Citations

Abstract Objective The goal is to provide an overview of artificial intelligence (AI) and machine learning (ML) methodology and appraisal tailored to clinicians and researchers in the headache field to facilitate interdisciplinary communications and research. Background The application of AI to the study of headache and other healthcare challenges is growing rapidly. It is critical that these findings be accurately interpreted by headache specialists, but this can be difficult for non‐AI specialists. Methods This paper is a narrative review of the fundamentals required to understand ML/AI headache research. Using guidance from key leaders in the field of headache medicine and AI, important references were reviewed and cited to provide a comprehensive overview of the terminology, methodology, applications, pitfalls, and bias of AI. Results We review how AI models are created, common model types, methods for evaluation, and examples of their application to headache medicine. We also highlight potential pitfalls relevant when consuming AI research, and discuss ethical issues of bias, privacy and abuse generated by AI. Additionally, we highlight recent related research from across headache‐related applications. Conclusion Many promising current and future applications of ML and AI exist in the field of headache medicine. Understanding the fundamentals of AI will allow readers to understand and critically appraise AI‐related research findings in their proper context. This paper will increase the reader's comfort in consuming AI/ML‐based research and will prepare them to think critically about related research developments.

Abstract BACKGROUND Brain metastases (BMs) are a significant cause of morbidity and mortality in metastatic breast cancer (MBC), yet prognostic and predictive biomarkers remain poorly understood. STUDY DESIGN 90 consecutive patients with breast cancer BMs who underwent surgical resection from 2006-2022 were retrospectively identified. Review of the electronic medical record and institutional cancer registry were completed to identify baseline clinical parameters and outcomes. BM mutation analysis was performed using a CLIA-certified targeted DNA mutation sequencing assay as part of routine clinical care. RESULTS All patients were female, median age at first BM resection was 57.9 years (range 27.0-81.3 years), and receptor subtypes included HR+/HER2- (n=27, 30.0%), HER2+ (n=33, 36.7%), and triple negative BC (TNBC; n=30, 33.3%). At time of first BM resection, most patients had a solitary BM (n=55, 61.1%), with anatomic locations including cerebellar (n=29, 32.2%), frontal (n=26, 28.9%), and parietal (n=17, 18.9%) sites. BM DNA mutation analysis (n= 20) revealed pathogenic alterations in TP53 (n=17, 85.0%), ERBB2 amplification (n=7, 35.0%), PIK3CA (n=4, 20.0%), BRCA1 (n=3, 15.0%), and BRCA2 (n=2, 10.0%). Most patients received postoperative radiation (n=82, 90.8%) and/or systemic therapy (n=67, 82.7%). Cerebellar location was significantly associated with the risk of developing leptomeningeal disease compared to other BM resection sites (HR = 2.75, p=0.02). Among all patients, median overall survival (OS) from first BM resection was 27.8 months (range 0.1-196.6 months). HR+/HER2- (median 37 months) or HER2+ (median 36 months) MBC was associated with significantly increased OS compared to TNBC (14.4 months) (p<0.01, log-rank test). CONCLUSIONS Patients with breast cancer BMs exhibited a median OS of >2 years following BM resection, and patients with TNBC had a significantly worse prognosis. Targetable mutations were observed in 40% of resected BM specimens, supporting the utility of molecular analysis of resected BMs to help guide clinical management.

Abstract Background Most breast biopsies are diagnosed as benign breast disease, with 1.5- to 4-fold increased breast cancer risk. Apart from pathologic diagnoses of atypical hyperplasia, few factors aid in breast cancer risk assessment of these patients. We assessed whether a 313–single nucleotide variation (formerly single-nucleotide polymorphism) polygenic risk score stratifies risk of benign breast disease patients. Methods We pooled data from 5 Breast Cancer Association Consortium case-control studies (mean age = 59.9 years), including 6706 participants with breast cancer and 8488 participants without breast cancer. Using logistic regression, we estimated breast cancer risk associations by self-reported benign breast disease history and strata of polygenic risk score, with median polygenic risk score category among women without benign breast disease as the referent. We assessed interactions and mediation of benign breast disease and polygenic risk score with breast cancer risk. Results Benign breast disease history was associated with increased breast cancer risk (odds ratio [OR] = 1.48, 95% confidence interval [CI] = 1.37 to 1.60; P < .001). Polygenic risk score increased breast cancer risk, irrespective of benign breast disease history (Pinteraction = .48), with minimal evidence of mediation of either factor by the other. Women with benign breast disease and polygenic risk score in the highest tertile had more than twofold increased odds of breast cancer (OR = 2.73, 95% CI = 2.41 to 3.09), and those with benign breast disease and polygenic risk score in the lowest tertile experienced reduced breast cancer risk (OR = 0.79, 95% CI = 0.70 to 0.91) compared with the referent group. Women with benign breast disease and polygenic risk score in the highest decile had a 3.7-fold increase (95% CI = 3.00 to 4.61) compared with those with median polygenic risk score without benign breast disease. Conclusion Breast cancer risks are elevated among women with benign breast disease and increase progressively with polygenic risk score, suggesting that optimal combinations of these factors may improve risk stratification.

11526 Background: There are no FDA-approved treatments for advanced cCS. Chemotherapy provides limited benefit. IDH1/2 mutations (m) occur in 50% of cCS. Both IDHm and wild-type cCS harbor epigenetic dysregulation. In preclinical models, HDAC + DNMT inhibition (i) suppressed growth by inducing apoptosis, reversing the hypermethylated state, and upregulating expression of interferon (IFN) response genes including PDL1(1). This prompted a phase 2 study of HDACi + DNMTi in cCS, which failed to meet the 1° endpoint of ORR. The majority of pts had a best response of stable disease, with a trend towards improved mPFS in IDHm (2). Here we report correlative analysis of NCI 10330. Methods: NCI 10330 is a single-arm, multicenter, phase 2 study of belinostat with SGI-110 or ASTX727 in advanced cCS. All pts were required to have pre- and on-treatment (tx) biopsies (Cycle 2, Day 3-5). Tissue was evaluated with whole exome sequencing (WES) and RNA sequencing (RNAseq). From RNAseq, Tumor Inflammation Signature Scores (TISS) and Tumor Microenvironment functional Gene Set Enrichment Analysis scores (TME GSEA) were calculated as the mean of the log2 normalized counts of 18 signature genes (Danaher et al. J Imm Can 2018) and pre-defined gene sets (3), respectively. Wilcoxon rank sum test was used to analyze TME GSEA between pre- and on-tx samples. Differential expression was significant if |FC| > 1.33, adjusted P <0.01. Results: 19 pts were treated; all received paired biopsies. WES was adequate in 7/19 (37%) pts, with IDHm identified in 3/7 (43%) pts. All pts were MSI-stable, TMB: 1.01-3.57 mut/Mb. RNAseq was adequate in 7/19 (37%) pts at pre-tx and 5/19 (26%) pts at on-tx. TISS ranges trended higher for pre-tx (8.77-10.14) vs on-tx (7.36-9.51) samples. TME GSEA identified significantly enriched tumor immune infiltration (p < 0.05) in pre- vs on-tx samples. Differential analysis identified several gene sets related to inflammation overexpressed in pre-tx samples only, including IFN response. On-tx samples were enriched in myogenesis and coagulation gene sets. Conclusions: This is the first study to describe transcriptomic changes following epigenetic tx in cCS. Contradictory to our preclinical data, HDACi + DNTMi resulted in low expression of inflammation. Prior studies reported that the immune infiltrate of CS at progression is immunosuppressive; higher immune infiltrate is correlated with worse outcomes (4). Loss of inflammation may have implications for disease stability experienced by most pts on NCI 10330. Further analyses are planned to correlate TME (pro- vs anti-inflammatory infiltrates) and outcomes. Analyses are limited by small sample size, highlighting the challenges in collecting adequate CS specimens. 1. Sheikh, Schwartz et al. Mol Cancer Ther 2021. 2. Lacuna et al. ASCO 2023: #11531. 3. Bagaev et al. Can Cell 2021. 4. Richert et al. J Bone Onc 2020. Clinical trial information: NCT04340843 .

BACKGROUND Obesity is a multifactorial disease with adverse health implications including insulin resistance (IR). In patients with obesity, the presence of high circulating levels of leptin, deemed hyperleptinemia, is associated with IR. Recent data in mice with diet-induced obesity (DIO) show that a partial reduction in leptin levels improves IR. Additional animal studies demonstrate that IL-4 decreases leptin levels. In rodents, resident adipose tissue eosinophils (AT-EOS) are the main source of IL-4 and are instrumental in maintaining metabolic homeostasis. A marked reduction in AT-EOS content is observed in animal models of DIO. These observations have not been explored in humans. METHODS We analyzed AT from individuals with obesity and age-matched lean counterparts for AT-EOS content, IL-4, circulating leptin levels, and measures of IR. RESULTS Our results show that individuals with obesity ( n = 15) had a significant reduction in AT-EOS content ( P < 0.01), decreased AT–IL-4 gene expression ( P = 0.02), and decreased IL-4 plasma levels ( P < 0.05) in addition to expected IR ( P < 0.001) and hyperleptinemia ( P < 0.01) compared with lean subjects ( n = 15). AT-EOS content inversely correlated with BMI ( P = 0.002) and IR ( P = 0.005). Ex vivo AT explants and in vitro cell culture of primary human mature adipocytes exposed to either IL-4 or EOS conditioned media produced less leptin ( P < 0.05). CONCLUSION Our results suggest that IL-4 acts as a link between EOS, AT, and leptin production. Future studies exploring this interaction may identify an avenue for the treatment of obesity and its complications through amelioration of hyperleptinemia. TRIAL REGISTRATION Clinicaltrials.gov NCT02378077 & NCT04234295.

Abstract Background We aimed to determine if pre-existing immunocompromising conditions (ICCs) were associated with the presentation or outcome of patients with acute coronavirus disease 2019 (COVID-19) admitted for pediatric intensive care. Methods Fifty-five hospitals in 30 US states reported cases through the Overcoming COVID-19 public health surveillance registry. Patients &lt;21 years admitted 12 March 2020–30 December 2021 to the pediatric intensive care unit (PICU) or high-acuity unit for acute COVID-19 were included. Results Of 1274 patients, 105 (8.2%) had an ICC, including 33 (31.4%) hematologic malignancies, 24 (22.9%) primary immunodeficiencies and disorders of hematopoietic cells, 19 (18.1%) nonmalignant organ failure with solid-organ transplantation, 16 (15.2%) solid tumors, and 13 (12.4%) autoimmune disorders. Patients with ICCs were older, had more underlying renal conditions, and had lower white blood cell and platelet counts than those without ICCs, but had similar clinical disease severity upon admission. In-hospital mortality from COVID-19 was higher (11.4% vs 4.6%, P = .005) and hospitalization was longer (P = .01) in patients with ICCs. New major morbidities upon discharge were not different between those with and without ICC (10.5% vs 13.9%, P = .40). In patients with ICCs, bacterial coinfection was more common in those with life-threatening COVID-19. Conclusions In this national case series of patients &lt;21 years of age with acute COVID-19 admitted for intensive care, existence of a prior ICCs were associated with worse clinical outcomes. Reassuringly, most patients with ICCs hospitalized in the PICU for severe acute COVID-19 survived and were discharged home without new severe morbidities.

Abstract Opportunities for genetic counselors to work in a variety of practice settings have greatly expanded, particularly in the laboratory. This study aimed to assess attitudes of genetic counselors working both within and outside of the laboratory setting regarding (1) the re‐wording and/or expansion of key measures of genetic counselors' competency, including practice‐based competencies (PBCs) and board examination, to include laboratory roles, (2) preparation and transferability of competencies developed in master's in genetic counseling (MGC) programs to different roles, (3) need of additional training for genetic counselors to practice in laboratory settings, and (4) preferred methods to obtain that training. An e‐blast was sent to ABGC diplomats ( N = 5458) with a link to a 29‐item survey with 12 demographic questions to compare respondents to 2021 NSGC Professional Status Survey (PSS) respondents. Statistical comparisons were made between respondents working in the laboratory versus other settings. Among 399 responses received, there was an oversampling of respondents working in the laboratory (52% vs. 20% in PSS) and in non‐direct patient care positions (47% vs. 25% in PSS). Most respondents agreed the PBCs were transferable to their work yet favored making the PBCs less direct patient care‐focused, expanding PBCs to align with laboratory roles, adding laboratory‐focused questions to the ABGC exam, and adding laboratory‐focused training in MGC programs. Most agreed requiring post‐MGC training would limit genetic counselors' ability to change jobs. Genetic counselors working in the laboratory reported being significantly less prepared by their MGC program for some roles ( p < 0.001) or how the PBCs applied to non‐direct patient care positions ( p < 0.001). Only 53% of all respondents agreed that NSGC supports their professional needs and others in their practice area, and genetic counselors working in the laboratory were significantly less likely to agree ( p = 0.002). These sentiments should be further explored.

Abstract Background Given the elevated prevalence and impact of overweight and the potential risk of non-communicable diseases among women of Somali background in high-income countries and recognising the potential positive impact of physical activity (PA) on these health conditions, it becomes imperative to focus on understanding the PA behaviour of this specific population. The objectives of this paper were twofold: firstly, to provide a comprehensive description of both objectively and subjectively measured PA level and sedentary time in a group of overweight women of Somali background in Norway, and secondly, to assess the effectiveness of a tailored, culturally sensitive, and community-based intervention in enhancing PA levels and reducing sedentary time. Methods 169 overweight women of Somali background in Norway were randomised by borough to either a lifestyle programme or a comparison group. The programme consisted of two sessions per week for 12 weeks, combining classroom discussion with graded group-based PA activity led by coaches, followed by monthly sessions over nine months. PA was measured objectively using the ActivPAL monitor and subjectively using the international PA questionnaire short form (IPAQ). Results The women took on average 6804 (SD = 3286) steps per day and were sedentary for 9.1 hours (SD = 3) per day at baseline. There were no differences between groups on any objectively measured PA variable at any timepoint. There were significant differences on vigorous intensity PA and total PA at the 12-month measurement session between the two groups using the IPAQ. Conclusion Despite observing initially low PA levels and high sedentary time at baseline and thus a considerable potential for intervention, the intervention failed in attaining an increase in objectively measured PA or reduction in sedentary time compared to the control condition. However, self-reported measures indicated success in these aspects. The potential threats to the programme's reliability and validity include contamination, low attendance rates, and high drop-out rates possible due to the COVID-19 pandemic. These challenges underscore the complexity of interventions in this demographic, emphasising the need for further exploration and refinement of study methodologies to effectively enhance PA levels and reduce sedentary time in immigrant women living in high-income countries. Trial registration : clinicaltrials.gov NCT04578067. Keywords: Immigrant Somali women, Lifestyle programme, Overweight, Physical activity, Sedentary time, ActivPAL

Background and objectives: In June 2022, the mRNA COVID-19 vaccination was recommended for young children. We examined clinical characteristics and factors associated with vaccination status among vaccine-eligible young children hospitalized for acute COVID-19. Methods: We enrolled inpatients 8 months to <5 years of age with acute community-acquired COVID-19 across 28 US pediatric hospitals from September 20, 2022 to May 31, 2023. We assessed demographic and clinical factors, including the highest level of respiratory support, and vaccination status defined as unvaccinated, incomplete, or complete primary series [at least 2 (Moderna) or 3 (Pfizer-BioNTech) mRNA vaccine doses ≥14 days before hospitalization]. Results: Among 597 children, 174 (29.1%) patients were admitted to the intensive care unit and 75 (12.6%) had a life-threatening illness, including 51 (8.5%) requiring invasive mechanical ventilation. Children with underlying respiratory and neurologic/neuromuscular conditions more frequently received higher respiratory support. Only 4.5% of children hospitalized for COVID-19 (n = 27) had completed their primary COVID-19 vaccination series and 7.0% (n = 42) of children initiated but did not complete their primary series. Among 528 unvaccinated children, nearly half (n = 251) were previously healthy, 3 of them required extracorporeal membrane oxygenation for acute COVID-19 and 1 died. Conclusions: Most young children hospitalized for acute COVID-19, including most children admitted to the intensive care unit and with life-threatening illness, had not initiated COVID-19 vaccination despite being eligible. Nearly half of these children had no underlying conditions. Of the small percentage of children who initiated a COVID-19 primary series, most had not completed it before hospitalization.