The power of large cohorts to improve cardiovascular risk prediction and prevention

This editorial refers to ‘Biomarker-based prediction of fatal and non-fatal cardiovascular outcomes in individuals with diabetes mellitus’, by P.M. Haller et al., https://doi.org/10.1093/eurjpc/zwad122. The importance of preventive cardiology is gaining attention in view of the still very high number of patients suffering from cardiovascular disease (CVD).1 An individual’s risk for CVD is routinely assessed by a variety of risk factors, which are combined into risk scores (e.g. the SCORE2 by the European Society of Cardiology).2,3 Recently identified biomarkers are related to the actual underlying pathology, which include, for example, (chronic) myocardial damage as well as inflammation. Hence, some examples of novel biomarkers are high-sensitivity cardiac troponin (hs-cTn), natriuretic peptides (NTproBNP), and high-sensitivity C-reactive protein (hs-CRP). These risk factors have the potential to further individualize preventive and therapeutic measures to reduce CVD risk. While the SCORE2 includes parameters like age, sex, lipids, systolic blood pressure, and smoking status, metabolic disorders also significantly increase CVD risk. Individuals with diabetes have a significantly greater risk for adverse CVD events compared to those without this condition.4,5 The importance of diabetes for preventive cardiology is underscored by the fact that this disorder may not be detected until CVD is already present.6 In addition, pre-diabetes, a condition much more prevalent than established diabetes, is already related with adverse cardiac remodelling thereby increasing CVD risk.7 Whether hs-cTn, NTproBNP, and hs-CRP perform equally well predicting CVD risk in individuals with and without diabetes is currently unclear. In a very comprehensive and elegant analysis, Haller et al. aimed to answer this clinically very relevant question in the current issue of the European Journal of Preventive Cardiology.8 For their analysis, the authors used harmonized data from the Biomarker for Cardiovascular Risk Assessment across Europe (BiomarCaRE) and MOnica Risk, Genetics, Archiving and Monograph (MORGAM) consortium. The primary endpoint of their analysis was a composite of fatal and non-fatal cardiovascular events (including composite fatal and non-fatal definite or possible myocardial infarction, unstable angina pectoris, unclassifiable death, cardiac revascularization, or fatal as well as non-fatal likely cerebral infarction during follow-up). Adjusted Cox regression models stratified by diabetes status were used to assess the relation between these biomarkers and the primary endpoint, with follow-up limited to 15 years. The models were compared using optimism-corrected C-indices and likelihood ratio test for nested models. In a second approach, penalized cubic splines were used to plot the relationship between the hs-cTn, NTproBNP, and hs-CRP with risk for CVD over time. The final analysis sample included 95 292 individuals. A total of 6090 of these individuals had diabetes. The authors reported that the presence of diabetes increased the risk for CVD by 2.5-fold. Importantly, individuals with diabetes had significantly higher concentrations of hs-cTn, NTproBNP, and hs-CRP compared to individuals without diabetes. Using nested models, Haller et al. showed that hs-cTn, NTproBNP, and hs-CRP were independently associated with the primary endpoint. Interestingly, using such a large cohort, the authors were able to show that the individual biomarker concentrations carried a similar risk for CVD independent of diabetes status. The importance of sex for preventive cardiology is increasingly recognized. In a very relevant sensitivity analysis, Haller et al. report that the risk from diabetes for fatal and non-fatal cardiovascular events and for fatal and non-fatal CAD-related events was higher in women compared with men. Nonetheless, there was no difference between sexes in terms of all-cause mortality. The authors also provide information on the association of hs-cTn, NTproBNP, and hs-CRP with the primary endpoint. While the general direction of the estimates was not influenced by sex, a significant interaction was reported for NTproBNP with the primary endpoint and hs-cTn with all-cause mortality. The observations are of great interest to explore sex-specific mechanisms of CVD. Using data from a large and comprehensively very well phenotyped study population of the general population, Haller et al. demonstrated that diabetes is an important risk factor for CVD. Furthermore, the underlying pathophysiology of CVD, which can be assessed using hs-cTn, NTproBNP, and hs-CRP, contributes to the risk for fatal and non-fatal CVD risk independent of diabetes status. The authors provide a very good overview of the biological mechanisms underlying these risk factors. With regard to the future of preventive cardiology in clinical practice, the presented manuscript highlights the importance to not just assess the classical risk factors but also the presence of diabetes. One can only speculate on whether hs-cTn, NTproBNP, and hs-CRP add significant predictive value when added to the SCORE2 parameters. The authors also acknowledge that there are currently no guidelines for the use of these (pathophysiological) parameters for risk prediction in clinical practice. However, one needs t to keep in mind that the current guidelines for CVD risk assessment apply throughout Europe (and possibly beyond), which includes countries with a high degree of heterogeneity in socioeconomic status and health care providers. The importance of the assessed risk factors in individuals with diabetes is that they experience the CVD event significantly earlier in life compared to study participants without diabetes. Hence, strategies to implement cardiovascular disease prevention need to screen patients with diabetes in greater detail to assess CVD risk and employ (non)pharmacological therapies. We have previously outlined the molecular mechanisms underlying the effect of exercise training to reduce CVD risk in patients with diabetes.4,9 There are some limitations of the presented work that are acknowledged by the authors. These include, for example, the lack of information with regard to the duration of the diabetes or the age of onset. In addition, no information is presented on pharmacological therapies to treat diabetes and CVD risk factors. The authors also recognize that by design, cross-sectional studies cannot be used to imply causality. Nonetheless, a major strength of the work by Haller et al. is the large sample size in combination with a clinically very relevant endpoint which provides potentially generalizable results for the European population. ‘Biomarker-based prediction of fatal and non-fatal cardiovascular outcomes in individuals with diabetes mellitus’ is a prime example of how large cohorts of the general population may be used to address very relevant clinical issues in preventive cardiology. Haller et al. demonstrate that individuals with diabetes need to be monitored very closely for their CVD risk. Moreover, in this high-risk population, novel pathophysiological biomarkers, which are surrogates for cardiomyocyte stretch and inflammation, may improve risk prediction. Whether these parameters may be used to individualize preventive treatment needs to be evaluated by future clinical trials.