| Sepsis is defined as life-threatening organ dysfunction caused by a dysregulated host response to infection.At present,there is no gold standard for clinical sepsis diagnosis.It can only be identified through a series of clinical symptoms and scores of suspected infected patients,which has caused great obstacles to the diagnosis,treatment and related studied.In this study,the differences in plasma and urine metabolic profiles of sepsis and non-sepsis patients in Intensive care units were studied using UPLC-Q-TOF/MS-based non-targeted metabolomics.And 64 potential biomarkers were screened and identified between the two groups,and is associated with the changes of metabolic pathways in sepsis patients.Binary Logistic regression was used to search for risk factors in traditional physiological indicators and potential biomarkers.Based on these risk factors,Lasso regression predictive model was established to evaluate their diagnostic ability,and self-validation was carried out.According to the occurrence of shock or not,sepsis patients were divided into two subgroups:septic shock group and non-septic shock group.We investigate the differences in metabolic profiles between the septic shock and the non-septic shock group to look for potential biomarkers that can distinguish them.Patients with sepsis were divided into control group and vancomycin group according to the use of antibiotics.The appropriate experimental procedure was established by urine metabolomics to find biomarkers that could evaluate the efficacy of vancomycin.Based on this process,the same metabolomics method was used to study the efficacy biomarkers of mesalazine in the mouse UC model induced by sodium dextran sulfate to verify the feasibility of this experimental process and help doctors rationally use drugs in the future.Research contents and results:（1）Basic information and physiological indicators of sepsis and control group in ICUs were collected and recorded,and clinical scores of patients were evaluated.The results showed that there were no significant differences between the two groups in age,gender,length of hospital stay and ICUs stay.Respiratory Rate,Alveolar-arterial oxygen differential pressure,C-reactive protein and procalcitonin were significantly increased in the sepsis group,while Ca2+and p H were significantly decreased in the sepsis group,indicating respiratory distress syndrome,acidosis,hypocalcemia and inflammation in the sepsis group.The APACHE-Ⅱscore of the sepsis group was significantly higher than that of control group,indicating that the situation of the sepsis group was more serious than that of control group and the prognosis was poor.The plasma metabolomics and lipidomics and urine metabolomics were studied based on UPLC-Q-TOF/MS.Fifty-five potential biomarkers from plasma and potential biomarkers from urine were screened out.These potential biomarkers were involved in seven metabolic pathways.Tryptophan metabolism and glycerolphospholipid metabolism have the greatest impact.（2）Based on binary Logistic regression analysis,5 risk factors for sepsis were found in the traditional physiological indicators,and 10 risk factors for sepsis were found in the potential biomarkers,and the correlation between them was studied.Lasso regression models were established by these risk factors to evaluate their diagnostic ability.In the Lasso regression predictive model of traditional physiological indicators,the AUC is 0.968,95%CI is 0.934-1.000,sensitivity is 80.6%,specificity is 100%.In the Lasso regression predictive model of potential biomarkers,the AUC is 0.986,95%CI is 0.963-1.000,sensitivity is 100%,specificity is 93.3%.The potential biomarkers have better diagnostic ability than traditional physiological indicators in the Lasso regression predictive model.In order to further increase the diagnostic ability of the model,we combined traditional physiological indicators and potential biomarkers to establish a Lasso regression predictive model.The AUC value of the combined Lasso regression model is 1.000,95%CI is 1.000-1.000,sensitivity is 100%,specificity is 100%,which was superior to the Lasso regression model established by traditional physiological indicators and potential biomarkers respectively.Finally,after self-verification,it was found that the AUCs of traditional physiological indicators combined with potential biomarkers in training set and test set were all more than 0.95,indicating that the predictive model was reliability.（3）According to the occurrence of shock or not,sepsis patients were divided into two subgroups:septic shock group and non-septic shock group.The information of sepsis patients was collected to compare the changes in basic information,physiological indicators and clinical scores between the two subgroups.Results show that the Alveolar-arterial oxygen differential pressure of septic shock group is significantly higher than that of non-septic shock.The scores of q SOFA,SOFA and APACHE-Ⅱin the septic shock group were significantly higher than those in the non-septic shock group,indicating that the situation and prognosis of the septic shock group were worse than non-septic shock group.The Glasgow Coma Scale of the septic shock group was significantly lower than that of non-septic shock because of shock.PE-Cer（D16:1（4E）/19:0）was found in plasma metabolomics and significantly higher in non-septic shock group than that of septic shock group,and it was significantly correlated with q SOFA（P<0.05）.By ROC analysis,the AUC value of PE-Cer（D16:1（4E）/19:0）is0.9471,95%CI is 0.8613-1.000,the sensitivity is 100%,and the specificity is 90.48%.It can be used to predict septic shock.（4）Patients with sepsis were divided into the control group and the vancomycin group according to the use of antibiotics.Physiological indicators and clinical scores were recorded.The appropriate experimental procedures were established based on urine metabolomics to find biomarkers that could evaluate the efficacy of vancomycin.The urea and APACHE-Ⅱscore were significantly increased in the vancomycin group,which suggests that the vancomycin group was more severe.Although we could not find the efficacy biomarkers of vancomycin.But the experiment process of metabolomic that find the efficacy biomarkers of drug in inflammatory diseases were established.Based on this process,the same metabolomics method was used to identify four biomarkers that could evaluate the efficacy of mesalazine in the mouse model of UC induced by sodium dextran sulfate.This indicates that this experimental process is feasible,and more biomarkers of drug efficacy for inflammatory diseases may be found through metabolomics to help doctors rationally use drugs in the future.Summary and Outlook:In this study,the potential biomarkers between the sepsis group and the control group were identified based on metabolomics,and the combined Lasso regression model was used to greatly improve the diagnostic ability of sepsis on the existing basis.Metabolomics in two subgroups of septic shock group and non-shock group identified a potential biomarker PE-Cer（D16:1（4E）/19:0）with good diagnostic ability.Finally,the experiment process of metabolomic that find the efficacy biomarkers of drug in inflammatory diseases were established.And the experimental process was found to be feasible after verification.However,few subjects were included in this study,and larger experiments should be conducted in the future to verify the potential biomarkers of sepsis.In the subgroup study,it is possible to expand the experimental scale,extend the observation time and collect samples at more time points in the future,in order to find suitable biomarkers for the diagnosis of septic shock.As for the experiment process of metabolomic that find the efficacy biomarkers of drug in inflammatory diseases,more samples related to inflammatory diseases can be collected in the future to verify the experimental process and find biomarkers of drug efficacy for more inflammatory diseases. |
PDF Full Text Request