| Background and ObjectiveSepsis is defined as a life-threatening organ dysfunction caused by the host’s unregulated response to infection.Sepsis is the main cause of morbidity and death in the intensive care unit(ICU),and it is also a major public health problem in the world.In the past ten years,despite the progress made in treatment,the fatality rate remains high.Due to the heterogeneous manifestations of sepsis and the complexity of pathogenicity,early diagnosis,timely control of the progress of sepsis,and effective treatment are still challenging.In order to reduce the mortality rate,an effective predictive method is still needed to guide specific treatment.Even if some clinical inflammatory markers,such as procalcitonin(PCT)and C-reactive protein(CRP),have certain accuracy in predicting sepsis,they still have great limitations Sex,and the fatality rate of sepsis remains high.Therefore,the identification and verification of reliable biomarkers for the diagnosis and prognosis of sepsis is a top priority.Metabolomics is an emerging discipline that studies endogenous small molecule compounds in the body.Nuclear magnetic resonance(NMR),gas chromatograph/mass spectrometer(GC/MS)and liquid chromatography/mass spectrometry(LC/MS)are commonly used methods in metabolomics.Changes in plasma metabolism are the main feature of sepsis[1].The basic research of our team found that the metabolites and intestinal flora of sepsis rats will change significantly[2,3]In this study,we used a non-targeted metabolomics method based on LC/MS to analyze the metabolites of 28-day,in-hospital,and 90-day sepsis survival and death patients,and screen potential differential metabolic biomarkers.In order to discover the different metabolites of the prognosis of patients with sepsis.Use metabolomics methods to detect blood metabolism spectrum of patients with sepsis,screen specific and sensitive metabolic markers,and conduct metabolic pathway analysis,and then combine the selected metabolites with statistically significant clinical indicators,In order to carry out the logistic regression prediction model.Explore sepsis-related metabolic processes and provide new evaluation indicators for the prognosis of sepsis.Materials and MethodsFrom March 2019 to September 2019,sepsis patients were included in the ICU of the First Affiliated Hospital of Zhengzhou University.According to the prognosis of patients in 28 days,hospital and 90 days,they are divided into survival at 28 days(28dS)and death group(28dD),hospital survival(HOS-survival)and death group(HOS-death),survival at 90 days(90dS)and death Group(90dD)three groups.Collect plasma samples from patients who meet the sepsis 3.0 diagnostic criteria,and collect clinical data,such as sequential organ failure assessment(SOFA)and Glasgow coma score(Glasgow coma score,24 hours after the diagnosis of sepsis)GCS),acute physiology and chronic health evaluation scoring system(APACHE-Ⅱ),blood routine,liver function indexes,renal function indexes and comprehensive ICU mortality rate.After pretreatment,the samples are subjected to metabolic profile testing,metabolite identification,screening and bioinformatics analysis,and further screening of pathways related to the onset and development of sepsis.Finally,binary logistic regression and receiver operating characteristic curve(receiver operating characteristic curve,ROC curve)analysis were performed to screen out potential marker metabolites with high specificity and sensitivity.Then,the selected metabolites were compared with statistically significant metabolites.The clinical indicators are combined to perform a logistic regression prediction model.ResultsFrom March 2019 to September 2019,96 patients with sepsis were included in the ICU of the First Affiliated Hospital of Zhengzhou University.There were 49 deaths at 28 days,with a case fatality rate of 51.0%;53 died in the hospital,with a case fatality rate of 55.2%;54 cases died at 90 days,with a case fatality rate of 56.3%.Compared with 28dD,the heart rate,serum cholinesterase,SOFA score,APACHE-Ⅱ score,respiratory support,and blood pressure drug levels of 28dS patients were significantly reduced,and the 24-hour urine output level was significantly increased;Compared with HOS-death,the heart rate,monocyte count,absolute value of lymphocytes,red blood cell volume distribution width,SOFA score,APACHE-Ⅱscore,respiratory support level,hematocrit,average hemoglobin content,The alkaline phosphatase level was significantly increased;Compared with 90dD,90dS patients’heart rate,red blood cell volume distribution width,thrombin time,SOFA score,APACHE-Ⅱ score,respiratory support,and blood pressure drug levels were significantly reduced,and serum cholinesterase and lipase levels were significantly increased.Metabolomics analysis:In the orthogonal partial least squares discriminant analysis(OPLS-DA)analysis model,when the variable importance in projection(VIP)>1 and P When the value is less than 0.05,potential differential metabolites are screened out.A total of 4 different metabolites in the patient’s hospital,13 in 28 days,and 27 in 90 days were identified.The kyoto encyclopedia of genes and genomes(KEGG)metabolic pathway analysis was carried out,and the 28dD and 28dS patients were screened for metabolic disorders mainly concentrated in tyrosine metabolism,α-linolenic acid metabolism,alanine metabolism,Aspartate metabolism and glutamate metabolism pathway.Metabolic disorders in patients with 90dD and 90dS mainly focus on arginine and proline metabolism,folic acid synthesis,tyrosine metabolism and arachidonic acid metabolic pathways.The 28-day prognosis prediction model for patients with sepsis has a prediction accuracy of 77.08%and an area under curve(AUC)of 0.881,indicating indole acetic acid,3-methylene indole,heart rate,and respiration.The application of support and vasopressor drugs has a high value in predicting the 28-day prognosis of patients with sepsis,with sensitivity and specificity of 75.51%and 78.72%,respectively.The prediction model for the in-hospital prognosis of patients with sepsis has a prediction accuracy of 72.92%and an area under the diagnostic curve AUC of 0.830,indicating the absolute value of lymphocytes,alkaline phosphatase(ALP),SOFA and L-α-amino-The 1H-pyrrole-1-hexanoic acid combination has a higher value in predicting the in-hospital prognosis of patients with sepsis.The sensitivity and specificity are 73.58%and 72.09%,respectively.The 90-day prognosis prediction model for patients with sepsis has a prediction accuracy of 80.21%,and the area under the diagnostic curve AUC is 0.892,indicating that the application of pyrrolidine,dopamine,heart rate,respiratory support,and blood pressure medicines combined to predict the 90-day prognosis of patients with sepsis.The value is higher,the sensitivity and specificity are 83.33%and 76.19%,respectively.ConclusionThis study used non-target metabolomics methods to identify blood metabolites and metabolic pathways in patients with sepsis,and further found that the changes in blood metabolites in patients with sepsis at 28 days,hospital and 90 days were related to the prognosis.The 28 days,hospital and 90 days prognosis prediction models of sepsis patients were constructed by using plasma marker metabolites and clinical indicators with statistical significance,and the model has certain prognosis efficacy,but more studies are needed to prove the effectiveness of the research results. |
