Citations

Pityriasis rubra pilaris (PRP) is a rare inflammatory skin disease with a poorly understood pathogenesis. Through a molecularly driven precision medicine approach and an extensive mechanistic pathway analysis in PRP skin samples, compared to psoriasis, atopic dermatitis, healed PRP, and healthy controls, we identified IL-1β as a key mediator, orchestrating an NF-κB–mediated IL-1β–CCL20 axis, including activation of CARD14 and NOD2. Treatment of three patients with the IL-1 antagonists anakinra and canakinumab resulted in rapid clinical improvement and reversal of the PRP-associated molecular signature with a 50% improvement in skin lesions after 2 to 3 weeks. This transcriptional signature was consistent with in vitro stimulation of keratinocytes with IL-1β. With the central role of IL-1β underscoring its potential as a therapeutic target, our findings propose a redefinition of PRP as an autoinflammatory keratinization disorder. Further clinical trials are needed to validate the efficacy of IL-1β antagonists in PRP.

AbstractBackgroundThe potential association of long‐term oral corticosteroid (OCS) use with adverse events (AEs) in patients with bullous pemphigoid (BP) is not well characterized in a real‐world setting.ObjectivesTo evaluate the effect of OCS use and treatment duration on the incident AEs in patients with BP.MethodsThis retrospective cohort study used medical and pharmacy claims and patient enrollment data from IQVIA PharMetrics® Plus from 2006−2021. Eligible patients had a diagnosis of BP between 1 January 2006 and 30 June 2020 (≥2 claims for BP ≥ 30 days apart). Patients in the OCS cohort also had a claim for OCS use, ≥7.5 mg daily dose of prednisone or equivalent, and no claims for a nonoral systemic corticosteroid (CS) at any time during the study period. A control cohort of patients with BP not using OCS was also selected. Relative risk ratios were estimated between the incidence of AEs in OCS users with different exposure durations (short‐term, <30 days; medium‐term, 30−90 days; long‐term, >90 days) and nonusers by Poisson regression, after adjusting for age, sex, prior drug use, and baseline Charlson Comorbidity Index score within the follow‐up period.ResultsAt the 1‐year follow‐up, long‐term OCS users had a higher incidence of infections, cataract, osteoporosis, heart failure, depression or anxiety, and diabetes compared with OCS nonusers (p < 0.05). Furthermore, compared with nonusers, long‐term users had increased 1‐year risks (risk ratio; 95% confidence interval) for heart failure (2.27; 1.50−3.44), diabetes (2.05; 1.30−3.24), osteoporosis (1.72; 1.23−2.41), and infection (1.50; 1.13−1.99). Long‐term OCS users also had increased 1‐year risks for heart failure (2.00; 1.21−3.28) and osteoporosis (1.70; 1.16−2.50) compared with short‐term OCS users.ConclusionsLong‐term OCS use (≥7.5 mg) in patients with BP was associated with an increased risk of AEs. Study findings demonstrate a need for steroid‐sparing options with an improved safety profile.

Abstract Aims Mobile devices such as smartphones and watches can now record single-lead electrocardiograms (ECGs), making wearables a potential screening tool for cardiac and wellness monitoring outside of healthcare settings. Because friends and family often share their smart phones and devices, confirmation that a sample is from a given patient is important before it is added to the electronic health record. Methods and results We sought to determine whether the application of Siamese neural network would permit the diagnostic ECG sample to serve as both a medical test and biometric identifier. When using similarity scores to discriminate whether a pair of ECGs came from the same patient or different patients, inputs of single-lead and 12-lead medians produced an area under the curve of 0.94 and 0.97, respectively. Conclusion The similar performance of the single-lead and 12-lead configurations underscores the potential use of mobile devices to monitor cardiac health.

AbstractBackgroundUniversal skin cancer screening for the Latino population is impractical.ObjectivesWe focused on melanoma education by conducting an online education campaign and collected baseline, postintervention, and 3‐month follow‐up surveys to assess sun‐protective behaviors.MethodsParticipants were recruited through (1) social media ads and (2) emails sent to Mayo Clinic Latino patients.ResultsTotal social media impressions were 508,442 and 413,007 in Spanish and English, respectively. There were 18,790 and 11,731 clicks for the Spanish and English ads, respectively, resulting in 175 baseline survey responses. From email distribution, 4151 emails translated to 219 baseline participants. Of the 394 patients who participated in the baseline survey, 89 completed the postintervention survey, and 57 completed the 3‐month follow‐up survey (28 from social media and 29 Mayo Clinic patients). Social media provides a unique opportunity for the medical community to promote melanoma education, distribute content quickly, and provide medical education in disparate healthcare areas. Our Spanish ad click‐through rate was 3.70% compared to 2.84% for English. The click‐to‐action rate was 0.57%, and the baseline‐to‐completion rate was 16.00%.ConclusionsOur study suggests social media could be a unique tool for distributing melanoma educational materials to Latino populations. The click‐through rates for this study were several‐fold higher than other studies' reported click‐through rates (0.4%–0.93%). It is important to develop melanoma educational materials that are culturally relevant and focused on Latino populations that can provide persistent improvements in mitigating melanoma risk behaviors such as sunscreen application and self‐skin examinations.

Background Delivery of dermatologic care through telemedicine was accelerated by the COVID-19 pandemic. We sought to analyze the teledermatology experience across Mayo Clinic's health care system to identify strengths and limitations of teledermatology. Methods Electronic health records of dermatology televisits were reviewed from multiple U.S. Mayo Clinic sites from January 2020 through January 2021. Results A total of 13,181 dermatology televisits were conducted in 6468 unique patients. Patients were primarily female (60.2%), and mean age of all patients was 34.1 years. Synchronous / live video conferencing visits were the most common (40.0%) telecare modality. Synchronous / live audio conferencing and asynchronous / store-and-forward visits comprised 33.0% and 27.0% of appointments. In total, 3944 televisits (29.9%) were successfully concluded via a single appointment. An in-person appointment was needed for 1693 patients (26.2%) after their initial televisit. For patients with a single televisit, synchronous / live video conferencing was the most common virtual modality (58.0% vs 32.2% of patients with multiple visits, p < 0.001). Patients needing in-person follow-up visits were slightly older than those who did not (mean [SD], 38.8 [22.3] vs 35.0 [23.6] years; p < 0.001) but without any sex-based difference. Around one-third of patients needed an in-person follow-up visit after their initial asynchronous / store-and-forward visit which was higher when compared with synchronous / live audio and video conferencing. Conclusion Single dermatology televisits effectively managed nearly one-third of patients who did not require in-person follow-up. An initial synchronous / live video conferencing was more likely to yield a single clinical encounter, whereas asynchronous / store-and-forward visits required more in-person follow-up. Future studies are required that focus on dermatology-specific cost, diagnoses, access, quality of care, and outcomes.

Introduction: Patients diagnosed with ischemic stroke routinely undergo prolonged cardiac telemetry (PCT) to evaluate for occult atrial fibrillation (AF), but this approach is limited by high cost and low yield. Clinical scores including CHA2DS2-VASc and CHARGE-AF can risk stratify patients but lack desired accuracy. Therefore, we aimed to evaluate an artificial intelligence (AI)-enabled ECG algorithm to predict AF detected by PCT after index stroke. Methods: This retrospective study included all adult patients with ischemic stroke evaluated at Mayo Clinic with baseline electrocardiogram (ECG) and PCT between 2018-2020. We recorded demographics, CHA2DS2-VASc score, CHARGE-AF score, stroke etiology, duration of PCT, and if PCT detected AF. ECGs were analyzed using the AI-ECG algorithm to determine AF likelihood. Stroke etiology was adjudicated using TOAST and embolic stroke of undetermined source (ESUS) definitions. The ability of the AI-ECG algorithm, CHA2DS2-VASc, and CHARGE-AF to predict AF detected by PCT was assessed via receiver operating characteristics analysis, calculating the area under curve (AUC) (c-statistic). Sensitivity and specificity analyses were performed for each tool using optimal cutoffs (using maximum Youden indices). Results: 863 patients were included; median age was 69 years, 496 (57.5%) were male, and 561 (65%) patients were categorized as ESUS. PCT detected AF in 85 (9.8%) patients. Median duration of cardiac telemetry was 30 (IQR 25-30) days. Sensitivity and specificity analyses are outlined in Table 1. The AI-ECG algorithm outperformed the CHA2DS2-VASc score (AUC 0.690 vs. 0.597, p=0.013), but was similar to the CHARGE-AF score (AUC 0.690 vs 0.651, p=0.233). Conclusions: Our AI-ECG algorithm demonstrated improved ability to predict AF detection on PCT after index stroke when compared to the CHA2DS2-VASc score, but was similar in performance to the CHARGE-AF score. All three methods had suboptimal accuracy.

AbstractBackgroundImmunotherapies have changed the landscape of late‐stage melanoma; however, data evaluating timely access to immunotherapy are lacking.MethodsA retrospective cohort study utilizing the National Cancer Database was conducted. Stage III and IV melanoma cases diagnosed between 2011 and 2018 that received systemic treatment with either immunotherapy or chemotherapy were included. Chemotherapy included BRAF/MEK inhibitors. Multivariable logistic regression models were utilized to evaluate factors associated with the likelihood of receiving immunotherapy as primary systemic treatment relative to chemotherapy; additionally, Cox proportional hazards models were utilized to incorporate time from diagnosis to primary systemic therapy into the analysis.ResultsThe study population was comprised of 14,446 cases. The cohort included 12,053 (83.4%) immunotherapy and 2393 (16.6%) chemotherapy cases. In multivariable logistic regression analysis, factors significantly associated with immunotherapy receipt included population density, circle distance, year of diagnosis, Breslow thickness, and cancer stage. Immunotherapy timing was evaluated using multivariable Cox regression analysis. Minorities were less likely to receive timely immunotherapy than non‐Hispanic Whites (HR 0.83, CI 0.74–0.93, p = 0.001). Patients at circle distances of 10–49 miles (HR 0.94, CI 0.89–0.99, p = 0.02) and ≥50 miles (HR 0.83, CI 0.77–0.90, p < 0.001) were less likely to receive timely immunotherapy.ConclusionPatients traveling ≥10 miles and minorities have a decreased likelihood of receiving timely immunotherapy administration for primary systemic treatment. Future research is needed to identify what barriers and approaches can be leveraged to address these inequities.

Background: Our team has previously developed convolutional neural networks (CNNs) to estimate age and sex from a 10-second, 12-lead ECG as indicators of patient wellness. Here, we develop an additional wellness network to estimate body mass index (BMI) from the ECG input signal. Aims: To evaluate the performance of neural networks trained to classify and directly estimate BMI. Methods: We identified all adult patients in the Mayo Clinic system who had a 10-second, 12-lead ECG within 15 days of a BMI measurement. Of the&nbsp;174,863 patients in the cohort, 2,895 were underweight (BMI < 18.5), 42,387 were of normal weight (BMI 18.5-24.9), and 57,939 were overweight only (BMI 25.0-29.9). Among patients categorized as obese, 39,788 had class I obesity (BMI 30.0-34.9), 18,978 had class II obesity (BMI 35.0-39.9), and 12,876 had class III obesity (BMI ≥ 40.0). The cohort was split into training, validation, and testing datasets in an 8:1:1 ratio to develop networks for BMI classification and a direct&nbsp;estimation of BMI. Results: In detecting overweight in the testing set, the model achieved an AUC of 0.86 (95% CI 0.85, 0.87). In detecting class I, class II, and class III obesity, the model yielded AUCs of 0.84 (95% CI 0.83, 0.85), 0.84 (95% CI 0.83, 0.85), and 0.85 (95% CI 0.85, 0.86), respectively. The model to estimate BMI achieved a mean absolute error of 3.70. Conclusions: AI-enabled ECGs demonstrate efficient detection of overweight patients and all classes of obesity. Further research is needed to determine if discrepancies between the traditionally measured BMI and the AI-ECG BMI reflect the presence or lack of metabolic abnormalities.