Development of a TaqMan Array card to target 21 purulent meningitis-related pathogens

Purulent meningitis (PM), which is often caused by bacterial or fungal infection, is a serious disease of the central nervous system and particularly life-threatening for children and newborns. The disease is characterized by acute onset, high fever, severe headache, vomiting, stiff neck, and high disability and mortality rates. But in infants and young children, these “classical “signs are often absent, they may be poor feeding, lethargy, disorientation or reduced conscious level. Approximately 10 to 20% of PM survivors develop post-PM neurological sequelae [1]. The common sequelae of PM include hydrocephalus, cerebral hemorrhage, epilepsy, paralysis, blindness, and mental retardation [2, 3]. Thus, timely and accurate treatments are critical to improve PM prognosis, and effective treatments for PM depend on an accurate and rapid identification of the causative bacteria or fungus [4, 5].

The commonly recognized gold standard to identify the causative bacteria for PM is the cerebrospinal fluid (CSF) culture method [6, 7]. However, the CSF culture method has many limitations. For instance, the procedure is too complex; the time for an affirmative result is too long (usually longer than 48 h); the method often fails if patients receive empiric antibiotic therapies [8, 9].

The direct Gram staining method can provide results rapidly and confirm a diagnosis of bacterial infection in 60–90% of patients, and a confirmed diagnosis from Gram staining depends on the bacterial concentration in CSF samples [7]. The Gram staining method may fail when the bacterial concentration is too low in CSF samples, and the sensitivity of the Gram staining method also varies widely for different microorganisms [10]. Although CT may be useful to diagnose PM, a low accessibility to CT scanners and the requirement of further laboratory tests limit its application [11]. The results of the method of MALDI-TOF MS were usually affected by culture media, cultivation conditions, or incubation times [12, 13]. A regular PCR combined with gel electrophoresis has a low sensitivity. The sequencing method is usually time-consuming, has a complex protocol, and is expensive [14‐16]. Luminex Verigene, a high-throughput platform, can detect multiple pathogens from one sample simultaneously but requires a relatively long time to collect results [17]. In 2015, the United States Food and Drug Administration approved the first multiplex PCR kit (FilmArray® system) as an auxiliary diagnostic method to identify causative pathogens of meningitis and encephalitis. The FilmArray® system [18, 19] can simultaneously detect 14 pathogens (6 bacteria, 7 viruses, and C. neoformans/C. gattii) from CSF samples in approximately 1 h. However, the system is very expensive ($305–$453/sample and $22–$32/pathogen) [20, 21].

A TaqMan array card (TAC), which combines the real-time PCR technology and microfluidics high-throughput technology, appears superior to those methods. A TAC has 384 wells on a microfluidic chip and thus can detect up to 48 targeted pathogens simultaneously using a small amount of a sample. The TAC technology has been used to identify respiratory pathogens [22], enteropathogens [23], bioterrorism pathogens [24, 25], neonatal infection [26], and drug resistance mutations [27]. Liu and colleagues compared the performance of a TAC, multiplex real-time quantitative PCR, and the PCR-Luminex platform in the identification of enteric pathogens [28], and they found that the TAC showed the best sensitivity and specificity. TAC assays were used to identify the causative microorganisms of respiratory disease outbreaks in the United States [29]. Although TAC systems have been used successfully to identify pathogens, a comparison of the sensitivity of a TAC versus the CSF culture method to identify pathogens in CSF samples from patients receiving empiric antibiotic treatments is still lacking. Moreover, whether TAC results could indicate PM severity remains unclear. Here, we aim to fill these knowledge gaps. In the current study, we developed a high-throughput PM-TAC, compared the sensitivity of the PM-TAC versus the CSF culture method to detect pathogens from CSF samples of children with PM after using antibiotics, and explored whether PM-TAC results could indicate PM severity.

In the current study, we developed a PM-TAC and optimized the PM-TAC assay. Our PM-TAC showed a LOD of 5 to 50 copies/reaction, except Streptococcus pneumoniae with a LOD of 100 copies/reaction, a sensitivity of 95%, and a specificity of 96%. These performance parameters are similar to the findings from a previous study [11]. Compared with the CSF culture method, which usually has a sensitivity of 81.3%, our PM-TAC was more sensitive [7, 10]. In addition to the superior sensitivity, our PM-TAC assay produced results fairly faster (in 3 h) at a lower cost ($126/sample) than the next generation sequencing (NGS) method (72 h of the assay time and $500/sample) [14]. Although the NGS technology has been improved continuously, it remains to be tedious and requires comprehensive bioinformatic analyses. All of those disadvantages of the NGS method limit its application in clinical practice. In contrast to the NGS method, it only took 3 h to collect results from our PM-TAC assay. Liu and colleagues have shown that the cost of TAC assay is much lower than that of ELISA and other methods [28] The estimated average cost of our PM-TAC assay for each targeted pathogen was approximately $6 and $144 for a whole panel ($6 x [21 PM-associated pathogens+ 3 controls] = $144), which appears affordable for economically underdeveloped regions and countries.

Compared with the CSF culture method, our PM-TAC assay was particularly more sensitive to detect pathogens after empiric antibiotic therapies. The positive rate of our PM-TAC was 53.3% (8/15), whereas it was only 13.3% (2/15) for the CSF culture method for CSF samples collected after empiric antibiotic therapies. These results indicate substantial clinical significance. The PM-TAC assay allows for detection of these pathogens which is independent of the use of antibiotics. The assay can then be used to assess the efficacy of antibiotic therapy in clearing the pathogen and the infection. Previous studies have demonstrated that the rate of inappropriate antibiotic use in the hospital setting is 30 to 50% because of the lacking an accurate identification of the causative bacteria [31, 32]. Every pathogen is sensitive to the specific antibiotics. The inappropriate antibiotic use before the pathogen was identified could induce drug resistant in bacteria. To be clinically effective, antibiotics that are specifically targeting the causative bacteria should be prescribed at the proper dosage.

In addition, we found that the CT of a pathogen from the PM-TAC assay appeared to indicate PM severity. Primary infections are usually pathogenic specific, and thus a pathogen load is associated with the infection severity. Our findings suggest that the PM-TAC assay could provide an effective reference to guide the therapeutic strategies for PM. For example, clinicians could adjust the dosage of antibiotic treatments bases on the PM-TAC results. Our PM-TAC system also allows multiplex detection. According to the clinical experiences of pediatricians, most cases of PM are caused by a single pathogen. In our current study, the 15 clinical CSF samples also showed a single-pathogen infection, which confirmed the clinical experiences.

There are limitations in our current study. There was a small sample size indeed in this study due to the difficulty to collect enough CSF in children clinically. Because many clinical examinations were performed so almost all cerebrospinal fluid were used up and this reduced greatly the sample size. Although the sample size was small, it could reflect a very meaningful phenomenon. Fortunately, we constructed simulated CSF samples being close to the clinical CSF sample based on previously reported studies [33], which made up for the small sample size. Moreover, we are also making efforts to collect new CSF samples from hospitals, but it will take a long time, and with the accumulation of sample size of CSF, we plan to do more in-depth related research in the future. The cost of PM-TAC is still prohibitively expensive for resource poor settings although it is cheaper compared with other molecular diagnostic techniques. Actually, the main cost is the microfluidic chip and at present, microfluidic versions have great potential applications in all aspects and have attracted much attention. At the same time, the production technology of microfluidic plate is developing rapidly. When the material and technology of TAC break through, the cost of this method will be further reduced. In addition, we are also increasing the number of pathogen detected, so that the detection cost of each pathogen will be lower. All of the targeted pathogens on the PM-TAC were selected mainly based on common clinical experiences. Therefore, the PM-TAC is unable to identify previously unidentified pathogens whereas the NGS method could. We will continuously update and expand the targeted pathogen panel on the PM-TAC following the most updated literature. Although the evolutionarily conserved domains were selected to target the pathogens, mutations in the targeted gene may cause false negative results because of the high specificity of the minor groove binder probes. In general, pathogenic bacteria and fungi have a low mutation rate, whereas virus often mutate quickly. Our future study will take gene mutations into consideration when we design a PM-TAC.

The PM-TAC in this study can detect 21 PM-associated pathogens in 3 h and showed a higher sensitivity than the CSF culture method particularly for CSF samples collected after empiric antibiotic therapies. The CT of a pathogen from the PM-TAC assay, which represents a pathogen concentration or load in CSF samples, indicated PM severity. Thus, the PM-TAC may be a promising tool for an accurate and rapid diagnosis of PM and thus facilitate a timely antibiotic treatment to improve PM prognosis.