Multivariate statistical analysis methods to investigate interindividual glucose dynamics for subjects with type 1 diabetes mellitus

Background: To measure the relationship between variability in HbA1c and microalbuminuria (MA) and diabetic retinopathy (DR) in the long term. Methods: A prospective case-series study, was conducted on 366 Type 1 Diabetes Mellitus patients with normoalbuminuria and without diabetic retinopathy at inclusion. The cohort was followed for a period of 12 years. The Cox survival analysis was used for the multivariate statistical study. The effect of variability in microangiopathy (retinopathy and nephropathy) was evaluated by calculating the standard deviation of HbA1c (SD-HbA1c), the coefficient of variation of HbA1c (CV-HbA1c), average real variability (ARV-HbA1c) and variability irrespective of the mean (VIM-HbA1c) adjusted for the other known variables. Results: A total of 106 patients developed diabetic retinopathy (29%) and 73 microalbuminuria (19.9%). Overt diabetic nephropathy, by our definition, affected only five patients (1.36%). Statistical results show that the current age, mean HbA1c, SD-HbA1c and ARV-HbA1c are significant in the development of diabetic retinopathy. Microalbuminuria was significant for current age, mean HbA1c, CV-HbA1c and ARV-HbA1c. Conclusions: By measuring the variability in HbA1c, we can use SD-HbA1c and ARV-HbA1c as possible targets for judging which patients are at risk of developing DR and MA, and CV-HbA1c as the target for severe DR.

Download Full-text

Influence of initial insulin dosage on blood glucose dynamics of children and adolescents with newly diagnosed type 1 diabetes mellitus

Pediatric Diabetes ◽

10.1111/pedi.12374 ◽

2016 ◽

Vol 18 (3) ◽

pp. 196-203 ◽

Cited By ~ 3

Author(s):

Yi Wang ◽

Chunxiu Gong ◽

Bingyan Cao ◽

Xi Meng ◽

Liya Wei ◽

...

Keyword(s):

Diabetes Mellitus ◽

Type 1 Diabetes ◽

Blood Glucose ◽

Type 1 Diabetes Mellitus ◽

Children And Adolescents ◽

Insulin Dosage ◽

Newly Diagnosed ◽

Glucose Dynamics

Download Full-text

Metabolomic profiling of acute diabetes complications in children with type 1 diabetes mellitus

Postępy Polskiej Medycyny i Farmacji ◽

10.5604/01.3001.0014.1586 ◽

2020 ◽

Vol 7 ◽

pp. 1-9

Author(s):

Beata Małachowska ◽

Wojciech Fendler

Keyword(s):

Diabetes Mellitus ◽

Type 1 Diabetes ◽

Statistical Analysis ◽

Decision Tree ◽

Type 1 Diabetes Mellitus ◽

Acute Type ◽

Metabolic Disturbances ◽

Negative Ionization ◽

Group Comparisons

Acute type 1 diabetes mellitus (T1DM) complications – diabetes ketoacidosis (DKA) and hypoglycemia (HG) – are dangerous not only as a threat to patients’ life but also because of their long-term sequelae. Aim: Evaluation of serum metabolic changes caused by episode of DKA and HG, that can be detected despite restoring parameters typically changed during the episodes. Selection of putative long-standing biomarkers of past episodes of DKA and HG. Materials and methods: Four groups of children with T1DM were recruited: patients after episode of DKA and HG, children with established T1DM (EDM) and patients with newly diagnosed diabetes without diabetes ketoacidosis (NDM). Serum samples were collected in three group-specific time points (since the hospital admission): 0h – 24h – 72h for DKA and NDM group and 0h – 12h – 48h for HG group. From EDM patients only one sample was collected during running routine laboratory tests. Patients were assigned to two batches: DKA-NDM-EDM (N = 20x3, N = 10x3, N = 10) and HG-EDM- -NDM (N = 10x3, N = 25, N = 15x3). All patients within the batches were matched based on age and sex. Metabolic fingerprinting was performed with LC- -QTOF-MS (Agilent 6550 iFunnel). Results: In DKA batch after technical filtering 248 metabolomic features out of 712 (in positive ionization) and 295 out of 652 (in negative ionization) were suitable for between-group comparisons. Statistical analysis selected 22 metabolic features as putative biomarkers of episodes of DKA occurrence in nearest 72h. Decision tree to diagnose past DKA episode, based on two best metabolites, achieved sensitivity of 95% (CI (confidence interval): 81.79–99.13%) and specificity of 80% (CI: 67.30–88.81%). In HG batch after technical filtering 359 metabolomic features out of 1006 (in positive ionization) and 374 out of 763 (in negative ionization) were suitable for between-group comparisons. Statistical analysis selected 9 metabolic features as putative biomarkers of episodes of DKA occurrence in nearest 48h. Decision tree to diagnose past HG episode, based on two best metabolites, achieved sensitivity of 90% (CI: 72.32–97.38%) and specificity 80% (CI: 68.39–88.26%). Conclusions: Metabolic disturbances caused by DKA may be traced in serum up to 72h after the episode and for hypoglycemia up to 48h.

Download Full-text