Miscues on the “lack of MEF2A mutations” in coronary artery disease

We would like to comment on the report by Weng et al. that claimed lack of MEF2A mutations in coronary artery disease (CAD) (1) and the accompanying commentary by Altshuler and Hirschhorn (2). They analyzed the exons and exon-intron boundaries of the MEF2A gene in 300 CAD patients and 300 elderly subjects by direct sequence analysis.

First, the new report actually found 1 patient with CAD who had a mutation (S360L) that is located in the critical transcriptional activation domain (3) and may affect the transactivation function of MEF2A. The mutation was not found in any control subject. Rather than following up its assessment experimentally, the investigators analyzed S360L quickly with computer software and claimed it to be benign, but computational analysis does not have any power to distinguish whether a variant is a disease-causing mutation or a polymorphism. The bias is quite apparent, and the title of the report, “Lack of MEF2A mutations in coronary artery disease,” is inaccurate and inappropriate. When the data from the study by Weng et al. is taken together with data from our previous report of MEF2A mutations in sporadic cases of CAD (4), there are 5 of 507 individuals with CAD with MEF2A mutations and 0 of 491 controls, which suggests a significant association between MEF2A mutations and CAD (P = 0.03, estimated by conservative Fisher exact test). The statistical analysis by Altshuler and Hirschhorn appears to be incorrect and the significance was inaccurately reported as P > 0.2 (2).

Second, the phenotyping work in this new report for categorization of CAD is grossly inadequate. The investigators used resting echocardiograms in the 5 subjects they classified as “controls” with the 21-bp deletion. This test does not provide any insight regarding the presence of significant coronary atherosclerosis. Coronary artery disease is frequently asymptomatic, and its diagnosis relies on performance of a coronary angiogram. In all of our studies, we have used the angiographic evidence of disease as critical criteria for classification as a control (4–6). Furthermore, it is impossible to classify young individuals as unaffected controls, and this was done in the case of several individuals, including the 45-year-old controls of families 2 and 3 and the 37-year-old and 42-year-old individuals in family 1 (III:5 and III:6). Virtually all of the “controls” in the report by Weng et al. (1) should have been properly defined as “uncertain phenotype” with respect to the presence or absence of CAD. It is quite unreasonable for any conclusion regarding cosegregation of the 21-bp deletion of MEF2A to be based on such poor phenotypic characterization.

Of note, the proband of family 1 had a transient ischemic attack (1), which raises the issue of atherosclerotic vascular disease, and only 2 other individuals in the family were old enough to be properly categorized — only 1 of these had an exercise stress test, and none had a coronary angiogram. Two individuals from the family without the 21-bp deletion are affected with CAD, but they may represent phenocopies, as CAD is very common (both parents of 1 affected individual, III:1, are normal, and the father of another affected individual, III:3, had a stroke and died at 46 years of age), and the disease can derive from the married-in spouses. In sum, careful analysis of family data suggests that the available clinical data are insufficient to make any conclusions and certainly does not support the author’s conclusion that the 21-bp deletion does not cosegregate with the disease and does not cause CAD.

Third, the authors mistakenly suggest that our 21-bp deletion in MEF2A was associated with premature CAD and myocardial infarction (MI) (5). This was not at all the case (5). The age of onset for the 21-bp deletion in the Wang et al. report ranges from 35 to 68 years for CAD and 40 to 80 years for MI (5). Reduced penetrance is well known for many human diseases with autosomal dominant inheritance patterns including cardiovascular diseases such as Brugada syndrome and long QT syndrome. The phenotypic expression of a mutation varies in different families as well as in different members of the same family. For example, 45% of mutation carriers can be silent carriers, and incomplete penetrance in families can be as low as 12.5% for Brugada syndrome (7). The molecular mechanisms for reduced or incomplete penetrance are not clear. Modifier genes, environmental factors, gene-gene interactions, and gene-environment interactions may play a role. For a complex disease such as CAD with involvement of numerous environmental factors, genetic factors, and interactions among these factors, incomplete penetrance should not be unexpected and may explain why the 4 individuals over 60 years of age and with the 21-bp deletion in the new report have not developed MI yet. The heterogeneity of European-American white individuals (who are known to be highly heterogeneous) may also be a contributing factor.

It is interesting that the 21-bp deletion is present in 0.15% of elderly subjects (1). The action of this deletion may be analogous to that of a variant of the cardiac sodium channel gene SCN5A, S1103Y, which is more prevalent in African Americans (13.2%) than in other populations (8). S1103Y cosegregates with cardiac arrhythmia in some families and is associated with arrhythmia in the African American population; but many carriers do not develop arrhythmias (8, 9).

Identification of mutations of MEF2A in CAD patients, but not in controls, is consistent with the hypothesis that MEF2A mutations are causative. In another 200 CAD patients, we have identified another potential mutation (T215A) (unpublished observations). Continued identification of MEF2A mutations in CAD population by other independent research groups and studies with knockout/knockin mice with MEF2A mutations will provide further supportive evidence that MEF2A is a disease-causing gene for CAD and reveal how MEF2A defects lead to atherosclerosis.

In conclusion, identification of the 21-bp deletion of MEF2A, which cosegregates with CAD, in 1 large family (5), identification of multiple mutations in other CAD patients and families (1, 4), and functional data demonstrating the deleterious effect of mutations on the function of the MEF2A protein (4, 5) provide evidence that MEF2A is a disease-causing gene for CAD. The new report by Weng et al. (1) does not refute our findings. It advances the field by helping to establish the incidence of the 21-bp MEF2A deletion and pointing out the possibility of incomplete penetrance. But without proper phenotyping work and experimental biological assessment, it is a misleading report that unfortunately suggests a negative bias and premature dismissal of an important biologic underpinning of CAD.