Duy Le, MD

Kjaergaard AD, Johansen JS, Bojesen SE, and Nordestgaard BG. Elevated Plasma YKL-40, Lipids and Lipoproteins, and Ischemic Vascular Disease in the General Population. Stroke. 2015

Mendelian Randomization is initially a difficult process to grasp. However, when broken down to its most essential parts, it actually makes good sense. The classic example used is the relationship between lower cholesterol levels (variable) and cancer (outcome). Does having lower cholesterol expose one to having cancer? –or is it the other way around, where having cancer leads to lower nutrition and lower levels of cholesterol? Or are there too many confounding factors to be able to tell if there really is a relationship between the two? Mendelian randomization employs a third party, in this case, ApoE2 (AKA the extra variable) which is a known variant of lower cholesterol levels, to help decipher the difference. If lower cholesterol is indeed the cause of cancer, then the cancer patients should have more ApoE2 alleles than the control group. Otherwise, the distribution should be similar in both groups.

                                 Confounding factors
                                  |                              |
                                  ˅                              ˅ 
Extra variable à original variable à outcome

In this model, the question is whether there is a true relationship between the original variable and the outcome. In order for this process to logistically flow: 

1) the extra variable must be reliable related to the original variable
2) the original variable and outcome are affected by confounding factors, but the extra variable cannot
3) The extra variable does not directly affect outcome, rather it is associated with the original variable only

Now that the rocket science has been explained, we go on to discuss Kjaergaard et al’s paper. In essence, the question is whether or not plasma YKL-40 levels are associated with elevated lipids, lipoproteins, and an increased risk of ischemic vascular disease. YKL-40 is 40 kDa plasma protein produced by lipid laden macrophages inside the vessel wall. More than 96,000 individuals in a Danish population were evaluated between 1977 to 2013. When comparing 91-100% vs. 0-33% YKL-40 percentile category, there was a 34% increase in triglycerides (TG). With multifactorial adjustment, HR’s were 1.99 (significant) for stroke, 1.85 (significant) for ischemic cardiovascular disease, 1.28 (non significant) for MI. The extra variable employed as the CHI3LI gene which is known to be associated with YKL-40 levels, yet not with TG levels or with risk of ischemic vascular disease.

When applying Mendelian randomization, the CHI3LI gene was associated with upwards to tripling the YKL-40 levels. Additionally, the CHI3LI gene when evaluating baseline characteristics was not shown to be significantly related to any of the characteristics. Thus, when applying this to the equation above for Mendelian randomization to hold true, there is 1) proof that the extra variable is related to the original variable 2)the extra variable was shown not to be associated with baseline characteristics (ie no confounding) 3) the extra variable was not shown to have a direct relationship with the outcome. All three are fulfilled, and one can claim the statement that “elevated YKL-40 is associated with a 34% increase in TG levels and 2-fold increase risk of ischemic stroke.”

Beyond the complexities of statistics and the data that claims there is an association between TKL-40 and ischemic stroke, what does this actually mean to the clinical physician? The first interesting point is that YKL-40 was shown to be significantly associated with ischemic stroke, but not myocardial infarction. Is there something about this marker than predicts it to be specific to neurologic ischemia? The authors hypothesize that astrocytes in the brain express this protein more in ischemic strokes and that it may potentially be a sensitive marker of small or covert cerebrovascular disease.

This research ultimately does trudge on generalized principles, which makes it difficult for the lay person to see what potential clinical applications may ensue. What role will checking WKL-40 in stroke patients have? Will it be used in lieu of lipids? Will it be a lab to check in the acute period where there is concern for stroke? Or will it be used as a primary stroke prevention screen? There are still many questions to answer on a wide open topic. Nonetheless, I would have been extremely interested to see a stratification of TOAST criteria and YKL-40 levels in conjunction to strokes. Ultimately, YKL-40 is expressed in macrophages in vessel walls. If we could find a correlation to a TOAST subtype, it would help elucidate the potential underlying mechanism in which YKL-40 has a relationship to increased ischemic stroke events.