The Use of Areas Under Curves in Diabetes Research

Obesity Research Center, St. Luke's/Roosevelt Hospital, Columbia University College of Physicians and Surgeons, New York, New York 10025, USA.
Diabetes Care (Impact Factor: 8.42). 03/1995; 18(2):245-50. DOI: 10.2337/diacare.18.2.245
Source: PubMed


Recently, several articles appearing in the diabetes literature have suggested that many investigators are unclear about a number of issues involving the use of areas under the curve (AUCs). This prompted us to reconsider issues in the calculation, use, meaning, and presentation of AUCs. We discuss five issues: 1) What is a curve and an area? 2) How should one graphically present a group's curve? 3) How should one calculate AUCs? 4) Should one subtract baseline values from outcome values before calculating AUCs? And 5) are AUCs the best way to combine multiple readings into a single index?

Download full-text


Available from: Xavier Pi-Sunyer, Dec 17, 2013
356 Reads
  • Source
    • "The general characteristics of the population studied are presented as means ± SD (Table 1). Responses of glucose and insulin observed during OGTT were examined as area under the curve above baseline across 120 minutes following the oral glucose load (AUC), calculated by the trapezoidal method [17]. Differences between carriers (CG) and non-carriers (CC) in the effect of melatonin (melatonin-placebo) on AUC for glucose and insulin in the morning and in the evening were analyzed by unpaired t-test and further analyses were performed by ANCOVA in which effects were adjusted for age and BMI (Table 1). "
    [Show abstract] [Hide abstract]
    ABSTRACT: Aims: The common MTNR1B genetic variant rs10830963 is associated with an increased risk of type 2 diabetes (T2D). To date, no experimental study has tested the effect of the MTNR1B variant on glucose metabolism in humans during exposure of the melatonin receptors to their ligand. The aim of this study was to investigate whether this MTNR1B variant influenced the effect of melatonin (5mg) on glucose tolerance assessed by an oral glucose tolerance test (OGTT; 75g) at different times of the day (morning and evening) as compared to a placebo. Methods: Seventeen normoglycemic women (24±6years; BMI 23.0±3.3kg/m(2)) completed the study (11 carriers of the risk allele [CG] and 6 noncarriers [CC]). Results: The effect of melatonin on glucose tolerance depended on the genotype. In the morning, the effect of melatonin (melatonin-placebo) on the glucose area under the curve (AUC) above baseline differed significantly (P=0.036) between the carriers and noncarriers. This effect of melatonin in the carriers was six times as large as that in the noncarriers. The MTNR1B SNP explained over one-quarter (26%) of the inter-individual differences in the effect of melatonin on glucose AUC. However, in the evening, the effect of melatonin on glucose AUC of the carriers and noncarriers did not differ significantly (P>0.05). Conclusions: MTNR1B rs10830963 risk variant worsens the effect of melatonin on glucose tolerance, suggesting the importance of genotyping and personalized recommendations, especially in people consuming food when melatonin levels are elevated. Large-scale studies in vulnerable populations are necessary to translate these results into real-world, clinically relevant recommendations.
    Metabolism: clinical and experimental 10/2015; DOI:10.1016/j.metabol.2015.08.003 · 3.89 Impact Factor
  • Source
    • "Glucose and insulin area under the curve (AUC) was determined using a trapezoid model (Allison et al. 1995). All participants stayed overnight at CHP to undergo the euglycemic clamp test the next morning. "
    [Show abstract] [Hide abstract]
    ABSTRACT: We examined the joint and independent associations between VAT and LF with insulin sensitivity (IS) and lipids in seventy-one obese adolescents (BMI > 95th, 14.9 ± 1.8 years). VAT was assessed by magnetic resonance imaging and LF was quantified by proton magnetic resonance spectroscopy. IS was evaluated by a 3-hour hyperinsulinemic (80 mU/m2/min)-euglycemic clamp. Independent associations between VAT and LF on metabolic variables were assessed in mutually adjusted multivariate models. The joint association between VAT and LF on metabolic variables was assessed by categorizing participants into a low VAT + low LF group (n=35), high VAT + low LF group (n=26), or high VAT + high LF group (n=10) based on a VAT median split (1.17kg) and high (≥5%) and low (<5%) LF. Both VAT and LF were independently associated with fasting insulin, 2-hour insulin, insulin AUC, IS, and triglycerides (P<0.05). Adolescents with high VAT + high LF had higher 2-hour glucose, glucose AUC, 2-hour insulin, triglycerides, and lower insulin sensitivity compared to adolescents with high VAT only (P<0.025 for all). In obese adolescents, VAT and LF were independently associated with insulin sensitivity and dyslipidemia, and the concomitant presence of VAT and LF is strongly associated with metabolic risk factors.
    Biochemistry and Cell Biology 11/2014; DOI:10.1139/bcb-2014-0064 · 2.15 Impact Factor
  • Source
    • "Plasma C-peptide and glucose concentrations were used to determine the ISR in response to the oral glucose load and the sensitivity of the β-cell response (ISR) to changes in plasma glucose by using a minimal model (25). This model provides an estimate of the total amount of insulin secreted in response to plasma glucose as a function of time (i.e., total ISR in pmol/min) and partitions this total response into a dynamic component (ISRdynamic), which represents the rapid release of a readily releasable pool of insulin secretory granules in response to the rate of increasing plasma glucose concentration, and a static component (ISRstatic), which represents the slower release of a reserve pool of insulin secretory granules in response to the ambient plasma glucose concentration (28). "
    [Show abstract] [Hide abstract]
    ABSTRACT: OBJECTIVE Nonnutritive sweeteners (NNS), such as sucralose, have been reported to have metabolic effects in animal models. However, the relevance of these findings to human subjects is not clear. We evaluated the acute effects of sucralose ingestion on the metabolic response to an oral glucose load in obese subjects.RESEARCH DESIGN AND METHODS Seventeen obese subjects (BMI 42.3 ± 1.6 kg/m(2)) who did not use NNS and were insulin sensitive (based on a homeostasis model assessment of insulin resistance score ≤2.6) underwent a 5-h modified oral glucose tolerance test on two separate occasions preceded by consuming either sucralose (experimental condition) or water (control condition) 10 min before the glucose load in a randomized crossover design. Indices of β-cell function, insulin sensitivity (SI), and insulin clearance rates were estimated by using minimal models of glucose, insulin, and C-peptide kinetics.RESULTSCompared with the control condition, sucralose ingestion caused 1) a greater incremental increase in peak plasma glucose concentrations (4.2 ± 0.2 vs. 4.8 ± 0.3 mmol/L; P = 0.03), 2) a 20 ± 8% greater incremental increase in insulin area under the curve (AUC) (P < 0.03), 3) a 22 ± 7% greater peak insulin secretion rate (P < 0.02), 4) a 7 ± 4% decrease in insulin clearance (P = 0.04), and 5) a 23 ± 20% decrease in SI (P = 0.01). There were no significant differences between conditions in active glucagon-like peptide 1, glucose-dependent insulinotropic polypeptide, glucagon incremental AUC, or indices of the sensitivity of the β-cell response to glucose.CONCLUSIONS These data demonstrate that sucralose affects the glycemic and insulin responses to an oral glucose load in obese people who do not normally consume NNS.
    Diabetes care 04/2013; 36(9). DOI:10.2337/dc12-2221 · 8.42 Impact Factor
Show more