Diagnostic challenges with accurate identification of Listeria monocytogenes isolates from food and environmental samples in South Africa

Background The 2017–2018 listeriosis outbreak in South Africa warranted testing for Listeria monocytogenes in food products and processing environments. Diagnostic tests are needed to accurately differentiate L. monocytogenes from other Listeria species. Objective The study assessed the performance of the commonly used tests in our setting to accurately identify L. monocytogenes. Methods The study was conducted in a public health laboratory in South Africa. Cultured isolates from food and environmental samples were tested both prospectively and retrospectively between August 2018 and December 2018. Isolates were phenotypically identified using tests for detecting β-haemolysis, Christie-Atkins-Munch-Peterson, alanine arylamidase (AlaA), mannosidase, and xylose fermentation. Listeria monocytogenes isolates were identified using automated systems, Microscan Walkaway Plus 96, Vitek® MS, Vitek® 2 and Surefast Listeria monocytogenes PLUS PCR. All results were compared to whole-genome sequencing results. Results β-haemolysis and Christie-Atkins-Munch-Peterson tests gave delayed positivity or were negative for L. monocytogenes and falsely positive for one strain of Listeria innocua. The AlaA enzyme and Colorex Listeria agar lacked specificity for L. monocytogenes identification. Based on a few phenotypic test results, an aberrant L. monocytogenes strain and Listeria seeligeri strain were reported. All automated platforms overcalled L. monocytogenes in place of other Listeria species. Conclusion No test was ideal in differentiating Listeria species. This is an issue in resource-limited settings where these tests are currently used. Newer technologies based on enzyme-linked immunosorbent assay and other molecular techniques specific to L. monocytogenes detection need to be investigated.


Introduction
The Analytical Profile Index Listeria test (BioMerieux, Marcy d'Etoile, France) fails in 10% -15% of identification cases. The main reason for this failure is due to the weak colour determinations. This is particularly applicable to the arylamidase test. The arylamidase enzyme, tested for in the popular Differentiation Innocua Monocytogenes (DIM) test, is supposed to be negative in L. monocytogenes and positive in other Listeria species. 12 Often a weak positive DIM result was considered a negative result, increasing the false positive L. monocytogenes determinations. 4 This might be due to the doubtfulness of the colour determinations by the reader of the test. Furthermore, false negative identification in atypical L. monocytogenes strains is also frequent. 4 Matrix-Assisted Laser Desorption Time of Flight (MALDI-TOF) mass spectrometry is a quick and easy methodology gaining popularity in several microbiology laboratories. However, MALDI-TOF reportedly misidentifies L. innocua as L. monocytogenes American Type Culture Collection (ATCC) strain and the L. seeligeri ATCC strain as L. monocytogenes or L. innocua. 4 Rychert et al. reported that Vitek ® Mass Spectrometer (MS) version 2.0 system (BioMerieux, Marcy d'Etoile, France) correctly identified only 76% (34/45) of L. monocytogenes to the species level and 9% (4/45) to the genus level, while in 15% (7/45) identification could not be finalised because split identification and re-testing were not performed. 13 The Vitek ® 2 system has also been reported to misidentify L. monocytogenes as L. innocua based on a negative reaction for phospholipase C in 1.4% (4/288) of a collection of isolates tested. 14 The instrument could not identify an L. monocytogenes strain and gave a species error in another study. 15 In a previous evaluation of the Vitek system, when genus level identification of various Listeria species was sought, the instrument had a sensitivity of 97.5%. 16 The Microscan Walkaway Si system (Siemens Healthcare Diagnostics, West Sacramento, California, United States) could not identify one L. monocytogenes ATCC strain BAA-751 during a comparative study with the Vitek ® 2 compact system. 17 The reason for this was potentially attributed to the limited number of Listeria species strains on the database.
During the investigation of the South African listeriosis outbreak in 2017-2018, four Listeria species (L. monocytogenes, L. innocua, L. welshimeri, and L. seeligeri) were detected from food samples and environmental swabs tested at the Infection Control Services Public Health Laboratory in Johannesburg, South Africa. This is similar to what has been described elsewhere in outbreak settings. 5 As a result, accurate discrimination of L. monocytogenes from other species is of critical importance. Whole-genome sequencing is a useful tool for confirmatory identification of L. monocytogenes and can be used as the reference standard test for comparing other tests. 18 Subsequent to the reported limitations of Listeria tests commonly utilised in most public health laboratories, particularly in low-and middle-income countries, the Infection Control Services Public Health Laboratory evaluated the performance of the commonly utilised phenotypic tests (conventional phenotypic tests and chromogenic media) for the identification of Listeria species in comparison to WGS results. The Infection Control Services Public Health Laboratory also compared the performance of the different automated diagnostic systems available in the institution for the identification of L. monocytogenes utilising known Listeria isolates characterised by WGS.
The study results will inform whether current tests are acceptable for future use and, if not, it will justify the evaluation of other technologies for accurate identification of L. monocytogenes.
Open Access

Ethical considerations
Only cultured isolates from food and environmental samples were utilised in this research. No isolates from animals or animal-derived samples were used. Therefore, no ethical clearance was required.

Study design and samples used
Data for this analysis were collected prospectively from August 2018 to December 2018 at the Infection Control Services Public Health Laboratory in Johannesburg. Isolates were cultured from food and environmental swabs of several food processing facilities across all of the provinces in South Africa during the listeriosis outbreak period. All Listeria isolates were identified on Vitek ® 2 (BioMerieux, Marcy-I'Etoile, France). The phenotypic tests were performed either to (1) confirm the initial identification from Vitek ® 2 or (2) to discriminate between Listeria species if two species identifications were given by Vitek ® 2. As a result, not all phenotypic tests were performed on all isolates. The accuracy of the conventional phenotypic tests to discriminate the four Listeria species (L. monocytogenes, L. innocua, L. welshimeri, and L. seeligeri) was assessed (Table 1). 3 The isolate identity (Vitek ® 2 and phenotypic testing) was confirmed by WGS.

Laboratory analyses
Beta (

Data analysis
The sensitivity, specificity, positive predictive value, and negative predictive value of the test methods for detecting L. monocytogenes were calculated. Data were collected on Excel spreadsheets (Microsoft, Redmond, Washington, United States) and analysis was performed using two-by-two tables. 20 Calculations were done as follows: •

Results
The phenotypic results of the Listeria species in comparison to the WGS results are summarised in the Online Supplementary Table 1.

Performance of the phenotypic tests for L. monocytogenes identification
The three phenotypic tests used to confirm the identification of L. monocytogenes by Vitek ® 2 and discriminate it from the other Listeria species were β-haemolysis, the CAMP test, and AlaA activity (Table 2).
Of the 39 L. monocytogenes isolates identified by WGS, all three phenotypic tests corroborated the WGS findings in 82% (32/39) of the isolates. β-haemolysis and the CAMP test were absent in 18% (7/39) of the isolates. Delayed positivity to both of these tests occurred at 72 h in 5.1% (2/39) of isolates. One L. innocua isolate was falsely positive to both of these tests.
Of the L. monocytogenes isolates, 18% (7/39) were falsely positive for AlaA on Vitek ® 2. All L. innocua isolates and the one L. seeligeri isolate were falsely negative for AlaA.

Performance of the automated systems in the identification of L. monocytogenes
All the systems overcalled L. monocytogenes in place of other species (Table 3). All Listeria species results on the various