Application Of The Prediction Model Of Warfarin Steady-state Dose In Anticoagulation After Valve Replacement Surgery

Objective: This study aims to analyze the relationship between demographic, clinical factors and stable dose of warfarin of patients with valve replacement operation. To research individual and cross impactation of the related gene(VKORC1 and CYP2C9) polymorphism on steady dose of warfarin. Using five warfarin steady dose prediction model to predict the steady state dose. of all patients and each subgroup. At the same time, analysis the prediction efficiency of the warfarin steady dose prediction model to predict postoperative 5 days whether enter the therapeutic window for analysis.Methods: Retrospective collection 210 patients with warfarin anticoagulation after valve replacement surgery from 2014 February to 2014 October in department of cardiovascular surgery of Tianjin Chest Hospital. To collect relevant demographic information, including age, gender, height, weight, body surface area(BSA), smoking history and other demographic factors; clinical factors, including hypertension, diabetes, atrial fibrillation, operation of cerebral stroke, cardiac function grade, combined drugs; collect after anticoagulation related data, including warfarin steady-state dosing, INR in postoperative 5 days.(1) the influence of the demographic factors and the clinical effect on the steady state dose of Hua Falin. According to the steady-state warfarin dose of patients, all the patients were divided into three groups: low dose group(less than or equal to 2.25mg/day), middle dose group 2.25-4.5mg/d, high dose group(more than or equal to 4.5mg/day). The demographic factors and clinical factors of the three groups were compared.(2) To study VKORC1 1173, VKORC1-1639, CYP2C9*2 and CYP2C9*3 in all the patients expressed distribution. Analysis of these genotype’s individual and cross effect on warfarin steady-state dosing and INR in postoperative 5 days.(3) Compare the effectiveness of the existing dose models at home and abroad, bycomparing the predictive absolute error(MAE) and the predicted percentage, find the most closed to the actual dose of warfarin. Use warfarin steady-state dosing model to predict that INR on postoperative 5 days whether to enter the therapeutic window. Results:(1) Age increased more reflected in the elderly population, in the adult population, age for steady-state warfarin dose effect is relatively weak. There is a weak positive correlation with weight factors(including height, weight, BSA) and steady state warfarin dose, more function in underweight and overweight individuals. Clinical factors and other demographic factors do not have a greater impact on the dose of warfarin.(2) This study found that in VKORC1-1639, the average steady-state dose of AG/GG group(4.02 + 0.882 mg/day) is higher than the AA group(3.09 + 0.728mg/day), and has statistical difference(P < 0.001). The average warfarin steady dose of CYP2C9*1*1(3.24 + 0.811mg/day) is higher than *1*2/*1*3 group(2.82 + 0.623mg/day), and with significant difference(P = 0.043). According to the WRI group, compare the average warfarin steady dose of three groups: WRI=0 group: 2.82 + 0.624mg/day, WRI=1 group: 3.12 + 0.735mg/day, group WRI=2: 4.02 + 0.883mg/day, and has a statistically significant difference(P < 0.001). This suggests that VKORC1-1639 and CYP2C9*3 have cross effects on warfarin stable dose.(3) Comparison of the INR of postoperative 5th day, VKORC1 AA group(1.72 + 0.536), AG/GG group(1.44 + 0.385), with statistical difference(P=0.015). On the 5 day after surgery, the proportion of patients in the treatment window in the AG/GG group was significantly lower than that in the AA group. This shows that VKORC1 has a greater impact on the INR of the 5th days after surgery. Comparison of INR of the postoperative 5th day, CYP2C9*1*1 group(1.68 + 0.522), *1*2/*1*3 group:(1.78 + 0.594), no statistical difference(P=0.452). The proportion of patients in the treatment window on the postoperative 5th day was no significant difference between the two groups(P=0.634). The proportion ofpatients in the treatment window on the postoperative 5th day in different WRI group respectively was WRI=0(52.9%), WRI=1(59.1%), and WRI=2(31.8%). The proportion of patients with WRI=2 after 5 days of the lower sensitivity of the group was significantly lower.(4) Only FDA guidance overestimate stable dose of warfarin, the other four models all underestimate the steady dose of warfarin. There were statistically significant differences between the five models(P < 0.001) in the predicted dose and the actual steady state dose. MAE minimum is the IWPC prediction model(0.65 + 0.558mg), secondly HUANG prediction model(0.67 + 0.524mg). Ideal prediction percentage of FDA guidance, IWPC dose model and HUANG prediction model were more than 50%, respectively, 57.4%, 58.5%, 59.6%. After the sub group analysis, the IWPC dose model and HUANG prediction model in the vast majority of subgroups in the ideal prediction of the percentage of more than 50%, only in the high dose group and high sensitivity group, the prediction performance is slightly worse. And other models in most of the subgroups the predictily ability is poor.(5) This study also applied subjects operating curve(ROC) to evaluate the prediction performance of the warfarin steady dose prediction model used in postoperative 5 days whether to enter the therapeutic window of warfarin in the application significance. The results indicated that the AUC prediction model of HUANG was: 0.602, other dose prediction model AUC were not more than 0.6. Conclusion:(1) Age take more effect on steady-state warfarin dose in the elderly. There is a weak positive correlation with weight factors(including height, weight, BSA) and steady state warfarin dose, more function in underweight and overweight individuals. Clinical factors and other demographic factors do not have a greater impact on the dose of warfarin.(2) VKORC1-1639 and CYP2C9 polymorphism affected the warfarin steady-state dose, in the present study-1639 VKORC1 have a larger influence than CYP2C9. At the same time they have cross effects on steady-state warfarin dose.(3) VKORC1-1639 had a greater impact on INR of postoperative 5th day, while CYP2C9 had little effect. The proportion of patients in the treatment window on the postoperative 5th day in WRI=2 was significantly lower than other groups.(4) Comprehensive evaluation, the ideal prediction of the percentage of more than 50% that IWPC model and HUANG model in the whole group and the majority of sub groups, only in the high dose group and low sensitivity group, the prediction performance is slightly worse. And other models in the whole group or most of the sub group, the predictive ability is poor.(5) Warfalin steady-state dose prediction model to predict whether the INR enter the treatment window on 5th day after surgery is poor.