| Objective:1.To develop an Ultra Performance Liquid Chromatography method for the determination of drug concentrations in children with epilepsy in Xinjiang,which was administrated of Lamotrigine(LTG)and applied to the individualized treatment of epileptic children.2.A total of 300 cases of children with epilepsy whose blood concentration reached steady in the People’s Hospital of the autonomous region were collected,and the factors affecting the blood concentration of the antiepileptic drug lamotriginein in children with epilepsy in Xin Jiang were analyzed.3.A genetic algorithm-backpropagation(GA-BP)propagation neural network prediction model for lamotrigine blood concentrations was constructed to initially provide concentration prediction for children who need to adjust the dose of lamotrigine or who cannot be monitored for lamotrigine blood concentrations,allowing rapid and effective dosing regimens to be developed in the clinic based on individualised situations.Methods: 1.Chromatographic conditions: column: Waters ACQUITY UPLC BEH C18 column(2.1mm×50.0 mm,1.7 μ m),mobile phase: acetonitrile-ammonium acetate solution(10 mol/L,p H adjusted to4.1 with formic acid)= 20:80,detection wavelength: 240 nm,column temperature: 30 ℃,flow rate: 0.2 m L/ min.The methodological validation of the LTG blood concentration assay was carried out under these conditions.2.A total of 300 cases of children with epilepsy who were regularly taking LTG at steady-state blood concentrations from October 2015 to October 2021 in the People’s Hospital of the Autonomous Region were retrospectively collected,and the children were grouped by age,gender,ethnicity and combined medication.The LTG dose and blood concentration of the grouped children were subjected to ANOVA analysis of variance;the other continuous variables were subjected to multiple linear regression analysis.3.The 300 children were divided into modelling group(240 cases)and model validation group(60 cases),and the variables were screened using three methods such as principal component analysis and multiple linear regression analysis.The variables were selected by trial-and-error method,and the selected variables were applied in the artificial neural network to predict blood drug concentrations.The prediction results of the models established by the variables in each group were compared,and the optimal model was selected,and the GA-BP neural network model was finally constructed to predict the blood concentration of epileptic children in Xin Jing who were taking LTG.Result: The method of determining LTG blood levels by UPLC has good specificity and other substances in the patient’s serum do not interfing the concentration determination.The retention time of LTG is 1.91 min.The linearity of LTG is good at the range of 2.50～40.00 μg/m L and the related coefficient correlation is0.9998,the linear regression equation is Y=6506X-5236.6,the limit of quantification is 2.5 ug/m L,and the lowest limit of detection is 1ug/m L;the intra-day and inter-day accuracries ranged from 97.80% to104.68%,the precision ranged from 0.58% to 1.56%,the intra-day and inter-day RSD values are less than15%,and the extraction recoveries are 97.98% ～ 103.70%.2.300 children with epilepsy were collected in the study,including 166 males and 134 females,aged(8.49 ± 4.01)years,weighing(32.01 ± 17.91)kg and standing(131.19 ± 25.82)cm,were included in the study.Height,age,sex,dose,combined medication,body mass index,albumin,glutathione,lymphocyte percentage,neutrophil percentage,red blood cell count,haemoglobin,body weight,white blood cell count,total protein,haemoglobin,and haematocrit were analysed.The results showed that the factors such as leukocyte count,platelet count,ethnicity,concentration-dose ratio in children aged 0 to 3 years and 6 to 16 years,and LTG combined with glucuronidation inhibitor group have statistically significant difference(all P < 0.05).While there was no statistically significant difference between factors such as gender,concentration-dose ratio in children aged3 to 6 years,LTG alone,and LTG combined with glucuronidation inducer group(all P > 0.05).3.The prediction model was established by GA-BP neural network for the screened variables,and 60 children with epilepsy in the model validation group were used to predict blood drug concentrations.The percentage of error values less than 15% is 93.33%,the mean prediction error value is 0.23%,the mean absolute error value is 3.92%,and the correlation coefficient between the predicted and measured values is 0.9864,the prediction result is more satisfactory.Conclusions:1.A UPLC method for the determination of LTG blood concentration is successfully constructed.2.Using multiple linear regression analysis,factors of white blood cell count,platelet count,age and combined medications that would affect the level of LTG drug concentration in patients are screened,and provided a reference for the subsequent screening of variables for the establishment of concentration prediction models.3.A GA-BP neural network is used to successfully construct a LTG blood concentration prediction The model suggested that the GA-BP neural network could be used to predict LTG blood concentration in children with epilepsy in Xinjiang,and could be used for individualized clinical drug administration. |
PDF Full Text Request