [0.31] vs 1.17 [0.33] μmol/L) ceramides compared to White adults (p<0.001). Ceramide scores among African American adults were not associated with LE8 scores (p=0.077) or ASCVD risk scores (r=-0.12, p=0.428). Adjusting for age and sex, African American adults had lower ceramide scores compared to White adults (mean [SD] 1.90 [1.97] vs 3.48 [2.88], p=0.006). Conclusions In our study comparing African American and White adults in the community, the association observed between ceramide scores and CVH or ASCVD risk scores in the African American cohort did not reach statistical significance. Interestingly, despite a more suboptimal cardiometabolic profile, African American adults had lower ceramide scores compared to White adults. Our findings suggest that ceramide profiles may not accurately reflect cardiovascular risk in African American adults, warranting further exploration of alternative sphingolipid biomarkers of risk for this population.

The T allele at rs7903146 in TCF7L2 increases the rate of conversion from prediabetes to type 2 diabetes. This has been associated with impaired β-cell function and with defective suppression of α-cell secretion by glucose. However, the temporal relationship of these abnormalities is uncertain. To study the longitudinal changes in islet function, we recruited 128 subjects, with 67 homozygous for the diabetes-associated allele (TT) at rs7903146 and 61 homozygous for the protective allele. Subjects were studied on two occasions, 3 years apart, using an oral 75-g glucose challenge. The oral minimal model was used to quantitate β-cell function; the glucagon secretion rate was estimated from deconvolution of glucagon concentrations. Glucose tolerance worsened in subjects with the TT genotype. This was accompanied by impaired postchallenge glucagon suppression but appropriate β-cell responsivity to rising glucose concentrations. These data suggest that α-cell abnormalities associated with the TT genotype (rs7903146) occur early and may precede β-cell dysfunction in people as they develop glucose intolerance and type 2 diabetes. Article Highlights

Background: Physical inactivity and poor adherence to healthy eating guidelines impact cardiovascular health (CVH) outcomes in African American (AAs). Our randomized controlled trial (RCT) testing the FAITH! App, an mHealth lifestyle intervention, showed efficacy in promoting ideal CVH, particularly diet and physical activity (PA) among AAs, warranting exploration of underlying factors. Psychosocial factors (e.g., self-regulation, self-efficacy, social support) are key to healthy behavior adoption. Hypothesis: The intervention group would have greater improvements in diet and PA-related psychosocial factors than the control group after using the FAITH! App. Methods: A 2-arm cluster RCT was conducted among AAs from 16 churches in Minnesota comparing the FAITH! App intervention to a delayed intervention control group. The FAITH! App included culturally relevant, CVH-focused education modules, diet/PA self-monitoring and a sharing board. The outcomes were change in diet and PA self-regulation and self-efficacy (primary) and social support and barriers (secondary) at 6 months post-intervention. The intervention effect was assessed using linear regression. Results: Sixty-eight participants were included (71% female; mean age [SD]: 54.7 [12.6] years) ( Table 1 ). The intervention group showed a significantly greater increase in diet self-regulation [planning/tracking nutrition] (+0.66, p<.03) and PA self-regulation [step count strategies] (+0.47, p=.003) than the control group. Within groups, the intervention group demonstrated significant improvements across diet and PA self-regulation measures, while the control group showed an improvement only in fruit/vegetable and grain intake. No effect was observed in self-efficacy, social support, or barriers. Conclusions: Our culturally tailored mHealth lifestyle intervention was associated with notable increases in diet and PA self-regulation, suggesting a potential mechanism by which the FAITH! App improves healthy eating and PA among AAs.

Portal hypertension (PHTN) is a severe complication of liver cirrhosis and is associated with intrahepatic sinusoidal remodeling induced by sinusoidal resistance and angiogenesis. Collagen type IV (COL4), a major component of basement membrane, forms in liver sinusoids upon chronic liver injury. However, the role, cellular source, and expression regulation of COL4 in liver diseases are unknown. Here, we examined how COL4 is produced and how it regulates sinusoidal remodeling in fibrosis and PHTN. Human cirrhotic liver sample RNA sequencing showed increased COL4 expression, which was further verified via immunofluorescence staining. Single-cell RNA sequencing identified liver sinusoidal endothelial cells (LSECs) as the predominant source of COL4 upregulation in mouse fibrotic liver. In addition, COL4 was upregulated in a TNF-α/NF-κB–dependent manner through an epigenetic mechanism in LSECs in vitro. Indeed, by utilizing a CRISPRi-dCas9-KRAB epigenome-editing approach, epigenetic repression of the enhancer-promoter interaction showed silencing of COL4 gene expression. LSEC-specific COL4 gene mutation or repression in vivo abrogated sinusoidal resistance and angiogenesis, which thereby alleviated sinusoidal remodeling and PHTN. Our findings reveal that LSECs promote sinusoidal remodeling and PHTN during liver fibrosis through COL4 deposition.