Health economic analysis of polygenic risk score use in primary prevention of coronary artery disease - A system dynamics model

Background

Primary prevention programs utilising traditional risk scores fail to identify all individuals who suffer acute cardiovascular events. We aimed to model the impact and cost effectiveness of incorporating a Polygenic risk scores (PRS) into the cardiovascular disease CVD primary prevention program in Australia, using a whole-of-system model.

Methods

System dynamics models, encompassing acute and chronic CVD care in the Australian healthcare setting, assessing the cost-effectiveness of incorporating a CAD-PRS in the primary prevention setting. The time horizon was 10-years.

Results

Pragmatically incorporating a CAD-PRS in the Australian primary prevention setting in middle-aged individuals already attending a Heart Health Check (HHC) who are determined to be at low or moderate risk based on the 5-year Framingham risk score (FRS), with conservative assumptions regarding uptake of PRS, could have prevented 2, 052 deaths over 10-years, and resulted in 24, 085 QALYs gained at a cost of $19, 945 per QALY with a net benefit of $724 million. If all Australians overs the age of 35 years old had their FRS and PRS performed, and acted upon, 12, 374 deaths and 60, 284 acute coronary events would be prevented, with 183, 682 QALYs gained at a cost of $18, 531 per QALY, with a net benefit of $5, 780 million.

Conclusions

Incorporating a CAD-PRS in a contemporary primary prevention setting in Australia would result in substantial health and societal benefits and is cost-effective. The broader the uptake of CAD-PRS in the primary prevention setting in middle-aged Australians, the greater the impact and the more cost-effective the strategy.

Primary prevention of cardiovascular disease in women

Ischemic heart disease is the primary cause of cardiovascular disease (CVD) mortality in both men and women. Strategies targeting traditional modifiable risk factors are essential - including hypertension, smoking, dyslipidemia and diabetes mellitus - particularly for atherosclerosis, but additionally for stroke, heart failure and some arrhythmias. However, challenges related to education, screening and equitable access to effective preventative therapies persist, and are particularly problematic for women around the globe and those from lower socioeconomic groups. The association of female-specific risk factors (e.g. premature menopause, gestational hypertension, small for gestational age births) with CVD provides a potential window for targeted prevention strategies. However, further evidence for specific effective screening and interventions is urgently required. In addition to population-level factors involved in increasing the risk of suffering a CVD event, efforts are leveraging the enormous potential of blood-based 'omics', improved imaging biomarkers and increasingly complex bioinformatic analytic approaches to strive toward more personalized early disease detection and personalized preventative therapies. These novel tactics may be particularly relevant for women in whom traditional risk factors perform poorly. Here we discuss established and emerging approaches for improving risk assessment, early disease detection and effective preventative strategies to reduce the mammoth burden of CVD in women.

Prediction of coronary artery disease using urinary proteomics

AIMS

Coronary artery disease (CAD) is multifactorial, caused by complex pathophysiology, and contributes to a high burden of mortality worldwide. Urinary proteomic analyses may help to identify predictive biomarkers and provide insights into the pathogenesis of CAD.

METHODS AND RESULTS

Urinary proteome was analysed in 965 participants using capillary electrophoresis coupled with mass spectrometry. A proteomic classifier was developed in a discovery cohort with 36 individuals with CAD and 36 matched controls using the support vector machine. The classifier was tested in a validation cohort with 115 individuals who progressed to CAD and 778 controls and compared with two previously developed CAD-associated classifiers, CAD238 and ACSP75. The Framingham and SCORE2 risk scores were available in 737 participants. Bioinformatic analysis was performed based on the CAD-associated peptides. The novel proteomic classifier was comprised of 160 urinary peptides, mainly related to collagen turnover, lipid metabolism, and inflammation. In the validation cohort, the classifier provided an area under the receiver operating characteristic curve (AUC) of 0.82 [95% confidence interval (CI): 0.78-0.87] for the CAD prediction in 8 years, superior to CAD238 (AUC: 0.71, 95% CI: 0.66-0.77) and ACSP75 (AUC: 0.53 and 95% CI: 0.47-0.60). On top of CAD238 and ACSP75, the addition of the novel classifier improved the AUC to 0.84 (95% CI: 0.80-0.89). In a multivariable Cox model, a 1-SD increment in the novel classifier was associated with a higher risk of CAD (HR: 1.54, 95% CI: 1.26-1.89, P < 0.0001). The new classifier further improved the risk reclassification of CAD on top of the Framingham or SCORE2 risk scores (net reclassification index: 0.61, 95% CI: 0.25-0.95, P = 0.001; 0.64, 95% CI: 0.28-0.98, P = 0.001, correspondingly).

CONCLUSION

A novel urinary proteomic classifier related to collagen metabolism, lipids, and inflammation showed potential for the risk prediction of CAD. Urinary proteome provides an alternative approach to personalized prevention.

Dissecting the mediating and moderating effects of depression on the associations between traits and coronary artery disease: A two-step Mendelian randomization and phenome-wide interaction study

Background

Depression is often present concurrently with coronary artery disease (CAD), a disease with which it shares many risk factors. However, the manner in which depression mediates and moderates the association between traits (including biomarkers, anthropometric indicators, lifestyle behaviors, etc.) and CAD is largely unknown.

Methods

In our causal mediation analyses using two-step Mendelian randomization (MR), univariable MR was first used to investigate the causal effects of 108 traits on liability to depression and CAD. The traits with significant causal effects on both depression and CAD, but not causally modulated by depression, were selected for the second-step analyses. Multivariable MR was used to estimate the direct effects (independent of liability to depression) of these traits on CAD, and the indirect effects (mediated via liability to depression) were calculated. To investigate the moderating effect of depression on the association between 364 traits and CAD, a cross-sectional phenome-wide interaction study (PheWIS) was conducted in a study population from UK Biobank (UKBB) (N=275,257). Additionally, if the relationship between traits and CAD was moderated by both phenotypic and genetically predicted depression at a suggestive level of significance (Pinteraction≤0.05) in the PheWIS, the results were further verified by a cohort study using Cox proportional hazards regression.

Results

Univariable MR indicated that 10 of 108 traits under investigation were significantly associated with both depression and CAD, which showed a similar direct effect compared to the total effect for most traits. However, the traits "drive faster than speed limit" and "past tobacco smoking" were both exceptions, with the proportions mediated by depression at 24.6% and 7.2%, respectively. In the moderation analyses, suggestive evidence of several traits was found for moderating effects of phenotypic depression or susceptibility to depression, as estimated by polygenic risk score, including chest pain when hurrying, reason of smoking quitting and weight change. Consistent results were observed in survival analyses and Cox regression.

Conclusion

The independent role of traits in CAD pathogenesis regardless of depression was highlighted in our mediation analyses, and the moderating effects of depression observed in our study may be helpful for CAD risk stratification and optimized allocation of scarce medical resources.

Opportunities and Challenges with Artificial Intelligence in Genomics

The development of artificial intelligence and machine learning algorithms may allow for advances in patient care. There are existing and potential applications in cancer diagnosis and monitoring, identification of at-risk groups of individuals, classification of genetic variants, and even prediction of patient ancestry. This article provides an overview of some current and future applications of artificial intelligence in genomic medicine, in addition to discussing challenges and considerations when bringing these tools into clinical practice.

Polygenic risk score and age: an extra help in the cardiovascular prevention of the young?

All major guidelines recommend assessing the risk of atherosclerotic cardiovascular disease (ASCVD) using risk scores. In fact, it has been shown that their use at the population level increases the accuracy of event prediction and facilitates the choice of strategies to be adopted in primary prevention. In fact, their use in clinical practice is far from optimal and their predictive ability on an individual level is not excellent. Our genetic heritage is substantially stable from birth and determines a ‘baseline risk’ on which external influences act. Genetic information therefore has the potential to be an early predictor of risk. Common diseases such as diabetes mellitus, ASCVD and neurodegenerative diseases are conditioned by different genetic variants with small individual effects, so that a reliable risk prediction requires careful examination of the aggregate impact of these multiple variants. The polygenic risk score (PRS) is a tool that potentially enables this complex assessment and provides a new opportunity to explore our risk of developing common diseases, including coronary artery disease (CAD). In the future, it is possible that a specific PRS could be used as an independent CAD screening tool, but this requires a detailed assessment of the practical implications, including the population to be investigated, and the consequent interventions that would then be offered.

Decision Support Tools for Coronary Artery Calcium Scoring in the Primary Prevention of Cardiovascular Disease Do Not Meet Health Literacy Needs: A Systematic Environmental Scan and Evaluation

A shared decision-making approach is considered optimal in primary cardiovascular disease (CVD) prevention. Evidence-based patient decision aids can facilitate this but do not always meet patients' health literacy needs. Coronary artery calcium (CAC) scans are increasingly used in addition to traditional cardiovascular risk scores, but the availability of high-quality decision aids to support shared decision-making is unknown. We used an environmental scan methodology to review decision support for CAC scans and assess their suitability for patients with varying health literacy. We systematically searched for freely available web-based decision support tools that included information about CAC scans for primary CVD prevention and were aimed at the public. Eligible materials were independently evaluated using validated tools to assess qualification as a decision aid, understandability, actionability, and readability. We identified 13 eligible materials. Of those, only one qualified as a decision aid, and one item presented quantitative information about the potential harms of CAC scans. None presented quantitative information about both benefits and harms of CAC scans. Mean understandability was 68%, and actionability was 48%. Mean readability (12.8) was much higher than the recommended grade 8 level. Terms used for CAC scans were highly variable. Current materials available to people considering a CAC scan do not meet the criteria to enable informed decision-making, nor do they meet the health literacy needs of the general population. Clinical guidelines, including CAC scans for primary prevention, must be supported by best practice decision aids to support decision-making.

Mortality and Cardiovascular Outcomes in Patients Presenting With Non-ST Elevation Myocardial Infarction Despite No Standard Modifiable Risk Factors: Results From the SWEDEHEART Registry

Background A significant proportion of patients with ST-segment-elevation myocardial infarction (MI) have no standard modifiable cardiovascular risk factors (SMuRFs) and have unexpected worse 30-day outcomes compared with those with SMuRFs. The aim of this article is to examine outcomes of patients with non-ST-segment-elevation MI in the absence of SMuRFs. Methods and Results Presenting features, management, and outcomes of patients with non-ST-segment-elevation MI without SmuRFs (hypertension, diabetes, hypercholesterolemia, smoking) were compared with those with SmuRFs in the Swedish MI registry SWEDEHEART (Swedish Web-System for Enhancement and Development of Evidence-Based Care in Heart Disease Evaluated According to Recommended Therapies; 2005-2018). Cox proportional hazard models were used. Out of 99 718 patients with non-ST-segment-elevation MI, 11 131 (11.2%) had no SMuRFs. Patients without SMuRFs had higher all-cause and cardiovascular mortality at 30 days (hazard ratio [HR], 1.20 [95% CI, 1.10-1.30], P<0.0001; and HR, 1.25 [95% CI, 1.13-1.38]), a difference that remained after adjustment for age and sex. SMuRF-less patients were less likely to receive secondary prevention statins (76% versus 82%); angiotensin-converting enzyme inhibitors/angiotensin receptor blockade (54% versus 72%); or β-blockers (81% versus 87%, P for all <0.0001), with lowest rates observed in women without SMuRFs. In patients who survived to 30 days, rates of all-cause and cardiovascular death were lower in patients without SMuRFs compared with those with risk factors, over 12 years. Conclusions One in 10 patients presenting with non-ST-segment-elevation MI present without traditional risk factors. The excess 30-day mortality rate in this group emphasizes the need for both improved population-based strategies for prevention of MI, as well as the need for equitable evidence-based treatment at the time of an MI.

Polygenic risk for type 2 diabetes, lifestyle, metabolic health, and cardiovascular disease: a prospective UK Biobank study

BACKGROUND

Few studies have examined associations between genetic risk for type 2 diabetes (T2D), lifestyle, clinical risk factors, and cardiovascular disease (CVD). We aimed to investigate the association of and potential interactions among genetic risk for T2D, lifestyle behavior, and metabolic risk factors with CVD.

METHODS

A total of 345,217 unrelated participants of white British descent were included in analyses. Genetic risk for T2D was estimated as a genome-wide polygenic risk score constructed from > 6 million genetic variants. A favorable lifestyle was defined in terms of four modifiable lifestyle components, and metabolic health status was determined according to the presence of metabolic syndrome components.

RESULTS

During a median follow-up of 8.9 years, 21,865 CVD cases (6.3%) were identified. Compared with the low genetic risk group, participants at high genetic risk for T2D had higher rates of overall CVD events, CVD subtypes (coronary artery disease, peripheral artery disease, heart failure, and atrial fibrillation/flutter), and CVD mortality. Individuals at very high genetic risk for T2D had a 35% higher risk of CVD than those with low genetic risk (HR 1.35 [95% CI 1.19 to 1.53]). A significant gradient of increased CVD risk was observed across genetic risk, lifestyle, and metabolic health status (P for trend > 0.001). Those with favorable lifestyle and metabolically healthy status had significantly reduced risk of CVD events regardless of T2D genetic risk. This risk reduction was more apparent in young participants (≤ 50 years).

CONCLUSIONS

Genetic risk for T2D was associated with increased risks of overall CVD, various CVD subtypes, and fatal CVD. Engaging in a healthy lifestyle and maintaining metabolic health may reduce subsequent risk of CVD regardless of genetic risk for T2D.

Neural network-based integration of polygenic and clinical information: development and validation of a prediction model for 10-year risk of major adverse cardiac events in the UK Biobank cohort

BACKGROUND

In primary cardiovascular disease prevention, early identification of high-risk individuals is crucial. Genetic information allows for the stratification of genetic predispositions and lifetime risk of cardiovascular disease. However, towards clinical application, the added value over clinical predictors later in life is crucial. Currently, this genotype-phenotype relationship and implications for overall cardiovascular risk are unclear.

METHODS

In this study, we developed and validated a neural network-based risk model (NeuralCVD) integrating polygenic and clinical predictors in 395 713 cardiovascular disease-free participants from the UK Biobank cohort. The primary outcome was the first record of a major adverse cardiac event (MACE) within 10 years. We compared the NeuralCVD model with both established clinical scores (SCORE, ASCVD, and QRISK3 recalibrated to the UK Biobank cohort) and a linear Cox-Model, assessing risk discrimination, net reclassification, and calibration over 22 spatially distinct recruitment centres.

FINDINGS

The NeuralCVD score was well calibrated and improved on the best clinical baseline, QRISK3 (ΔConcordance index [C-index] 0·01, 95% CI 0·009-0·011; net reclassification improvement (NRI) 0·0488, 95% CI 0·0442-0·0534) and a Cox model (ΔC-index 0·003, 95% CI 0·002-0·004; NRI 0·0469, 95% CI 0·0429-0·0511) in risk discrimination and net reclassification. After adding polygenic scores we found further improvements on population level (ΔC-index 0·006, 95% CI 0·005-0·007; NRI 0·0116, 95% CI 0·0066-0·0159). Additionally, we identified an interaction of genetic information with the pre-existing clinical phenotype, not captured by conventional models. Additional high polygenic risk increased overall risk most in individuals with low to intermediate clinical risk, and age younger than 50 years.

INTERPRETATION

Our results demonstrated that the NeuralCVD score can estimate cardiovascular risk trajectories for primary prevention. NeuralCVD learns the transition of predictive information from genotype to phenotype and identifies individuals with high genetic predisposition before developing a severe clinical phenotype. This finding could improve the reprioritisation of otherwise low-risk individuals with a high genetic cardiovascular predisposition for preventive interventions.

FUNDING

Charité-Universitätsmedizin Berlin, Einstein Foundation Berlin, and the Medical Informatics Initiative.

Preventing unnecessary imaging in patients suspect of coronary artery disease through machine learning of electronic health records

Aims

With the ageing European population, the incidence of coronary artery disease (CAD) is expected to rise. This will likely result in an increased imaging use. Symptom recognition can be complicated, as symptoms caused by CAD can be atypical, particularly in women. Early CAD exclusion may help to optimize use of diagnostic resources and thus improve the sustainability of the healthcare system. To develop sex-stratified algorithms, trained on routinely available electronic health records (EHRs), raw electrocardiograms, and haematology data to exclude CAD in patients upfront.

Methods and results

We trained XGBoost algorithms on data from patients from the Utrecht Patient-Oriented Database, who underwent coronary computed tomography angiography (CCTA), and/or stress cardiac magnetic resonance (CMR) imaging, or stress single-photon emission computerized tomography (SPECT) in the UMC Utrecht. Outcomes were extracted from radiology reports. We aimed to maximize negative predictive value (NPV) to minimize the false negative risk with acceptable specificity. Of 6808 CCTA patients (31% female), 1029 females (48%) and 1908 males (45%) had no diagnosis of CAD. Of 3053 CMR/SPECT patients (45% female), 650 females (47%) and 881 males (48%) had no diagnosis of CAD. On the train and test set, the CCTA models achieved NPVs and specificities of 0.95 and 0.19 (females) and 0.96 and 0.09 (males). The CMR/SPECT models achieved NPVs and specificities of 0.75 and 0.041 (females) and 0.92 and 0.026 (males).

Conclusion

Coronary artery disease can be excluded from EHRs with high NPV. Our study demonstrates new possibilities to reduce unnecessary imaging in women and men suspected of CAD.

Coronary artery disease burden in women poorly explained by traditional risk factors: Sex disaggregated analyses from the BioHEART-CT study

BACKGROUND AND AIMS

Targeting the modifiable risk factors for coronary artery disease (CAD) has substantial impact at the community level. However, it is not uncommon for individuals to present with atherosclerosis related events without identified risk factors. We examined sex differences in the association of risk factors and atherosclerotic burden assessed by CT coronary angiography (CTCA).

METHODS

We analysed clinical and imaging data in 1002 individuals in the BioHEART cohort.

RESULTS

45% were female, 35% had no CAD identified. Median coronary calcium score was 9.9 Agatston units (IQR: 0-146), and median Gensini Score was 3.5 (IQR: 0-11.5). 26% had a calcified plaque predominant phenotype, and 18% had a non-calcified plaque predominant phenotype. There were no sex differences in the prevalence of risk factors. However, there were notable sex differences in the adjusted associations of risk factors with CAD. Age and hypercholesterolaemia (OR 1.56, 95% CI 1.03-2.36, p = 0.04 in males, and OR 1.75, 95% CI 1.09-2.78, p = 0.02 in females) were associated with the presence of CAD in both genders (p < 0.05). Diabetes and smoking were associated with presence of CAD, calcified CAD, and non-calcified plaque in males (p < 0.05) but not females. In women, none of the standard modifiable risk factors were associated with the amount of plaque present when adjusted for age, BMI, and family history of premature CAD.

CONCLUSIONS

CTCA provides an important opportunity for improving the stratification of cohorts to assess underlying biology and risk. We demonstrate sex-specific differences in associations of risk factors with atherosclerosis burden.

Metabolic Signatures in Coronary Artery Disease: Results from the BioHEART-CT Study

Despite effective prevention programs targeting cardiovascular risk factors, coronary artery disease (CAD) remains the leading cause of death. Novel biomarkers are needed for improved risk stratification and primary prevention. To assess for independent associations between plasma metabolites and specific CAD plaque phenotypes we performed liquid chromatography mass-spectrometry on plasma from 1002 patients in the BioHEART-CT study. Four metabolites were examined as candidate biomarkers. Dimethylguanidino valerate (DMGV) was associated with presence and amount of CAD (OR) 1.41 (95% Confidence Interval [CI] 1.12-1.79, p = 0.004), calcified plaque, and obstructive CAD (p < 0.05 for both). The association with amount of plaque remained after adjustment for traditional risk factors, ß-coefficient 0.17 (95% CI 0.02-0.32, p = 0.026). Glutamate was associated with the presence of non-calcified plaque, OR 1.48 (95% CI 1.09-2.01, p = 0.011). Phenylalanine was associated with amount of CAD, ß-coefficient 0.33 (95% CI 0.04-0.62, p = 0.025), amount of calcified plaque, (ß-coefficient 0.88, 95% CI 0.23-1.53, p = 0.008), and obstructive CAD, OR 1.84 (95% CI 1.01-3.31, p = 0.046). Trimethylamine N-oxide was negatively associated non-calcified plaque OR 0.72 (95% CI 0.53-0.97, p = 0.029) and the association remained when adjusted for traditional risk factors. In targeted metabolomic analyses including 53 known metabolites and controlling for a 5% false discovery rate, DMGV was strongly associated with the presence of calcified plaque, OR 1.59 (95% CI 1.26-2.01, p = 0.006), obstructive CAD, OR 2.33 (95% CI 1.59-3.43, p = 0.0009), and amount of CAD, ß-coefficient 0.3 (95% CI 0.14-0.45, p = 0.014). In multivariate analyses the lipid and nucleotide metabolic pathways were both associated with the presence of CAD, after adjustment for traditional risk factors. We report novel associations between CAD plaque phenotypes and four metabolites previously associated with CAD. We also identified two metabolic pathways strongly associated with CAD, independent of traditional risk factors. These pathways warrant further investigation at both a biomarker and mechanistic level.