A predictive model for lack of partial clinical remission in new-onset pediatric type 1 diabetes.

TitleA predictive model for lack of partial clinical remission in new-onset pediatric type 1 diabetes.
Publication TypeJournal Article
Year of Publication2017
AuthorsMarino KR, Lundberg RL, Jasrotia A, Maranda LS, Thompson MJ, Barton BA, Alonso LC, Nwosu BUdoka
JournalPLoS One
Volume12
Issue5
Paginatione0176860
Date Published2017
ISSN1932-6203
KeywordsAdenosine Triphosphatases, Adolescent, Age Factors, Anion Transport Proteins, Autoantibodies, Biomarkers, Body Mass Index, Child, Child, Preschool, Diabetes Mellitus, Type 1, Female, Glycated Hemoglobin A, Humans, Male, Models, Biological, Models, Statistical, Prognosis, Remission Induction, Risk, ROC Curve, Sex Factors, Treatment Outcome
Abstract

IMPORTANCE: >50% of patients with new-onset type 1 diabetes (T1D) do not enter partial clinical remission (PCR); early identification of these patients may improve initial glycemic control and reduce long-term complications.

AIM: To determine whether routinely obtainable clinical parameters predict non-remission in children and adolescents with new-onset T1D.

SUBJECTS AND METHODS: Data on remission were collected for the first 36 months of disease in 204 subjects of ages 2-14 years with new-onset type 1 diabetes. There were 86 remitters (age 9.1±3.0y; male 57%), and 118 non-remitters (age 7.0±3.1y; male 40.7%). PCR was defined as insulin-dose adjusted hemoglobin A1c of ≤9.

RESULTS: Non-remission occurred in 57.8% of subjects. Univariable analysis showed that the risk for non-remission was increased 9-fold in patients with 4 diabetes-associated auto-antibodies (OR = 9.90, p = 0.010); 5-fold in patients <5 years old (odds ratio = 5.38, p = 0.032), 3-fold in those with bicarbonate of <15 mg/dL at diagnosis (OR = 3.71, p = 0.008). Combined estimates of risk potential for HC03 and the number of autoantibodies by multivariable analysis, adjusted for BMI standard deviation score, showed HC03 <15 mg/dL with a clinically significant 10-fold risk (OR = 10.1, p = 0.074); and the number of autoantibodies with a 2-fold risk for non-remission (OR = 1.9, p = 0.105). Male sex and older age were associated with decreased risk for non-remission. A receiver-operating characteristic curve model depicting sensitivity by 1-specificity for non-remission as predicted by bicarbonate <15 mg/dL, age <5y, female sex, and >3 diabetes-associated autoantibodies had an area under the curve of 0.73.

CONCLUSIONS: More than 50% of children and adolescents with new-onset T1D do not undergo partial clinical remission and are thus at an increased risk for long-term complications of diabetes mellitus. A predictive model comprising of bicarbonate <15 mg/dL, age <5y, female sex, and >3 diabetes-associated autoantibodies has 73% power for correctly predicting non-remission in children and adolescents with new-onset T1D. Early identification of these non-remitters may guide the institution of targeted therapy to limit dysglycemia and reduce the prevalence of long-term deleterious complications.

DOI10.1371/journal.pone.0176860
Alternate JournalPLoS ONE
PubMed ID28459844
PubMed Central IDPMC5411061