Fasting insulin model helps identify prediabetes.

BOSTON -- Prediabetes can be accurately and easily identified in the clinical setting using a predictive model based on fasting insulin quartiles, a study has shown.

Given its clinical utility, the model may be a practical alternative to existing methods for prediabetes screening, Jennal L. Johnson reported at the Fifth Annual World Congress on the Insulin Resistance Syndrome.

Glucose tolerance testing and insulin clamp studies, which are considered the standard for assessing glycemic status and insulin sensitivity, are costly and impractical for screening in a busy clinical setting. Assessing fasting glucose values alone is not an acceptable alternative, because doing so may miss patients with impaired glucose tolerance who do not have fasting glucose elevations, noted Ms. Johnson, an endocrinology nurse practitioner and certified diabetes educator with Endocrinology Associates of Phoenix.

In an effort to identify clinical and/or laboratory variables that might be predictive of prediabetes, Ms. Johnson and her colleagues conducted a retrospective analysis of 1,036 patients who underwent 2-hour glucose testing for suspected prediabetes or diabetes between 2000 and 2004.

For each patient, the investigators recorded demographic data (including weight, height, sex, and age) and laboratory data (including fasting blood glucose, fasting insulin, 2-hour postprandial glucose, 2-hour postprandial insulin, total cholesterol, HDL cholesterol, LDL cholesterol, triglycerides, and systolic and diastolic blood pressure). Fasting blood glucose and fasting insulin levels were used to calculate the homeostasis model assessment for insulin resistance (HOMA-IR) index.

Of the 1,036 patients included in the review, 71 were identified as having diabetes, based on their fasting glucose or 2-hour postprandial glucose levels, and were excluded from the analysis. Of the remaining 965 patients, 287 were identified as having prediabetes, based on a fasting glucose of 100-125 mg/dL or a 2-hour postprandial glucose of 140-200 mg/dL.

The investigators then generated multivariate logistic regression models and tested their predictive value. "Because fasting glucose was identified as a primary variable [in defining prediabetes], we excluded it to determine if other variables could predict prediabetes," Ms. Johnson said. Also, 2-hour postprandial glucose and insulin values were excluded from the models because of the impracticality of the testing, and in light of the study goal of determining if prediabetes could be diagnosed without these variables, she noted.

In the first of two multivariate models, patients with a fasting insulin level in the fourth quartile (mean value 25 microU/mL) were five times more likely to have prediabetes than those with a fasting insulin level in the first quartile (mean value 5 microU/mL). Patients with second-and third-quartile fasting insulin levels (8 and 12 microU/mL, respectively) were two and three times more likely to have prediabetes than those with first-quartile values. All three differences were statistically significant. Older age and increased diastolic blood pressure contributed modestly to this model, Ms. Johnson said.

A second model assessing the potential predictive value of the HOMA-IR index showed that this variable and older age were only modestly predictive of prediabetes, she said.

Although body mass index did not predict prediabetes, it did predict insulin quartiles--a finding that warrants further analysis, Ms. Johnson noted.

The findings suggest that the fasting insulin predictive model, if validated through additional studies, could potentially increase the number of early prediabetes diagnoses, and in so doing, "afford an opportunity for intervention, possibly preventing or delaying the onset of type 2 diabetes," she said.

BY DIANA MAHONEY

New England Bureau