Development of disease progression model in korean patients with type 2 diabetes mellitus

Abstract

Type 2 diabetes mellitus (T2DM) is a progressive disease. Decline in β-cell function, increase in insulin resistance with concomitant increase in fasting plasma glucose are the natural sequelae of T2DM. Although some pharmacotherapies for patients with type 2 diabetes mellitus have shown to slow the rate of progression or improve disease status, none of them has been proven to halt disease progression. Nevertheless, based on meta-analysis of existing data, it is hoped that more intensive treatment compared to standard glycemic control would significantly reduce coronary events without a parallel increase in mortality and morbidity. In such regards, disease progression of T2DM might be controlled by adopting proper treatment strategies. Pharmacometrics is a cutting-edge area of clinical pharmacology which is defined as the science that quantifies drug action, disease progression and patient information to aid efficient drug development, regulatory decision, and treatment in actual clinical practice. A disease model, represented by baseline disease status, natural disease history and drug (and placebo) response, describes the time course of disease progression, as measured by biomarkers or clinical outcomes, segregated from the drug response. The purpose of this study is to develop a disease progression model for Korean patients with T2DM and to apply the developed model as a supportive tool to designing an optimal treatment regimen in this clinical population using NONMEM.A total of 347 new or established patients who visited Severance hospital for the first time from 2006 to 2012 were retrospectively investigated after screening 888 patients. The patients were prescribed 0 to 3 kinds of oral hypoglycemic agents (OHAs) at first visit according to their disease status by endocrinology specialists. The primary endpoints measured were fasting plasma glucose (FPG, mmol/L) and glycated hemoglobin (HbA1c, %). We performed the analyses according to the intention-to-treat principle. Disease progression was modeled using the natural history and drug effect influencing the baseline (“offset”) or the progress rate (“slope”), or both. The constant, linear, simple Emax, sigmoid Emax and power models were evaluated to describe the treatment effect. The concept of effect compartment was applied to explain the delay in time course between pharmacokinetic steady state and pharmacodynamic steady state. The number of OHAs was incorporated into the structural model. Next, stepwise covariate model building was performed. The relationship between the structural parameters and the classes of the OHAs were investigated in this step. The final model was evaluated by the visual predictive check (VPC) for 1000 simulations. The median age of eligible patients (185 males, 162 females) was 61 years (19 – 85). Metformin was most frequently prescribed followed by sulfornylureas. The greater the number of OHA classes prescribed, the faster were the dropout rates. The number of patients without drug prescription was 19 and these patients were excluded from the population analysis because they were not suitable candidates for a placebo effect group. A total of 328 patients were prescribed 1 to 3 classes of OHAs. The number of OHAs positively correlated with the level of FPG or HbA1c acquired from initial laboratory tests. Linear models for offset effect and constant model for slope effect were incorporated into the final disease progression model for FPG and HbA1c. FPG and HbA1c increased naturally 0.197 mmol/L/year (3.55 mg/dL/year) and 0.181%/year, respectively. The amplitude of offset effect increased with the number of prescribed drugs in both FPG and Hb1Ac, though for HbA1c, there were little differences between 2 drugs and 3 drugs. The equilibration half-lives for offset effect were 14 days and 35 days for FPG and HbA1c, respectively. Therefore, it took around 55 to 69 days and 140 to 175 days for FPG and HbA1c, respectively to reach pharmacokinetic and pharmacodynamic equilibrium state. The disease progression rates of the Korean patients with type 2 diabetes mellitus who visited Severance hospital were similar to those of western patients, with rates for fasting plasma glucose and glycated hemoglobin 0.039 to 0.31 mmol/L and 0.07 to 0.24 %/L, respectively. Fasting plasma glucose decreased faster than glycated hemoglobin at the beginning of treatment. Treatment effect decreased inversely proportional to diabetes duration. The greater the number of prescribed drugs, the bigger was the offset effect. I will build disease progression model with greater utility after undertaking further studies.