Citations

ABSTRACT Clinical worsening events are increasingly recognized as a meaningful outcome in pulmonary arterial hypertension (PAH). We applied machine‐learning models to real‐world data to identify clinical features that may predict clinical worsening events in PAH. Data were obtained retrospectively from the electronic health records of adults diagnosed with PAH at Mayo Clinic locations (January 2015–December 2019). Machine‐learning LASSO regularized logistic regression models were developed to analyze the association between 100 clinical features and occurrence of clinical worsening events. In total, 455 patients were included (mean age 62.1 years; 59.3% female). Of these, 232 (51.0%) experienced a clinical worsening event after a median (quartile 1, quartile 3) of 10.9 (4.1, 21.4) months. The best‐performing model had an area under the curve of 0.78, sensitivity of 78%, and specificity of 59%. The model identified 11 clinical features associated with clinical worsening events in PAH with a non‐zero coefficient: baseline age, body mass index (BMI), creatinine level, red cell (erythrocyte) distribution width (RDW), mean pulmonary arterial pressure, pulmonary vascular resistance, sodium concentration, and QT interval, along with dyspnea, number of clinic visits, and other forms of heart disease within a 1‐month window before a clinical worsening event. Baseline RDW and BMI, and number of clinic visits within the 1‐month window were statistically significantly associated with clinical worsening events. Our study used machine‐learning models to identify clinical features associated with risk of clinical worsening events in PAH. Automation of risk prediction could lead to earlier therapeutic intervention to optimize patient outcomes.

ABSTRACT Pulmonary arterial hypertension (PAH) is a progressive disease with significant morbidity and mortality. Due to nonspecific symptoms, diagnosis can be challenging and subject to substantial delays. Using data from Mayo Clinic′s electronic health records, we looked at causes of delayed diagnosis and whether earlier diagnosis means better outcomes. This retrospective cohort study included adults with PAH confirmed by right heart catheterization (RHC) between 2015 and 2019. Univariate and multivariate analyses evaluated the association of 229 clinical and laboratory features with time to PAH diagnosis and survival outcomes. Early diagnosis was defined as < 6 months and a delayed diagnosis as ≥ 22.2 months from an eligible event. Survival probability was determined using Kaplan‐Meier analysis . The study enrolled 160 patients. The most common initial PAH symptoms were dyspnea (42.5%) and fatigue (16.9%). It took a median of 7, 13, 26, and 123 days from eligible event to the first X‐ray, electrocardiogram, echocardiogram, and RHC, respectively. Factors most closely associated with delayed diagnosis were chronic obstructive pulmonary disease, normal creatinine, high systolic blood pressure, acute respiratory infection, normal iron, normal diastolic blood pressure, and being male aged 18.5 to < 25 years. Compared with delayed diagnosis patients (22.6%), more early diagnosis patients saw a pulmonologist (35.3%) before another provider. After 5 years, 73% of patients in the early diagnosis and 31% in the delayed diagnosis group were still alive. This study identified factors associated with a delayed diagnosis of PAH and found that earlier diagnosis and treatment initiation were associated with significantly improved survival.

Background: Pulmonary hypertension (PH) is a life-threatening disease affecting up to 1% of the global population. Diagnosis can be challenging and is often delayed due to the need for advanced imaging and invasive procedures. The use of artificial intelligence applied to ECGs (ECG-AI) has been shown to detect subtle patterns in voltage-time data and may be a valuable tool for the early detection of PH. Hypothesis and Purpose: To evaluate the performance of a previously trained, ECG-AI algorithm to detect PH (ECG-AI PH) using real-world data (RWD) collected in a multicenter, validation study. A joint primary hypothesis required sensitivity (Sn), specificity (Sp), positive predictive value (PPV), and negative predictive value (NPV) to exceed the null values of 76%, 70%, 10% and 90%. Study Design and Methods: This retrospective validation study was conducted at 5 geographically diverse U.S. health systems. Adult subjects were eligible for inclusion if they had a 12-lead ECG paired with an echocardiogram (Echo) in which tricuspid regurgitation velocity (TRV) was recorded, following presentation with dyspnea. Patients were classified according to echocardiographic criteria as either PH (PH+, TRV >3.4 m/s) or controls (PH-, TRV ≤2.8 m/s) to simulate the real-world use of ECG-AI, where a positive result could lead to a follow-up Echo. The study database was locked before processing the digital ECGs with ECG-AI PH. Performance was also estimated in a subset of subjects that later had a right heart catheterization using mPAP ≥ 20mmHg as the definition of PH. Results: A total of 14281 subjects (53% female, 63% aged 50+ years) met the inclusion criteria, including 3019 PH+ cases and 11262 PH- controls (Figure). Sn and Sp were 84.0% (95% CI: 82.6%, 85.3%) and 71.7% (95% CI: 70.9%, 72.6%), respectively. The positive and negative predictive values were 18.9% (95% CI: 18.4%, 19.4%) and 98.3% (95% CI: 98.1%, 98.4%), respectively, at 7.3% prevalence. Each endpoint met pre-defined performance criteria. In the subset of 1683 subjects with a follow up RHC, performance remained robust (Sn 85% (1155/1358); Sp 46% (151/325)). Conclusion: While ECG-AI PH was first developed as an investigational tool to detect PH, continued development as software as a medical device for clinical use demonstrated that the algorithm retained strong performance to detect PH in diverse, non-overlapping clinical settings and patient populations.

Introduction: Pulmonary hypertension (PH) is a life-threatening, progressive disease with non-specific symptoms, often leading to delayed diagnosis. Early identification of World Health Organization Group 1 (Pulmonary Arterial Hypertension, PAH) and Group 4 (Chronic Thromboembolic Pulmonary Hypertension, CTEPH) is essential, as effective therapies can improve outcomes. Hypothesis: An electrocardiogram-based AI algorithm for PH detection (ECG-AI PH) may enable earlier diagnosis and reduce healthcare utilization, including hospitalizations and procedures. Methods: Retrospective analysis used a de-identified data platform of >7M clinical records from a multistate integrated health system. Adult precapillary PH patients (mPAP >20 mmHg, PVR >2 WU, PCWP ≤15 mmHg) were identified as PAH or CTEPH based on ICD codes, use of approved therapies, or surgical interventions (for CTEPH) between 2002 and 2024. ECG-AI PH was applied to ECGs within 30 days of diagnostic right heart catheterization, using a 5:1 randomly sampled PH-negative control cohort. Training set patients were excluded. Clinical event frequency was compared between two intervals: from first possible PH symptom (dyspnea, syncope, chest pain, fatigue, lower limb swelling) to diagnosis, and from symptom onset to first positive ECG-AI PH prediction. Results: A total of 1882 PAH and 359 CTEPH patients met inclusion criteria. Of these, 1340 PAH and 258 CTEPH patients had symptom codes prior to diagnosis. Both groups showed prolonged intervals from symptom onset to diagnosis, with multiple diagnostic procedures and hospitalizations (Figure). ECG-AI PH performance evaluation on the test set included 647 PAH and 152 CTEPH patients. ECG-AI PH achieved AUCs of 0.90 and 0.89 for PAH and CTEPH, sensitivities of 80.3% and 76.8%, and specificities of 83.4% and 82.4%. Among those tested, 576 PAH and 95 CTEPH patients had a positive ECG-AI PH prediction after symptom onset but before diagnosis. Compared to the current patient journey, the interval between initial symptoms and a positive output from ECG-AI PH was shorter and had fewer diagnostic tests/visits. Conclusion: ECG-AI PH demonstrated strong performance in detecting PAH and CTEPH. It may reduce diagnostic delays, support earlier PH-focused screening (e.g., echocardiograms evaluating the right heart), enable earlier intervention, and reduce pre-diagnosis healthcare burden, benefitting both patient outcomes and healthcare system efficiency.

Background: Immunotherapy-based regimens have expanded the treatment landscape for unresectable hepatocellular carcinoma (uHCC); however, real-world data are limited. Methods: This retrospective, observational study used data from electronic medical records from Mayo Clinic sites across the United States. Patients with uHCC who initiated a first-line (1L) systemic therapy between June 2020–October 2022 with ≥2 follow-up visits were included. Treatment patterns, overall survival (OS), and post-index gastrointestinal (GI) bleeding were assessed by GI bleeding risk defined by Child–Pugh Class B or C, pre-index GI bleeding, uncontrolled hypertension, or significant varices and band ligation. Results: Of 186 included patients, 68.8% had GI bleeding risk and 31.2% did not. Atezolizumab plus bevacizumab was the most common 1L systemic therapy in patients with or without GI bleeding risk (72.7% and 29.3%, respectively). Median OS (95% confidence interval) with atezolizumab plus bevacizumab was 12.8 (8.0–19.3) months and not reached in patients with and without GI bleeding risk, respectively. OS rates with atezolizumab plus bevacizumab in patients with or without GI bleeding risk, respectively, were 52.3% and 70.6% at 12 months, 41.6% and 57.8% at 18 months, and 34.6% and 51.3% at 24 months. Post-index GI bleeding with atezolizumab plus bevacizumab occurred in 19.4% and 5.9% of patients with and without GI bleeding risk, respectively. Conclusions: During this study period, atezolizumab + bevacizumab was the most common 1L therapy for patients with uHCC, regardless of GI bleeding risk. OS rates with atezolizumab + bevacizumab were lower in patients with versus without GI bleeding risk. Findings highlight the unmet need for guidance on characteristics-driven treatment decisions.

e16167 Background: Guidelines recommend systemic therapy for uHCC ineligible for locoregional therapy. Until the approval of tremelimumab + durvalumab for uHCC, in Oct 2022, 1L systemic therapies in the US included multikinase inhibitors and atezolizumab (A) + bevacizumab (B). It is crucial to understand factors driving real-world treatment decisions and outcomes in uHCC. Methods: Data were collected from electronic medical records from US Mayo Clinic sites. Pts with uHCC who initiated a 1L systemic therapy between Jun 2020–Oct 2022 (A + B was the only approved 1L immunotherapy (IO)-based regimen) with ≥2 follow-up visits, were included. Index was the date of first 1L systemic therapy. Pts were assessed by risk of post-index gastrointestinal (GI) bleeding (Child-Pugh class B or C, pre-index GI bleeding, uncontrolled hypertension [HTN] [≥2 anti-HTN drugs or ≥2 vital entries of SBP > 140 mmHg or DBP > 90 mmHg]), or significant varices and band ligation). Treatment patterns, overall survival (OS), and post-index GI bleeding were assessed. Results: A total of 186 pts met the inclusion criteria. Most common etiologies of liver disease were MASH (42.7%) and hepatitis C virus (40.2%). Pre-index GI bleeding was reported in 29.0% of pts, of which 46.3% occurred within 6 months (mo) pre-index; 62.4% of pts had EGD, of which 72.4% had varices. There were 128 (68.8%) pts with GI bleeding risk and 58 (31.2%) pts without GI bleeding risk. The most common 1L systemic therapies for pts with GI bleeding risk were A + B (72.7%) and A only (9.4%), while for pts without GI bleeding risk, A + B (29.3%), nivolumab + ipilimumab (13.8%), and B + chemotherapy (12.1%) were the most common. Median OS and OS rates at 12, 18, and 24 mo were lower, and post-index GI bleeding rate was higher in pts with vs without GI bleeding risk (Table). For pts treated with A + B, median OS and OS rates at 12, 18, and 24 mo were lower in pts with vs without GI bleeding risk (Table). Conclusions: In this network, prior to the approval of other 1L IO-based regimens for uHCC, many pts received A + B, despite risk of post-index GI bleeding. Median OS with A + B was shorter in pts with vs without GI bleeding risk. Data highlight the complexity of uHCC and the unmet need for guidance on characteristics-driven treatment decisions. © 2025 American Society of Clinical Oncology, Inc. Reused with permission. This abstract was accepted and previously presented at the 2025 ASCO Gastrointestinal Cancers Symposium. All rights reserved. Table. Full cohort A + B cohort All ptsN=186 GI bleeding risk N=128 No GI bleeding risk N=58 All pts N=110 GI bleeding risk N=93 No GI bleeding risk N=17 Median follow-up, mo 20.6 20.9 20.4 21.3 21.6 20.3 Median OS, mo 14.4 11.4 40.3 13.9 12.8 Not reached OS rates, % 12 mo 55.2 48.9 68.9 55.2 52.3 70.6 18 mo 45.3 38.1 61.1 44.1 41.6 57.8 24 mo 39.4 31.8 56.8 37.1 34.6 51.3 Post-index GI bleed, % 17.2 23.4 3.4 17.3 19.4 5.9

572 Background: Guidelines recommend systemic therapy for uHCC ineligible for locoregional therapy. Until the approval of tremelimumab + durvalumab for uHCC, in Oct 2022, 1L systemic therapies in the US included multikinase inhibitors and atezolizumab (A) + bevacizumab (B). It is crucial to understand factors driving real-world treatment decisions and outcomes in uHCC. Characteristics, treatment patterns, and outcomes were assessed in US pts with uHCC treated with 1L systemic therapy. Methods: Data were collected from electronic medical records from US Mayo Clinic sites. Pts with uHCC who initiated an NCCN-recommended 1L systemic therapy between Jun 2020–Oct 2022 (A + B was the only approved 1L immunotherapy (IO)-based regimen) with ≥2 follow-up visits, were included. Index was the date of first 1L systemic therapy. Pts were assessed by risk of post-index gastrointestinal (GI) bleeding (pts with Child-Pugh class B or C, pre-index GI bleeding, uncontrolled hypertension [HTN] [≥2 anti-HTN drugs or ≥2 vital entries of SBP >140 mmHg or DBP >90 mmHg]), or significant varices and band ligation). Treatment patterns, overall survival (OS), and post-index GI bleeding were assessed. Results: A total of 186 pts met the inclusion criteria. Most common etiologies of liver disease were metabolic dysfunction-associated steatohepatitis (42.7%) and hepatitis C virus (40.2%). Pre-index GI bleeding was reported in 29.0% of pts, of which 46.3% occurred within 6 months (mo) pre-index; 62.4% of pts had EGD, of which 72.4% had varices. There were 128 (68.8%) pts with GI bleeding risk and 58 (31.2%) pts without GI bleeding risk. The most common 1L systemic therapies for pts with GI bleeding risk were A + B (72.7%) and A only (9.4%), while for pts without GI bleeding risk, A + B (29.3%), nivolumab + ipilimumab (13.8%), and B + chemotherapy (13.8%) were the most common. Median OS and OS rates at 12, 18, and 24 mo were lower, and post-index GI bleeding rate was higher in pts with vs without GI bleeding risk (Table). For pts treated with A + B, median OS and OS rates at 12, 18, and 24 mo were lower in pts with vs without GI bleeding risk (Table). Conclusions: In this network, prior to the approval of other 1L IO-based regimens for uHCC, many pts received A + B, despite risk of post-index GI bleeding. Median OS with A + B was shorter in pts with vs without GI bleeding risk. Data highlight the complexity of uHCC and the unmet need for guidance on characteristics-driven treatment decisions. Full cohort A + B cohort All ptsN=186 GI bleeding risk N=128 No GI bleeding risk N=58 All pts N=110 GI bleeding risk N=93 No GI bleeding risk N=17 Median follow-up, mo 20.6 20.9 20.4 21.3 21.6 20.3 Median OS, mo 14.4 11.4 40.3 13.9 12.8 Not reached OS rates, % 12 mo 55.2 48.9 68.9 55.2 52.3 70.6 18 mo 45.3 38.1 61.1 44.1 41.6 57.8 24 mo 39.4 31.8 56.8 37.1 34.6 51.3 Post-index GI bleed, % 17.2 23.4 3.4 17.3 19.4 5.9

PURPOSE Teclistamab is initiated with a step-up dosing (SUD) schedule to mitigate the risk of cytokine release syndrome (CRS) and immune effector cell–associated neurotoxicity syndrome (ICANS). Early teclistamab users commonly received SUD in a hospital setting. This study aimed to evaluate safety and health care resource utilization (HRU) in real-world patients with multiple myeloma who initiated teclistamab SUD in an outpatient setting. METHODS This was a retrospective study using Mayo Clinic's electronic medical records from October 26, 2022, to October 31, 2023. Patient characteristics were summarized for all patients treated with teclistamab and separately for patients who started SUD outpatient. SUD pattern, safety, HRU, and post-SUD dosing schedule were described in patients with complete SUD. RESULTS At data cutoff, 65 patients received ≥1 teclistamab dose, including 58 patients who initiated SUD outpatient (median age, 69.2 years; male, 63.8%; White, 89.7%). Among 57 patients who completed SUD in an outpatient setting, all received premedications on the days of teclistamab administrations per label recommendation; 18 (31.6%) developed CRS (13 grade 1, four grade 2, and one grade 4) and two developed ICANS (one each with grade 2 and 4). All CRS and ICANS resolved with supportive care and all patients continued treatment. Eighteen patients were admitted to the hospital for CRS treatment, with a median CRS-related hospital stay of 2 days per admission. Most (60%) doses during SUD required <1 hour clinic time between administration and checkout. Post-SUD, clinic time for treatment doses decreased to <30 minutes for most doses (82%). CONCLUSION Outcomes of this study support outpatient administration as a safe and feasible option for teclistamab SUD to potentially reduce HRU and improve patient experiences.

e13156 Background: Breast Cancer (BC) patients are classified as HER2-positive (IHC3+ or IHC2+/ISH+) or HER2-negative (IHC0, IHC1+ and 2+/ISH-). Within the HER2-negative category, patients who meet HER2-low (IHC1+ or IHC2+/ISH-) criteria can benefit from HER2 targeted therapy like trastuzumab deruxtecan. Patients who meet IHC0 criteria with incomplete and faint membrane staining in >0 but ≤10% of tumor cells — defined as “HER2 Ultra-low”, may also benefit from HER2 targeted therapy. This study aimed to assess the prevalence of “HER2 Ultra-Low” expression based on re-scored HER2 IHC biopsy slides. Methods: This study utilized EHR data from all three US Mayo Clinic campuses. A total of 300 patients with advanced BC (stages III-IV), and clinically documented HER2 IHC0 status between Jan 2017 and Jan 2023, were identified. One biopsy slide per patient (IHC assay: VENTANA Pathway anti-HER2/neu (4B5) rabbit monoclonal primary antibody) was digitized and scanned at 40x magnification. Two Mayo Clinic pathologists independently scored and reported HER2 status according to 2023 ASCO-CAP guidelines and included tumor stain percentage for each slide. Patients were considered “HER2 Ultra-Low” if their slide was scored as IHC0 with >0% but ≤10% staining. Cohen’s κ was used to quantify agreement between pathologists. Results: The re-scored patients (n=300) had a mean age of 57.8 years (SD=13.5), with ~97% having IHC0 documentation within 30 days of the reviewed biopsy slide. A total of 285 (95%) patients remained classified as IHC0 by at least one pathologist. Of these, 60% (n=171) were determined as “HER2 Ultra-low” by at least one pathologist. The rate of “HER2 Ultra-Low” per pathologist ranged from 43% (n=121) to 45% (n=104). The overall inter-pathologist concordance was 57%; patients with no observable HER2 IHC staining accounted for majority of the concordant patients (Table). Conclusions: About 3 in 5 advanced BC patients originally classified as HER2 IHC0 met the “HER2 Ultra-Low” criteria by at least one pathologist, implying that significant number of additional patients may benefit from HER2 targeted therapy. The relatively low concordance between HER2 expert pathologists in identifying lower levels of HER2 expression suggests the need for increased precision enabled by digital tools, leveraging AI, training for community practice pathologists, and continuing to follow best practice recommendations in determining HER2 IHC status. [Table: see text]

PURPOSE Immune checkpoint inhibitors (ICIs) have revolutionized cancer treatment, yet their use is associated with immune-related adverse events (irAEs). Estimating the prevalence and patient impact of these irAEs in the real-world data setting is critical for characterizing the benefit/risk profile of ICI therapies beyond the clinical trial population. Diagnosis codes, such as International Classification of Diseases codes, do not comprehensively illustrate a patient's care journey and offer no insight into drug-irAE causality. This study aims to capture the relationship between ICIs and irAEs more accurately by using augmented curation (AC), a natural language processing–based innovation, on unstructured data in electronic health records. METHODS In a cohort of 9,290 patients treated with ICIs at Mayo Clinic from 2005 to 2021, we compared the prevalence of irAEs using diagnosis codes and AC models, which classify drug-irAE pairs in clinical notes with implied textual causality. Four illustrative irAEs with high patient impact—myocarditis, encephalitis, pneumonitis, and severe cutaneous adverse reactions, abbreviated as MEPS—were analyzed using corticosteroid administration and ICI discontinuation as proxies of severity. RESULTS For MEPS, only 70% (n = 118) of patients found by AC were also identified by diagnosis codes. Using AC models, patients with MEPS received corticosteroids for their respective irAE 82% of the time and permanently discontinued the ICI because of the irAE 35.9% (n = 115) of the time. CONCLUSION Overall, AC models enabled more accurate identification and assessment of patient impact of ICI-induced irAEs not found using diagnosis codes, demonstrating a novel and more efficient strategy to assess real-world clinical outcomes in patients treated with ICIs.