Comparison of the genexpert enterovirus assay (GXEA) with real-time one step RT-PCR for the detection of enteroviral RNA in the cerebrospinal fluid of patients with meningitis

Enteroviruses (EVs) are the most common cause of aseptic meningitis in children and adults and may cause up to 90% of aseptic meningitis cases [1]. The rapid and accurate diagnosis of human enteroviral infections can reduce the use of antibiotics, duration of hospitalization, and financial costs [2-6]. Methods involving the amplification of nucleic acids have replaced traditional culture-based methods as the gold standard for the detection of EVs in cerebrospinal fluid (CSF) because of their increased sensitivity and speed. Routine diagnostic methods using real-time reverse transcription polymerase chain reaction (RT-PCR) have been developed over the past 15 years in an attempt to improve experts’ ability to detect EVs in CSF [7]. The latest development in the diagnosis of enteroviral meningitis is the GeneXpert Enterovirus Assay (GXEA; Cepheid, Sunnyvale, CA), a fully automated real-time multiplex RT-PCR assay. GXEA is the only assay approved by the U.S. Food and Drug Administration for the qualitative detection of enteroviral RNA in CSF. The GeneXpert system is a closed, unit-dose, molecular, microfluidics instrument that performs extraction, processing, and real-time RT-PCR, as well as the detection of nucleic acid targets. The system uses single-use cartridges that contain all of the reagents required for sample processing and PCR [8-10].

We compared the speed and reliability of GXEA with that of real-time one step RT-PCR (RTo-PCR), a method described in a previous publication by Verstrepen et al., that is routinely used for the detection of enteroviral RNA in CSF [11].

To determine the detection range of RTo-PCR using a TaqMan-formatted probe, we obtained five reference strains belonging to different serotypes (enterovirus 71, coxsackievirus B2, echovirus 30, coxsackievirus A24, and poliovirus 1) from the American Type Culture Collection (ATCC) and used these strains to determine the detection limit and positive control of RTo-PCR.

To assess the applicability of GXEA to detect EVs in clinical specimens with a low viral load, from June to September of 2008, we collected 109 CSF specimens from patients with aseptic meningitis. We had approval of an Institutional Review Board (IRB) through the Korean National Institute of Health. We analyzed these specimens by GXEA according to the manufacturer’s instructions using 140 μL of each sample. For the RTo-PCR assay, viral RNA was extracted using silica-coated magnetic beads based on Boom’s method in combination with an automatic liquid handling machine (TECAN, Männedorf, Switzerland) [12]. The TaqMan format RTo-PCR for the detection of EVs was performed according to the previously optimized reaction conditions [11]. Both assays targeted the 5′ noncoding region of the EV genome, which is commonly used in routine diagnostic assays because of its highly conserved genetic identity [11,13].

GXEA and RTo-PCR detected the presence of enterovirus in test samples from five reference strains, of different serotypes, with the limits of detection ranging from 2 to 0.05 TCID₅₀/mL for GXEA and from 1 to 0.01 TCID₅₀/mL for RTo-PCR (data not shown). Next, we compared the detection efficiency of the recently developed GXEA molecular diagnostic system, which is based on the concept of a “lab on a chip,” with that of RTo-PCR, using the CSF samples collected in 2008 from patients with meningitis. Of the 109 clinical samples assayed by GXEA and RTo-PCR, 66 and 43 were determined to be positive, respectively (Table 1). All of the samples found to be positive by RTo-PCR were also detected by GXEA, with no false positives. In the detection of enteroviral RNA from CSF, the sensitivities/specificities of RTo-PCR and GXEA were 65%/100% and 100%/100%, respectively. The GXEA results were in agreement with those produced by RTo-PCR in 78.9% of cases.

To compare the detection efficiencies of the assays used in this study, we used the mean Ct values from the positive samples; specifically, we divided the values from the GXEA-positive samples by those from the RTo-PCR-positive samples, and the values from the GXEA-positive samples by those from the RTo-PCR-negative samples (Figure 1). The mean Ct values from the GXEA- and RTo-PCR-positive samples (GXEA(+)/TaqMan(+)) and those from the GXEA-positive, RTo-PCR-negative samples (GXEA(+)/TaqMan(−)) were 32.38 and 34.85, respectively. According to our results, GXEA is more sensitive than RTo-PCR. This can be explained by the distribution of the Ct values for the positive specimens, which is an indirect measure of the viral load, as shown in Figure 1.

The sensitivity of GXEA was higher than that of RTo-PCR, and GXEA did not produce discordant positive results when conduced on the RTo-PCR-positive specimens. The mean Ct values for the GXEA-only positive samples (GXEA(+)/TaqMan(−)) were about 2.5 fold higher than those for the samples found to be positive by both assays (GXEA(+)/TaqMan(+)); thus, GXEA is at least 10 times more sensitive than routine RTo-PCR. The discrepancy observed in this study amomg the qualitative results for the CSF samples with a low viral load (n = 20) indicates a higher than expected incidence of CSF specimens from patients with viral meningitis with viral titers below the detection limit of routine molecular assays. Evidently, qualitative tests for the detection of EVs can be influenced by the detection limit of the molecular assay being used.

The GXEA system produced a lower Ct value than RTo-PCR. This may be explained by the fact that GXEA utilizes single-use cartridges that contain all of the reagents required for sample processing. Thus, the entire amount of extracted RNA is used in GXEA, while, in RTo-PCR, the amount of RNA used may be 10-fold lower. The sensitivity of an assay depends in part on the total amount of RNA used.