Author + information
- A.J. Marian, MD∗ ()
- Center for Cardiovascular Genetics, Brown Foundation Institute of Molecular Medicine, The University of Texas Health Science Center and Texas Heart Institute, Houston, Texas
- ↵∗Reprint requests and correspondence:
Dr. Ali J. Marian, Center for Cardiovascular Genetics, The Brown Foundation Institute of Molecular Medicine, The University of Texas Health Sciences Center, 6770 Bertner Street, Suite C900A, Houston, Texas 77030.
“If epidemiologists are compared with fishermen, causality is the big fish. It is elusive to find, difficult to catch, and claims to have measured it are often exaggerated.” Burgess et al. (1) were correct: establishing a cause-and-effect relationship is always challenging. In its purest definition, analogous elements of the Koch’s postulates of causality have to be fulfilled (2). In its simplistic definition, the cause has to precede the effect and be necessary and sufficient for phenotype expression, albeit with a variable degree of expressivity.
Yet, these extreme definitions of causality are seldom germane in genetics. Very few genetic variants are fully penetrant, independent of the genetic backgrounds in which they operate (3,4). Fully penetrant genetic variants are rare in the population, as well as in each genome, and typically cause familial diseases that exhibit Mendelian patterns of inheritance. Strength of evidence for their causality depends not only on the penetrance but also family size and structure (number of informative meiosis). Typically, a genetic variant exhibits incomplete penetrance and the family size is not large enough to ascertain cosegregation unambiguously. Therefore, even in single gene disorders that exhibit Mendelian patterns of inheritance, definite identification of the causal gene or variant remains difficult. The advent of massively parallel nucleic acid sequencing technologies and the resultant availability of large databases of human genetic variants have illustrated overabundance of the sequence variants in each genome, including in genes previously implicated in human diseases (5,6). The plethora of genetic variants renders unequivocal identification of the true disease-causing variant, even in a single gene disorder, a formidable task.
On the other end of the genetic causality spectrum, the vast majority of genetic variants that exhibit low penetrance and exert negligible or modest effect sizes (2,7). Expectedly, such variants do not exhibit Mendelian patterns of inheritance but might show aggregation in families or cases with the phenotype of interest. Evidence of their role in susceptibility to disease typically originates from genetic epidemiological studies, including genome-wide association studies (GWAS), showing an excess burden of such variants in cases as compared to controls. During the last decade or so, GWAS and candidate gene studies have identified a very large number of common variants, defined as a population minor allele frequency of >1%, associated with cardiovascular phenotypes. As of February 20, 2015, the last update of the GWAS catalog, 15,396 single nucleotide polymorphisms (SNPs) are described in association with various complex traits. Pertinent to this editorial is the association of serum uric acid levels with more than 40 SNPs in at least 30 genes.
Clinically, an association between serum uric acid levels and gout as well as hypertension, type II diabetes mellitus (T2DM), kidney disease, coronary heart disease (CHD), and heart failure (HF) has been recognized and substantiated in numerous modern epidemiological studies (8). Consequently, serum uric acid is considered a risk factor for cardiovascular and metabolic diseases and their clinical outcomes (8).
Conventional epidemiological studies are primarily observational and subject to confounding effects of multiple factors, as well as potential biases in study design and reverse causation. Consequently, the findings of epidemiological data, whether conventional or GWAS, are vulnerable to confounders and seriously limited in inferring causality.
Considering the shortcomings of the conventional epidemiological studies, in this issue of the Journal, Keenan et al. (9) sought to determine the causal role of serum uric acid in CHD, T2DM, HF, and ischemic stroke. To reduce effects of potential confounders, the authors used genetic variants associated with serum uric acid levels as instrumental variables. The rationale for using genotypes as instrumental variables, referred to as Mendelian randomization (MR) (10), is that genetic variants are assorted randomly during meiosis, except for variants that are in close genetic proximity (linkage disequilibrium). Therefore, confounders are to be distributed randomly among 3 genotypes of each SNP.
Accordingly, Keenan et al. (9) analyzed the causal role of 28 SNPs individually and 14 SNPs exclusively associated with serum uric acid levels (nonpleiotropic), collectively, as a genetic risk score (GRS) in cardiometabolic syndromes. To support validity of the approach, they showed that genetically determined increased serum uric level was associated with increased risk of gout, a well-established phenotypic consequence of hyperuricemia. However, SNPs were neither individually nor collectively as a GRS associated with T2DM, CHD, ischemic stroke, or HF. Hence, the authors speculate that lowering serum uric acid levels is not expected to improve risk of cardiometabolic syndromes.
This well-designed and meticulously performed study benefits from a large study population sample size, albeit comprised of ethnically mixed populations with considerable cultural and environmental differences. The findings show that a modest shift in serum uric acid levels is sufficient to increase risk of gout but not risk of CHD, T2DM, ischemic stroke, or HF. The rs12498742 SNP at the SLC2A9 locus had the largest effect size on serum uric acid levels, which was only 0.37 mg/dl, and associated with risk of gout but, once again, not cardiovascular nor metabolic syndromes. The small effect size of the genetic variants, typical in genetic studies of complex traits (11), renders genotypes as weak instrumental variables. The 14-SNP GRS, however, shifted serum uric acid levels by 1.4 mg/dl. Despite the relatively larger effect size compared to individual SNPs, GRS was associated with gout but not the selected cardiometabolic phenotypes.
GRS, while benefiting from a larger effect size, is based on numerous assumptions not supported by biological data. GRSs are particularly speculative when derived from a diverse group of genes with multifarious functions and in genetically and environmentally heterogeneous populations. Unfortunately, data on serum uric acid levels and the clinical phenotypes were not consistently available in the present study populations to assess whether serum uric levels, independent of the genetic variants, were associated with the clinical phenotypes.
The present study’s null results regarding serum uric acid and CHD echo the findings of a recently published large-scale MR study in a homogeneous Danish population (12). Yet another MR study, performed in a population of European origin, with a study design similar to the present one, concluded that each 1 mg/dl increase in genetically predicted uric acid concentration was causally associated with cardiovascular death and sudden cardiac death (13).
To quote Johann Wolfgang von Goethe, “It's in the anomalies that nature reveals its secrets”(14). A notable example of nature’s anomalies on uric acid metabolism is Lesch-Nyhan syndrome, an X-linked disease due to mutations in HPRT (hypoxanthine phosphoribosyltransferase 1) gene. It manifests with severe hyperuricemia since birth, gout, renal stones, and neurological impairment but not CHD, T2DM, HF, or ischemic stroke. Perhaps, despite lifelong exposure, the relatively short lifespan of patients with Lesch-Nyhan syndrome masks expression of CHD. A few other rare forms of single gene disorders also cause severe hyperuricemia but do not express as these cardiovascular diseases. Collectively, these rare anomalies of nature do not support a causal role for uric acid in CHD or ischemic heart disease but there are scant data to make a firm conclusion.
Various randomized clinical trials are ongoing to test beneficial effects of lowering serum uric acid levels on various cardiovascular phenotypes, with negative results thus far (15). A beneficial effect, if observed, would not necessarily indicate a causal role, as the benefits might result from interventions modulating various biological effects of uric acid and the xanthine oxidase metabolic pathway (16), analogous to the beneficial effects of inhibiting the renin-angiotensin-aldosterone system in HF.
Causality in genetics is seldom deterministic, as it is in rare large families with single gene disorders. It is commonly probabilistic, as in complex traits. Because hyperuricemia does not meet either the strict Koch’s postulates or simple definition of causality, it is not a deterministic cause of the aforementioned cardiovascular phenotypes. However, uric acid and the metabolic pathway that generates it are involved in a diverse array of biological functions (15), possibly contributing to pathogenesis of cardiovascular and metabolic phenotypes, rendering it pathogenic but not causal. The results of ongoing randomized clinical trials might shed some light on serum uric acid’s pathogenic role in cardiovascular and metabolic syndromes.
↵∗ Editorials published in the Journal of the American College of Cardiology reflect the views of the authors and do not necessarily represent the views of JACC or the American College of Cardiology.
Dr. Marian is supported in part by grants from the National Institutes of Health, National Heart, Lung, and Blood Institute (NHLBI, R01 HL088498, and R34 HL105563), Leducq Foundation (14 CVD 03), Roderick MacDonald Foundation (13RDM005), TexGen Fund from Greater Houston Community Foundation and George and Mary Josephine Hamman Foundation.
- 2016 American College of Cardiology Foundation
- Burgess S.,
- Butterworth A.,
- Malarstig A.,
- Thompson S.G.
- Marian A.J.
- Marian A.J.,
- Belmont J.
- Marian A.J.
- Keenan T.,
- Zhao W.,
- Rasheed A.,
- et al.
- Palmer T.M.,
- Nordestgaard B.G.,
- Benn M.,
- et al.
- Kleber M.E.,
- Delgado G.,
- Grammer T.B.,
- et al.
- ↵Goethe JW. Johan Wolfgang von Goethe quotes. Available at: http://www.bestquotes4ever.com/quote/it-s-in-the-65771. Accessed December 4, 2015.
- Givertz M.M.,
- Anstrom K.J.,
- Redfield M.M.,
- et al.