Metagenomic Next-generation Sequencing of Cerebrospinal Fluid for the Diagnosis of Central Nervous System infections in HIV-infected patients

Background HIV-infected patients have extremely low immunity and various opportunistic infections. Early diagnosis and treatment of these pathogens is critical for patients with HIV infection, especially those with central nervous system (CNS) infections. Metagenomic next generation sequencing (mNGS) has the advantage of identifying a broad range of pathogens and was suggested as a promising tool in the clinical diagnosis for infectious diseases. The clinical application of mNGS in the diagnosis of CNS infections in patients infected with HIV remains inadequately characterized.Methods We retrospectively analyzed data from 22 patients with suspected central nervous system infections who underwent both mNGS and conventional methods including culture, PCR, X-pert/RIF and antigen testing to explored the utility of mNGS in clinical diagnostic microbiology of CNS infections in HIV-infected patients.Results A total of 22 patients participated in the study between June 2018 and May 2019. The consistency of positive percentage of mNGS compared to clinical diagnosis was significantly higher than that of conventional methods (86.36% vs. 45.21%). The proportion of co-infections in mNGS positive samples was significantly higher than that in traditional methods (40.91% vs. 14.39%). Sixteen Extra Pathogens in 14 cases identified by metagenomic NGS only, 6 pathogens affected clinical reasoning and 7 pathogens guided antimicrobial therapy.Conclusions MNGS is a powerful diagnostic method for identifying pathogens in central nervous system infections and provide actionable information in some cases. MNGS technology has positive significance for the diagnosis and clinical treatment of central nervous system infection in HIV-infected patients.


Introduction
HIV infection is a global epidemic [1]. As of 2018, approximately 37.9 million people worldwide were infected with HIV [2]. In 2017, approximately 770,000 people died from diseases caused by HIV infection [2]. The HIV virus attacks human immune cells, causing low immunity. Central nervous system infections are destructive syndromes that remain undiagnosed in most cases [3] . Furthermore, some patient groups, such as those infected with HIV, who often show subtle or non-specific symptoms and signs, are at high risk for encephalitis, accidental infection or unusual pathogens , and are considered to have more serious consequences. [4]. For many patients, the traditional methods requires extensive, often negative, series of tests using a combination of culture, serological, antigen, and molecular methods, leading to delayed or missed diagnoses and is achieved in only 50-80% of cases [5,6]. Failure to diagnose patients with central nervous system diseases in time often results in delayed, inadequate, inappropriate treatment, and increased a high cost burden to the healthcare. This applies in particular to CNS infections, where rapid, accurate identification of pathogens and rapid initiation of antimicrobial treatment are potentially lifesaving.
Metagenomic next generation sequencing (mNGS) is a rapidly evolving molecular technology with great potential to provide a direct, unbiased analysis of the microbial composition of clinical samples without reliance on traditional culture or targeted molecular tests [7,8]. Cerebrospinal fluid (CSF) is particularly suitable for NGS due to its sterility in healthy individuals. Several recent published reports describing the usefulness of mNGS of CSF to identify known and unsuspected pathogens and to discover new microorganisms [9][10][11]. However, there have been no 5 reports of studies on cerebrospinal fluid (CSF) samples from HIV-infected individuals. CNS infection is more common in patients with severe immunosuppression of CD3 + 4 + T cell counts <200/ul than in immunocompetent hosts. In immunosuppressed individuals, opportunistic infections (OIs) is the most probably diagnostic considerations and multiple etiologies can coexist. [12].An indepth understanding of the various etiologies of CNS disease in HIV-infected patients is important for proper assessment and management . Recently, we explored the application of mNGS of CSF in the diagnosis of CNS infections in patients with severe immunosuppression.

Study Patients
Twenty-two HIV-infected patient with a diagnosis of encephalitis or meningitis were recruited from Infectious Diseases Center of Guangzhou Eighth People's Hospital between June 2018 and May 2019. Criteria for diagnosis of encephalitis/meningitis are based on relevant guidelines and/or consensus [13,14]. Cerebrospinal fluid specimens were tested by clinical microbiology and mNGS respectively. All samples had complete clinical microbiological data. The study was conducted in accordance with the Helsinki declaration and approved by the local institutional review committee. Written informed consent was obtained from All participants.

mNGS of CSF
DNA Extraction, library construction and sequencing 0.5-3mL CSF sample from patient was collected according to standard procedures.1.5ml microcentrifuge tube with 0.5ml sample, 1g 0.5mm glass bead were placed on a vortex mixer on a horizontal platform and stirred violently for 6 30min at 2,800-3,200 rpm. The 0.3mL sample was isolated into a new 1.5mL microcentrifuge tube and DNA was extracted using TIANamp Micro DNA Kit (DP316, TIANGEN BIOTECH) according to the manufacturer's recommendations. [15]. DNA libraries were constructed by DNA fragmentation, end-repair, splicing and PCR amplification. Agilent 2100 was used to control the quality of DNA libraries. The qualified libraries were sorted by using the BGISEQ-50 platform [16].
High-quality sequencing data wre obtained by subtracting the low quality sequencing data [17], and then the Burrows-Wheeler Alignment method was used to compute the subtraction of human host sequences mapped to the human reference genome (hg19) [16]. By removing the low complexity reading data and comparing the four microbial genome databases of virus, bacteria, fungi and parasites, the remaining data were classified. The databases were downloaded from NCBI (ftp://ftp.ncbi.nlm.nih.gov/genomes/). It contains 4152 genome-wide sequences of viral taxa, 3446 bacterial genomes or scaffolds, 206 fungi associated with human infections, and 140 parasites associated with human disease. The number of unique alignment reads (SMRN) and SDSMRN was calculated (SDSMRN=SMRN* 20000000/ (Clean Reads number)). Using BED Tools to calculate the coverage ratio and depth of each microorganism [18].

Criteria of Positive mNGS Result
To reduce the influence of potential contamination, we used the following criteria for positive results of CSF mNGS: 1.
1.Bacteria (excluding mycobacteria) and parasites: mNGS identified a microbe (species level) whose SDSMRN of the ≥10 that ranked among the top 5 for bacteria, virus or parasites than that of any other microorganisms.

2.
Fungi : mNGS identified a microorganism (species level) with a score of 5-fold higher for a specific pathogen (SDSMRN) than any other fungus due to its low biomass in DNA extraction [19].

3.
Mycobacterium tuberculosis (MTB): the result was considered positive when the species or genus level detected by mNGS at least 1 read owing to the difficulty in DNA extraction and low probability of contamination [20]. In total, NGS detected 5 bacterial species, of which M. tuberculosis (2cases) was the most common, 20 viral species, of which (EBV)(9) and Human betaherpes virus 5(CMV)(4) were the most common, 6 fungal species, both of which were Cryptococcus (2)and Talaromyces marneffei (2), and 1 parasitic species, of which was the Toxoplasma gondii. ( Figure 1-A) Nine co-infections were detected, of which Human gammaherpesvirus 4 (EBV) (7cases) and cytomegalovirus(3cases) was the most common (Figure 1-B). The percentage of co-infection in mNGS positive specimens was notably higher than that of conventional methods positive specimens. (40.91% vs. 14.39%).  For mNGS identified pathogens, the initial empirical diagnosis was often confirmed by mNGS (10.53%) or modified (21%), while only 16% were considered unreliable (unsupported) and 10% uncertain.

Clinical Effect
The mNGS detected pathogens but not detected by conventional methods included: 3 bacterial species,4 viral species, 3 fungal species, and 1 parasitic species. Among the pathogens identified by the metagenomic NGS alone, 6 pathogens the clinician declared that the results of mNGS approvingly influenced their clinical reasoning and 7 pathogens guided antimicrobial therapy ( Table 3).  [23,10,24]. For these reasons, metagenomics ought to be considered earlier in immunocompromised patients.
Nevertheless, challenges remain, such as contamination or background microbial genes being detected in samples, as reported by other studies [25][26][27] . Due to the high sensitivity ,contaminations may derived from the following sources: some of which is those with no previous report, and other pathogens that exist in the environment, skin flora or the reagents [28]. The most common bacteria detected by mNGS were Propionibacterium, followed by Cutibacterium and Acidovorax.

Conclusion
In conclusion, this study demonstrated mNGS has great potential value in the Concordance Between mNGS and conventional methods for Pathogen Detection Figure 3 The cocordance of initial diagnosis and mNGS