Back grey_arrow_rt.gif
HBA1c may sharpen cardiovascular-event prediction in diabetics
  July 26, 2011 | Steve Stiles AHA
Boston, MA - Future cardiovascular risk in men and women with diabetes varies according levels of hemoglobin A1c (HbA1c) and so might potentially be used to improve CV-event prediction-compared with lumping all diabetics together as one high-CV-risk group-when used alongside other risk markers in diabetics, suggest analyses based on two large cohorts followed for a decade or more [1].
The findings are consistent with abundant evidence that "not all diabetics are at high risk of future vascular events" and suggest "that the use of HbA1c levels as part of overall CV-disease risk scores may improve predictive ability in diabetic patients, whose HbA1c levels are routinely measured in clinical practice," write the authors, led by Dr Nina P Paynter (Brigham and Women's Hospital, Boston, MA).
The group's analysis, based on >24 000 participants in the Women's Health Study (WHS) and >11 000 men in the Physician's Health Study II (PHS II) and their diabetic subgroups, was published online July 25, 2011 in the Archives of Internal Medicine.
It has implications for management of CV risk factors in diabetics, the authors observe. For example, a way to stratify diabetics into higher- and lower-risk categories could potentially allow more targeted therapy with statins, which might be avoided in those unlikely to benefit from them.
An accompanying editorial [2] explains that guidelines from the National Cholesterol Education Program Adult Treatment Panel III (NCEP ATP III) consider diabetes a coronary heart disease (CHD) risk equivalent-that is, all diabetics "are categorized into the highest CHD risk category (along with persons with a history of myocardial infarction or ischemia, other clinical atherosclerotic diseases, and persons with estimated CHD risk >20%) and are subject to relatively aggressive cholesterol-lowering goals and treatment thresholds."
The current study, however, has a few limitations, according to editorialist Dr Mark J Pletcher (University of California, San Francisco). For example, glycated hemoglobin levels weren't greatly elevated among the cohorts' diabetics, making the analysis less relevant for those with poorly controlled diabetes.
Still, "there is little doubt that the primary findings by Paynter et al are true," he writes, in that "10-year CVD risk, especially in women and in young men, is often lower than 20%, and allowing downward reclassification for some diabetic individuals using a risk equation that included an indicator variable for diabetes and/or HbA1c measurements should improve the overall accuracy of 10-year CVD risk prediction for diabetic patients compared with the NCEP ATP III algorithm."
In fact, Paynter told heartwire, their algorithm incorporating HbA1c levels "reclassified around 19% of the [diabetic] women studied and less than 1% of [diabetic] men studied to low CV risk," that is, a <5% risk of vascular events-compared with the algorithm based on NCEP ATP III criteria.
She and her colleagues created their models incorporating HbA1c levels based on outcomes of the overall WHS and PHS II cohorts. They were based on variables from NCEP ATP III and the Reynolds risk score (RRS).
They then tested the tools in much smaller subgroups of aged-matched WHS and PHS II diabetics, 685 and 563 patients respectively, comparing them with CV-risk stratification according to the currently recommended system that casts diabetes as a CHD risk equivalent, that is, corresponding to a >20% 10-year CHD risk. In the WHS cohort, the predictive tools using HbA1c levels improved the C statistic (which reflects risk stratification power) by 0.177 (p<0.001) compared with the model based on diabetes as a CV risk equivalent. The effect was weaker but still significant in the PHS II cohort, with a C-statistic improvement of 0.039 (p=0.04).
Moreover, the models that used HbA1c levels improved the C statistic better than a model that considers diabetes as a "yes" or "no" variable (p=0.03) rather than as a CHD risk equivalent.
On the other hand, Pletcher questioned whether the risk-stratification models would realistically be used to exclude some diabetics from statin therapy. "Will it really benefit diabetic persons with low or medium short-term risk to withhold statins?" Their long-term risk is elevated, regardless, "and we know that atherosclerotic damage accumulates early in life," he writes.
"Whether to consider downward reclassification of individuals with diabetes (using HbA1c or otherwise) and less intensive CV-disease prevention efforts for these patients is a decision that will have to take such nuances into account."
Paynter reports receiving funding for this study from F Hoffmann-La Roche; disclosures for the coauthors are listed in the paper. WHS and PHS II were supported by the National Institutes of Health and the Donald W Reynolds Foundation. Pletcher had no disclosures.
  icon paper stack View Older Articles   Back to Top