Response to Metabolic Syndrome and Early Death: Extending the Discussion on Heterogeneity
We thank Colagiuri et al1 for their comments on our editorial.2 As they correctly highlight, we understated the complexity of the National Cholesterol Education Program (NCEP) Adult Treatment Panel III classification of the metabolic syndrome (MetS). However, the maximum number of possible combinations for the NCEP MetS actually exceeds the number proposed by Colagiuri et al.1 Under the NCEP classification, high blood pressure (BP) is defined as elevations in systolic or diastolic BP.3 Thus, the maximum number of distinct MetS phenotypes derivable using the NCEP definition is 27. The relevance of distinguishing between MetS phenotypes derived using systolic or diastolic BP is outlined below. If one were to additionally consider treatment with medication as an alternative way of defining raised BP, lipids, or glucose, the number of possible MetS categories approaches 150.
Perhaps more important than the number of possible MetS phenotypes is the way in which these phenotypes correlate and how this impacts the use of standard MetS definitions when used in epidemiological studies. To illustrate this point, we calculated pairwise partial correlations for each of the 27 NCEP MetS categories using adult data from National Health and Nutrition Examination Survey 2003–2004 (N=1821).4 The spectrum of possible pairwise correlations ranges from −1 to 1. If the relationships between MetS categories were completely random, one would expect a mean correlation of r=0, with the tails of the distribution extending to its extremes (ie, −1 and 1). If each of the MetS categories perfectly negatively or positively predicted each other, all of the pairwise correlations would be r=−1 or r=1, respectively. Neither a perfect nor a low level of mean correlation is desirable. The former suggests that only 1 of all of the categories provides unique information; the limitations of the latter are described below. The majority of pairwise correlations are >0, but <0.50 (mean: 0.44; SD ±0.18), and only 7% of the categories correlate at >0.75. The pairwise correlations for the phenotypes that differ only by systolic or diastolic BP are also relatively low (r=0.33 to 0.56).
Another way of interpreting these observations is to consider that if any 2 of the 27 MetS categories were randomly selected, on average only ≈19% (ie, the square of the pairwise mean r) of individuals classified with the MetS using the first category would also be classified using the second category. This low level of overlap between NCEP MetS categories indicates that this definition of the MetS does not characterize a specific pathophysiological entity.
The large number of distinct, heterogeneous MetS phenotypes derivable using standard categorical classifications may help explain why these classifications often predict disease outcomes with low sensitivity.5 Undoubtedly, the desire for simple scoring methods for clustered cardiovascular risk will persist; whether simple scores can be designed to predict future disease with comparable sensitivity to the individual component traits remains to be determined.
Source of Funding
P.W.F. was supported in part by funding from the Västerbotten’s local health authority (VLL).
Colagiuri S, Lee C, Huxley R, Woodward M. Metabolic syndrome and early death. Hypertension. 2007; 50: e5.
Franks PW, Olsson T. Metabolic syndrome and early death: getting to the heart of the problem. Hypertension. 2007; 49: 10–12.
National Health and Nutrition Examination Survey Data: 2003–2004. U.S. Department of Health and Human Services, Centers for Disease Control and Prevention, National Center for Health Statistics: Hyattsville, MD. Available at: http://www.cdc.gov/nchs/about/major/nhanes/nhanes2003-2004/nhanes03_04.htm. Accessed May 15, 2007.