Characterizing the Field of Mendelian Randomization Studies for Cardiovascular Disease

Cardiovascular disease (CVD) is a leading cause of morbidity and mortality worldwide [1]. Much of our understanding of CVD comes from large-scale observational cohort studies like the Framingham Heart Study [2]. These studies have revealed correlations between lifestyle factors (e.g., smoking, diet, or physical inactivity), clinical measurements (e.g., blood pressure or cholesterol levels), and subsequent development of heart disease and stroke that guide current clinical practice. However, observational studies pose certain limitations to establish causality. First, unobserved variables can lead to confounding in the association signal. For example, individuals who smoke may differ from nonsmokers in ways that go beyond nicotine consumption, such as diet or stress levels. This means the observed association between smoking and CVD, though likely driven in part by the direct effects of tobacco, may be complicated by broader lifestyle and socioeconomic factors. And second, certain processes of CVD pathophysiology can influence lifestyle and/or clinical measurements leading to reverse causation. For instance, early symptoms or diagnoses might alter an individual’s behaviors, leading to erroneous inferences about which came first. While creative study designs (e.g., prospective cohorts) and statistical techniques (e.g., multivariable regression, propensity-score matching) can mitigate these limitations, they cannot fully eliminate the challenges of non-randomized data. Consequently, observational studies cannot determine the causal status of risk factors. This gap has incentivized the development of alternative methodologies that can inform causality.Over the last 2 decades, genome-wide association studies have revealed hundreds of genetic associations with lifestyle factors and clinical measurements. Mendelian randomization (MR) capitalizes on these associations to avoid non-random effects in association studies. More precisely, relying on the fact that genetic variants are inherited randomly at conception it uses single-nucleotide polymorphisms (SNPs), or sets of SNPs, as instrumental variables for exposures [3]. For example, if a particular SNP robustly associates with low-density lipoprotein (LDL) cholesterol, researchers can treat this genotype-based risk factor proxy much like a randomized treatment assignment. Because these genotypes are fixed before birth and generally unaffected by confounding factors, MR analyses more convincingly test whether a change in an exposure causally affects the outcome. If genetically elevated LDL correlates with increased CVD risk, beyond confounding or reverse causation, this strongly implies that changes in LDL cholesterol causally affect disease incidence. Overall, MR has emerged as a crucial tool to study CVD etiology.Given the potential of MR studies, there has been significant interest in the research community to evaluate the causal relationship of lifestyle factors and clinical measurements in CVD risk. Guo et al. [4] presents a sweeping overview of how researchers have employed MR techniques in CVD contexts over the last 2 decades. Such bibliometric and visualization studies are a valuable tool for fellow researchers to understand the shape and needs of a rapidly expanding field such as MR for CVD. For this, the authors performed a comprehensive literature search in the Web of Science Core Collection (WoSCC), selecting English-language articles published between 2003 and 2024 that included both “mendelian randomization” and “cardiovascular disease” as relevant search terms. This starting point yielded 1,870 articles. The authors then analyzed these references using specialized bibliometric software packages such as VOSviewer and CiteSpace. VOSviewer enabled mapping of co-authorship networks and keyword co-occurrences, while CiteSpace facilitated the detection of papers or topics that gained citation momentum over specific intervals. Importantly, the authors recorded how many articles each institution produced, which countries housed the most prolific research groups, and the key themes or risk factors commonly studied in MR-based CVD analyses. By synthesizing these data, they developed detailed visualizations, such as cluster diagrams, heat maps, and timeline plots, to illustrate changes in publication rates, topic emphasis, and inter-laboratory connections. Their analyses revealed a steady rise in publication counts, particularly in the second half of the 2010s onward. This surge coincides with the broader availability of large-scale genetic data, including the UK Biobank, and the refinement of MR analytical methods that address pitfalls like pleiotropy. Geographically, the USA, UK, and China were identified as leaders in this domain, both in terms of absolute publication numbers and robust collaborative ties. Networks revealed that institutions such as the University of Bristol, the University of Cambridge, and the University of Copenhagen often appear at the core of multiple collaborations, reflecting their longstanding commitments to genetic epidemiology. A notable pattern emerged in the keyword cluster analysis: earlier MR papers in CVD frequently focused on classical biomarkers, for example, HDL or LDL cholesterol. Over time, investigators expanded their focus to incorporate other metabolic and inflammatory traits (e.g., C-reactive protein, interleukin-6, or obesity measures), as well as exploring how genetic instruments for lifestyle factors (e.g., smoking behaviors, alcohol consumption, or education level) can impact CVD risk.The author’s work presents a significant contribution to our broad understanding of the MR CVD domain. However, certain limitations of bibliometric studies are not discussed and should be acknowledged. For instance, bibliometric data (e.g., citation counts) can reflect popularity or controversy, not just significance of findings and methodological rigor. Hence, highly cited work may be widely debated or contain findings that are later refined or disputed. Similarly, meta-data in bibliometric studies are limited. This hinders the author’s ability to evaluate research trends in relevant topics such as population-specific MR studies.MR has enabled significant contributions to our understanding of CVD etiology, particularly causal relationships of traditional CVD risk factors such as blood lipids [5], obesity [6], or smoking [7]. Importantly, the results from Guo and colleagues[4] highlight a growing interest in evaluating the role of nontraditional CVD risk factors in disease using MR. Further efforts in this area can reveal novel pathophysiological pathways of CVD risk to inform preventive strategies or therapeutic targets. Moreover, the documentation of research hotspots and collaborative networks presented in Guo et al. [4] can guide new researchers in forging effective partnerships or training opportunities.As scientific interest in MR continues to expand, we can anticipate ongoing refinements and a deepening integration of genetic methods, ultimately improving how we combat the global burden of CVD. The study by Guo and colleagues provides a temporal description of the field, showing topics with increased research interest growth and clusters of highly cited researchers in the field.The author has no conflicts of interest to declare.Ben Omega Petrazzini is supported by the Clarendon Fund Scholarship and Uruguay’s National Agency of Research and Innovation.Ben Omega Petrazzini was responsible for drafting the original version of the manuscript and for revising and editing it, incorporating feedback, and making necessary improvement.