3-Feature Classification Scheme For Ketosis-Prone Diabetes And Predictive Value Of Ketosis-Prone Diabetes Classification Schemes

Part 1 Three-feature classification scheme establishmentObjective: The aim of the study is to establish a three-feature classification scheme and explore the predictive value of A(antibody) B(β-cell function)and the new classification scheme inβ-cell function change between onset and 3 years follow-up.Design: Cross-section and case-control study.Subjects and Methods: Consecutive KPD patients (n=166) followed more than 1 year were noted general data, detected autoantibodies associated withβ-cell and HLA-DQ genotyping. The preservedβ-cell function was defined by the FCP=333pmol/L after 1 year follow-up. Finding the best cutoff for baseline FCP and BMI. Using Logistic regression to find the best affected features among some baseline features. The pick-up feature plus AB is ABC classification scheme. Compared the FCP value between onset and 3 years follow-up with AB and ABC. Finally, estimating the predictive value of the two classification schemes inβ-cell function change.Results: The best cutoff is 266pmol/L for baseline FCP and 22.3Kg/m~2 for BMI. The baseline FCP, BMI and antibody are the best affected features under the judge of 1 year follow-up FCP. The maximal FCP changed value between 1 to 2 year follow-up is in A-B- group. The FCP changed curve is more smooth in A+B- and A-B+ groups, but more abrupt in A+B+,A-B- groups. C+ and C- groups in A-B+ had the maximal FCP changed value in all ABC groups.Conclusion: BMI is the optimal third feature for AB classification scheme. AB classification scheme may have the predictive value inβ-cell function change for KPD patients.Part 2 Comparison accuracy and predictive value of classification schemes for ketosis-prone diabetesObjective: To explore accuracy and predictive value of classification schemes for one year follow-upβ-cell function in ketosis-prone diabetes.Design: Cross-section and case-control study.Subjects and Methods: If the one year follow-up FCP=333pmol/L, theβ-cell function was defined as preserved, if the one year follow-up FCP＜333pmol/L, theβ-cell function was defined as present. First of all, comparison AB classification scheme and other 1-feature or 2-feature classification schemes. Insulin treatment (Ⅰ) and HLA-DQ gene(G) are the two third-feature for AB classification scheme, so-called ABI and ABG classification Schemes. Comparison AB, ABC and the mentioned 3-feature classification schemes. Finally, Comparison AB and ABC classification schemes in DKA onset patients.Results: AB classification scheme is the most accuracy one in all 1-feature and 2-feature classification schemes. Sensitivity, Yueden's index and ROC AUC are higher in AB classification scheme than in others. ABC classification has the more higher sensitivity, specificity, Yueden' index and ROC AUC than AB scheme. Comparison AB and ABC classification schemes shared the same results in DKA onset patients.Conclusion: AB scheme is more accuracy than other one or two features schemes. ABC may have more accuracy and predictive value than AB scheme, but the further study must be followed.