When a disease is well-understood, medical science speaks of a “cause” of that disease. An example is rabies, which is caused by the rabies virus. When the pathogenesis of a disease is poorly understood, medical science instead speaks of “risk factors” for that disease. Approximately 300 risk factors for atherosclerosis have been described . This huge number can only mean one thing: despite bold statements that if everyone took enough statins atherosclerosis would disappear, mainstream medical science still does not know the cause of the number one killer on the planet. Recently, a new risk factor, trimethylamine-N-oxide (TMAO), has been described and proposed to be a cause of atherosclerosis.
In “Intestinal Microbial Metabolism of Phosphatidylcholine and Cardiovascular Risk” , published April 2013 in the New England of Journal of Medicine, Tang et al. showed that ingestion of eggs resulted in increased plasma and urinary TMAO levels. This increase was suppressed by antibiotics which decreased gut flora and was restored after antibiotics were stopped. The authors then showed that increased TMAO levels were associated with an increased risk of adverse cardiovascular events. Along with elevated TMAO levels, patients with adverse cardiovascular events were significantly older, had higher rates of hypertension, diabetes, and previous myocardial infarctions. Therefore, the investigators created a statistical model which showed that TMAO remained a significant predictor of major adverse cardiovascular events after adjustment for traditional risk factors and other baseline covariates. Using these data, the authors also created a “low-risk” subgroup in which TMAO levels still predicted adverse cardiovascular outcomes. In their discussion, the authors speculated that TMAO increases the formation of fatty streaks.
Because atherosclerosis has so many risk factors, statistical models and multivariate analyses are necessary. (Mark Twain might call them necessary evils.) As medical scientists, we should no doubt celebrate when a new biomarker is added to the family of cardiovascular risk factors.
Challenge to Recently Published Histopathologic Evidence
A biomarker is categorically different from an etiologic cause of disease, and a much higher bar must be set to establish the cause of a disease. There must be strong evidence and a plausible pathogenic mechanism. In “Intestinal microbiota metabolism of L-carnitine, a nutrient in red meat, promotes atherosclerosis” , the same Cleveland Clinic research group expanded on their findings by feeding carnitine to apoE knockout mice to increase plasma TMAO levels, and then sought histopathologic evidence of atherosclerosis. The paper contains four small photomicrographs of histopathologic sections of the aortic root, which purportedly demonstrate atherosclerotic plaque, and also reports quantitative data for atherosclerotic plaque lesion surface area. The authors also reported data showing that TMAO inhibits reverse transport of cholesterol.
Exactly what the photomicrographs in this May 2013 paper published in Nature Medicine demonstrate is unknown to me, but the lesions do not resemble any human lesion. Photomicrographs of lesions in this model, which more closely resemble human atherosclerotic plaques, have been published elsewhere [4, 5]. In this age of molecular biology, how many scientific journal reviewers will have sufficient knowledge of histopathology to critically examine those photomicrographs? I suspect that most readers will not be aware of how vulnerable the evidence for atherosclerosis is in the paper and will simply accept the authors’ conclusions. Even if the same lesions in the mouse aortic root were fatty streaks, fatty streaks do not necessarily develop into atherosclerotic plaques . Fatty streaks routinely resolve without sequelae . For a fatty streak to develop into an atherosclerotic plaque, some cofactor must be present. Alternatively, it may be that the cofactor is sufficient, and the fatty streak is not necessary, etiologically, as in the case of atherosclerotic plaques, which develop in synthetic arteriovenous grafts without pre-existing fatty streaks. 
This is not the first time interpretation of published photomicrographs has been in error. A photomicrograph of composer George Gershwin’s fatal brain tumor was published in 1979  in a leading journal of histopathology, the American Journal of Surgical Pathology. When I read the paper years later, I realized the photomicrograph did not support the accepted diagnosis and wrote a commentary in 2001 with an alternative interpretation of his medical history and final diagnosis . Surely, long before then, someone must have recognized that the photomicrograph in the American Journal of Surgical Pathology pointed to an alternative diagnosis. Whether due to inertia or lack of courage, no one bothered to correct the literature.
How do fatty streaks resolve? Normally, endothelial cells form a barrier to prevent transport of plasma proteins into the vessel wall. The proper hemodynamic milieu, pulsatile laminar flow, is necessary for normal endothelial morphology and adequacy of this barrier function. In a benchmark paper published in 1999 in the Journal of the American Medical Association, Malek et al. described this protective endothelial barrier as hemodynamic shear stress (tangential frictional force) of approximately 15 dynes/cm2 or greater.  In the presence of disturbed blood flow, such as abrasive high shear stress or vulnerable low shear stress, gaps develop between endothelial cells, and the transport of all plasma proteins, not just lipoproteins, increases across the endothelium. These proteins undergo pinocytosis by intimal dendritic cells and macrophages, with a resultant increase in cytoplasmic volume, forming foam cells and a fatty streak. Foam cells are motile and disgorge their cytoplasmic contents into lymphatics by transcytosis of endosomes. If the hemodynamic milieu overlying the fatty streak changes with aging and progressive vascular stiffening, the influx of plasma components decreases, and the fatty streak can disappear.
Fatty streaks are universally present from infancy in humans, therefore, it should be no surprise that there exists a mechanism by which blood vessels handle molecules which cross into the intima. What would be interesting to me would be if plasma components caused destructive inflammation in a blood vessel. It is the accumulation of macrophage and dendritic foam cells, which gives rise to the common conception that atherosclerosis is an inflammatory disease. But fatty streaks are no more an inflammatory lesion than are cholesterolosis of the gall bladder, xanthelasma of the eyelid, or lipid storage diseases such as Niemann-Pick disease, all of which are characterized by the presence of macrophage foam cells.
I predict that trimethylamine-N-oxide will have a clinical impact similar to homocysteine. There will be a brief period of interest in the molecule which will quickly fade. At the height of the interest in homocysteine, we ran three to five homocysteine tests per day at our hospital pathology laboratory. Now we run one to two a week.
5. Won KJ, Seo SH, Hwu Y, et al. In vivo real-time vessel imaging and ex vivo 3D reconstruction of atherosclerosis plaque in apolipoprotein E-knockout mice using synchrotron radiation microscopy. International Journal of Cardiology 2007; 114(2): 166-171.
Last updated: 2014-12-30