Framingham Underestimates CVD Risk in HIV-Infected Persons

Daniel M. Keller, PhD

October 30, 2013

BRUSSELS, Belgium — The Framingham model for global cardiovascular disease (CVD) risk may significantly underestimate the likelihood of CV events in HIV-infected patients, say researchers, who believe they have developed a better model, derived from the Data Collection on Adverse Events of Anti-HIV Drugs (D:A:D) study, which also considers CD4 cell counts and administration of antiretroviral agents.

Author of the new work, Nina Friis-Møller, MD, PhD, senior consultant and associate professor at Odense University Hospital, Denmark, said the Framingham model can be recalibrated to predict CVD risk in the HIV-infected population. But "our analyses suggest that risk equations developed from the D:A:D data set are superior in HIV-positive persons, in particular for the accuracy of prediction in subgroups," she reported here at the 14th European AIDS Conference.

As people are living longer with HIV, a better predictive model may alert clinicians to those patients who need more aggressive CVD risk-mitigation interventions, she observed.

Andrew Carr, MBBS, PhD, head of HIV immunology and infectious diseases at St. Vincent's Hospital and professor of medicine at the University of New South Wales in Sydney, Australia, told Medscape Medical News it makes sense to derive a risk model using the specific population of interest. "I think the Framingham equation was shown a decade ago to be a very reasonable place to start," he noted, adding that a recent publication from the Veterans Administration cohort suggests that multiplying a Framingham score by about 1.5 might generate a more realistic estimate. "The data that were shown today suggest that the D:A:D equation gives…a risk that more closely matches the actual risk when you follow the patients…for quite a number of years," he commented.

HIV-Specific Factors Alter CVD risk

The D:A:D model is based on observations in 32,663 HIV-positive persons from 20 European countries and Australia who were free of CVD at study entry and who provided full information on cardiovascular risk factors. Participants were followed from study entry to a first cardiac event, 6 months after the last clinic visit, or February 2011.

The rate of a composite CVD end point, comprising myocardial infarction (MI), stroke, coronary artery revascularization, carotid endarterectomy, or any CVD death, was assessed over 186,364.5 person-years.

Full and reduced D:A:D models were developed to calculate estimated 5-year risks of CVD.

The data that were shown today suggest that the D:A:D equation gives…a risk that more closely matches the actual risk when you follow the patients…for quite a number of years." Andrew Carr, MBBS, PhD

There were 1010 composite CVD events, giving a rate of 5.42/1000 person-years. Almost half the events were MIs (493/1010), followed by stroke (295/1010), angioplasty (129/1010), coronary artery bypass, carotid endarterectomy, and other events.

Patients who developed CVD tended to have more of the traditional cardiovascular risk factors at baseline compared with those without CVD. The group developing CVD was older (47 vs 39 years), contained more men (87.5% vs 74%), had a higher prevalence of diabetes (10.4% vs 2.8%), more smokers, higher total cholesterol, higher systolic blood pressure, and lower high-density-lipoprotein cholesterol.

The final full D:A:D model included traditional CVD risk factors as well as CD4 counts, combination antiretroviral therapy (cART), cumulative exposure to protease- and nucleoside-reverse-transcriptase inhibitors (NRTI), and current use of abacavir. The reduced model does not include use of ART.

Cardiovascular Event Hazard Ratios

HIV-Specific Risk Factor Per Unit D:A:D Full Model D:A:D Reduced Model
CD4 cell count 2-fold higher 0.89 0.89
Abacavir current exposure   1.47
Protease-inhibitor cumulative exposure year 1.05
NRTI cumulative exposure year 1.03

Dr. Friis-Møller said that the D:A:D models were significantly better risk predictors compared with the Framingham model (P < .001), providing more accurate 5-year risk prediction for certain patient subgroups — for example, those with diabetes mellitus.

Besides predicting individual risk, the models are also good for estimations of population levels of risk for research purposes, she noted.

She cautioned, however, that generalizability of the findings still requires external independent validation in further cohorts of HIV-positive persons.

Dr. Carr said the Framingham risk score is derived from people living in a town in the northeastern United States and probably has very few people with HIV in the database. The results that the D:A:D and Framingham models give are statistically different, "but they didn't appear numerically huge," he pointed out. "So I think, for me, the take-home message is not which one you use, as long as you use something, as long as you estimate risk."

One limitation of the D:A:D equation is that it estimates only 5-year risk because it does not have long enough follow-up to assess 10-year risk, as the Framingham score does. Nonetheless, he said he uses the D:A:D because "it's a desktop calculator like the Framingham, so it's something you can do in clinic with a patient sitting next to you."

Dr. Carr said primary-care physicians with a high case load of HIV patients may find the D:A:D useful, but those with only a few HIV patients might avoid using it, because "it's just another layer of confusion and complexity."

A D:A:D 5-year estimated risk calculator is available here.

The study oversight committee had representatives from academia, the patient community, the Food and Drug Administration, European Medicines Agency, and a consortium of industry organizations: AbbVie, Boehringer Ingelheim, Bristol-Myers Squibb, Gilead, ViiV, Merck, Pfizer, Hoffmann-LaRoche, and Janssen. Dr. Friis-Møller reports no relevant financial relationships. Dr. Carr has received funds for research support from Gilead and MSD; has served as a consultant and on advisory boards for Gilead, MSD, and ViiV; and has received lecture and travel support from Gilead, MSD, Roche, and ViiV.

European AIDS Clinical Society 14th European AIDS Conference. Abstract PS1/3, presented October 17, 2013.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.