Eight proteins that may serve as potential therapeutic targets for treating heart failure have been identified in a meta-analysis of population-based proteomic studies. This study may not only inform new heart failure treatments, wrote the authors of the analysis, but their methods can provide a roadmap for discovering drug targets in other diseases using proteomic and genomic data.
Heart failure is a growing cause of hospitalizations and deaths in the United States. In 2014, there were an estimated 80,000 deaths in the US from heart failure and another 230,000 deaths from heart failure with another co-morbidity. Moreover, a study published in the American Journal of Medicine in 2020 found that the mean predicted 10-year risk of heart failure increased from two percent to three percent of the population between 1999 and 2016. Despite being highly prevalent, the mechanisms of heart failure are incompletely understood.
To elucidate the causes of heart failure and reveal potential therapeutic targets, Thomas Lumbers from the University College London and collaborators used data from four population-based studies as part of the SCALLOP (Systematic and Combined AnaLysis of Olink Proteins) consortium in a meta-analysis to uncover therapeutic targets for heart failure. Through the examination of 90 cardiovascular proteins in the plasma of 3,019 participants (among whom there were 732 heart failure events), a total of 44 proteins were observationally associated with heart failure.
An additional dataset from a separate study of 30,000 individuals was used to identify 75 proteins with one or more cis-genetic instruments, and the overlap between the 44 and 75 proteins from the respective cohorts resulted in 40 heart-failure-associated proteins available for evaluation with the epidemiological technique of Mendelian randomization. A total of 120 combinations of instrument selection parameters were evaluated to improve the precision of the derived causal estimates.
Combining genetic variation with circulating protein levels for insights into pathophysiology
Mendelian randomization is a technique developed to utilize the wealth of available genetic information as a kind of natural randomized clinical trial. Genes are randomly assorted during meiosis; differences between the different parental alleles mimic a randomized clinical trial — with cases and controls replaced by a difference in the alleles distributed to the offspring. These differences will influence the level of the circulating protein in question, serving as a life-long exposure to the individual which can then be connected to the phenotype of interest (that is, the presence or absence of disease).
As an example, a genetic variant associated with higher LDL cholesterol levels that is also associated with a higher risk of coronary heart disease would provide supportive evidence for a causal effect of LDL cholesterol on coronary heart disease.
Each individual’s genome, with its millions of individual variants mapped into haplotype blocks, together with phenotypic data relating to the disease being studied, can then be combined with data from hundreds or thousands of other individuals. This wealth of genomic data and phenotype data enables a statistical analysis of particular genetic instruments in terms of the relative concentration of a collection of circulating plasma proteins. This provides a shortlist of proteins that are highly likely to be causative in the disease phenotype of interest, which then can be manipulated through pharmaceutical intervention to mitigate the disease in question.
Interest in using Mendelian randomization is growing rapidly. A proportional search for the term ‘Mendelian randomization’ via the tool PubMed by Year yields the following graph as a proxy for the popularity of this method, with some 1,288 uses in 2021 and 899 in 2020.
Identification of proteins with likely causality in heart failure
Using this technique, Lumbers and colleagues identified eight proteins with strong evidence of causality for heart failure: three risk factor proteins, CSF-1, Gal-3, and KIM-1; and five protective proteins, ADM, CHI3L1, CTSL1, FGF-23, and MMP-12. The ChEMBL public drug discovery database and a clinical trial registry were consulted for ongoing drug development and estimated druggability of the proteins involved. The only protein not rated for druggability is KIM-1, and the remaining seven targets are of interest for future drug development.
One protein target, fibroblast growth factor 23 (FGF-23) already has an on-market approved therapy for X-linked hypophosphatemia. In addition, colony-stimulating factor-1 (CSF-1) and matrix metallopeptidase-12 (MMP-12) are in early Phase I/II clinical trials for various disease indications including asthma, hypertension, and stomach neoplasms. Importantly the two protein targets adrenomedullin (ADM) and galactin-3 (Gal-3) provide confirmatory evidence for the development and evaluation of pharmaceuticals targeting these proteins that are currently in clinical trials for heart failure.
The authors concluded their paper by saying, “Proteomewide studies incorporating both direct association with target outcomes and genetic-based inference through [Mendelian randomization] are likely to provide important new tools for therapeutic target discovery and prioritization.”