Expanding antibiotic, vaccine, and diagnostics development and access to tackle antimicrobial resistance.

The increasing number of bacterial infections globally that do not respond to any available antibiotics indicates a need to invest in-and ensure access to-new antibiotics, vaccines, and diagnostics. The traditional model of drug development, which depends on substantial revenues to motivate investment, is no longer economically viable without push and pull incentives. Moreover, drugs developed through these mechanisms are unlikely to be affordable for all patients in need, particularly in low-income and middle-income countries. New, publicly funded models based on public-private partnerships could support investment in antibiotics and novel alternatives, and lower patients' out-of-pocket costs, making drugs more accessible. Cost reductions can be achieved with public goods, such as clinical trial networks and platform-based quality assurance, manufacturing, and product development support. Preserving antibiotic effectiveness relies on accurate and timely diagnosis; however scaling up diagnostics faces technological, economic, and behavioural challenges. New technologies appeared during the COVID-19 pandemic, but there is a need for a deeper understanding of market, physician, and consumer behaviour to improve the use of diagnostics in patient management. Ensuring sustainable access to antibiotics also requires infection prevention. Vaccines offer the potential to prevent infections from drug-resistant pathogens, but funding for vaccine development has been scarce in this context. The High-Level Meeting of the UN General Assembly in 2024 offers an opportunity to rethink how research and development can be reoriented to serve disease management, prevention, patient access, and antibiotic stewardship.

Alternating low-rank tensor reconstruction for improved multiparametric mapping with cardiovascular MR Multitasking.

PURPOSE: To develop a novel low-rank tensor reconstruction approach leveraging the complete acquired data set to improve precision and repeatability of multiparametric mapping within the cardiovascular MR Multitasking framework. METHODS: A novel approach that alternated between estimation of temporal components and spatial components using the entire data set acquired (i.e., including navigator data and imaging data) was developed to improve reconstruction. The precision and repeatability of the proposed approach were evaluated on numerical simulations, 10 healthy subjects, and 10 cardiomyopathy patients at multiple scan times for 2D myocardial T1/T2 mapping with MR Multitasking and were compared with those of the previous navigator-derived fixed-basis approach. RESULTS: In numerical simulations, the proposed approach outperformed the previous fixed-basis approach with lower T1 and T2 error against the ground truth at all scan times studied and showed better motion fidelity. In human subjects, the proposed approach showed no significantly different sharpness or T1/T2 measurement and significantly improved T1 precision by 20%-25%, T2 precision by 10%-15%, T1 repeatability by about 30%, and T2 repeatability by 25%-35% at 90-s and 50-s scan times The proposed approach at the 50-s scan time also showed comparable results with that of the previous fixed-basis approach at the 90-s scan time. CONCLUSION: The proposed approach improved precision and repeatability for quantitative imaging with MR Multitasking while maintaining comparable motion fidelity, T1/T2 measurement, and septum sharpness and had the potential for further reducing scan time from 90 s to 50 s.

Differentiation between descending thoracic aortic diseases using machine learning and plasma proteomic signatures.

BACKGROUND: Descending thoracic aortic aneurysms and dissections can go undetected until severe and catastrophic, and few clinical indices exist to screen for aneurysms or predict risk of dissection. METHODS: This study generated a plasma proteomic dataset from 75 patients with descending type B dissection (Type B) and 62 patients with descending thoracic aortic aneurysm (DTAA). Standard statistical approaches were compared to supervised machine learning (ML) algorithms to distinguish Type B from DTAA cases. Quantitatively similar proteins were clustered based on linkage distance from hierarchical clustering and ML models were trained with uncorrelated protein lists across various linkage distances with hyperparameter optimization using fivefold cross validation. Permutation importance (PI) was used for ranking the most important predictor proteins of ML classification between disease states and the proteins among the top 10 PI protein groups were submitted for pathway analysis. RESULTS: Of the 1,549 peptides and 198 proteins used in this study, no peptides and only one protein, hemopexin (HPX), were significantly different at an adjusted p < 0.01 between Type B and DTAA cases. The highest performing model on the training set (Support Vector Classifier) and its corresponding linkage distance (0.5) were used for evaluation of the test set, yielding a precision-recall area under the curve of 0.7 to classify between Type B from DTAA cases. The five proteins with the highest PI scores were immunoglobulin heavy variable 6-1 (IGHV6-1), lecithin-cholesterol acyltransferase (LCAT), coagulation factor 12 (F12), HPX, and immunoglobulin heavy variable 4-4 (IGHV4-4). All proteins from the top 10 most important groups generated the following significantly enriched pathways in the plasma of Type B versus DTAA patients: complement activation, humoral immune response, and blood coagulation. CONCLUSIONS: We conclude that ML may be useful in differentiating the plasma proteome of highly similar disease states that would otherwise not be distinguishable using statistics, and, in such cases, ML may enable prioritizing important proteins for model prediction.

Sex differences in coronary atherosclerotic plaque activity using 18F-sodium fluoride positron emission tomography.

INTRODUCTION: There are sex differences in the extent, severity, and outcomes of coronary artery disease. We aimed to assess the influence of sex on coronary atherosclerotic plaque activity measured using coronary 18F-sodium fluoride (18F-NaF) positron emission tomography (PET), and to determine whether 18F-NaF PET has prognostic value in both women and men. METHODS: In a post-hoc analysis of observational cohort studies of patients with coronary atherosclerosis who had undergone 18F-NaF PET CT angiography, we compared the coronary microcalcification activity (CMA) in women and men. RESULTS: Baseline 18F-NaF PET CT angiography was available in 999 participants (151 (15%) women) with 4282 patient-years of follow-up. Compared to men, women had lower coronary calcium scores (116 [interquartile range, 27-434] versus 205 [51-571] Agatston units; p = 0.002) and CMA values (0.0 [0.0-1.12] versus 0.53 [0.0-2.54], p = 0.01). Following matching for plaque burden by coronary calcium scores and clinical comorbidities, there was no sex-related difference in CMA values (0.0 [0.0-1.12] versus 0.0 [0.0-1.23], p = 0.21) and similar proportions of women and men had no 18F-NaF uptake (53.0% (n = 80) and 48.3% (n = 73); p = 0.42), or CMA values > 1.56 (21.8% (n = 33) and 21.8% (n = 33); p = 1.00). Over a median follow-up of 4.5 [4.0-6.0] years, myocardial infarction occurred in 6.6% of women (n = 10) and 7.8% of men (n = 66). Coronary microcalcification activity greater than 0 was associated with a similarly increased risk of myocardial infarction in both women (HR: 3.83; 95% CI:1.10-18.49; p = 0.04) and men (HR: 5.29; 95% CI:2.28-12.28; p < 0.001). CONCLUSION: Although men present with more coronary atherosclerotic plaque than women, increased plaque activity is a strong predictor of future myocardial infarction regardless of sex.

Downward myocardial creep during stress PET imaging is inversely associated with mortality.

PURPOSE: The myocardial creep is a phenomenon in which the heart moves from its original position during stress-dynamic PET myocardial perfusion imaging (MPI) that can confound myocardial blood flow measurements. Therefore, myocardial motion correction is important to obtain reliable myocardial flow quantification. However, the clinical importance of the magnitude of myocardial creep has not been explored. We aimed to explore the prognostic value of myocardial creep quantified by an automated motion correction algorithm beyond traditional PET-MPI imaging variables. METHODS: Consecutive patients undergoing regadenoson rest-stress [82Rb]Cl PET-MPI were included. A newly developed 3D motion correction algorithm quantified myocardial creep, the maximum motion at stress during the first pass (60 s), in each direction. All-cause mortality (ACM) served as the primary endpoint. RESULTS: A total of 4,276 patients (median age 71 years; 60% male) were analyzed, and 1,007 ACM events were documented during a 5-year median follow-up. Processing time for automatic motion correction was < 12 s per patient. Myocardial creep in the superior to inferior (downward) direction was greater than the other directions (median, 4.2 mm vs. 1.3-1.7 mm). Annual mortality rates adjusted for age and sex were reduced with a larger downward creep, with a 4.2-fold ratio between the first (0 mm motion) and 10th decile (11 mm motion) (mortality, 7.9% vs. 1.9%/year). Downward creep was associated with lower ACM after full adjustment for clinical and imaging parameters (adjusted hazard ratio, 0.93; 95%CI, 0.91-0.95; p < 0.001). Adding downward creep to the standard PET-MPI imaging model significantly improved ACM prediction (area under the receiver operating characteristics curve, 0.790 vs. 0.775; p < 0.001), but other directions did not (p > 0.5). CONCLUSIONS: Downward myocardial creep during regadenoson stress carries additional information for the prediction of ACM beyond conventional flow and perfusion PET-MPI. This novel imaging biomarker is quantified automatically and rapidly from stress dynamic PET-MPI.

Reproducibility of artificial intelligence-enabled plaque measurements between systolic and diastolic phases from coronary computed tomography angiography.

OBJECTIVES: Current coronary CT angiography (CTA) guidelines suggest both end-systolic and mid-diastolic phases of the cardiac cycle can be used for CTA image acquisition. However, whether differences in the phase of the cardiac cycle influence coronary plaque measurements is not known. We aim to explore the potential impact of cardiac phases on quantitative plaque assessment. METHODS: We enrolled 39 consecutive patients (23 male, age 66.2 ± 11.5 years) who underwent CTA with dual-source CT with visually evident coronary atherosclerosis and with good image quality. End-systolic and mid- to late-diastolic phase images were reconstructed from the same CTA scan. Quantitative plaque and stenosis were analyzed in both systolic and diastolic images using artificial intelligence (AI)-enabled plaque analysis software (Autoplaque). RESULTS: Overall, 186 lesions from 39 patients were analyzed. There were excellent agreement and correlation between systolic and diastolic images for all plaque volume measurements (Lin's concordance coefficient ranging from 0.97 to 0.99; R ranging from 0.96 to 0.98). There were no substantial intrascan differences per patient between systolic and diastolic phases (p > 0.05 for all) for total (1017.1 ± 712.9 mm3 vs. 1014.7 ± 696.2 mm3), non-calcified (861.5 ± 553.7 mm3 vs. 856.5 ± 528.7 mm3), calcified (155.7 ± 229.3 mm3 vs. 158.2 ± 232.4 mm3), and low-density non-calcified plaque volume (151.4 ± 106.1 mm3 vs. 151.5 ± 101.5 mm3) and diameter stenosis (42.5 ± 18.4% vs 41.3 ± 15.1%). CONCLUSION: Excellent agreement and no substantial differences were observed in AI-enabled quantitative plaque measurements on CTA in systolic and diastolic images. Following further validation, standardized plaque measurements can be performed from CTA in systolic or diastolic cardiac phase. CLINICAL RELEVANCE STATEMENT: Quantitative plaque assessment using artificial intelligence-enabled plaque analysis software can provide standardized plaque quantification, regardless of cardiac phase. KEY POINTS: • The impact of different cardiac phases on coronary plaque measurements is unknown. • Plaque analysis using artificial intelligence-enabled software on systolic and diastolic CT angiography images shows excellent agreement. • Quantitative coronary artery plaque assessment can be performed regardless of cardiac phase.

Progression of non-obstructive coronary plaque: a practical CCTA-based risk score from the PARADIGM registry.

OBJECTIVES: No clear recommendations are endorsed by the different scientific societies on the clinical use of repeat coronary computed tomography angiography (CCTA) in patients with non-obstructive coronary artery disease (CAD). This study aimed to develop and validate a practical CCTA risk score to predict medium-term disease progression in patients at a low-to-intermediate probability of CAD. METHODS: Patients were part of the Progression of AtheRosclerotic PlAque Determined by Computed Tomographic Angiography Imaging (PARADIGM) registry. Specifically, 370 (derivation cohort) and 219 (validation cohort) patients with two repeat, clinically indicated CCTA scans, non-obstructive CAD, and absence of high-risk plaque (≥ 2 high-risk features) at baseline CCTA were included. Disease progression was defined as the new occurrence of ≥ 50% stenosis and/or high-risk plaque at follow-up CCTA. RESULTS: In the derivation cohort, 104 (28%) patients experienced disease progression. The median time interval between the two CCTAs was 3.3 years (2.7-4.8). Odds ratios for disease progression derived from multivariable logistic regression were as follows: 4.59 (95% confidence interval: 1.69-12.48) for the number of plaques with spotty calcification, 3.73 (1.46-9.52) for the number of plaques with low attenuation component, 2.71 (1.62-4.50) for 25-49% stenosis severity, 1.47 (1.17-1.84) for the number of bifurcation plaques, and 1.21 (1.02-1.42) for the time between the two CCTAs. The C-statistics of the model were 0.732 (0.676-0.788) and 0.668 (0.583-0.752) in the derivation and validation cohorts, respectively. CONCLUSIONS: The new CCTA-based risk score is a simple and practical tool that can predict mid-term CAD progression in patients with known non-obstructive CAD. CLINICAL RELEVANCE STATEMENT: The clinical implementation of this new CCTA-based risk score can help promote the management of patients with non-obstructive coronary disease in terms of timing of imaging follow-up and therapeutic strategies. KEY POINTS: • No recommendations are available on the use of repeat CCTA in patients with non-obstructive CAD. • This new CCTA score predicts mid-term CAD progression in patients with non-obstructive stenosis at baseline. • This new CCTA score can help guide the clinical management of patients with non-obstructive CAD.

Latent Coronary Plaque Morphology From Computed Tomography Angiography, Molecular Disease Activity on Positron Emission Tomography, and Clinical Outcomes.

BACKGROUND: Assessments of coronary disease activity with 18F-sodium fluoride positron emission tomography and radiomics-based precision coronary plaque phenotyping derived from coronary computed tomography angiography may enhance risk stratification in patients with coronary artery disease. We sought to investigate whether the prognostic information provided by these 2 approaches is complementary in the prediction of myocardial infarction. METHODS: Patients with known coronary artery disease underwent coronary 18F-sodium fluoride positron emission tomography and coronary computed tomography angiography on a hybrid positron emission tomography/computed tomography scanner. Coronary 18F-NaF uptake was determined by the coronary microcalcification activity. We performed quantitative plaque analysis of coronary computed tomography angiography datasets and extracted 1103 radiomic features for each plaque. Using weighted correlation network analysis, we derived latent morphological features of coronary lesions which were aggregated to patient-level radiomics nomograms to predict myocardial infarction. RESULTS: Among 260 patients with established coronary artery disease (age, 65±9 years; 83% men), 179 (69%) participants showed increased coronary 18F-NaF activity (coronary microcalcification activity>0). Over 53 (40-59) months of follow-up, 18 patients had a myocardial infarction. Using weighted correlation network analysis, we derived 15 distinct eigen radiomic features representing latent morphological coronary plaque patterns in an unsupervised fashion. Following adjustments for calcified, noncalcified, and low-density noncalcified plaque volumes and 18F-NaF coronary microcalcification activity, 4 radiomic features remained independent predictors of myocardial infarction (hazard ratio, 1.46 [95% CI, 1.03-2.08]; P=0.03; hazard ratio, 1.62 [95% CI, 1.04-2.54]; P=0.02; hazard ratio, 1.49 [95% CI, 1.07-2.06]; P=0.01; and hazard ratio, 1.50 (95% CI, 1.05-2.13); P=0.02). CONCLUSIONS: In patients with established coronary artery disease, latent coronary plaque morphological features, quantitative plaque volumes, and disease activity on 18F-sodium fluoride positron emission tomography are additive predictors of myocardial infarction.

Is typical angina still prognostically important? The influence of "treatment bias" upon prognostic assessments.

BACKGROUND: Since typical angina has become less frequent, it is unclear if this symptom still has prognostic significance. METHODS: We evaluated 38,383 patients undergoing stress/rest SPECT myocardial perfusion imaging followed for a median of 10.9 years. After dividing patients by clinical symptoms, we evaluated the magnitude of myocardial ischemia and subsequent mortality among medically treated versus revascularized subgroups following testing. RESULTS: Patients with typical angina had more frequent and greater ischemia than other symptom groups, but not higher mortality. Among typical angina patients, those who underwent early revascularization had substantially greater ischemia than the medically treated subgroup, including a far higher proportion with severe ischemia (44.9% vs 4.3%, P < 0.001) and transient ischemic dilation of the LV (31.3% vs 4.7%, P < 0.001). Nevertheless, the revascularized typical angina subgroup had a lower adjusted mortality risk than the medically treated subgroup (HR = 0.72, 95% CI: 0.57-0.92, P = 0.009) CONCLUSIONS: Typical angina is associated with substantially more ischemia than other clinical symptoms. However, the high referral of patients with typical angina patients with ischemia to early revascularization resulted in this group having a lower rather than higher mortality risk versus other symptom groups. These findings illustrate the need to account for "treatment bias" among prognostic studies.

Machine and deep learning models for accurate detection of ischemia and scar with myocardial blood flow positron emission tomography imaging.

BACKGROUND: Quantification of myocardial blood flow (MBF) is used for the noninvasive diagnosis of patients with coronary artery disease (CAD). This study compared traditional statistics, machine learning, and deep learning techniques in their ability to diagnose disease using only the rest and stress MBF values. METHODS: This study included 3245 rest and stress rubidium-82 positron emission tomography (PET) studies and matching diagnostic labels from perfusion reports. Standard logistic regression, lasso logistic regression, support vector machine, random forest, multilayer perceptron, and dense U-Net were compared for per-patient detection and per-vessel localization of scars and ischemia. RESULTS: Receiver-operator characteristic area under the curve (AUC) of machine learning models was significantly higher than those of traditional statistics models for per-patient detection of disease (0.92-0.95 vs. 0.87) but not for per-vessel localization of ischemia or scar. Random forest showed the highest AUC = 0.95 among the different models compared. On the final hold-out set for generalizability, random forest showed an AUC of 0.92 for detection and 0.89 for localization of perfusion abnormalities. CONCLUSIONS: For per-vessel localization, simple models trained on segmental data performed similarly to a convolutional neural network trained on polar-map data, highlighting the need to justify the use of complex predictive algorithms through comparison with simpler methods.

Comparative predictors of mortality among patients referred for stress single-photon emission computed tomography versus positron emission tomography myocardial perfusion imaging.

BACKGROUND: There is currently little information regarding the usage and comparative predictors of mortality among patients referred for single-photon emission computed tomography (SPECT) versus positron emission tomography (PET) myocardial perfusion imaging (MPI) within multimodality imaging laboratories. METHODS: We compared the clinical characteristics and mortality outcomes among 15,718 patients referred for SPECT-MPI and 6202 patients referred for PET-MPI between 2008 and 2017. RESULTS: Approximately two-thirds of MPI studies were performed using SPECT-MPI. The PET-MPI group was substantially older and included more patients with known coronary artery disease (CAD), hypertension, diabetes, and myocardial ischemia. The annualized mortality rate was also higher in the PET-MPI group, and this difference persisted after propensity matching 3615 SPECT-MPI and 3615 PET-MPI patients to have similar clinical profiles. Among the SPECT-MPI patients, the most potent predictor of mortality was exercise ability and performance, including consideration of patients' mode of stress testing and exercise duration. Among the PET-MPI patients, myocardial flow reserve (MFR) was the most potent predictor of mortality. CONCLUSIONS: In our real-world setting, PET-MPI was more commonly employed among older patients with more cardiac risk factors than SPECT-MPI patients. The most potent predictors of mortality in our SPECT and PET-MPI groups were variables exclusive to each test: exercise ability/capacity for SPECT-MPI patients and MFR for PET-MPI patients.

From bench to bedside: Improving the clinical safety of GalNAc-siRNA conjugates using seed-pairing destabilization.

Preclinical mechanistic studies have pointed towards RNA interference-mediated off-target effects as a major driver of hepatotoxicity for GalNAc-siRNA conjugates. Here, we demonstrate that a single glycol nucleic acid or 2'-5'-RNA modification can substantially reduce small interfering RNA (siRNA) seed-mediated binding to off-target transcripts while maintaining on-target activity. In siRNAs with established hepatotoxicity driven by off-target effects, these novel designs with seed-pairing destabilization, termed enhanced stabilization chemistry plus (ESC+), demonstrated a substantially improved therapeutic window in rats. In contrast, siRNAs thermally destabilized to a similar extent by the incorporation of multiple DNA nucleotides in the seed region showed little to no improvement in rat safety suggesting that factors in addition to global thermodynamics play a role in off-target mitigation. We utilized the ESC+ strategy to improve the safety of ALN-HBV, which exhibited dose-dependent, transient and asymptomatic alanine aminotransferase elevations in healthy volunteers. The redesigned ALN-HBV02 (VIR-2218) showed improved specificity with comparable on-target activity and the program was reintroduced into clinical development.

Initial Invasive or Conservative Strategy for Stable Coronary Disease.

BACKGROUND: Among patients with stable coronary disease and moderate or severe ischemia, whether clinical outcomes are better in those who receive an invasive intervention plus medical therapy than in those who receive medical therapy alone is uncertain. METHODS: We randomly assigned 5179 patients with moderate or severe ischemia to an initial invasive strategy (angiography and revascularization when feasible) and medical therapy or to an initial conservative strategy of medical therapy alone and angiography if medical therapy failed. The primary outcome was a composite of death from cardiovascular causes, myocardial infarction, or hospitalization for unstable angina, heart failure, or resuscitated cardiac arrest. A key secondary outcome was death from cardiovascular causes or myocardial infarction. RESULTS: Over a median of 3.2 years, 318 primary outcome events occurred in the invasive-strategy group and 352 occurred in the conservative-strategy group. At 6 months, the cumulative event rate was 5.3% in the invasive-strategy group and 3.4% in the conservative-strategy group (difference, 1.9 percentage points; 95% confidence interval [CI], 0.8 to 3.0); at 5 years, the cumulative event rate was 16.4% and 18.2%, respectively (difference, -1.8 percentage points; 95% CI, -4.7 to 1.0). Results were similar with respect to the key secondary outcome. The incidence of the primary outcome was sensitive to the definition of myocardial infarction; a secondary analysis yielded more procedural myocardial infarctions of uncertain clinical importance. There were 145 deaths in the invasive-strategy group and 144 deaths in the conservative-strategy group (hazard ratio, 1.05; 95% CI, 0.83 to 1.32). CONCLUSIONS: Among patients with stable coronary disease and moderate or severe ischemia, we did not find evidence that an initial invasive strategy, as compared with an initial conservative strategy, reduced the risk of ischemic cardiovascular events or death from any cause over a median of 3.2 years. The trial findings were sensitive to the definition of myocardial infarction that was used. (Funded by the National Heart, Lung, and Blood Institute and others; ISCHEMIA ClinicalTrials.gov number, NCT01471522.).

Incremental prognostic value of stress phase entropy over standard PET myocardial perfusion imaging variables.

PURPOSE: Phase analysis can assess left ventricular dyssynchrony. The independent prognostic value of phase variables over positron emission tomography myocardial perfusion imaging (PET-MPI) variables including myocardial flow reserve (MFR) has not been studied. The aim of this study was to explore the prognostic value of phase variables for predicting mortality over standard PET-MPI variables. METHODS: Consecutive patients who underwent pharmacological stress-rest 82Rb PET study were enrolled. All PET-MPI variables including phase variables (phase entropy, phase bandwidth, and phase standard deviation) were automatically obtained by QPET software (Cedars-Sinai, Los Angeles, CA). Cox proportional hazard analyses were used to assess associations with all-cause mortality (ACM). RESULTS: In a total of 3963 patients (median age 71 years; 57% male), 923 patients (23%) died during a median follow-up of 5 years. Annualized mortality rates increased with stress phase entropy, with a 4.6-fold difference between the lowest and highest decile groups of entropy (2.6 vs. 12.0%/year). Abnormal stress phase entropy (optimal cutoff value, 43.8%) stratified ACM risk in patients with normal and impaired MFR (both p < 0.001). Among three phase variables, only stress phase entropy was significantly associated with ACM after the adjustment of standard clinical and PET-MPI variables including MFR and stress-rest change of phase variables, whether modeled as binary variables (adjusted hazard ratio, 1.44 for abnormal entropy [> 43.8%]; 95%CI, 1.18-1.75; p < 0.001) or continuous variables (adjusted hazard ratio, 1.05 per 5% increase; 95%CI, 1.01-1.10; p = 0.030). The addition of stress phase entropy to the standard PET-MPI variables significantly improved the discriminatory power for ACM prediction (p < 0.001), but the other phase variables did not (p > 0.1). CONCLUSION: Stress phase entropy is independently and incrementally associated with ACM beyond standard PET-MPI variables including MFR. Phase entropy can be obtained automatically and included in clinical reporting of PET-MPI studies to improve patient risk prediction.

Mitigating bias in deep learning for diagnosis of coronary artery disease from myocardial perfusion SPECT images.

PURPOSE: Artificial intelligence (AI) has high diagnostic accuracy for coronary artery disease (CAD) from myocardial perfusion imaging (MPI). However, when trained using high-risk populations (such as patients with correlating invasive testing), the disease probability can be overestimated due to selection bias. We evaluated different strategies for training AI models to improve the calibration (accurate estimate of disease probability), using external testing. METHODS: Deep learning was trained using 828 patients from 3 sites, with MPI and invasive angiography within 6 months. Perfusion was assessed using upright (U-TPD) and supine total perfusion deficit (S-TPD). AI training without data augmentation (model 1) was compared to training with augmentation (increased sampling) of patients without obstructive CAD (model 2), and patients without CAD and TPD < 2% (model 3). All models were tested in an external population of patients with invasive angiography within 6 months (n = 332) or low likelihood of CAD (n = 179). RESULTS: Model 3 achieved the best calibration (Brier score 0.104 vs 0.121, p < 0.01). Improvement in calibration was particularly evident in women (Brier score 0.084 vs 0.124, p < 0.01). In external testing (n = 511), the area under the receiver operating characteristic curve (AUC) was higher for model 3 (0.930), compared to U-TPD (AUC 0.897) and S-TPD (AUC 0.900, p < 0.01 for both). CONCLUSION: Training AI models with augmentation of low-risk patients can improve calibration of AI models developed to identify patients with CAD, allowing more accurate assignment of disease probability. This is particularly important in lower-risk populations and in women, where overestimation of disease probability could significantly influence down-stream patient management.

Unsupervised learning to characterize patients with known coronary artery disease undergoing myocardial perfusion imaging.

PURPOSE: Patients with known coronary artery disease (CAD) comprise a heterogenous population with varied clinical and imaging characteristics. Unsupervised machine learning can identify new risk phenotypes in an unbiased fashion. We use cluster analysis to risk-stratify patients with known CAD undergoing single-photon emission computed tomography (SPECT) myocardial perfusion imaging (MPI). METHODS: From 37,298 patients in the REFINE SPECT registry, we identified 9221 patients with known coronary artery disease. Unsupervised machine learning was performed using clinical (23), acquisition (17), and image analysis (24) parameters from 4774 patients (internal cohort) and validated with 4447 patients (external cohort). Risk stratification for all-cause mortality was compared to stress total perfusion deficit (< 5%, 5-10%, ≥10%). RESULTS: Three clusters were identified, with patients in Cluster 3 having a higher body mass index, more diabetes mellitus and hypertension, and less likely to be male, have dyslipidemia, or undergo exercise stress imaging (p < 0.001 for all). In the external cohort, during median follow-up of 2.6 [0.14, 3.3] years, all-cause mortality occurred in 312 patients (7%). Cluster analysis provided better risk stratification for all-cause mortality (Cluster 3: hazard ratio (HR) 5.9, 95% confidence interval (CI) 4.0, 8.6, p < 0.001; Cluster 2: HR 3.3, 95% CI 2.5, 4.5, p < 0.001; Cluster 1, reference) compared to stress total perfusion deficit (≥10%: HR 1.9, 95% CI 1.5, 2.5 p < 0.001; < 5%: reference). CONCLUSIONS: Our unsupervised cluster analysis in patients with known CAD undergoing SPECT MPI identified three distinct phenotypic clusters and predicted all-cause mortality better than ischemia alone.

Does Coronary Plaque Morphology Matter Beyond Plaque Burden?

PURPOSE OF REVIEW: Imaging of adverse coronary plaque features by coronary computed tomography angiography (CCTA) has advanced greatly and at a fast pace. We aim to describe the evolution, present and future in plaque analysis, and its value in comparison to plaque burden. RECENT FINDINGS: Recently, it has been demonstrated that in addition to plaque burden, quantitative and qualitative assessment of coronary plaque by CCTA can improve the prediction of future major adverse cardiovascular events in diverse coronary artery disease scenarios. The detection of high-risk non-obstructive coronary plaque can lead to higher use of preventive medical therapies such as statins and aspirin, help identify culprit plaque, and differentiate between myocardial infarction types. Even more, over traditional plaque burden, plaque analysis including pericoronary inflammation can potentially be useful tools for tracking disease progression and response to medical therapy. The identification of the higher risk phenotypes with plaque burden, plaque characteristics, or ideally both can allow the allocation of targeted therapies and potentially monitor response. Further observational data are now required to investigate these key issues in diverse populations, followed by rigorous randomized controlled trials.

Development and validation of ischemia risk scores.

BACKGROUND: The likelihood of ischemia on myocardial perfusion imaging is central to physician decisions regarding test selection, but dedicated risk scores are lacking. We derived and validated two novel ischemia risk scores to support physician decision making. METHODS: Risk scores were derived using 15,186 patients and validated with 2,995 patients from a different center. Logistic regression was used to assess associations with ischemia to derive point-based and calculated ischemia scores. Predictive performance for ischemia was assessed using area under the receiver operating characteristic curve (AUC) and compared with the CAD consortium basic and clinical models. RESULTS: During derivation, the calculated ischemia risk score (0.801) had higher AUC compared to the point-based score (0.786, p < 0.001). During validation, the calculated ischemia score (0.716, 95% CI 0.684- 0.748) had higher AUC compared to the point-based ischemia score (0.699, 95% CI 0.666- 0.732, p = 0.016) and the clinical CAD model (AUC 0.667, 95% CI 0.633- 0.701, p = 0.002). Calibration for both ischemia scores was good in both populations (Brier score  < 0.100). CONCLUSIONS: We developed two novel risk scores for predicting probability of ischemia on MPI which demonstrated high accuracy during model derivation and in external testing. These scores could support physician decisions regarding diagnostic testing strategies.

Decline in typical angina among patients referred for cardiac stress testing.

OBJECTIVE: To evaluate temporal trends in the prevalence of typical angina and its clinical correlates among patients referred for stress/rest SPECT myocardial perfusion imaging (MPI). PATIENTS AND METHODS: We evaluated the prevalence of chest pain symptoms and their relationship to inducible myocardial ischemia among 61,717 patients undergoing stress/rest SPECT-MPI between January 2, 1991 and December 31, 2017. We also assessed the relationship between chest pain symptom and angiographic findings among 6,579 patients undergoing coronary CT angiography between 2011 and 2017. RESULTS: The prevalence of typical angina among SPECT-MPI patients declined from 16.2% between 1991 and 1997 to 3.1% between 2011 and 2017, while the prevalence of dyspnea without any chest pain increased from 5.9 to 14.5% over the same period. The frequency of inducible myocardial ischemia declined over time within all symptom groups, but its frequency among current patients (2011-2017) with typical angina was approximately three-fold higher versus other symptom groups (28.4% versus 8.6%, p < 0.001). Overall, patients with typical angina had a higher prevalence of obstructive CAD on CCTA than those with other clinical symptoms, but 33.3% of typical angina patients had no coronary stenoses, 31.1% had 1-49% stenoses, and 35.4% had ≥ 50% stenoses. CONCLUSIONS: The prevalence of typical angina has declined to a very low level among contemporary patients referred for noninvasive cardiac tests. The angiographic findings among current typical angina patients are now quite heterogeneous, with one-third of such patients having normal coronary angiograms. However, typical angina remains associated with a substantially higher frequency of inducible myocardial ischemia compared to patients with other cardiac symptoms.

Increasing frequency of dyspnea among patients referred for cardiac stress testing.

OBJECTIVE: To assess the frequency, change in prevalence, and prognostic significance of dyspnea among contemporary patients referred for cardiac stress testing. PATIENTS AND METHODS: We evaluated the prevalence of dyspnea and its relationship to all-cause mortality among 33,564 patients undergoing stress/rest SPECT-MPI between January 1, 2002 and December 31, 2017. Dyspnea was assessed as a single-item question. Patients were divided into three temporal groups. RESULTS: The overall prevalence of dyspnea in our cohort was 30.2%. However, there was a stepwise increase in the temporal prevalence of dyspnea, which was present in 25.6% of patients studied between 2002 and 2006, 30.5% of patients studied between 2007 and 2011, and 38.7% of patients studied between 2012 and 2017. There was a temporal increase in the prevalence of dyspnea in each age, symptom, and risk factor subgroup. The adjusted hazard ratio for mortality was higher among patients with dyspnea vs those without dyspnea both among all patients, and within each chest pain subgroup. CONCLUSIONS: Dyspnea has become increasingly prevalent among patients referred for cardiac stress testing and is now present among nearly two-fifths of contemporary cohorts referred for stress-rest SPECT-MPI. Prospective study is needed to standardize the assessment of dyspnea and evaluate the reasons for its increasing prevalence.