Evaluation of the Novel CE-IVD-Marked Multiplex PCR QIAstat-Dx Meningitis/Encephalitis Panel

The diagnosis of infectious meningitis and encephalitis relies mostly on specific PCR and culturing methods, but commercial syndromic panel assays are bringing a change in diagnostics. With multiplexed analysis, the identification of the pathogen is potentially faster, and less sample material is needed. ABSTRACT Central nervous system (CNS) infections such as meningitis and encephalitis are life-threatening conditions that demand hospital care and prompt identification of the causative agent. Since 2015, there has been only one CE-IVD-marked rapid multiplexed diagnostic assay in cassette format for bacterial and viral detection from cerebrospinal fluid (CSF): the BioFire FilmArray meningitis/encephalitis (ME) panel. In the beginning of 2022, Qiagen introduced the QIAstat-Dx meningitis/encephalitis panel. It is a CE-IVD-marked multiplex PCR cassette test intended for the identification of suspected infectious meningitis, encephalitis, or meningoencephalitis caused by bacterial, viral, or fungal pathogens. In this study, we evaluated patient and quality control samples using the QIAstat-Dx meningitis/encephalitis panel and compared the results to those of the BioFire FilmArray meningitis/encephalitis panel and reference methods (current routine analysis methods in our laboratory, PCR, or cultivation). The combined positive percent agreement between the two panel assays was 100%, and the negative percent agreement was 94%. We further compared specifically herpes simplex virus 1 (HSV-1), HSV-2, and varicella-zoster virus (VZV) dilution series using six commercial herpesvirus assays, including the two cassette tests. The results suggested that real-time PCR methods (with separate extraction) were the most sensitive methods. When comparing the cassette tests, the BioFire FilmArray meningitis/encephalitis panel produced more positive results than the QIAstat-Dx meningitis/encephalitis panel in the herpesvirus analyses. IMPORTANCE The diagnosis of infectious meningitis and encephalitis relies mostly on specific PCR and culturing methods, but commercial syndromic panel assays are bringing a change in diagnostics. With multiplexed analysis, the identification of the pathogen is potentially faster, and less sample material is needed. The novel QIAstat-Dx meningitis/encephalitis panel assay is intended for the rapid identification of pathogens from cerebrospinal fluid for suspected central nervous system (CNS) infection, which is a life-threatening condition and difficult to diagnose. We studied the performance of this panel assay using patient samples and dilution series of selected viruses. The evaluation data for this novel meningitis/encephalitis panel assay are useful for other clinical laboratories and organizations using or considering using this test.

the most common cause of fatal sporadic encephalitis. HSV encephalitis is an emergency situation with a mortality rate of around 70% if left untreated, and even when treated, the mortality rate is high, and long-term effects may occur (7)(8)(9).
Distinguishing different central nervous system infections based only on symptoms is difficult. Therefore, prompt laboratory analysis is needed to identify possible pathogens and guide treatment. Microbiological laboratory analyses from cerebrospinal fluid (CSF) samples traditionally include bacterial culturing, microscopy and antigen tests (10,11). Viral pathogens are often identified using singleplex real-time PCR methods, which have generally shown high sensitivity and specificity (11,12). Also, metagenomic sequencing methods have been developed for the identification of CNS infections (13).
To further improve targeted molecular identification, rapid diagnostics have been developed. The aims are to reduce the time of PCR processes by abolishing manual steps, decrease the sample volume needed, and improve diagnostics by combining several PCR tests into one multiplex PCR panel test. There are now two CE-IVDmarked rapid diagnostic products available (in cassette format) for meningitis and/or encephalitis multiplex PCR testing from CSF samples: the BioFire FilmArray meningitis/encephalitis (ME) panel (FA/ME), which has been available since 2015, and the novel QIAstat-Dx meningitis/encephalitis panel (QIA/ME). FA/ME includes 14 viral, bacterial, and fungal pathogens, while QIA/ME includes 15 pathogens (see Table 1 for more details on the panels).
Multiplex tests have been welcomed with high hopes of reducing the time needed for specific diagnosis, the amount of CSF needed, empirical antibiotic use, and overall costs, etc., and some positive outcomes have already been reported (14)(15)(16). Multiplex panel assays have also raised concerns about their use and the interpretation of their results (17,18), and thus, the evaluation of the new options is of great importance. Comparisons of the two panels will give valuable information on the performances of these assays and practical knowledge for clinical laboratories interpreting the results of panel assays. In this study, the performance of the newly released QIA/ME assay was evaluated by comparing it with the performances of FA/ME and the reference methods Evaluation of the QIAstat-Dx ME Assay Microbiology Spectrum used in our laboratory. The study was further continued by focusing specifically on HSV-1, HSV-2, and VZV detection performance by comparing six different commercial assays using dilution series.

RESULTS
Results for patient samples. We analyzed the CSF samples in parallel using QIA/ ME and FA/ME. FA/ME was considered a comparator method in this analysis. The positive percent agreement (PPA) and negative percent agreement (NPA) of QIA/ME were 100% for all included pathogens except S. pneumoniae ( Table 2). The NPA for S. pneumoniae was 98.7%, and this was due to one sample that was positive by our diagnostic reference method and QIA/ME but negative by FA/ME. Unfortunately, there was not enough sample material for further discrepancy analyses.
The performances of the QIA/ME and FA/ME tests were similar when positive results were compared to the results of the diagnostic reference methods (see Tables 4 and  5). The overall PPAs were 95.1% for QIA/ME and 93.8% for FA/ME. The lowest single PPA (50%) was seen for human herpesvirus 6 (HHV-6) (both QIA/ME and FA/ME versus the reference method); however, there were only two positive samples included in the evaluation. One sample was weakly positive by the reference method (quantification cycle [C q ] value of 39.3), but both QIA/ME and FA/ME gave a negative test result, which was deemed a false-negative result. The other HHV-6-positive sample was a doubly positive sample with additional cytomegalovirus (CMV) positivity. This was correctly identified as doubly positive by FA/ME, but QIA/ME lacks CMV identification.
The diluted samples (Table 3) performed well: all diluted bacterial samples gave the expected results according to the reference method. Two diluted samples, one positive for HSV-1 and another positive for HSV-2, were negative by both QIA/ME and FA/ME. We reanalyzed these samples using the reference method, which identified HSV-1 (C q value of 28.1) and HSV-2 (C q value of 33.2) respectively. The unidentified HSV samples decreased the PPAs to 90% for HSV-1 and 91.7% for HSV-2 (Tables 4 and 5).
Results of herpesvirus dilutions. To demonstrate the performance of the QIA/ME assay near the limit of detection, we prepared a dilution series of HSV-1, HSV-2, and VZV ( Table 6). As expected, less-multiplexed real-time PCR methods with separate extraction (especially RealStar assays) detected much more dilute samples than the panel assays, and the viral target was detected even in the most dilute samples. Overall, FA/ME performed better than QIA/ME with herpesvirus dilutions. FA/ME seemed to perform even better than the Simplexa HSV 1&2 direct assay, especially with HSV-2 dilutions ( Table 6). The Simplexa HSV 1&2 direct assay targets only HSV-1 and -2 and is also considered a rapid diagnostic  assay. Allplex Meningitis-V1 is a multiplex assay but with a separate extraction step, such as in the other two real-time PCR methods. This assay also performed well, detecting only slightly fewer positives than artus HSV-1/2 and RealStar HSV with HSV-1 and -2.
C q values are shown in Table 6. However, C q values cannot be used for quantitation for samples in qualitative assays, but the values are a tool to estimate if the sample is near the limit of detection or strongly positive.
Results for external quality control samples. QCMD (Quality Control for Molecular Diagnostics) 2021 central nervous system I (viral meningitis and encephalitis) external quality assessment (EQA) proficiency samples were used to evaluate the performances of the QIA/ME and FA/ME tests for viral targets (HSV, VZV, enterovirus, and parechovirus). Both tests performed well (Table 7) and had 100% consistency with the expected results.
User experience and device functionality of QIA/ME. Rapid diagnostics have deliberately been designed to be user-friendly, and QIA/ME is not an exception: the handling of the assay cassette is even faster and easier than with FA/ME as the user pipettes the sample directly into the cassette, which is then loaded into the QIAstat-Dx analyzer. Using the QIAstat-Dx analyzer was as easy as using the FilmArray analyzer. The user does not need any extra programs or computers as the results can be read directly from the analyzer's touch screen, or the results can be sent to the laboratory information management system (LIS) (not tested in this study). Clear advantages of QIA/ME are the reported C q values and PCR amplification curves.
The total error rate of QIA/ME in this study was 6.5% (3 errors and 4 failed results from 108 samples run, including dilutions and quality control [QC] samples). The FA/ ME error rate was 0% (total of 115 samples run). There were three errors with QIA/ME during the study; two of them were due to QIAstat-Dx analyzer problems. One error  Evaluation of the QIAstat-Dx ME Assay Microbiology Spectrum was due to a test cassette problem. These three samples were rerun in another analyzer and other test cassettes, and they received the expected results in the second run (all three were positive for HSV-1). In addition, four patient samples gave failed results with QIA/ME. Three of these samples were rerun and received the same results as those of the FA/ME and reference methods (two negative samples and one HSV-1-positive sample). The fourth failed sample was H. influenzae positive by the reference method and FA/ME. Unfortunately, there was not enough CSF left to rerun this sample with QIA/ME.

DISCUSSION
Infectious meningitis, encephalitis, or meningoencephalitis is always a severe and life-threating condition. There are many possible causes, including bacterial, viral, or fungal agents. The rapid identification of the infectious agent is essential for the selection of the appropriate treatment, and multiplex panel assays aim to answer this need. In this study, we evaluated the recently released CE-IVD-marked QIAstat-Dx meningitis/encephalitis panel (QIA/ME). The results of the patient sample analyses indicated that the new QIA/ME assay and the BioFire FilmArray meningitis/encephalitis (FA/ME) assay are comparable and perform quite similarly.
In this study, with a relatively small sample size, the positive and negative percent agreements of QIA/ME and FA/ME (Tables 3 to 5) were slightly better than the specificity and sensitivity values seen in previous studies evaluating FA/ME (17)(18)(19)(20)(21)(22)(23)(24). The PPA for QIA/ME (compared to the reference methods) was 95% in our study, and for example, Tansarli and Chapin (18) estimated a 90% mean sensitivity (and a 97% mean specificity) for FA/ME in their meta-analysis.
Since CNS infection is a severe condition with considerable mortality and morbidity, there should not be false-positive or -negative results by the assays. However, previous studies have drawn attention to false-positive and -negative results using FA/ME (17,18). In our study, no false-positive results were observed, but a few false-negative results were detected. Previous reports have found false-negative results mostly for HSV-1, HSV-2, enteroviruses, and Cryptococcus gattii/C. neoformans using FA/ME (17,18,22). In this study, QIA/ME was unable to identify HSV-1-, HSV-2-, and HHV-6-positive samples (one each). These same samples were also negative by FA/ME, which may suggest similar problems with the identification of herpesviruses. In addition, we noted one false-negative result for S. pneumoniae using FA/ME. However, S. pneumoniae has previously been linked mainly to false-positive results (18,19,23).  Further studies using dilution series suggested that FA/ME might be more sensitive than QIA/ME for the detection of herpes simplex viruses. There was a clear difference in the results of HSV-1 and HSV-2 dilution series by QIA/ME and FA/ME (Table 6). Yet this experiment was prepared with only one representative strain, and the abilities of the assays to detect different strains may vary. The dilution series experiment was prompted by previous reports on the lack of sensitivity of the FA/ME assay for the detection of herpesviruses (12,13,17; L. Mannonen, R. Loginov, T. Holma, J. Antikainen, unpublished data).
In the dilution series experiment, we tested other commercial assays in addition to QIA/ME and FA/ME. For instance, we used the multiplexed Allplex Meningitis-V1 assay from Seegene. To our knowledge, there are no such previous reports on the Allplex Meningitis-V1 panel assay. Overall, one or two target PCR methods are still probably the most common methods used to detect herpesviruses (12). artus from Qiagen and RealStar from Altona were included in the evaluation of the dilution series of HSV-1, HSV-2, and VZV, and they performed better than the multiple-target PCR panel assays, as expected. The fewest differences between the assays were seen for VZV detection.
Published data are accumulating for FA/ME, making it clear that it cannot be used as a stand-alone test. Also, both FA/ME and QIA/ME have several limitations listed in their instructions for use (BioFire FilmArray ME panel CE-IVD instruction booklet [BioFire Diagnostics, bioMérieux LLC] and QIAstat-Dx ME panel instructions for use [handbook], January 2022 [Qiagen GmbH]). One of the limitations is the inability to identify all viral or bacterial strains with sufficient sensitivity. For instance, neither of the tests identifies all HSV-1 strains, and the sensitivity for HSV-1, which is one of the most important pathogens, is compromised. Furthermore, it should be noted that the analytical verification of QIA/ME has shown a potential risk of false-negative results for HSV-1 when S. pneumoniae is also present in the sample. Previous studies have suggested that precautions should be taken when diagnosing a patient based solely on FA/ME results and that additional confirmatory testing is still needed (17,23,24). These studies suggest the need for confirmatory testing, especially for HSV-1, HSV-2, and enteroviruses. The results of our study also support the need for further confirmatory testing for at least  a Results are interpreted as positive, negative, or borderline. The numbers below the results are C q values obtained from replicate analyses (1 to 3 replicates were run for each dilution). The BioFire FilmArray ME (FA/ME) assay does not give C q values. Dark shading indicates a positive result. Light shading indicates results that are in the detection limit area and where the result was positive in only one replicate. QIA/ME, QIAstat-Dx ME panel. b Invalid internal control result. The assay does not make the interpretation. The sample was not rerun.
HSV-1 and HSV-2 when using QIA/ME. This finding naturally needs further studies of QIA/ME with a wider range of samples. However, rapid methods have great potential in the diagnostics of meningitis and/ or encephalitis, and these assays are likely most useful in emergency departments. There are reports of clearly positive impacts of the use of panel assays for meningitis and/or encephalitis diagnostics, for instance, in reducing the time to diagnosis and shortening empirical antibiotic use (14,15). QIA/ME is also a rapid assay with a run time of 1 h and with the availability of an LIS for fast and reliable reporting to the clinic. Both QIA/ME and FA/ME need only 200 mL of a CSF sample, which is important with the usually low sample volume. However, in cases of an error or a failed result, the laboratory must rerun the sample, and more sample is then needed. This makes the reliability of the assays and analyzers critical. In this study, FA/ME was more reliable than QIA/ME, with 0% and 6.5% error rates, respectively.
Interestingly, some laboratories have also reported an increase in the prevalence of findings when using the FA/ME test (25). As a fair number of meningitis or encephalitis patients are left without a final diagnosis of the causative agent, it might be beneficial to test more patients with rapid molecular assays, but it may also add some false-positive results. For instance, the instructions for use for QIA/ME warn that Cutibacterium acnes or Mycoplasma genitalium may give a false-positive Mycoplasma pneumoniae result, and similarly, Haemophilus haemolyticus may be falsely identified as Haemophilus influenzae. Also, not all findings may be equally significant in the panel assays, and it is of the utmost importance to consider all aspects of the patient's situation when making clinical decisions. Although HHV-6 can cause severe CNS disease, especially in immunocompromised hosts, a positive HHV-6 finding may also result from subclinical reactivation of the latent virus or chromosomal integration and may not be relevant for the CNS infection (26)(27)(28)(29). Furthermore, one can detect only those pathogens that are involved in the panel and not potential new or rare causes, which could be found, for example, by bacterial culture. Thus, rapid diagnostics cannot replace all of the current methods since they may have insufficient sensitivity, and the panels lack many pathogens that potentially cause meningitis or encephalitis. Moreover, bacterial antibiotic sensitivity cannot be fully determined by rapid diagnostic methods.
Ideally, rapid testing for the most common microbes causing CNS infections simultaneously improves patient care and outcomes and reduces the overall costs of patient treatment. However, the most important thing is the quality of the test. Here, we have shown that there is still room for improvement in CSF diagnostics using rapid cassette tests. Highly multiplexed assays are usually more difficult to design so that the sensitivity for all targets would be optimal. Therefore, more targeted, less multiplexed panels in a rapid cassette format for the diagnosis of CNS infections might make sense.
Study limitations. This is a retrospective single-center study in which frozen residual CSF samples from clinical diagnostics were used. Testing of CNS infection assays are usually limited by the small sample size, and the samples are often prechosen based on availability, which was also true for this study. We aimed to evaluate the performance of QIA/ME with as wide of a range of positive CSF samples as possible, and this study does not represent the real-life situation in a clinical laboratory, where the majority of CSF samples are negative. Based on the availability of residual samples, this study lacked some bacterial pathogen targets completely, and some pathogen targets had only a few representatives. With the preselected samples and with the priority of having enough sample material, this study had a rather narrow-range C q value sample group, which does not reflect the real clinical situation (Table 3). Thus, the material lacked more challenging samples with low viral loads.
Conclusions. The recently established QIA/ME assay is a rapid diagnostic multiplex PCR assay that significantly reduces the time to a result and the amount of a CSF sample needed when meningitis or encephalitis is suspected. QIA/ME performed very similarly to FA/ME, and further studies are expected to confirm and widen these first observations of the performance of this assay. Panel assays such as QIA/ME and FA/ME provide an option for the rapid diagnostics of central nervous system infections in clinical laboratories, but users should interpret the results carefully in the context of individual patients.

MATERIALS AND METHODS
Panel assays and reference methods. The QIAstat-Dx meningitis/encephalitis (ME) panel (QIA/ME; Qiagen GmbH, Hilden, Germany) was used according to the manufacturer's instructions with QIAstat-Dx analyzer 1.0 (Qiagen GmbH) and QIAstat-Dx Application software 1.4.0.
The BioFire FilmArray meningitis/encephalitis (ME) panel (FA/ME; BioFire Diagnostics, bioMérieux LLC, Salt Lake City, UT, USA) was used according to the manufacturer's instructions with the BioFire FilmArray Torch instrument and software (BioFire Diagnostics, bioMérieux LLC). All of the pathogens included in QIA/ ME and FA/ME and the reference methods used (PCR methods) are listed in Table 1. For culturing, the clinical samples were cultured at the Helsinki University Hospital Laboratory using standard methods (30).
Samples and patients. Patient samples used in this study were residual frozen CSF samples sent to the Helsinki University Hospital Laboratory from 2017 to 2022. The samples were sent for meningitis and/ or encephalitis diagnostics. The main selection criterion was that there was enough residual CSF to carry out both the QIA/ME and FA/ME tests (200 mL is required for each analysis). Second, we aimed to have representative samples for most of the pathogens included in the two panels (see Table 1 for the list of pathogens). We did not have samples for Escherichia coli K1, Streptococcus pyogenes (only for QIA/ME), and Mycoplasma pneumoniae (only for QIA/ME). The number of samples in each pathogen category was limited by the sample volume. Thus, the number of positive samples was increased by diluting positive patient samples in residual negative CSF. Altogether, 23 samples were diluted, most of which were bacterial samples (dilution factors are shown in Table 3). The residual negative CSF samples were defined negative using the FA/ME assay; in addition, the diluted samples were reanalyzed using the appropriate reference method.
In total, 63 positive CSF samples were analyzed (positive by the reference method). The median age of the patients was 47 years, with a range from a few weeks to 91 years. A total of 57% (n = 36) of the positive samples were from male patients, and 43% (n = 27) were from female patients. In addition, 16 clinical samples that were negative by FA/ME were analyzed. These negative samples were used for the calculation of the negative percent agreement of QIA/ME versus FA/ME. Besides these negative samples, samples with a positive result for one target (or two targets in the case of a doubly positive result) in the panel were considered negative for all other targets in the statistical calculations.
Dilution series. Dilution series of HSV-1, HSV-2, and VZV were tested in parallel with six different PCR methods. The dilution series were prepared from leftover cutaneous samples by diluting these into residual negative CSF (defined as negative by FA/ME). A dilution series of 1:10 2 to 1:10 7 was constructed for each virus. The dilution series were analyzed using QIA/ME and FA/ME. In addition, we used Simplexa HSV 1&2 direct (DiaSorin Molecular LLC, Cypress, CA, USA) and Simplexa VZV direct (DiaSorin Molecular LLC) assays with the Liaison MDX instrument according to the manufacturer's instructions. Dilution series were also analyzed using the Allplex Meningitis-V1 assay (Seegene Inc., Seoul, Republic of Korea), the artus HSV-1/2 LC PCR and VZV LC PCR kits (Qiagen GmbH), and RealStar HSV PCR kit 1.0 and VZV PCR kit 1.0 (Altona Diagnostics GmbH, Hamburg, Germany). For these tests, DNA extraction was performed by using the automated NucliSENS easyMAG system (bioMérieux, Boxtel, Netherlands) with a 200-mL starting sample volume, using a generic protocol, and eluting the sample into 25 mL. The Allplex Meningitis-V1 assay was used with the CFX96 Dx system (Bio-Rad Laboratories, Hercules, CA, USA) and Seegene Viewer (V3.24.00) analysis software. The results of the artus assays were analyzed using the LightCycler 2.0 system (Roche, Basel, Switzerland), and the results of the RealStar assays were analyzed using the Applied Biosystems 7500 real-time PCR system (ABI7500; Thermo Fisher Scientific, Waltham, MA, USA).
External quality control samples. The QCMD (Quality Control for Molecular Diagnostics) 2021 central nervous system I (viral meningitis and encephalitis) EQA program was purchased from QCMD (Glasgow, Scotland, UK).
Statistical calculations. Positive percent agreement (PPA) was calculated using the equation 100% Â TP/ (TP 1 FN), where TP is a true-positive sample and FN is a false-negative sample. Negative percent agreement (NPA)was calculated using the equation 100% Â TN/ (TN 1 FP), where TN is a true-negative sample and FP is a false-positive sample.
Ethics statement.