Development and validation of an advanced fragment analysis‐based assay for the detection of 22 pathogens in the cerebrospinal fluid of patients with meningitis and encephalitis

Background Meningitis and encephalitis (ME) are central nervous system (CNS) infections mainly caused by bacteria, mycobacteria, fungi, viruses, and parasites that result in high morbidity and mortality. The early, accurate diagnosis of pathogens in the cerebrospinal fluid (CSF) and timely medication are associated with better prognosis. Conventional methods, such as culture, microscopic examination, serological detection, CSF routine analysis, and radiological findings, either are time‐consuming or lack sensitivity and specificity. Methods To address these clinical needs, we developed an advanced fragment analysis (AFA)‐based assay for the multiplex detection of 22 common ME pathogens, including eight viruses, 11 bacteria, and three fungi. The detection sensitivity of each target was evaluated with a recombinant plasmid. The limits of detection of the 22 pathogens ranged from 15 to 120 copies/reaction. We performed a retrospective study to analyze the pathogens from the CSF specimens of 170 clinically diagnosed ME patients using an AFA‐based assay and compared the results with culture (bacteria and fungi), microscopic examination (fungi), polymerase chain reaction (PCR) (Mycobacterium tuberculosis), and Sanger sequencing (virus) results. Results The sensitivity of the AFA assay was 100% for 10 analytes. For Cryptococcus neoformans, the sensitivity was 63.6%. The overall specificity was 98.2%. The turnaround time was reduced to 4‐6 hours from the 3‐7 days required using conventional methods. Conclusions In conclusion, the AFA‐based assay provides a rapid, sensitive, and accurate method for pathogen detection from CSF samples.


| INTRODUC TI ON
Meningitis and encephalitis (ME) are severe central nervous system (CNS) infections with high morbidity and mortality because of the difficulty in achieving a prompt diagnosis and receiving timely therapy. 1 CNS infections are mainly caused by pathogens including bacteria, mycobacteria, fungi, viruses, and parasites. 2 Cerebrospinal fluid (CSF) culture is necessary to diagnose ME. 3 CSF culture is the gold standard for the diagnosis of bacterial infections, though it is timeconsuming. Microscopic examination, blood culture, skin biopsy, and serum inflammatory markers are additional diagnostic tools that might aid in etiological diagnoses. If the causative pathogen is not clear, the clinician will preliminary determine the type of pathogen (bacteria, virus, or fungus) according to the patient's clinical manifestations and the cellular and chemical parameters of the CSF. All of these laboratory tests require a certain CSF volume. However, the methods listed above are time-consuming and generally have low sensitivity or specificity. In recent decades, the epidemiology and treatment strategies for meningitis have changed considerably, especially because of the introduction of conjugate vaccines such as the vaccines for pneumococcal, meningococcal, and Haemophilus influenza type b. [4][5][6][7][8] Therefore, the early diagnosis of ME has become even more imperative. Doctors may sometimes perform comprehensive anti-infection therapy, including antibiotics and antiviral and antifungal medications, immediately for cases that lack a definitive pathogen diagnosis if the patients are critically ill. However, most such treatments are ineffective, and certain drugs might be harmful to patients. Therefore, there is an urgent need for a rapid, sensitive, and accurate method that can detect a greater number of target pathogens from a small CSF volume.
According to population-based studies in China, the incidence of acute bacterial meningitis ranges from 12.4 to 19.2 cases/100 000 for children aged <5 years. [9][10][11] The primary pathogens of bacterial meningitis are Neisseria meningitidis, H. influenza, and Streptococcus pneumonia. [12][13][14] China has the second-highest prevalence of tuberculosis (TB) infection worldwide. China and another 21 high-burden countries account for 80% of the tuberculosis cases and approximately 22% of multidrug-resistant tuberculosis cases worldwide. 15,16 Tuberculous meningitis (TBM) is the most severe form of extrapulmonary tuberculosis (EPTB) and causes exceptionally high mortality and morbidity. [17][18][19] Viruses are the major cause of aseptic meningitis.
Human enteroviruses (HEVs) are a common cause of acute meningitis with a summer-fall season peak. Yihong Xie et al reported a 5-year study on acute ME in Guangxi, China. Their study revealed that enterovirus (31.5%) is the most common pathogen, followed by Japanese encephalitis (28.3%), mumps (23.2%), measles (5.1%), herpes simplex virus, rubella, cytomegalovirus, varicella zoster virus and EB virus. 20 Among the fungi, Cryptococcus neoformans is the most common cause of fungal meningitis.
Advanced fragment analysis (AFA) is a technique that provides an alternative high-throughput, multiplexed, quantitative gene ex- Software is used to analyze the data and determine the size and genotype. Traditional detection methods, such as culture, require at least 2-3 days. Currently, commonly used molecular detection methods, such as RT-PCR, can only detect 1-2 pathogens in one experiment, but AFA can detect up to 40 pathogens in 4-6 hours.
Thus, AFA has an absolute advantage in terms of the detection time and number of pathogens. AFA has been successfully applied in many fields, such as subtype classification of pediatric acute lymphoblastic leukemia, detection of multiple viruses for community-acquired pneumonia, Helicobacter pylori identification, and virulence and resistance analyses. [21][22][23] In the present study, the AFA technology was used to design primers for 22 target pathogens based on national epidemiological data. The most commonly reported pathogens causing ME, including eight viruses, 11 bacteria, and three fungi, were selected as detec- In this retrospective study, residual CSF specimens were collected and tested at the West China Hospital of Sichuan University.
The results of the AFA-based assay were compared with those of conventional culture for bacteria and yeast, PCR for M. tuberculosis, and Sanger sequencing for viruses.

| Study design and case definition
This study was a retrospective study of 170 patients admitted to the hospital for the first time due to ME. Meningitis was defined as an infection localized to the subarachnoid space sparing the brain parenchyma and was characterized by a fever, headache, nausea, vomiting, meningeal irritation, and alterations in the CSF. 24 Encephalitis was defined as the presence of an inflammatory process in the brain associated with clinical evidence of neurologic dysfunction. 25 The ME patients

| Clinical specimens
Specimens meeting the following inclusion criteria were selected: A CSF specimen was collected by lumbar puncture (LP) with adequate residual volume of uncentrifuged CSF (≥200 µL) left over from standard care testing for bacterial and yeast culture, and the specimen was enrolled within 7 days of collection for testing (<5 days frozen for nucleic acid extraction and 2 days for final testing). Each residual specimen collected for the study was assigned a unique number corresponding to our laboratory tests for the AFA assay. Thus, the authors had access to information that could not identify individual participants during or after data collection, including comparator PCR and patient demographic and clinical data, such as patient general information,

| Primer design
Target-specific primers were designed based on the alignment Information for all of the primers is listed in Tables S1 and S2. All primers were synthesized by Sangon Biotech (Shanghai, China).

| AFA-based multiplex assay
An RT-PCR mixture containing 4.5 µL of premixed solution, 0.5 µL of an RT-PCR enzyme and UDG enzyme mixture, and 5 µL of sample or positive control or negative control was added to a final volume of 10 µL/reaction. PCR amplification was performed using the ABI Verity 96 Thermal Cycler. The cycling conditions are listed in Table   S3.
Polymerase chain reaction products were prepared for capillary electrophoresis (CE) and fragment analysis using the 3500 Genetic Analyzer (ABI, USA) following the manufacturer's protocols. Next, 1.0 µL of PCR product was added to 10 µL of Hi-Di formamide solution along with 0.25 µL of GeneScan™-500 LIZ™ Size Standard (ABI, Foster City, CA, USA). The mixture was added to a 96-well plate, which was loaded onto a 3500 Genetic Analyzer for CE and fragment separation. The fragment size was used for target identification. For all targets, the assay was considered positive when the signal strength of the fluorescent dye was above 500 relative fluorescence units (RFU), undetermined for a signal strength between 300 and 500 RFU, and negative for a signal strength <300 RFU. If the signal strength was in the undetermined region, a repeat test was performed; if the test result was still in the undetermined zone, it was deemed positive. The corresponding amplicon sizes of the pathogens are listed in Tables S1 and S2.
In addition, each panel incorporated three reference genes, including B2M and RNaseP to monitor the quality of the extracted mRNA and DNA, respectively. For the B2M gene, the assay was designed to amplify mRNA around the second and third intron-exon junction to ensure that the mRNA was amplified. Additionally, an internal control (IC) was included as a quality control for the RT-PCR reaction. The One-Step RT-PCR Kit used for multiplex pathogen detection was obtained from Health Gene Technologies Co., Ltd.

| AFA ME panel testing
The AFA ME panel test consisted of nucleic acid extraction (50 minutes), reverse transcription and nucleic acid amplification (140 minutes), and fragment analysis (50 minutes), as shown below. All of the above steps plus the manual operation time (approximately 30 minutes) required 4-6 hours.

| Sensitivity of the AFA-based assay
For sensitivity studies, a recombinant plasmid was used. The RT-PCR products were extracted from a 1% agarose gel and then purified using a Gel Extraction Kit D2500 (OMEGA Bio-Tek, Norcross, GA, USA). The purified PCR product was ligated to the pMD® 18-T Simple Vector, which was used to transform E. coli (DH5a). White colonies were picked and inoculated in LB medium containing ampicillin and then were cultured overnight at 37°C. PCR was used to identify the cloned bacteria. The plasmid was extracted from E. coli using a Plasmid Mini Kit I D6943 (OMEGA Bio-Tek). The extracted plasmid was digested with EcoRI and HindIII. The positive recombinant plasmid was sequenced (Sangon Biotech) and identified using the BLAST tool of NCBI. After successful construction of the recombinant plasmid, each target pathogen recombinant was measured via twofold serial dilutions in PBS. The copy numbers were determined using the following formula: number of copies = (DNA amount * 6.022 × 10 23 )/ (DNA length * 1 × 10 9 * 660); number of copies = (ng * number/ mole)/(bp * ng/g * g/mole of bp). The diluted plasmid was used for the determination of the limit of detection.
For specificity, we used four clinically isolated strains (Enterobacter cloacae, Klebsiella pneumoniae, Pseudomonas aeruginosa, and Candida albicans) that often appear in the CSF of ME patients and were not included in the AFA panel. We extracted nucleic acids and then mixed samples of those four pathogens with CMV DNA in PBS, followed by detection using the AFA assay.

| Comparator testing
Bacterial and fungi cultures were performed on each specimen enrolled. Each sample was inoculated onto a blood agar plate, a chocolate agar plate, and a Sabouraud agar plate as well as in brain heart infusion broth (Autobio, Zhengzhou, China). These plates and broth were then incubated overnight at 37°C (except for the Sabouraud agar plate, which was incubated at 25°C) and in 5% CO 2 for For M. tuberculosis detection, DNA was extracted from patients with clinically diagnosed tuberculous meningitis (TBM) using the MagNA Pure LC 2.0 automated system with the total nucleic acid isolation high-performance kit (Roche Diagnostics, Indianapolis, IN, USA). Next, the samples were subjected to RT-PCR for MTB DNA using a commercial kit (Qiagen, Hilden, Germany) that was previously included in the diagnostic criteria for TBM. 19,27 For viral detection, samples for sequencing were extracted using AFA nucleic acid. All of the clinically confirmed positive samples were sequenced by Sangon Biotech, and then, the data were compared with the information of NCBI to assess whether the requirements were met.

| Results and discrepant analysis
The AFA assay result was considered true positive (TP) or true negative (TN) only when it agreed with the result from the comparator method. Additionally, the discrepancy investigation for our study relied heavily on additional clinical information about the subjects whose specimens were tested in this evaluation.

| Calculations and statistical analysis
Sensitivity and PPA were calculated as 100 × [TP/(TP + FN)], and specificity and NPA were calculated as 100 × [TN/(TN + FP)]. As described previously, PPA and NPA were calculated in the same manner as those for the sensitivity and specificity, respectively. The terms "PPA" and "NPA" are used instead of "sensitivity" and "specificity" to indicate that a non-gold standard assay (eg, PCR) was used for the original comparator analysis. This analysis referenced the study by Leber et al 28 concerning the evaluation of the BioFire FilmArray Meningitis/Encephalitis Panel.

| Specificity, accuracy, and study of the AFAbased assay
The primer specificity of the multiplex AFA-based assay was verified by Sanger sequencing. Twenty-two clinically confirmed pathogens causing ME were used to evaluate the accuracy of the assay. The multiplex study used the positive clinical samples, and positive controls consisted of recombinant plasmids. The multiplex study of clinical samples showed specific peaks for all three references (Hu_RNA, Hu_DNA, and IC), the positive controls, and pathogens in the panel.
The AFA-based multiplex assay showed 100% agreement with the Sanger sequencing. The four clinically isolated strains (E. cloacae, K. pneumoniae, P. aeruginosa, and C. albicans) did not produce any signals that would indicate nonspecific amplification when they were mixed with HCMV (figures shown in Figure S1A-E).
The recombinant plasmids were used to determine the limit of detection. The cutoff value was 500 RFU for positivity. The limit of detection for each target in the current assay is listed in S4.

| Clinical specimen demographics
We acquired a total of 170 retrospective CSF specimens that were clinically diagnosed with ME. The age distribution included 159 (93.5%) adults aged 16 years and 11 (6.5%) pediatric of patients aged <16 years.

| Laboratory examination of the CSF, serum, and radiological findings
Regarding the CSF analyses, the median CSF opening pressure was showed meningeal enhancement. The MRI (53.2%) abnormal rate was slightly higher than that of CT (30.7%), especially for the bacterial, tuberculous, and fungi groups. The laboratory examinations of CSF and sera and the radiological findings are shown in Table S6.

| Summary of the AFA panel findings for the clinical samples
The AFA ME panel detected at least one potential pathogen in 50 of the 170 specimens that were tested, shown as a positivity rate of 29.4% in Table 1. The highest detection rates were in the pediatric groups.  Table 2.
Codetections were observed in two specimens representing 1.2% of the specimens and 4% of the positive specimens (2/50), as shown in Table 1

| D ISCUSS I ON
With the worldwide change in the epidemiology of ME pathogens and universal use of antibiotics, as well as the application of multivalent combination vaccines, the clinical presentation of many ME cases is nonspecific, making a definitive etiologic diagnosis challenging. 31,32 The diagnosis of ME infections requires consideration of the most likely causative agents based on exposure, geography, and season as well as an understanding of the optimal diagnostic test and highest-yield clinical specimen or testing. 33,34 It is particularly important to develop rapid detection reagents that could detect many pathogens in one PCR for ME patients who use only a small CSF sample value.
It is generally accepted that early, accurate medication is correlated with a better prognosis of patients. A previous study has shown that delayed treatment significantly increases the risk of a fatal outcome, with a relative increase in mortality of 12.6% per hour of delay. 35 The delay is primarily associated with difficulties in recognizing ME due to the absence of typical symptoms in many cases. 31,32 Regarding conventional methods, CSF culture is still the gold standard for the diagnosis of CNS infections, especially for bacteria and fungi. However, the yield of CSF cultures in suspected ME cases is low, especially if the patients have received antibiotics. IgM is the most widely used test for HSV, CMV, and VZV. 43 However, this antibody has high cross-reactivity with other clinically relevant viruses and related vaccines. 44 Regarding the detection of viruses, molecular testing has improved sensitivity and is faster than culture; thus, this technique has become the standard of care for many viral CNS infections, including HSV, EV, and human parechovirus infections. 45 In addition, blood culture and histopathologic examination are complementary methods for the diagnosis of ME.
ME patient demographic data. In our study, 170 patients were clinically diagnosed with ME. These patients were mainly of the Han ethnicity (88.8%), followed by the Tibetan ethnicity (9.4%).
We found an interesting phenomenon in that the consistent with our data in this study (3/14 [21.4%]) and our group's previous study. [46][47][48] The most common clinical features and signs in those patients were headache, fever, sinusitis or otitis, seizures, and confusion. Only fever and headache occurred in more than half of the patients, and some patients were finally diagnosed with ME with no obvious clinical manifestations. In addition to CSF culture, microscopic examination, and other CSF-related laboratory tests, radiological findings were a good supplement to the diagnosis of ME.  Table S7). The patient who was positive for Staphylococcus detection was a 23-year-old man who had normal CSF parameters and a final diagnosis of viral ME (patient 63 in Table S7). The patient with A. baumanii detection was a 29-year-old man with a final diagnosis of TBM (patient 108 in Table S7

| AFA panel viral targets
Viral detection using the AFA assay was lower than the detection of bacterial targets. The sequencing comparator method confirmed the AFA assay results in 12 (75%) of these cases. The four inconsistent viral results were associated mainly with EBV infection (3/4 75%).
The calculated PPA was very good for all targets (100%), except for HSV-2 and HSV-6, which were not detected; thus, their PPA could not be calculated. The inconsistent result for the EV-positive case was obtained for a 61-year-old male with a final diagnosis of TBM.
His clinical data, presentation, symptoms, and laboratory findings all confirmed the clinician's diagnosis (patient 12 in Table S7). For the three inconsistent EBV-positive patients, two of three of their clinical and laboratory data did not support the AFA results. Thus, we speculated that the clinical significance of detecting EBV in the CSF was not clear and that the test most likely detected latent virus from cells that were present.
The above data show that the AFA assay has high sensitivity and specificity for the detection of viral pathogens and has a high application value, especially for laboratories without conventional PCR detection.

| AFA panel fungal targets
Seven C. neoformans-positive specimens were detected via the AFA assay, all of which were confirmed by culture and CSF Jincheng ink staining. The calculated NPA was 100%, and the PPA was 63.6%, which was lower than that of the FilmArray panel (100%).
For four cases of cryptococcals, CSF culture or CSF Jincheng ink staining was positive and the AFA was negative; all four patients had a final diagnosis of cryptococcal meningitis. Four patients with CSF Jincheng ink staining were positive several times, and for two of them, cultures of the CSF specimens collected at other periods were positive. A previous study reported similar results. 49 This finding may be due to the fungal cell walls being more difficult to break, leading to low nucleic acid extraction efficiency.
Additional data for all of the results with associated laboratory data and a final clinical diagnosis are presented in Table S7 in the Supporting Information.
After the comparative analysis, our study still had unresolved inconsistent results (n = 7), including four bacterial results and three viral results. We suspected that the main reasons for these unre- In conclusion, the AFA-based assay is a rapid, sensitive, and specific method for detecting pathogens in CSF.
The method can also be employed as a supplement to the traditional methods for diagnosing ME. Accurate identification of causative pathogens causing ME will improve patient management and epidemiological investigations.

ACK N OWLED G M ENTS
We thank all of the participants in this study.

DATA ACCE SS I B I LIT Y
All relevant data are included within the article and its Supporting Information files.