APSC CME Webinar: How Should We Apply ‘Omics’ on Cardiovascular Diseases to Clinical Medicine?

Share on email
Share on print
Share on whatsapp

Asian Pacific Society of Cardiology (APSC) CME Webinar: How Should We Apply ‘Omics’ on Cardiovascular Diseases to Clinical Medicine?

Date: 12 Jul 2021 (Monday)

Time: 5.30-6.30pm (GMT +8)

Hosted by: Asian Pacific Society of Cardiology (APSC)


  • Clin A/Prof Jack Tan
    • Deputy Head & Senior Consultant, Department of Cardiology, National Heart Centre Singapore
    • Senior Consultant NHSC Cardiology, Sengkang General Hospital
    • President, International Society of Cardiovascular Pharmacotherapy
    • President, APSC Singapore
  • Prof Faiez Zannad
    • Emeritus Professor of Universite de Lorraine, Therapeutics and Cardiology
    • Centre d Investigation Clinique Inserm and CHU Nancy
    • Chairman, CardioVascular Clinical Trialists Forum, France


  • Dr Kaoru Ito: How to utilise genomic information to realise the precision medicine for cardiovascular disease
    • Team Leader, Laboratory for Cardovascular Genomics and Informatics, RIKEN Center for Integrative Medical Sciences, Japan
  • Prof Seung-Pyo Lee: Multimodal data to discover novel high-risk group patients
    • Professor, Department of Internal Medicine (Cardiology), Seoul National University Hospital, South Korea
  • Assoc Prof Doreen Tan: CYP2C19 testing: the chicken or the egg?
    • Cardiology Specialist Pharmacist
    • Associate Professor, Department of Pharmacy, National University of Singapore


  • Prof Koji Hasegawa
    • Director of Translational Research, National Hospital Organisation Kyoto Medical Centre, Japan
  • Prof Mei-Ling Cheng
    • Director of Metabolomics Care Laboratory, Chang Gung University, Taoyuan City, Taiwan
  • Prof Roger Foo
    • Director, NUHS Cardiovascular Disease Translational Research Programme
    • Deputy Director, Cardiovascular Research Institute (CVRI)
    • Senior Consultant, National University Heart Centre, Singapore


  • Understanding of genetic, genomics, and proteomics (‘omics’) involved in pathophysiologic mechanisms of CVS diseases
  • Understand potential applications of genomic approaches to facilitate risk stratification and intervention to prevent or treat CVS diseases in clinical setting

How to utilise genomic information to realise the precision medicine for cardiovascular disease, by Dr Kaoru Ito

Genetic Risk Score as a disease biomaker

Genetic risk is a risk factors that we are constantly being exposed to and can be accumulated. It must be useful to be able to quantify it. CVS disease and genetic risk:

Monogenic disorder

  • One variant can be a biomarker for diagnosis
  • Example: channelopathies, cardiomyopathies, familial hypercholesterolemia

Polygenic disorder

  • Multiple variants are taken into account
  • Genetic Risk Scores (GRS)
  • Examples: CAD, AF

GWAS and GRS calculation

GWAS = genome-wide association study (GWAS)

  • Risk allele
  • P-value
  • Effect size (can be transformed into odds ratio)

GRS and odds ratio for disease development

  • Predictive performance of individual variants is low in polygenic disorders, hence need GRS
  • GRS is the sum of multiple variants with a specific weights

GRS and disease prevalence

  • GRS indicates accumulation of genetic risk for a disease
  • Low score = low risk
  • High score= high risk
  • Khera AV et al. Nat Genet 2018; 50:1219-1224
  • AF, T2DM, breast cancer: prevalence increase exponentially with increase in GRS

GRS in CVS disease

Application of GRS:

  • Disease risk prediction
  • Disease diagnosis
  • Disease prognosis
  • Therapeutic

In CVS (polygenic)

  • CAD and AF: common diseases (polygenic disorders)
  • CAD: with metaGRS, give good prediction of CAD. Can use GRS to predict disease. Also can be used as a therapeutic indicator (MI-GENES clinical trial)

In CVS (monogenic)

  • GRS not for just common diseases, but can also use for monogenic disorders
  • Exampels: DCM, HCM
  • GWAS for cardiac parameters
  • GRS for risk stratification in DCM (2020)
  • GRS for risk stratification in HCM (2021): successfully predicted clinical events

Application for East Asians

Population specificity in GRS – GRS performs better in similar group of patients

GWAS mainly for Europeans countries, non-Europeans under-resourced. Can we use the GWAS data from European countries for East Asians?

  • Trans-ancestry metaGWAS can improve GRS
  • GRS derived from Trans-ancestry metaGWAS shows better performance than using either GWAS alone –> results show can use trans-ancestry

CAD-GRS predicts CVS mortality


CAD-GRS papers: no additive effects of CAD-GRS to clinical risk scores

Additive effect of CAD-PRS to clinical risk score only observed in younger age group (<55yo)

  • GRS derived from GWAS has a potential to aid 4 points
  • Inconsistency in the reported GRS results draws attention to reproducibility of GRS
  • More efforts to improve the performance and robustness of GRS
  • Need to establish… (prospective cohort studies and RCTs)

Discussion and Q&A

Prof Roger Foo: At the end of the journey, to have the genome to stratify the populations and decide on treatment algorithm

Dr Ito: population specificity in GRS, we should prepare population specific GWAS result, before we can boost the power of GRS, using European GWAS results. Also because of population specificity, if we use European GWAS to developing GRS of non-European population, not as strong as developing own population GWAS

GRS is a continuous variable, difficult to define a cut-off value. But consider the nature of GRS, top 10% has high odds ratio compared to the remaining group. We need further RCTs to determine cut-off value. Not a binary.

Prof Lee: A piece of data to share, ethnicity specific for GRS. GRS develop in Koreans does not work well in UK biobank.

Specific cut-off? younger population can benefit more than older population, as the older population will have gotten more effects from the environment. Our data points towards better performance for patients under 50yo, while it loses its predictive value over 60yo.

Prof Jack: any GRS to be used in Asians?

Prof Lee: Korea no.

Prof Zannad: Issue of GRS and biomakers in general, is to make them to be actionable. We are still very much far away from it. The reproducibility is the real issue and extremely important. We have many papers on the genetic markers but never have been able to be reproduced. Including bio-assays reproducibility. Weight of genetics is really important as well.

Prof Jack: we need to work harder in Asia to generate the data that will be specific and relevant tot he Asian populations

Multimodal data to discover novel high-risk group patients, by Prof Seung-Pyo Lee

There is a need for precision medicine worldwide, due to the increasing cost of medicine.

Different cooking (analytics): How we can use our electronic health records to better understand the outcomes of these patients?

PCA = principal component analysis

  • To aggregate all of these data to explain some trends
  • Cluster the patients into the different variables
  • Then we can use these different clusters and check mortality data, resulting in different mortality patterns
  • A step towards accurate outcome prediction

Kwak S, Won S, Lee SP et al. Circ Cardiovasc imaging 2020

How to use phenotyping data to predict outcome a little bit more accurately. (on top of the current standards and methods)

Otto CM & Nishimura RA et al. Circulation 2021: most recent AS guidelines

Random forest (a type of machine learning) to segregate these patients to those who did not do well and those who did. What was striking was that after ranking all these data, the conventional parameters (e.g. LVEF, gender) are not of high importance. In contrast, certain parameters turned out to be important (e.g. ECV%).

This random forest is also good to find out the threshold, by doing a partial plot to plot the relationship between mortality and the parameters. With these threshold, we can separate out those with low risk and those with high-risk.

Simple AS-CMR score for post-AVR mortality

Clinical data to pinpoint AS for high risk group – using electronic health records and analyse them in a different way. We can use these insights from this data to pinpoint high risk groups to determine who needs an aortic valve replacement early.

Different ‘materials’ (data)

Search of novel high risk groups for common disease

Nordestgaard BG and Chapman MJ et al. Eur Heart J 2013

Khera AV and Kathiresan S et al. Nat Genetic 2018

High genetic risk: with lifestyle changes, we can get half reduction in 10-year coronary event rate. This also translates to: if we can know what are the genetic risk, we can offer the right advice to patients.

PRS for genetic prediction of metabolic syndrome (a high risk for CVS disease), hence we want to make a PRS to predict for MS. Construction of PRS for MS using the different phenotypes such as BP, cholesterol, diabetes

Novel High Risk Group to predict ASCVD: PRS-ASCVD vs Framingham score. Genetic predictor is better for younger population. But once an individual reaches 60yo, we lose the significance. But those with high PRS-ASCVD and Framingham risk scores, it redicted the CVS mortality similarly.

Discussion and Q&A

Prof Koji: aortic stenosis (AS) – in near future, situation will change due to the introduction of the TAVIR or TAVI.

Prof Lee: only minority group of patients on TAVIR as they are only introduced in 2011.

Prof Koji: genetic risk score for ASCVD. Cost-benefit and reproducibility. Sequencing is becoming cheaper to do and easily available. it is becoming closer to us. For reproducibility, if we cut off the 10% high-risk, and the younger age, we can see a high level of reproducibility, and should be significant data. However we still need prospective studies (take many years). Good that we can begin to incorporate into clinical practice, especially for high-risk and young patients.

Prof Lee: Issue is how to construct the GRS. This opens the forum to collaborate between Asian countries to construct the GRS.

Prof Jack: How should clinicians approach this in the future?

Prof Cheng: top-ranking is age. Co-morbidities and GRS are also important. What should be the cut-off for the age and co-morbidities? We cannot change the genes and age. How to prevent?

Prof Lee: Age cut-off at around 70-75yo (when we plot against mortality), to consider for aortic valve replacement. As for GRS, difficult to explain, better to use PRS

CYP2C19 testing: the chicken or the egg, by Assoc Prof Doreen Tan

Fact #1 ? – RCTs support the use of ticagrelor or prasugrel over clopidogrel for patients undergoing PCI?

TRITON-TIMI 38 and PLATO trials, hence we want to use ticagrelor –> also see American and European guidelines

Context 1.1: ticagrelor and prasugrel lead to more bleeding

  • PRASFIT-ACS: a Japanese study, using 1/3 dose of prasugrel: Lowered ischemic numerically; same bleeding
  • PHILO study: lowered ischemic and bleeding numerically
  • TICO-KOREA: safety endpoint study; ticagrelor has more bleeding

Asian guidelines: does not recommend ticagrelor as much as Western guidelines

Korean guidelines: more strongly recommend ticagrelor

Context 1.2: Real-World studies

Summary: fact #1 is true, but RCT results are not always reproducible in the real world. Guidelines do not cite database studies and only results from RCTs and meta-analyses.

Fact #2: Genotype guided selection of anti-platelets has shown better ischemic and bleeding outcomes than standard of care?

What is CYP2C19?

  • Clopidogrel need 2C19 to convert to active
  • LOF of this gene will result in losing effectiveness of clopidogrel
  • GOF of this gene will result in higher chance of bleeding
  • Look at CPIC guidelines


Evidence pyramid: RCTs are the basis to build upon


  • When experiencing a LOF for CYP2C19, we may get better outcomes if we use ticagrelor.
  • If there is no LOF of the CYP2C19 gene, clopidogrel is likely to get the same results as ticagrelor or prasugrel
  • Prevalence of CYP2c19 in various ethnic groups: always at least 50% and above in Asians
  • We need to think about patient characteristics: not all patients will be in RCTs

Fact #3: why RCTs findings may not be reproducible in real world?

  • External validity is limited, females, asians always smaller proportions
  • RCTs is one singular intervention, but in the real world, patients are complicated.

Potential: we can use aggregate information from multiple trials and registries, electronic health records and large digital data warehouses


Prof Jack: how should trials evolve over time?

Prof Zannad: RCTs are reductionist and not global. There are issues with globalisation and diversity, and ethnically diverse groups are usually not well represented as well. There is a need to call for action to all trials to get better ethnic representation. Huge gap in need for training in some countries.

Share via

Share on email
Share on facebook
Share on whatsapp
Share on telegram
Share on twitter
Share on linkedin

Also worth reading