We here report on our real-life experience of the implementation of the FilmArray ME panel into a diagnostic laboratory in a German university hospital setting. We implemented a simple sample selection strategy in our workflow: FilmArray ME analysis was restricted to CSF-samples with a high pretest probability of infectious (bacterial) meningitis. Decision was made in our laboratory based on gram stain and/or urgent suspicion of infectious meningitis communicated by clinicians, e.g. neither the clinical decision for lumbar puncture was affected, nor was explicit ordering of the assay permissible. Using this selection criteria we selected 171/4623 CSF specimens for additional syndromic molecular testing. Overall we observed positive results in 56/171 (32%), which is higher than in other published studies [6, 7, 9-12, 34-38] and substantially higher compared to the multicenter evaluation study [6] and three other studies [7, 32, 33], that reported positivity rates of under 15% (p<0.0001). In addition, we could increase the amount of bacterial pathogens to 53% of the specimens, again this is more than in other published studies that observed positive ranges for bacterial pathogens between 15-37% [6, 7, 32, 33]. Indeed, bacteria were the most abundant pathogens detected in our study (53.57 % of positive samples), followed by viruses detected in 48.21 % of positive samples. In contrast, in the comparator studies viruses were detected most frequently (83.8 %-54.46 % of positive samples) followed by bacteria (37.62 % - 15.21% of positive samples, p<0.0001).
In concordance with other studies [6, 7, 9-12, 34-38], S. pneumoniae (17/30) was the most frequent bacterial pathogen detected by the FilmArray ME assay. Additionally, in 10/30 specimens rapid detection of N. meningitides (n=5), H. influenzae (n=2) or L. monocytogenes (n=3) was important for optimal clinical management including rapid implementation of post exposure treatments in contact persons (antibiotic prophylaxis and/or vaccination).
Analytically we observed good concordance (52/56 positive results could be confirmed) between the FilmArray ME panel results and our comparator assays. Only in two of the samples the FilmArray ME panel yielded a discordant negative result compared to the local reference methods. Namely, in each of one sample S. pneumoniae and Parechovirus were detected by specific in-house PCRs. Although it might be crucial to detect a bacterial or viral meningitis in the individual case, our data confirm the high negative predictive value of the multiplex PCR assay that has been emphasized by others [33].
In four samples, potentially false positive results of FilmArray ME panel occurred, including three samples tested positive for bacterial pathogens by multiplex PCR that remained culture negative. Among the reasons for failure of primary culture is antibiotic administration prior to lumbar puncture [39]. In the three disconcordant samples, bacterial growth inhibitors (such as antibiotics) were found only in the two samples positive for E. coli and S. pneumoniae. One sample positive for S. agalactiae in FilmArray E panel showed no inhibition of growth in B. subtilis inhibition assay. Notably, culture negative yet PCR positive infectious meningitis cases caused by S. agalactiae have been described previously, underlining the importance of molecular diagnostic for optimization of patient management [40-42]. Otherwise, it cannot be excluded that in our sample, FilmArray ME detection of S.agalactiae was indeed false positive and the communication of the test result might have led to a dispensable antibiotic administration.
In our study, one sample remained disconcordant for HSV-1. Beside the probable useless administration of Acyclovir following a false positive HSV-1 result, one should also consider the questionable relevance of Herpesviridae detection in CSF in general. As these viruses feature life-long latency after primary infection, subclinical reactivation with replication can be seen in conjunction with different underlying clinical conditions [43, 44]. Thus, in a worst case scenario, Herpesviridae detection in multiplex PCR testing might lead to delayed diagnosis of the actual underlying disease, as described by Gomez et al for a patient suffering from tuberculous meningitis [14].
The herpesvirus HHV-6 is included in the FilmArray ME panel. This virus exhibits not only latency and also has the potential of chromosomal integration. Therefore, careful interpretation of HHV-6 detection in FilmArray ME panel is mandatory. The clinical diagnosis should generally not be made by molecular detection of HHV-6 in CSF alone [45, 46]. As this study focused on the implementation of the assay into the laboratory workflow, no clinical data was analyzed. The relevance of HHV-6 detection in our study remains unclear in its consequences for patient management and outcome.
An obvious limitation of any syndromic molecular panel testing approach is the limited number of pathogens included into the panel. Even though the FilmArray ME panel includes a broad range of pathogens, overreliance on negative results might be crucial especially in geographical regions with unusual etiologic agents of meningitis this might yield to ineffective testing [7]. In our study, we detected pathogens that are not included in the panel by comparator assays in ten samples: Bacillus sp., CoNS, S. aureus, and K. pneumoniae grew on culture media, whereas BK virus and Streptococcus spp. were detected by PCR. Albeit the questionable clinical relevance of Bacillus sp., Streptococcus spp. and CoNS in a CSF sample, detection of S. aureus, K. pneumoniae and BK-Virus may account for serious infections. This underlines, that reasonable and effective CNS diagnostic should combine classical cultural and molecular methods, rather than focus on one test system alone.
It has been proposed recently that implementation of the FilmArray ME Panel in routine diagnostic may help for cost saving in direct antimicrobial utilization and might decrease diagnostic costs even with uncontrolled routine availability of the assay [47, 48].
However, both Naccache et al and Radmard et al reported massive overutilization of the test, when no eligibility criteria was implemented [32, 33]. In the latter study, more than one third of samples analyzed were taken from patients without suspicion for infectious meningitis/encephalitis. The authors advice against potential overreliance of test results and suggest restriction strategies within the scope of diagnostic stewardship programs [49]. Corroborate findings were reported by Tan et al, who describe a potential overutilization of the assay in children, but yet point out the benefits of early diagnosis of a viral etiology, especially in terms of antibiotic usage [50].
Eichinger et al. reported their findings of implementation of the FilmArray ME panel as a POCT in children with suspected meningitis in a children hospital [51]. They highlight that though the availability of rapid diagnostic tests enhances administration of specific treatment and thus reduces inadequate usage of antibiotics, a structured approach in clinical implementation of the assay is needed. Taken together, there is agreement that restriction of the assay might be helpful for enhancing its clinical utility. In a point-counterpoint discussion [15] the pros and cons of syndromic testing approaches in CSF specimens in general have been summed up. Despite their contradictory positions on the use of the FilmArray ME panel in diagnostic laboratory routine, the authors agree on the fact that it ensures reduced turn-around time of molecular results and therefore might be beneficial for patient outcome. Yet, the appropriate patient population for testing still has to be identified and more data on the performance of the assay on clinical specimens are needed to evaluate the ideal approach for testing.
Our implementation approach is independent from approval by infectious disease specialists and it does not involve any ordering modifications. The latter opportunities have been proposed for diagnostic stewardship programs [49], however, implementation of such criteria might slow-down the whole workflow. Since one major benefit of molecular multiplex testing is the fast generation of results with high negative predictive values [33], any retardation might also be discussed controversially. Therefore we believe that our restriction approach can be an easy and effective alternative.
Limitations of the study: We report our real-life experience of the implementation of the FilmArray ME panel into our laboratory workflow, no sample randomization was done. Clinical specimens were tested prospectively, yet our selection strategy might exhibit a selection bias. Notably, sample selection strategy was mainly based on gram-stain abnormalities ensuring on-spot decision making, but not representing the ideal method for leucocyte detection. Furthermore, we did not assess all samples by all comparator assays, in particular for samples that were not selected for FilmArray ME analysis, diagnostic was performed according to orders and no additional assays were performed. Thus, we cannot rule out, that pathogens have been overlooked in some samples. Moreover, no clinical data was analyzed, hence effects on patient outcome or antibiotic utilization in our study remain unclear. Nevertheless we believe that our data gain new insights in how syndromic panel testing may be implemented into laboratory routine and therefore may help to identify the ideal approach to ensure its clinical utility.