The Value Of High-Throughput Sequencing In The Diagnosis Of Infectious Diseases Of The Central Nervous System

Objective:1.To investigate the sensitivity and specificity of high-throughput sequencing in the diagnosis of the central nervous system(CNS)viral infections,tuberculous meningitis,purulent meningitis and cryptococcal meningitis.2.To analyze the role of different number of unique reads in distinguishing infectious diseases of central nervous system.3.To analyze the differences in the detection results of mNGS in different stages of diseases.4.To investigate the differences of NGS results of different infected sites when different samples were selected.Materials and Methods:From November 2016 to February 2019,201 cases were prospectively enrolled.According to their diagnoses,they were classifed into four different groups:CNS viral infections(n=70),tuberculous meningitis(n=39),purulent meningitis(n=38)and intracranial fungal infection(n=24).Either cerebrospinal fluid(CSF)or blood samples was collected for DNA extraction and library preparation during routine diagnosis and treatment.Quality qualified libraries were sequenced on the BGISEQ-500/50 platform.Sequencing data were generated after biological information analysis,and the pathogen was interpreted according to the clinical information of patients.The diagnoses can be made according to the patients’ clinical information and other auxiliary examinations.The sensitivity,specificity,positive predictive value and negative predictive value of NGS for the diagnosis of infectious diseases of CNS were calculated,and thereby to compare the accuracy of different unique reads number criteria in the diagnosis of CNS infectious diseases by NGS.Results:1.When it was defined as the number of unique reads>2,the area under ROC curve of diagnosing CNS viral infections by NGS was the largest(0.651)in patients with confirmed CNS viral infections.The positive consistent percentage,the negative consistent percentage and the total consistent percentage were 40.8%,89.3%and 76.1%,respectively.When the infection involved the meninges only,the positive rate of NGS in CSF group(100%)was higher than that in the blood group(60%).Meanwhile,the number of unique reads and gene coverage in CSF group were higher than those in the blood group(p<0.05).When the infection involved the brain parenchyma only,the number of unique reads and gene coverage of NGS in brain tissue were higher than those in the CSF.2.When it was defined as the number of mapping reads number(MRN)of mycobacterium tuberculosis complex(MTC)≥ 1,the area under ROC curve of diagnosing tuberculous meningitis by NGS was the largest(0.610)in patients with confirmed and probable tuberculous meningitis.The positive consistent percentage,the negative consistent percentage and the total consistent percentage were 25.6%,96.3%and 82.6%,respectively.In patients with tuberculous meningitis diagnosed by gold standard,the sensitivity,specificity,positive predictive value and negative predictive value of NGS were 60%,96.3%,33.3%and 98.7%,respectively.3.When it was defined as the number of unique reads>5,the area under ROC curve was the largest(0.892)in patients with purulent meningitis diagnosed by gold standard.And the sensitivity,specificity,positive predictive value and negative predictive value were 83.3%,95.1%,55.6%and 98.7%,respectively.4.When it was defined as the number of unique reads>2,the sensitivity,specificity,positive predictive value and negative predictive value of NGS in the diagnosis of cryptococcal meningitis were 75%,99.47%,90%and 98.43%,respectively.Conclusions:1.The number of unique reads>2 should be used as the positive standard for NGS in the diagnosis of CNS viral infection.It is necessary to apply NGS for CSF in the early stage of CNS viral infections.When the infection involves the meninges only,the value of using CSF for NGS was is recommended as compared with using blood.2.When Mycobacterium tuberculosis complex was detected by NGS,it is suggested that at least one read mapped onto the species or genera could be used as the positive result.The diagnostic performance of NGS for predicting tuberculous meningitis needs to be further improved.At present,NGS can be used as an auxiliary diagnostic method and as one of the exclusive diagnostic methods.3.It is suggested that the number of unique reads>5 should be used as the positive standard for NGS in the diagnosis of purulent meningitis.The NGS results of purulent meningitis are multiple and the differentiation of colonization from infection is a challenge in the implementation of NGS tests.4.NGS was found to serve as a useful tool for the identification of Cryptococcus species,which is helpful to manage cryptococcal meningitis.We recommend that NGS,along with India ink staining,culture methods,and CrAg,should be used together and thereby improve the diagnostic accuracy of cryptococcal meningitis.5.Although NGS has a general diagnostic ability for CNS viral infections and tuberculous meningitis,it is undoubtedly a meaningful test in the current predicament of clinical diagnosis.NGS has good diagnostic ability for purulent meningitis and cryptococcal meningitis.Clinicians should synthetically identify the pathogens based on clinical evidence and high-throughput sequencing results.