INTRODUCTION
Infections of the upper and lower respiratory tract can be caused by a diversity of pathogens, both viral and bacterial. Community-acquired respiratory tract infections are a leading cause of hospitalization and responsible for substantial morbidity and mortality, especially in infants, the elderly, and immunocompromised patients. The etiological agent in such infections differs greatly according to season and age of patient, with highest prevalences being those of respiratory syncytial virus (RSV) in children and influenza virus in adults. Rapid microbiological diagnosis of a respiratory infection is important to ensure appropriate antimicrobial therapy and for the effective implementation of isolation precautions (
1).
In the last decade, many conventional diagnostic methods such as culture and antigen detection assays have been replaced by molecular assays for diagnosing respiratory tract infections. Multiplex real-time PCR assays have been developed and implemented for routine diagnostic application, detecting a wide variety of pathogens (
2–7). These assays have shown high sensitivity and specificity, but the limited number of fluorophores that can be used per reaction resulted in the need to run several real-time PCR assays to cover a broad range of relevant pathogens. Commercial assays using multiplex ligation-dependent probe amplification (MLPA), a dual priming oligonucleotide system (DPO), or a microarray technology were developed to overcome this problem and are able to detect up to 19 viruses simultaneously (
8,
9). All applications mentioned require nucleic acid extraction prior to amplification. For routine diagnostics, these methods are most suited for batch-wise testing, with a turnaround time of ∼6 to 8 h. To decrease the time to result and enable random access testing, syndromic diagnostic assays have been developed. These assays combine nucleic acid extraction, amplification, and detection in a single cartridge per sample and are suitable for decentralized or even point-of-care testing (POCT) with a time to result of <2 h.
A novel rapid diagnostic, cartridge-based assay for the detection of respiratory tract pathogens using the ePlex system (
Fig. 1) was developed by GenMark Diagnostics, Inc. (Carlsbad, CA). The ePlex respiratory pathogen panel (RP panel) is based on electrowetting technology, a digital microfluidic technology by which droplets of sample and reagents can be moved efficiently within a network of contiguous electrodes in the ePlex cartridge, enabling rapid thermal cycling for a short time to result. Following nucleic acid extraction and amplification, detection and identification are performed using the eSensor detection technology (
Fig. 2), as previously applied in the XT-8 system (
10).
In the current study, the performance of the syndromic RP panel was compared to those of laboratory-developed real-time PCR assays, using clinical specimens previously submitted for diagnosis of respiratory pathogens.
RESULTS
The 323 positive clinical specimens contained a total of 464 respiratory pathogens as detected by laboratory-developed real-time PCR assays (
Table 1). As shown in
Table 2, the 57 nonnasopharyngeal (non-NPS) specimens comprised 69 of the total 464 respiratory pathogens. Testing all samples with the RP panel resulted in an overall agreement for 452 (97.4%) targets from 311 specimens, prior to discrepant analysis. Of the specimens containing a single pathogen, the detected targets were concordant in 209/217 specimens. For samples with coinfection, the same pathogens could be identified in 77/81, 22/22, and 3/3 in the case of 2, 3, and 4 pathogens present, respectively. Eight of 12 discordant targets (PCR
+/RP
−) had a positive result with threshold cycle (
CT) values of >35 (
Fig. 3). Retesting with a third assay confirmed 10 of 12 real-time PCR-positive targets being human bocavirus (hBoV;
n = 3), rhinovirus (RV;
n = 2), parainfluenza virus type 2 (PIV2;
n = 1)), human coronavirus (hCoV) OC43 (
n = 1), hCoV 229E (
n = 1), hCoV HKU1 (
n = 1), and human metapneumovirus (hMPV;
n = 1). The two unresolved PCR
+/RP
− results consisted of two hMPV-positive samples (
CT values of 33.2 and 38.3).
The RP panel yielded a positive result in 17 specimens, where the laboratory-developed test (LDT) remained negative (PCR
−/RP
+), including 15 additional pathogens previously undetected by LDT in the 323 positive specimens and one influenza A H1N1 2009 virus that was detected as influenza A virus by LDT (
Table 1). Seven of these 15 additional targets could be confirmed, including three of RV/enterovirus (EV) (all confirmed as RV), two of PIV4, and one each of hBoV and hCoV NL63.
One of the selected negative samples tested positive for human adenovirus (hAdV) in the RP panel but could not be confirmed by discrepant testing. All other negative specimens tested negative in the RP panel as well.
Both Middle East respiratory syndrome coronavirus (MERS-CoV) isolates could be detected by the RP panel. By testing a 10-fold dilution series of both isolates, it was shown that MERS-CoV with a CT value of <30 in the laboratory-developed real-time PCR assay could be detected using the RP panel, while detection with a CT value of >30 was achievable but was not reproducible in every instance.
Of the 12 specimens from the Quality Control for Molecular Diagnostics (QCMD) 2016 Respiratory II Pilot external quality assessment (EQA) study panel, 10 were detected in full agreement with the content as reported by QCMD (
Table 3). The 2 false-negative tested specimens both contained hCoV NL63, of which one was a coinfection in an hMPV-positive sample. Both specimens had been tested with the laboratory-developed real-time PCR assay as well and were found positive for hCoV NL63, both with
CT values of 37.4.
The Qnostics evaluation panel consisted of 17 samples, including 15 different respiratory pathogens and one negative sample (
Table 3). The RP panel detected 15 of the specimens in agreement with the content, whereas hAdV type 1 and
Chlamydophila pneumoniae were not detected. Real-time PCR detection of these specimens was performed to confirm the presence of the respective pathogen in the specimen and was found positive for both hAdV (
CT value of 31.4) and
C. pneumoniae (
CT value of 35.4).
DISCUSSION
The performance of the ePlex RP panel was assessed by retrospective testing of 343 clinical respiratory specimens (obtained in 2009 to 2016) comprising five different types of specimens. Although the RP panel had been CE
in vitro diagnostic (CE-IVD) cleared for detection of respiratory pathogens from NPS swabs only, we included a range of alternate sample types that can be obtained and tested for respiratory pathogens in the diagnostic setting. By including a total of 57 respiratory non-NPS specimens with different pathogens (
Table 2), it was shown that the RP panel was able to accurately detect the pathogen(s) in the different types of specimens, as the assay showed 100% concordance with LDT. For sputum samples, preprocessing with Sputasol was introduced after the initial 6 tested specimens, since 1 false-negative result was found, which was resolved on retesting with Sputasol pretreatment. Further studies need to determine the frequency of preprocessing of sputum samples before efficiently running the RP panel.
Specimens for inclusion in this study were previously tested at two different sites, using both their own systems and validated assays. Although the initial setups of the LDT assays were the same (
11,
12), minor adjustments of the assays and the use of different PCR platforms may affect the performance of the LDTs and therefore were a limitation of this study.
Comparison of the results from the RP panel with the results from the routine multiplex real-time PCR showed an agreement of 97.4% in 464 pathogens tested.
Figure 3 shows that PCR
+/RP
− targets are mainly targets with a low viral or bacterial load (based on
CT values). Analysis of real-time PCR results with
CT values of <35 provided an agreement between the RP panel and laboratory-developed real-time PCR of 99.1%. Specimens containing pathogens with a high load (
CT value of <30) were all detected correctly by the RP panel, independently of the type of specimen, type of pathogen, or the number of different pathogens in a specimen. This finding is in line with earlier evaluation of the GenMark XT-8 system using the same eSensor principle of detection (
10). With 29 concordant targets with a
CT value between 35 to 40 and 4 targets with a
CT value of >40, the RP panel showed good detection rates with regard to lower viral or bacterial loads as well (
Fig. 3).
Although the performance of the RP panel appeared to be excellent using the tested specimens in this study, for PIV4 (n = 2) and C. pneumoniae (n = 0) the number of clinical specimens that could be analyzed was too low for a proper assessment of the assay, which was a limitation of this study.
In 14 different specimens, the RP panel identified 15 pathogens that had not been detected by routine testing (PCR
−/RP
+). In addition, one influenza A virus detected by LDT could be detected as influenza A H1N1 2009 virus by the RP panel. One of the selected negative samples was shown to contain an hAdV, while all other PCR
−/RP
+ targets were detected as copathogens to other positive targets in the samples. All the PCR
−/RP
+ targets were found in samples obtained from 1 institute. Discrepant analysis on the eluates of these samples was performed using the LDT of the other institute, where seven of the additionally detected pathogens could be confirmed by discrepant testing. These samples showed a relatively low viral load based on the mean
CT value found (33.3), probably around the limit of detection of the initial LDT. It was unclear whether the 8 unresolved PCR
−/RP
+ targets are false positive or the results of more efficient detection of multipathogen infections by the eSensor technology (
10).
A small number of LDT-negative specimens (n = 20) was included in this study since the main objective of this study was to determine the performance of the RP panel in detecting respiratory pathogens. Although this is a limitation of the current study, we believe that this issue will be addressed extensively in upcoming prospective clinical studies.
Owing to the lack of clinical specimens containing MERS-CoV, dilutions of two different culture isolates were tested in this study, of which dilutions with CT values of <30 as shown by the laboratory-developed real-time PCR assay could be detected consistently. It should be noted that the real-time PCR assay has been developed for research use and has not yet been validated for clinical use.
Assessment of the RP panel using EQA samples from QCMD and Qnostics showed results that are in line with the results obtained from clinical specimens. A total of 4 targets included in the EQA samples could not be detected using the RP panel, showing CT values of >35 (n = 3) and 31.4 (n = 1) when tested by real-time PCR.
The RP panel on the ePlex system enables rapid testing and can be used as a diagnostic system in either a laboratory or a decentralized setting that is closer to the patient. The assay turned out to be rapid and straightforward to perform. Compared to routine testing, hands-on time of the RP panel was very low (<2 min), whereas the hands-on time of the routine testing was about 30 to 45 min, depending on the nature and number of samples tested. The overall run time of the platforms was also in favor of the ePlex system, as it takes approximately 90 min for nucleic acid extraction, amplification, hybridization, and detection, whereas routine testing takes up to 2 h and 45 min using different systems and multiple real-time PCR assays in multiplex. An important advantage of the ePlex system is the possibility of random access testing, compared to batch-wise testing in the current diagnostic real-time PCR approach. With a relatively short turnaround time and the potential to randomly load and run up to 24 specimens, the ePlex system is very suitable for testing STAT samples, which require immediate testing. In contrast to LDTs, where CT values represent a quantitative indicator, the ePlex system generates qualitative results only. The CT value is dependent on many different factors such as sample type and course of infection and can therefore differ greatly, even within a single patient. Hence, a qualitative result, e.g., identification of the pathogen, is the major factor for patient management.
The costs of reagents per sample are relatively high for ePlex compared to LDT. However, when taking into account the hands-on time of technicians and the clinical benefit of more rapid results, the assay will most likely be more cost-effective. Studies evaluating a rapid diagnostic assay for respiratory pathogens, such as the FilmArray respiratory panel (BioFire Diagnostics, Salt Lake City, UT), have already shown the impact of rapid diagnostics for respiratory pathogens, since it decreased the duration of antibiotic use, the length of hospitalization, and the time of isolation, delivering financial savings (
13,
14). Although the RP panel on the ePlex system has the same potential, clinical studies remain to be conducted to fulfill this potential.
In conclusion, this study shows excellent performance of the GenMark ePlex RP panel in comparison to laboratory-developed real-time PCR assays for the detection of respiratory pathogens from multiple types of clinical specimens and EQA samples. The system provides a large amount of useful diagnostic data within a short time frame, with minimal hands-on time, helping to reduce laboratory costs for labor and deliver a faster result to the clinician in order to aid in appropriate antimicrobial therapy. Therefore, this syndrome-based diagnostic assay could be used as rapid diagnostic testing in many different settings.