 |
|
|
| |
Glycated Hemoglobin A1C as Screening for Diabetes Mellitus in HIV Infected Individuals
|
| |
| |
Reported by Jules Levin
CROI 2011 March 2 Boston
Benjamin Eckhardt, Robert S. Holzman, Candice K. Kwan, Jonathan Baghdadi, Judith A. Aberg New York University School of Medicine at Bellevue Hospital, New York, NY
ABSTRACT
Background: The American Diabetes Association now recommends glycated hemoglobin A1C (A1C) screening for diagnosis of diabetes (DM)1. It has been reported that A1C levels underestimate hyperglyce-mia in HIV infected persons2,3,4. Both increased mean corpuscular volume (MCV) and Abacavir use have been independently associated with A1C-glycemia discordance. We examined the performance of A1C screening for incident diabetes in a group of HIV infected non-diabetics and examined the effect of an-tiretroviral regimen on the association of A1C and fasting blood glucose (FBG).
Methods: We surveyed a group of HIV infected patients attending the Bellevue Virology Clinic and collect-ed A1C and FBG levels as well as additional clinical data. The Receiver Operating Characteristic (ROC) curve was used to assess screening test performance. The effect of treatment regimen on the relationship be-tween A1C and FBG was assessed by multiple linear regressions. Patients were included unless they had known DM, were pregnant, on corticosteroids, recently admitted or transfused, or had underlying hemo-globinopathy.
Results: From October 2009 through April 2010, 490 patients had an A1C test and 395 met inclusion crite-ria. Twenty-two of the 395 were new diabetics based on FBG ≥126 mg/dl. Using a cut-off of A1C ≥ 6.5%, A1C had a sensitivity of 40.9% and specificity of 97.5% for identification of incident diabetes, thus screening with A1C would have only diagnosed 9 of the 22. The ROC analysis suggested that a cutoff of 5.8% would have been optimal in this population, with sensitivity of 88.8% (20/22 diagnosed) and specificity of 77.5%. MCV greater than the median (p <.001) and current use of a protease inhibitor (p <.001) significantly in-creased the slope while NNRTI use (p =0.04) and particularly Efavirenz (p=0.02) significantly decreased the slope of the linear regression of FBG on A1C. Tenofovir and Raltegravir, previously unstudied agents, did not significantly alter the slope or y-intercept of the line.
Conclusions: Among HIV-infected, non-diabetic patients, A1C screening using ADA criteria is insensitive, though highly specific for new cases of DM. The ROC curve and the optimum cutpoint we found, 5.8%, was similar to the 5.9% optimum noted in a report based on NHANES data5.
|
| |
|
|
|
|
|
