Opportunistic AI-derived adiposity measures from coronary artery calcium scans predict new-onset type 2 diabetes in adults without obesity or hyperglycemia: insights from the AI-CVD study within MESA

Introduction: Coronary artery calcium (CAC) scans contain more information than currently reported. The AI-CVD initiative aims to extract actionable opportunististic infomration from a CAC scan to maximize its predictive value beyond the CAC score. We previously reported new AI-CVD algorithms applied to CAC scans for opportunistic measurement of bone mineral density (BMD), cardiac chamber volumes, left ventricular mass, liver steatosis, emphysema and other imaging biomarkers. In this report, we investigate the incremental value of these biomarkers on top of the CAC score for prediction of incident all cardiovascular disease (CVD) events. Methods: We applied AI-CVD to CAC scans from 5798 asymptomatic individuals (52% female, age 62±10 years) in the Multi-Ethnic Study of Atherosclerosis. Liver fat was estimated as the liver attenuation index (LAI) using the percentage of voxels below 40 HU. Phantomless BMD for three consecutive thoracic vertebrae (T2-T4) was calculated using the mean HU. Emphysema was estimated using the percentage of voxels within the entire lung in the field of view below -950 HU. We used Kaplan–Meier cumulative incidence curves to evaluate the incremental prognostic value of opportunistically derived biomarkers beyond the Agatston CAC score, using the highest quartile of risk per predictor. Results: A total of 1173 CVD accrued over 19 years follow-up (median [IQR]: 17.7 [12.9-18.5] years). The top quartile of CAC, LAI, and emphysema were defined as >90, >50.4% voxels below 40 HU, > 4.1% below -950 HU, respectively. The bottom quartile of BMD was defined as <130.1 mg/cc. Individuals in the highest risk quartile of all 4 measures (n=44) experienced 69.0% (95% CI: 57.3%-80.1%) incidence of all CVD events. While individuals with a high Agatston score alone experienced 47.9% (44.6%-51.4%) incidence of CVD events over 19 years. Both low BMD and high LAI revealed incremental CVD risk on top of high CAC scores, while high emphysema measurements did not. Conclusion: Applying AI to CAC scans can extract opportunistic incremental risk information for early detection of patients at risk of CVD events. The clinical utility of incorporating LAI, BMD, and emphysema and other opportunistic findings in CAC scans as part of the AI-CVD initiative to improve CVD risk prediction warrants further investigations.

IntroductionAbout one-third of adults in the USA have some grade of hepatic steatosis. Coronary artery calcium (CAC) scans contain more information than currently reported. We previously reported new artificial intelligence...

Background: Coronary artery calcium (CAC) scans contain more actionable information than the Agatston CAC score. We have previously shown in the Multi-Ethnic Study of Atherosclerosis (MESA) that AI-enabled left atrial (LA) volumetry in CAC scans (AI-CAC) enabled prediction of atrial fibrillation (AF) as early as one year. Furthermore, we have recently shown adding AI-CAC LA volumetry to CHA2DS2-VASc risk score improved stroke prediction in MESA. In this study we evaluated the performance of AI-CAC LA volumetry versus LA measured by human experts using cardiac magnetic resonance imaging (CMRI) for predicting AF and stroke, and compared them with CHARGE-AF risk score, Agatston score, and NT-proBNP. Methods: We used 15-year outcomes data from 3552 asymptomatic individuals (52.2% women, age 61.7±10.2 years) who underwent both CAC scans and CMRI in the MESA baseline examination. We have applied the AutoChamberTM (HeartLung.AI, Houston, TX) component of AI-CAC to 3552 CAC scans. CMRI LA volume was previously measured by human experts. Data on NT-proBNP, CHARGE-AF risk score and the Agatston score were obtained from MESA. Discrimination was assessed using the time-dependent area under the curve (AUC). Results: Over 15 years follow-up, 562 cases of AF and 140 cases of stroke accrued. The AUC for 15-year AF prediction by AI-CAC LA volume (0.801) was comparable to CMRI LA volume (0.797) and significantly higher than Agatston CAC Score (0.687) and NT-proBNP (0.704). Similarly, the AUC for 15-year stroke prediction for AI-CAC volumetry (0.761) was comparable to CMRI volumetry (0.751) and significantly higher than NT-proBNP (0.631) and Agatston CAC Score (0.646). AI-CAC LA volume outperformed CHARGE AF over 1-3 years for incident AF (p<0.02), but not for subsequent years. AI-CAC significantly improved the continuous Net Reclassification Index (NRI) for prediction of AF and stroke when added to CHARGE-AF risk score (0.28, 0.21), NT-proBNP (0.43, 0.37), and Agatston score (0.69, 0.41) respectively (p for all<0.0001). Conclusion: LA volumetry measured by the AutoChamber component of AI-CAC and CMRI LA volume measured by human experts similarly predicted incident AF and stroke over 15 years, and outperformed NT-proBNP and Agatston CAC Score. AI-CAC LA volumetry outperformed CHARGE-AF and NT-proBNP for short-term (1-3 years) AF prediction. Further studies to investigate the clinical utility of AI-CAC LA volumetry for AF and stroke prediction are warranted.

AimThe long-term value of coronary artery calcium (CAC) scanning has not been studied extensively in symptomatic patients, but was evaluated by us in 644 consecutive patients referred for stable chest pain.MethodsWe excluded patients with a history of cardiovascular disease and with a CAC score of zero. CAC scanning was done with a 16-row MDCT scanner. Endpoints were: (a) overall mortality, (b) mortality or non-fatal myocardial infarction and (c) the composite of mortality, myocardial infarction or coronary revascularisation. Revascularisations within 1 year following CAC scanning were not considered.ResultsThe mean age of the 320 women and 324 men was 63 years. Follow-up was over 8 years. There were 58 mortalities, while 22 patients suffered non-fatal myocardial infarction and 24 underwent coronary revascularisation, providing 104 combined endpoints. Cumulative 8‑year survival was 95% with CAC score <100, 90% in patients with CAC score >100 and <400, and 82% with CAC score ≥400 Agatston units. Risk of mortality with a CAC score >100 and ≥400 units was 2.6 [95% confidence interval (CI) 1.23–5.54], and 4.6 (95% CI 2.1–9.47) respectively. After correction for clinical risk factors, CAC score remained independently associated with increased risk of cardiac events.ConclusionsRisk increased with increasing CAC score. Patients with CAC >100 or ≥400 Agatston units were at increased risk of major adverse cardiac events and are eligible for preventive measures. CAC scanning provided incremental prognostic information to guide the choice of diagnostic and therapeutic options in many subjects evaluated for chest pain.

Determinants of Coronary Calcium Conversion Among Patients With a Normal Coronary Calcium Scan: What Is the “Warranty Period” for Remaining Normal?

Focused, low tube potential, coronary calcium assessment prior to coronary CT angiography: A prospective, randomized clinical trial

AI-enabled cardiac chambers volumetry in coronary calcium scans (AI-CAC) predicts heart failure and outperforms NT-proBNP: The Multi-Ethnic Study of Atherosclerosis

Long-term prognostic value of coronary artery calcium scanning, coronary computed tomographic angiography and stress myocardial perfusion imaging in patients with suspected coronary artery disease

Background: The AI-CVD initiative aims to extract opportunistic screening information from coronary artery calcium (CAC) scans to maximize cardiovascular disease prediction beyond the traditional risk factors and the Agatston CAC score. Hypothesis: In 2024, the American Heart Association introduced the PREVENT heart failure (HF) risk score based on age, sex, systolic blood pressure, body mass index, glomerular filtration rate (GFR), diabetes, smoking, and anti-hypertensive medication consumption. We sought to compare PREVENT HF vs. AI-CVD risk scores for predicting HF in the Multi-Ethnic Study of Atherosclerosis (MESA). Method: AI-CVD platform is a collection of deep learning models targeting various componenets of a CAC scan (see figure 1). We applied AI-CVD to 4,554 CAC scans of asymptomatic MESA participants aged 45–84 years (46.9% male). We used selected AI-CVD outputs included cardiac chamber volumes, thoracic skeletal muscle volume and density, epicardial fat volume, percentage of lung emphysema (&lt;950 HU), and percentage of liver fat (&lt;40 HU). Clinical data comprised demographic and anthropometric characteristics, laboratory results, lifestyle factors, and electrocardiogram parameters. Embedded feature selection methods were applied to identify the most important predictors of HF. The AI-CVD risk score for incident HF was developed using FasterRisk, an interpretable machine learning technique. We then compared the performance of PREVENT HF vs. AI-CVD using the area under the receiver operating curve (AUC) and DeLong’s test for predicting HF. Results: After a median follow-up of 17.7 (IQR: 13.0-18.5) years, 265 (5.8%) cases were diagnosed with HF. Age, GFR, hypertension, anti-hypertensive medication consumption, smoking, microalbuminuria, diabetes, left atrial volume, ratio of left ventricle to right ventricle volume, left ventricular mass, CAC score, epicardial fat volume, and emphysema were selected features for predicting HF. The AUC for AI-CVD (AUC: 0.84 [95% CI:0.82-0.87]) was significantly (P &lt; 0.001) higher than for PREVENT HF (AUC: 0.77, 95% CI: 0.74-0.81) for 10-year HF prediction. Conclusion: By integrating AI-generated opportunistic screening biomarkers from CAC scans with clinical data, the AI-CVD risk score significantly outperformed the PREVENT risk score for HF prediction in MESA participants over 10 years.

Introduction: New innovations in AI allow opportunistic detection of non-coronary features on coronary artery calcium (CAC) scans, enabling screening for a range of conditions, and improved cardiovascular disease (CVD) prediction. Myosteatosis, excessive fat infiltration into skeletal muscle, is increasingly recognized as a marker of systemic metabolic dysfunction and can be quantified in CT using the mean attenuation of skeletal muscle. We evaluated AI-measured myosteatosis in thoracic skeletal muscle for predicting future atrial fibrillation (AF), heart failure (HF), and total CVD. Methods: We used baseline CAC scans and 15-year follow-up data from 5,489 asymptomatic participants (47.8% male) in the Multi-Ethnic Study of Atherosclerosis (MESA). Myosteatosis was operationally defined as the lowest quartile of thoracic skeletal muscle mean attenuation (males&lt;33 Hounsfield Units (HU) and females&lt;27 HU). Hazard ratios [HR] for bottom vs top quartile of mean muscle CT density were evaluated using proportional hazards regression models adjusted for CVD risk factors, inflammatory markers, and social determinants of health. Results: Myosteatosis was associated with worse outcomes in both sexes: HRs in males were 4.59 (95% CI, 3.52–5.99) for AF, 8.46 (4.61–15.52) for HF, and 3.56 (2.89–4.37) for total CVD, with corresponding HRs in females of 4.68 (3.48–6.29), 8.01 (3.62–17.72), and 4.37 (3.42–5.57), respectively. After full adjustment, associations remained significant for HF (1.93 [1.31–2.82]), AF (1.78 [1.26–2.50]), and total CVD (1.44 [1.09–1.91]) in males, and for AF (1.69 [1.17–2.45]) and total CVD (1.75 [1.29–2.39]) in females. Individuals in the top quartile of CAC (&gt;89.5 HU) who also had myosteatosis had greater 15-year incidence of AF (45.4%) and HF (21.8%) than those in either group alone (CAC, AF: 29%, HF: 9.5%; myosteatosis, AF: 20.9%, HF: 5.3%). Conclusion: Thoracic skeletal myosteatosis in CAC scans is an independent predictor of AF, HF, and total CVD over 15 years. Improving clinical outcomes through the detection of myosteatosis, and other opportunistic findings in CAC scans as part of the AI-CVD initiative, merits further investigation.

Rationale and objectivesWe previously reported a novel manual method for measuring bone mineral density (BMD) in coronary artery calcium (CAC) scans and validated our method against Dual X-Ray Absorptiometry (DEXA). Furthermore, we have developed and validated an artificial intelligence (AI) based automated BMD (AutoBMD) measurement as an opportunistic add-on to CAC scans that recently received FDA approval. In this report, we present evidence of equivalency between AutoBMD measurements in cardiac vs lung CT scans. Materials and methodsAI models were trained using 132 cases with 7649 (3 mm) slices for CAC, and 37 cases with 21918 (0.5 mm) slices for lung scans. To validate AutoBMD against manual measurements, we used 6776 cases of BMD measured manually on CAC scans in the Multi-Ethnic Study of Atherosclerosis (MESA). We then used 165 additional cases from Harbor UCLA Lundquist Institute to compare AutoBMD in patients who underwent both cardiac and lung scans on the same day. ResultsMean±SD for age was 69 ± 9.4 years with 52.4% male. AutoBMD in lung and cardiac scans, and manual BMD in cardiac scans were 153.7 ± 43.9, 155.1 ± 44.4, and 163.6 ± 45.3 g/cm3, respectively (p = 0.09). Bland-Altman agreement analysis between AutoBMD lung and cardiac scans resulted in 1.37 g/cm3 mean differences. Pearson correlation coefficient between lung and cardiac AutoBMD was R2 = 0.95 (p < 0.0001). ConclusionOpportunistic BMD measurement using AutoBMD in CAC and lung cancer screening scans is promising and yields similar results. No extra radiation plus the high prevalence of asymptomatic osteoporosis makes AutoBMD an ideal screening tool for osteopenia and osteoporosis in CT scans done for other reasons.

e24056 Background: Many cancer (ca) survivors exhibit signs of insulin resistance (IR), an important risk factor for the development of coronary artery disease (CAD). Paramount in survivorship care is prevention of cardiovascular disease. Coronary artery calcium (CAC) scans offer a risk assessment of cardiovascular (CV) disease before cardiac damage has occurred. We investigated how IR affects CAC scores in cancer survivors. We hypothesized that CAC scores differ significantly between insulin-sensitive- and -resistant cancer survivors. Methods: We enrolled 90 cancer survivors of a large community hospital from March 2021 to January 2022 into this pilot study. Patients were subdivided into three groups: insulin-sensitive (IS), insulin-resistant/prediabetic and insulin-resistant/diabetic. Patients were tested for fasting insulin, -glucose, HgbA1c and lipids. Patients without evidence for prediabetes or diabetes also underwent an oral glucose tolerance test (oGTT). All patients received a CAC scan. Results: 32 patients were IS, 29 patients were IR/prediabetic and 29 patients were IR/diabetic. 17 CAC scans in the IS group, 6 CAC scans in the IR/prediabetic group and 5 CAC scans in the IR/diabetic group showed an Agatston score of 0. The p-value between the three groups was statistically significant ( p=0.005) where as the IR/prediabetic- and the IR/diabetic group did not differ statistically from each other. The mean MESA 10-year CHD risk with CAC was 7.8. There was a highly significant difference between the 3 groups: the IS group had a mean of 5.3, the IR/prediabetic group had a mean of 7.3, and the IR/diabetic group had a mean of 11.0 ( p &lt; 0.001). The two IR groups did not differ statistically (p=0.076). Conclusions: Our study showed that IR including prediabetes significantly increases the MESA 10-yr. CHD Risk with CAC in cancer survivors. Survivors with IR also have less frequent zero CAC scores than insulin-sensitive survivors. Survivors disproportionately exhibit insulin resistance, partly due to the association of certain types of cancer with IR. This trial highlights the importance of screening survivors for IR. Survivors diagnosed with IR should be screened for CAD more frequently than the general population. CAC scans are an inexpensive and efficient way of screening asymptomatic cancer survivors for CAD.

Whether a coronary artery calcium (CAC) scan provides added value to coronary computed tomographic angiography (CCTA) in emergency department patients with acute chest pain remains unsettled. We sought to determine the value of CAC scan in patients with acute chest pain undergoing CCTA. In the multicenter Rule Out Myocardial Infarction using Computer-Assisted Tomography (ROMICAT) II trial, we enrolled low-intermediate risk emergency department patients with symptoms suggesting acute coronary syndrome (ACS). In this prespecified subanalysis of 473 patients (54±8 years, 53% men) who underwent both CAC scanning and CCTA, the ACS rate was 8%. Overall, 53% of patients had CAC=0 of whom 2 (0.8%) developed ACS, whereas 7% had CAC>400 with 49% whom developed ACS. C-statistic of CAC>0 was 0.76, whereas that using the optimal cut point of CAC≥22 was 0.81. Continuous CAC score had lower discriminatory capacity than CCTA (c-statistic, 0.86 versus 0.92; P=0.03). Compared with CCTA alone, there was no benefit combining CAC score with CCTA (c-statistic, 0.93; P=0.88) or with selective CCTA strategies after initial CAC>0 or optimal cut point CAC≥22 (P≥0.09). Mean radiation dose from CAC acquisition was 1.4±0.7 mSv. Higher CAC scores resulted in more nondiagnostic CCTA studies although the majority remained interpretable. In emergency department patients with acute chest pain, CAC score does not provide incremental value beyond CCTA for ACS diagnosis. CAC=0 does not exclude ACS, nor a high CAC score preclude interpretation of CCTA in most patients. Thus, CAC results should not influence the decision to proceed with CCTA, and the decision to perform a CAC scan should be balanced with the additional radiation exposure required. http://www.clinicaltrials.gov. Unique identifier: NCT01084239.

Background: Coronary artery calcium (CAC) scans contain valuable information beyond the Agatston score which is used for coronary artery disease prediction only. We have previously reported AI-enabled cardiac chambers volumetry in CAC scans (AI-CAC) predicts incident atrial fibrillation (AF), heart failure (HF), and stroke in the Multi-Ethnic Study of Atherosclerosis (MESA). Here we compare the distribution of cardiac chambers volumes vs. risk categories of ASCVD pooled cohorts’ equation (PCE) and PREVENTTM risk scores. Methods: We applied the AutoChamberTM component of AI-CAC to 5830 individuals (52.2% women, age 61.7±10.2 years) without known CVD that were previously obtained for CAC scoring at the baseline examination of MESA. We calculated 10-year estimated risk from the PCE and PREVENT Risk Scores based on 4 categories of risk: &lt;5%, 5-7.5%, 7.5-20%, &gt;20% using baseline risk factors. The PREVENT total CVD base model was used in analysis, which excludes urinary albumin to creatinine ratio and social depravity index. We compared the distribution of the quartiles of left atrial (LA) and left ventricle (LV) volumes to categories of both risk scores. We defined enlarged cardiac chambers as the top quartile of LA (&gt;82.7 cc) and LV (&gt;136.5 cc) volume, which corresponded to 33% and 21.1% incidence of all-CVD events over 10 years (CHD, HF, AF, stroke, CVD deaths), respectively. LA and LV volumes were standardized by adjusting for body surface area (BSA). Results: A substantial portion of cases categorized by PREVENT as low risk (10-year risk &lt;5%) had enlarged cardiac chambers (Figure 1). In females the lowest category of PREVENT had 10.6% enlarged LA cases and 26.6% enlarged LV volume cases. In males, the lowest category of PREVENT had 13.7% enlarged LA cases and 29.4% enlarged LV volume cases. Similarly, in low risk PCE, females had 12.8% enlarged LA and 24.9% enlarged LV, and males had 12.8% enlarged LA and 24.6% enlarged LV. Conclusion: In this multi-ethnic longitudinal population study, a substantial portion of cases classified as low-risk by PCE and PREVENT risk scores had enlarged LA and LV volumes detected by AI-CAC, putting them at risk for future HF, AF, and stroke.

Background: The AI-CVD initiative aims to extract all useful opportunistic screening information from coronary artery calcium (CAC) scans and combines them with traditional risk factors to create a stronger predictor of cardiovascular diseases (CVD). These measurements include cardiac chambers volumes (left atrium (LA), left ventricle (LV), right atrium (RA), right ventricle (RV), and left ventricular mass (LVM)), aortic wall and valvular calcification, aorta and pulmonary artery volumes, torso visceral fat, emphysema score, thoracic bone mineral density, and fatty liver score. We have previously reported that the automated cardiac chambers volumetry component of AI-CVD predicts incident atrial fibrillation (AF), heart failure (HF), and stroke in the Multi-Ethnic Study of Atherosclerosis (MESA). In this report, we examine the contribution of other AI-CVD components for all coronary heart disease (CHD), AF, HF, stroke plus transient ischemic attack (TIA), all-CVD, and all-cause mortality. Methods: We applied AI-CVD to CAC scans of 5830 individuals (52.2% women, age 61.7±10.2 years) without known CVD that were previously obtained for CAC scoring at MESA baseline examination. We used 10-year outcomes data and assessed hazard ratios for AI-CVD components plus CAC score and known CVD risk factors (age, sex, diabetes, smoking, LDL-C, HDL-C, systolic and diastolic blood pressure, hypertension medication). AI-CVD predictors were modeled per standard deviation (SD) increase using Cox proportional hazards regression. Results: Over 10 years of follow-up, 1058 CVD (550 AF, 198 HF, 163 stroke, 389 CHD) and 628 all-cause mortality events accrued with some cases having multiple events. Among AI-CVD components, CAC score and chamber volumes were the strongest predictors of different outcomes. Expectedly, age was the strongest predictor for all outcomes except HF where LV volume and LV mass were stronger predictors than age. Figure 1 shows contribution of each predictor for various outcomes. Conclusion: AI-enabled opportunistic screening of useful information in CAC scans contributes substantially to CVD and total mortality prediction independently of CAC score and CVD risk factors. Further studies are warranted to evaluate the clinical utility of AI-CVD.