Dataset Information

Discriminatory ability of simple OGTT-based beta cell function indices for prediction of prediabetes and type 2 diabetes: the CODAM study.

ABSTRACT:

Aims/hypothesis

The hyperglycaemic clamp technique and the frequently sampled IVGTT are unsuitable techniques to assess beta cell function (BCF) in large cohorts. Therefore, the aim of this study was to evaluate the discriminatory ability of simple OGTT-based BCF indices for prediction of prediabetes (meaning impaired fasting glucose and/or impaired glucose tolerance) and type 2 diabetes.

Methods

Glucose metabolism status was assessed by 2 h 75 g OGTT at baseline (n = 476, mean age 59.2 years, 38.7% women) and after 7 years of follow-up (n = 416) in the Cohort on Diabetes and Atherosclerosis Maastricht (CODAM) study (1999-2009). Baseline plasma glucose, insulin and C-peptide values during OGTTs were used to calculate 21 simple indices of BCF. Disposition indices (BCF index × Matsuda index), to compensate for the prevailing level of insulin resistance, were calculated for the BCF indices with the best discriminatory abilities. The discriminatory ability of the BCF indices was estimated by the area under the receiver operating characteristics curve (ROC AUC) with an outcome of incident prediabetes (n = 73) or type 2 diabetes (n = 60 and n = 18 cases, respectively, in individuals who were non-diabetic or had normal glucose metabolism at baseline).

Results

For incident prediabetes (n = 73), all ROC AUCs were less than 70%, whereas for incident type 2 diabetes, I₃₀/I₀, CP₃₀/CP₀, ΔI₃₀/ΔG₃₀, ΔCP₃₀/ΔG₃₀ (where I, CP and G are the plasma concentrations of insulin, C-peptide and glucose, respectively, at the times indicated), and corrected insulin response at 30 min had ROC AUCs over 70%. In at-baseline non-diabetic individuals, disposition indices ΔI₃₀/ΔG₃₀, ΔCP₃₀/ΔG₃₀ and corrected insulin response at 30 min had ROC AUCs of over 80% for incident type 2 diabetes. Moreover, these BCF disposition indices had significantly better discriminatory abilities for incident type 2 diabetes than the Matsuda index alone.

Conclusions/interpretation

BCF indices reflecting early-phase insulin secretion have the best ability to discriminate individuals who will develop prediabetes and type 2 diabetes. Of these, ΔCP₃₀/ΔG₃₀, often referred to as the C-peptidogenic index, performed consistently well.

SUBMITTER: den Biggelaar LJ

PROVIDER: S-EPMC6518926 | biostudies-literature |

REPOSITORIES: biostudies-literature

ACCESS DATA

Similar Datasets

Project description:ObjectiveThe purpose of this study is to investigate the independent influencing factors of the transition from normal population to prediabetes, and from prediabetes to diabetes, and to further construct clinical prediction models to provide a basis for the prevention and management of prediabetes and diabetes.Materials and methodsThe data for this study were based on clinical information of participants from the Health Management Center of Peking University Shenzhen Hospital. Participants were classified into normal group, prediabetes group, and diabetes group according to their functional status of glucose metabolism. Spearman's correlation coefficients were calculated for the variables, and a matrix diagram was plotted. Further, univariate and multivariate logistic regression analysis were conducted to explore the independent influencing factors. The independent influencing factors were used as predictors to construct the full-variable prediction model (Full.model) and simplified prediction model (Simplified.model).ResultsThis study included a total of 5310 subjects and 22 variables, among which there were 1593(30%) in the normal group, 3150(59.3%) in the prediabetes group, and 567(10.7%) in the diabetes group. The results of the multivariable logistic regression analysis showed that there were significant differences in 9 variables between the normal group and the prediabetes group, including age(Age), body mass index(BMI), systolic blood pressure(SBP), urinary glucose(U.GLU), urinary protein(PRO), total protein(TP), globulin(GLB), alanine aminotransferase(ALT), and high-density lipoprotein cholesterol(HDL-C). There were significant differences in 7 variables between the prediabetes group and the diabetes group, including Age, BMI, SBP, U.GLU, PRO, triglycerides(TG), and HDL.C. The Full.model and Simplified.model constructed based on the above influencing factors had moderate discriminative power in both the training set and the test set.ConclusionAge, BMI, SBP, U.GLU, PRO, TP, and ALT are independent risk factors, while GLB and HDL.C are independent protective factors for the development of prediabetes in the normal population. Age, BMI, SBP, U.GLU, PRO, and TG are independent risk factors, while HDL.C is an independent protective factor for the progression from prediabetes to diabetes. The Full.model and Simplified.model developed based on these influencing factors have moderate discriminative power.

Project description:Historically, a number of typing methods have been evaluated for Staphylococcus aureus strain characterization. The emergence of contemporary strains of community-associated S. aureus, and the ensuing epidemic with a predominant strain type (USA300), necessitates re-evaluation of the discriminatory power of these typing methods for discerning molecular epidemiology and transmission dynamics, essential to investigations of hospital and community outbreaks. We compared the discriminatory index of 5 typing methods for contemporary S. aureus strain characterization. Children presenting to St. Louis Children's Hospital and community pediatric practices in St. Louis, Missouri (MO), with community-associated S. aureus infections were enrolled. Repetitive sequence-based PCR (repPCR), pulsed-field gel electrophoresis (PFGE), multilocus sequence typing (MLST), staphylococcal protein A (spa), and staphylococcal cassette chromosome (SCC) mec typing were performed on 200 S. aureus isolates. The discriminatory index of each method was calculated using the standard formula for this metric, where a value of 1 is highly discriminatory and a value of 0 is not discriminatory. Overall, we identified 26 distinct strain types by repPCR, 17 strain types by PFGE, 30 strain types by MLST, 68 strain types by spa typing, and 5 strain types by SCCmec typing. RepPCR had the highest discriminatory index (D) of all methods (D?=?0.88), followed by spa typing (D?=?0.87), MLST (D?=?0.84), PFGE (D?=?0.76), and SCCmec typing (D?=?0.60). The method with the highest D among MRSA isolates was repPCR (D?=?0.64) followed by spa typing (D?=?0.45) and MLST (D?=?0.44). The method with the highest D among MSSA isolates was spa typing (D?=?0.98), followed by MLST (D?=?0.93), repPCR (D?=?0.92), and PFGE (D?=?0.89). Among isolates designated USA300 by PFGE, repPCR was most discriminatory, with 10 distinct strain types identified (D?=?0.63). We identified 45 MRSA isolates which were classified as identical by PFGE, MLST, spa typing, and SCCmec typing (USA300, ST8, t008, SCCmec IV, respectively); within this collection, there were 5 distinct strain types identified by repPCR. The typing methods yielded comparable discriminatory power for S. aureus characterization overall; when discriminating among USA300 isolates, repPCR retained the highest discriminatory power. This property is advantageous for investigations conducted in the era of contemporary S. aureus infections.