A stitch in time: easily measured biomarkers to enhance cardiovascular risk prediction in primary prevention

This commentary refers to ‘Low-density lipoprotein cholesterol, C-reactive protein, and lipoprotein(a) universal one-time screening in primary prevention: the EPIC-Norfolk study’, by J.M. Kraaijenhof et al., https://doi.org/10.1093/eurheartj/ehaf209 and the discussion piece ‘Biomarkers predict cardiovascular outcomes in European cohort’, by H. Xing and Y. Su, https://doi.org/10.1093/eurheartj/ehaf486. Current practice in assessing atherosclerotic cardiovascular (ASCVD) risk and addressing the associated modifiable risk factors continues to fall short.1 A fundamental issue with our current guidelines based on 10-year risk models is that they miss a core principle, which is that for individual patients, knowledge early in life that they have a specific issue with a modifiable biomarker, whether it be low-density lipoprotein [LDL] cholesterol, high-sensitivity C-reactive protein [hsCRP], and lipoprotein(a) [Lp(a)], should serve as motivation for improved lifestyle change, diet, exercise, and smoking cessation much earlier in life than is currently done. In our recently published study, we focused on how these easily measured modifiable risk factors in primary prevention can help identify individuals at risk for ASCVD events later in life.2 While the concerns raised by Xing and colleagues are valid and important, we argue that their impact on our findings would be minimal. First, LDL cholesterol levels in our study were estimated using the Friedewald equation rather than obtained through direct measurement. Detailed reporting of triglyceride distributions or comparisons between LDL cholesterol estimation methods was outside the scope of our work. However, this limitation is unlikely to have materially affected our findings. Reassuringly, triglyceride levels were generally low, with fewer than 0.5% of participants exceeding the Friedewald threshold of 4.5 mmol/L. This is supported by the strikingly similar results observed in the Women’s Health Study (WHS), where LDL cholesterol was measured using a direct assay.3 Second, our statistical modelling approach followed established methodology and was substantiated by our full model, accounting for the other two biomarkers. Since Lp(a) levels are largely genetically determined and stable throughout life, and hsCRP levels show strong longitudinal consistency, comparable to blood pressure,4 intermittent fluctuations due to transient inflammation are typically not adjusted for. Any variability would bias results towards the null, suggesting that the observed associations in both WHS and EPIC-Norfolk would, if anything, underestimate the true impact of inflammation on long-term risk. Regarding potential confounding by anti-inflammatory treatment, findings from the WHS demonstrated that the initiation of statin therapy did not alter cardiovascular risk prediction, even across more than 30 years of follow-up transitioning from a pre-statin to a statin-treated era.3 Additionally, more detailed smoking status information would likely result in only modestly improved risk estimations compared with a simpler measure of current smoking status.5