Author + information
- Mark D. DeBoer, MD, MSc, MCR∗ (, )
- Matthew J. Gurka, PhD,
- Jessica G. Woo, PhD and
- John A. Morrison, PhD
- ↵∗Division of Pediatric Endocrinology, University of Virginia, P.O. Box 800386, Charlottesville, Virginia 22908
The long-term ability of the metabolic syndrome (MetS) to predict cardiovascular disease (CVD) has been limited by the binary nature of traditional MetS criteria and by discrepancies among African Americans, who have low rates of MetS classification despite higher rates of death from CVD (1). We previously used confirmatory factor analysis to formulate MetS severity z-scores for adolescents (2) and adults (3) that place differential weights on the individual MetS components to account for variation in how MetS is manifest by sex and racial/ethnic group. Our goal was to assess the ability of these scores to determine long-term risk for CVD.
The Princeton Lipid Research Cohort Study followed white and black (30.5%) individuals (55.5% female) over 3 phases: 1) the LRC (Lipid Research Clinic) (1973 to 1976) evaluated MetS measures on students in grades 1 to 12 (4); 2) the PFS (Princeton Follow-Up Study) (1998 to 2003) evaluated complete MetS measures and reported CVD status on 629 LRC participants (4); and 3) the PHU (Princeton Health Update) (2010 to 2014) assessed CVD outcomes via phone interviews and National Death Index query on 354 cohort members. CVD was classified as self-reported myocardial infarction, coronary artery bypass, other heart surgery, coronary revascularization procedure (angioplasty, stent placement), or stroke. MetS severity z-scores were calculated from each individual’s measures of body mass index z-score (children/adolescents) or waist circumference (adults), systolic blood pressure, fasting triglycerides, and fasting glucose, based on equations specific to sex and racial/ethnic subgroup from LRC and PFS visits. Mean MetS z-scores were compared based on participants’ CVD diagnosis by the PFS or PHU. Logistic regression and receiver-operating characteristic (ROC) curves were used to evaluate the ability of MetS severity scores to predict future CVD.
MetS severity z-scores during childhood (LRC, mean 12.9 years of age) were lowest among those who never developed CVD, highest among those with early CVD (PFS, mean 38.4 years of age) and intermediate among those with later CVD (PHU, mean 49.6 years of age) (Figure 1). In predicting future CVD, ROC curves revealed that childhood MetS severity z-scores had areas under the curve of 0.91 and 0.65 by PFS and PHU, respectively, while MetS z-scores at PFS had area under the curve of 0.84 for subsequent CVD by PHU.
Using logistic regression, each 1.0 increase in childhood MetS severity z-scores carried elevated odds ratios of 9.8 and 2.4 for incident CVD by PFS and PHU, respectively (p < 0.001 and p < 0.05). When change in MetS severity z-score from LRC to PFS was added to baseline LRC z-score in the model, this carried a further elevated odds ratio of 3.4 for incident CVD between PFS and PHU (p < 0.01).
The long-term health consequences of obesity—including CVD—underscore the need for clinical tools to assist in risk prediction to target at-risk individuals for preventive therapy. We found that a sex- and race/ethnicity-specific MetS severity z-score may serve as such a tool in assisting disease prediction in 2 ways: 1) baseline MetS severity scores in childhood and in mid-adulthood predicted later CVD diagnosis; and 2) the change in score during the interval from childhood to adulthood was associated with future disease, even after adjustment for baseline scores. In this sense, this score overcomes limitations of traditional MetS criteria, which are based on individuals having abnormalities in ≥3 of the individual MetS components and are thus unable to assess for changes in MetS over time within an individual (besides its presence/absence)—and are unable to assess risk related to component values just below the population-based cutoff.
This score is associated with risk for CVD and may serve as a marker of the degree of the severity of metabolic derangements behind MetS. Such a score—potentially calculated automatically in an electronic health record system—could enable tracking changes in a given individual’s MetS severity, both to assess response to specific therapies and to identify ominous increases in MetS severity as a marker of risk and a trigger for further intervention. Future research is needed to determine clinically useful cutoffs of particularly elevated risk and whether this score improves CVD risk prediction above traditional criteria on a sex and race/ethnic basis.
Please note: This work was supported by National Institutes of Health grants 5K08HD060739 to Dr. DeBoer, U54GM104942 to Dr. Gurka, 1R21DK085363 to Drs. DeBoer and Gurka, 1R01HL120960 to Drs. DeBoer and Gurka, and National Heart, Lung, and Blood Institute N01HV22914; a University of Virginia Children’s Hospital Grant-in-Aid (Charlottesville, Virginia) to Dr. DeBoer; a Cincinnati Children's Hospital Medical Center Heart Institute Research Core grant (Cincinnati, Ohio); a Schmidlapp Women’s Scholar’s Award (Cincinnati, Ohio) to Dr. Woo; and American Heart Association grant 9750129 (Chicago, Illinois) to Dr. Morrison. All authors were independent of these funding agencies. Drs. DeBoer and Gurka contributed equally to this work.
- American College of Cardiology Foundation