Comparison of the culture method with multiplex PCR for the confirmation of Legionella spp. and Legionella pneumophila.

AIMS: The detection and enumeration of Legionella spp. in water samples are typically performed via a cultural technique standardized in ISO 11731. This method is time-consuming (up to 15 days), and the specificity of the confirmation step is questionable. This study proposes the use of multiplex polymerase chain reaction (PCR) to confirm presumptive Legionella colonies directly from the culture plate; this shortens the response time by 2-5 days while still reporting results in colony forming units (CFU). METHODS AND
RESULTS: Two laboratories analysed a total of 290 colonies to compare the confirmation step of Legionella spp. and Legionella pneumophila in accordance with ISO 11731 by culture growth and agglutination vs multiplex PCR. Discordant results were resolved by the swiss national reference laboratory. The data were evaluated following ISO 16140 and showed that the PCR-technique had higher specificity.
CONCLUSIONS: The confirmation of Legionella spp., L. pneumophila and L. pneumophila serogroup 1 by multiplex PCR allows detection of positive colonies more rapidly and with higher specificity. SIGNIFICANCE AND IMPACT OF THE STUDY: The study highlights a possibility to shorten the response time significantly during the enumeration of Legionella spp. and achieving a higher specificity while adhering to the legally recognized reporting in CFU.

Introduction

Legionella spp. are gram‐negative bacteria that can cause legionellosis in humans after being inhaled together with aerosols containing them. The symptoms of legionellosis can vary from a flu‐like Pontiac fever to Legionnaires' disease—an atypical and acute pneumonia (Stojek and Dutkiewicz 2006) Therefore, Legionella spp. represent a significant health risk, especially for the elderly and people with a weakened immune system or underlying morbidities (Collins et al. 2015).

Legionella spp. tend to colonize artificial water systems with temperatures between 20 and 45°C and can reach high concentrations. Systems with the potential for aerosol formation, such as hot and cold water systems, cooling towers, evaporation condensers and whirlpools, are considered to be particularly risky (Bartram 2007). In response to these risks, national health authorities and drinking water legislations in Europe recommend regular testing for Legionella spp. and prescribe legal maximum values (e.g., Bundesministerium der Justiz und für Verbraucherschutz, 2016; Eidgenössische Departement des Innern, 2016). Moreover, the operators of public bath and shower facilities are obliged to perform checks, leading to an increased number of samples for analytical testing laboratories.

For the testing of environmental samples, the culture method (ISO 11731:2017) represents the standard technique for the detection and quantification of viable Legionella spp. in Europe (Collins et al. 2015). Results can take up to 15 days (ISO 11731:2017) after sampling, making the culture method a relatively slow technique. This limitation arises from the growth behaviour of Legionella, whose growth can be quickly inhibited by the presence of other microorganisms (Leoni and Legnani, 2001). Further limitations include poor reproducibility of the culture method (Boulanger and Edelstein 1995; Yáñez et al. 2005; in't Veld, 2017) as well as its inability to detect viable but non‐culturable cells (VBNC) that occur under specific environmental conditions (Shih and Lin, 2006). As an alternative to the culture method, the development of molecular methods, such as quantitative real‐time polymerase chain reaction (real‐time PCR), has been described by various authors (Wilson et al. 2003; Yáñez et al. 2005; Yang et al. 2010). At present, different studies have compared the detection and quantification of Legionella in different water samples by the culture method and real‐time PCR (Villari et al. 1998; Bonetta et al. 2010; Collins et al. 2015; Farhat et al. 2018). They were able to show that real‐time PCR can be used as a supplementary enumeration technique to screen for negative samples, thus, enabling results the same day. However, results in colony forming units (CFU) and genome units (GU) only show a weak correlation and therefore, the values cannot be directly converted (Collins et al. 2015). National legislations define the legal maximum value for Legionella in CFU. Therefore, PCR‐positive samples still need to be tested using the culture method to prove conformity with legislation.

When applying the culture method, it is almost impossible to distinguish different Legionella spp. based on the appearance of the colonies on agar media. Presumptive colonies should consequently be confirmed, according to ISO 11731:2017, by subculture on BCYE Agar with l‐cysteine (buffered charcoal yeast extract with α‐ketoglutarate) and BCYE‐Cys‐Agar (BCYE Agar without l‐cysteine). Blood agar, nutrient agar or tryptone soy agar can be used instead of BCYE‐Cys agar. This subculture step requires two to five additional days. Further confirmation of Legionella and identification of the species and serogroup can be done using commercially available latex agglutination kits, which allow for rapid identification of Legionella pneumophila and other clinical relevant strains with high specificity (Reyrolle et al. 2004; Helbig et al. 2006). One further advantage is that the method requires no expensive analytical equipment. However, the confirmation procedure based on subculture is laborious. Our experience shows that the agglutination test can lead to uncertain results and misinterpretations in cases where the reaction is delayed. The following alternative methods are routinely applied in laboratories to confirm presumptive colonies:

Polymerase chain reaction

PCR methods are an interesting alternative for the confirmation of presumptive Legionella colonies. In recent years, several PCR‐based methods aimed at detecting and quantifying of Legionella in water have been described. Most published PCR systems typically target the 5S and 16S rRNA genes for identification of the genus Legionella and the macrophage infectivity potentiator (mip) gene for the specific detection of L. pneumophila strains (Villari et al. 1998; Bonetta et al. 2010; Collins et al. 2015; Farhat et al. 2018). The ribosomal 23S–5S spacer region's DNA‐sequence information was used to develop a genus specific detection and identification system for Legionella spp., using PCR (Herpers et al. 2003). PCR methods can provide information on the serogroup as well as additional information. For example, by sequencing the PCR amplicon with subsequent sequence analysis, e.g., in a BLAST algorithm search, the strain under investigation can be identified at the species or even strain level.

Matrix‐assisted laser desorption/ionization time‐of‐flight

Matrix‐assisted laser desorption/ionization time‐of‐flight mass spectrometry (MALDI‐TOF‐MS) has recently emerged as a rapid bacterial spp. identification method. The disadvantages are the high cost associated with instrument acquisition. The method is therefore only suitable for high sample throughput. It has been shown that most bacterial spp. exhibit specific protein profiles (Hsieh et al. 2008). The method was further developed for the identification of Legionella spp. and L. pneumophila. The technique allows to perform a full analysis from a single colony in only a few minutes, thus, providing an inexpensive (after the initial investment costs) and rapid screening of a large number of colonies within a short time (Gaia et al. 2011). Generally, MALDI‐TOF is simpler to use than PCR, and results can be obtained faster. MALDI‐TOF cannot identify Legionella strains at the serogroup level but is a reliable tool for the rapid identification of Legionella strains at the species level (Moliner et al. 2010). However, colonies which have grown together can no longer be directly identified by MALDI‐TOF without subculturing.

Other methods

A range of other methods can also be used for the identification of presumptive colonies, such as fluorescent in situ hybridization (Baudart et al. 2015), Raman spectroscopy (Li et al. 2015), pulsed field gel electrophoresis, multilocus sequence typing and whole genome sequencing (Zourob 2008). These methods are typically labour‐intensive and require dedicated instrumentation and expertise. To our knowledge, these techniques are, therefore, not routinely applied to the identification of presumptive Legionella colonies. Technical improvements could change this in the future.

The confirmation step based on the non‐growth of presumptive Legionella spp. on only one culture medium, as described in ISO 11731:2017, is prone to errors. Alternative confirmation methods, such as multiplex PCR, should be considered for routine application to achieve a higher specificity and reduce the time expenditure by 2–5 days. This study aims to evaluate multiplex PCR for colony confirmation directly from the culture plate. Two different PCR systems—a commercial one (Biotecon Diagnostics, Potsdam, Germany) and an in‐house method—were applied by the two laboratories in this comparison study.

Materials and methods

Laboratories

Industrielle Werke Basel is the water supplier for the region of Basel, Switzerland. Its quality control laboratory is ISO 17025 (STS 0211) accredited and henceforth referred to as laboratory 1.

The Cantonal Laboratory of Zurich (Kantonales Labor Zurich) is responsible for enforcing the Federal Act on Food and Utility Articles in the Canton Zurich, Switzerland. It is ISO 17025 (STS 0172) accredited and will in the following be referred to as laboratory 2.

Samples, sample exchange and storage conditions

Between July 2019 and April 2020, 90 samples of presumptive Legionella colonies were collected in laboratory 1 and 200 samples in laboratory 2. All samples were analysed according to the confirmation procedure of the culture method and by PCR confirmation in the laboratory of origin. Furthermore, all samples from laboratory 1 and 50 samples from laboratory 2 were sent to the other laboratory for confirmation with PCR. Hence, a total of 140 samples were additionally subjected to an interlaboratory comparison for both PCR systems.

For confirmation with PCR, the presumptive colonies were stored up to 6 months to analyse several samples in one PCR run and pass on a larger number of samples between the laboratories. Both laboratories agreed on the following storage conditions: the presumptive colonies must be labelled clearly, and the culture plates must be stored upside‐down at 2–8°C. Additionally, presumptive colonies can be transferred into 0·5 Tris‐EDTA (TE) buffered water and stored at 2–8°C for a maximum of 14 days. The storage conditions ensured that both laboratories could perform their whole PCR protocol, starting with the lysis step.

Confirmation via culture method

The isolation of Legionella spp. from the water samples was performed according to ISO 11731:2017. For filtration, 0·45‐μm nitrocellulose membrane filters were used. The specific procedure was chosen according to the water type and the resulting concentration of accompanying flora. For water with low accompanying flora, direct plating and membrane filtration without washing were used. For water with high accompanying flora direct plating, membrane filtration without washing and if necessary, dilutions were used. Both laboratories used BCYE Agar and BCYE+AB (BCYE Agar with antibiotics) or GVPC Agar (Glycine‐Vancomycin‐Polymyxin‐B‐Cycloheximide‐Agar). Independent of the first step of the procedure, the inoculated plates were incubated for 7–10 days at 37 ± 1°C. All colonies with a characteristic morphology according to ISO 11731:2017 were typed as presumptive colonies. One to three colonies of each type were randomly chosen and subcultured on BCYE Agar and BCYE‐Cys‐Agar or Columbia Blood Agar. Colonies that grew only on BCYE Agar were counted as Legionella spp. For further identification of L. pneumophila serogroup 1 and L. pneumophila serogroup 2–14, a latex agglutination test (Thermo Fisher Diagnostic Switzerland, DR800 M) was conducted.

Commercial PCR kit (laboratory 1)

Laboratory 1 used the microproof® Legionella Quantification LyoKit (Biotecon Diagnostics GmbH, R 602 45) combined with the foodproof® StarPrep Three (Biotecon Diagnostics GmbH, S 400 18) for DNA extraction.

A small amount of the presumptive colony was taken with a sterile dee and resuspended in either 50‐ or 100‐µl lysis buffer. Pseudomonas aeruginosa (ATCC 27853) was used as negative control. Legionella pneumophila serogroup 1 (ATCC 33152) was used as positive control. The samples and controls were lysed at 98 ± 2°C for 5 min, briefly vortexed and centrifuged at 13 000  for 2 min.

For each sample and control, 5‐µl DNA‐extract was diluted with 20‐µl 0·5 TE buffer and added to the lyophilized master mix. The PCR strips were centrifuged for 0.5 min at 200 x or 5 s at 2000 x . PCR was performed with a CFX96 (Bio‐Rad, Potsdam, Germany; 1855195).

Legionella spp. was detected in the HEX channel via a specific region in the rDNA operon sequence, of which three copies are present in the Legionella genome. Legionella pneumophila was detected in the FAM channel via the mip gene and strains belonging to serogroup 1 by amplifying the wzm gene in the ROX channel. All three genes have been shown to allow specific detection of the respective targets (Mérault et al. 2011). The kit provides synthetic DNA as part of the lyophilized master mix that works as an internal control. It therefore allows monitoring of successful DNA amplification in each reaction tube by using a CY5‐probe. The kit was validated by the manufacturer in accordance with ISO/TS 12869: 2012. All parameters met the criteria set by the aforementioned norm. The validation data are available from the manufacturer.

Real‐time PCR protocol was as follows: pre‐incubation for 4 min at 37°C and incubation for 5 min at 95°C followed by 50 cycles of denaturation for 5 s at 90°C, annealing for 60 s at 60°C and extension for 60 s at 72°C. Data were collected after each annealing step. CFX Manager Software Industrial Diagnostic Edition ver. 3.0 was set to automatic baseline setting with activated baseline drift correction. For each sample, the result was given as C value.

As a bacterial colony typically contains billions of cells, amplification is achieved already after a few cycles. Therefore, samples with a C value ≤ 25 were rated as positive. Samples with a C value > 25 were regarded as potentially cross‐contaminated and consequently rated as negative. For a successful PCR run, the positive control had to be rated positive, whereas the negative control had to be rated negative for all three parameters/channels. Furthermore, all samples rated as negative, needed positive internal controls to exclude PCR inhibition, to be counted as a validated result. PCR‐inhibition was observed in only one sample. In this case, a successful retest was performed, otherwise a 1 : 5 dilution (v/v) of the DNA extract with PCR water would have been applied (as recommended by the kit manufacturer) to prevent PCR inhibition (e.g., due to DNA‐overload or presence of inhibiting substances).

In‐house PCR (laboratory 2)

Presumptive colonies (as described above) for Legionella spp. were picked and resuspended in 100‐μl DNA‐free pure water. For the DNA extraction, the tubes containing the samples were heat‐treated at 95°C for 15 min.

Real‐time PCR was performed for the qualitative analysis of Legionella spp. and L. pneumophila with the Rotor Gene Q6000 (Qiagen Switzerland, 9001720). Five microliter of the DNA‐sample were added to a PCR master mix of total 20 μl, which consisted of 6·9‐µl 0·5 TE buffer, specific primers 3 × 0·1 µl, 20 μmol l−1 (LegP F/R, Legio F/R, TEcol1553), specific Probes 3 × 0·1 µl, 50 μmol l−1 (Legio Fam, LegPROX, Tecol573Hex) and 12·5‐µl Sensifast (Bioline, London, UK, BIO‐86020). The following primers and probes were used for the detection: Detection of Legionella spp. using the 23S‐5S spacer region by primer Legio F (5′‐CTA ATT GGC TGA TTG TCT TGA C‐3′), primer Legio R (5′‐GGC GAT GAC CTA CTT TCG‐3′) and probe LegioFam (5′‐CGA ACT CAG AAG TGA AAC‐3′) (Herpers et al. 2003). Detection of L. pneumophila using the mip gene by primer LegP F (5′‐AAA GGC ATG CAA GAC GCT ATG‐3′, primer LegP R (5′‐TGT TAA GAA CGT CTT TCA TTT GCT G‐3′) and probe LegPRox (5′‐TGG CGC TCA ATT GGC TTT AAC CGA‐3′) (Welti et al. 2003). Detection of Escherichia coli using the elongation factor Tu (tuf) by primer TEcol1553 (5′‐TGG GAA GCG AAA ATC CTG‐3′), primer TEcol1754 (5′‐CAG TAC AGG TAG ACT TCT G‐3′) and probe TEcol573Hex (5′‐AACTGGCTGGCTTCCTGG‐3′) (Maheux et al. 2011). Thus, the FAM channel detects Legionella spp. and the ROX channel L. pneumophila. The internal control E. coli is detected in the HEX channel. DNA of E. coli can be added to the isolated DNA and can serve as an inhibition control. The in‐house PCR was validated by laboratory 2 according to Swiss national standard (Schweizerische Akkreditierungsstelle 2017).

Every real‐time PCR run contained a standard range of L. pneumophila DNA to optionally calculate a standard curve, a positive control (L. pneumophila ATCC 33152) and a negative control sample (0·5 TE buffer).

The two step cycling program for the PCR consisted of initial denaturation for 5 min at 95°C followed by 45 cycles of 5 s at 95°C for denaturation and 15 s at 55°C for annealing and elongation.

The PCR results were analysed with the Rotor‐Gene Q Series Software 2.0.2 according to the Operators Instructions (Rotor‐Gene® Q User Manual, ver. 2.0, 2009). The threshold was set automatically with ‘Auto‐Find Threshold’ to get the C value of each sample. For this analysis process, the ‘Dynamic Tube Normalization’ option was activated as well as ‘Slope Correct’, ‘Linear Scale’ and ‘Outlier Removal’ if required. Samples with a Legionella spp. C value ≤ 28 and a L. pneumophila C value ≤ 25 were rated as positive. Samples with higher C values were rated as negative.

This in‐house developed PCR‐method does not deliver information on serogroups.

Review by the reference laboratory

Samples with divergent results regarding the culture method and PCR were analysed by the Swiss reference laboratory (Centro Nazionale di Referenza per Legionella in Bellinzona) for verification. The corresponding colonies from the stored plates were first recultivated on BCYE Agar and then analysed using MALDI‐TOF. The reference laboratory uses the Vitek MS system from bioMérieux with a validated SARAMIS database and in‐house spectra. Legionella spp. and L. pneumophila positive colonies, respectively, were further examined using agglutination tests.

Statistical methods

The four‐field test described in ISO 16140:2003 was used to compare the qualitative methods assessing the following performance indicators: specificity, sensitivity and accuracy. The Kappa concordance index (ISO 16140:2003; Schweizerische Akkreditierungsstelle 2017) was chosen for statistical analysis, which shows the degree of agreement between the two test methods.

Results

Comparison of culture method and commercial PCR kit

Laboratory 1 analysed 90 samples. The results can be seen in Table 1.

Table 1

Comparison of culture method and commercial PCR kit (laboratory 1)

Total = 90	Legionella spp. (sample IDs)‡	Legionella pneumophila (sample IDs)	Legionella pneumophila serogroup 1 (sample IDs)
Culture* + and PCR† +	31	15	5
Culture − and PCR −	55	69	85
Culture + and PCR −	2 (ID 061, 025)	0 (no samples)	0 (no samples)
Culture − and PCR +	2 (ID 004, 034)	6 (ID 003, 012, 013, 032, 033, 034)	1 (ID 003)

Culture describes confirmation on specific agar. + and − describe the final confirmation result.

PCR describes confirmation using PCR. + and − describe the final confirmation result.

For samples with divergent PCR and culture method result, sample IDs are given (for more information, see Table 2).

Table 2

Comparison of samples with divergent results using the commercial PCR kit and culture method

Sample ID	Culture method (ISO 11731:2017)	Commercial PCR kit (laboratory 1)	MALDI‐TOF and agglutination test (Swiss reference laboratory)
003	Legionella spp.	Legionella pneumophila serogroup 1* (Legionella spp.)†	Legionella anisa
004	No Legionella spp.	Legionella spp.	Not analyzed
012	Legionella spp.	Legionella pneumophila serogroup 2–14	Legionella pneumophila serogroup 2–14
013	Legionella spp.	Legionella pneumophila serogroup 2–14	Legionella pneumophila serogroup 2–14
025	Legionella spp.	No Legionella spp.	Bradyrhizobium spp.
032	Legionella spp.	Legionella pneumophila serogroup 2–14	Legionella pneumophila serogroup 2–14
033	Legionella spp.	Legionella pneumophila serogroup 2–14	Not analyzed
034	No Legionella spp.	Legionella pneumophila serogroup 2–14	Not analyzed
061	Legionella spp.	No Legionella spp.	Not analyzed

Original result.

Result after recultivation (just for information purpose, was not included in the statistical evaluation).

For Legionella spp., the method comparison (PCR vs culture) showed an accuracy of 96%, a specificity of 97%, a sensitivity of 94% and a statistical agreement between the two methods of 0·90, as assessed by the Kappa concordance index. For L. pneumophila, the method comparison showed an accuracy of 93%, a specificity of 92%, a sensitivity of 100% and a statistical agreement between the two methods of 0·79. For L. pneumophila serogroup 1, the method comparison showed an accuracy and specificity of 99%, a sensitivity of 100% and a statistical agreement between the two methods of 0·90.

Nine of the 90 samples showed a difference between the result of the culture method and PCR. Five of the nine samples were successfully recultivated (plating on BCYE, up to 5 days at 37°C) for analysis at the Swiss reference laboratory using MALDI‐TOF analysis and agglutination tests. Table 2 shows an overview of the samples with divergent PCR and culture results together with the results of the reference laboratory. For all recultivated samples, the Swiss reference laboratory confirmed the PCR result. One sample initially showed a discrepancy between culture and PCR. After recultivating and re‐testing the sample, all methods agreed. For the statistical evaluation, the diverging first result has been used in the method comparisons.

Comparison of culture method and in‐house PCR

Laboratory 2 analysed a total of 200 samples. The results of method comparison (PCR vs culture) for Legionella spp. and L. pneumophila are presented in Table 3. For Legionella spp., an accuracy of 96%, a specificity of 100%, a sensitivity of 94% and a statistical agreement of 0·91 were obtained. For L. pneumophila, an accuracy of 97%, a specificity of 94%, a sensitivity of 99% and a statistical agreement between test methods of 0·94 were obtained. Thirteen of the 200 samples showed a difference between the culture method and the PCR result. All samples were successfully recultivated for analysis in the Swiss reference laboratory. Table 4 combines an overview of the samples with divergent PCR and culture results with the reference laboratory results using MALDI‐TOF analysis and agglutination tests. For all but one sample, the Swiss reference laboratory confirmed the PCR result. For the one divergent sample, the culture method analysed Legionella spp., the PCR method analysed L. pneumophila and the reference laboratory reported the colony identity as Mycobacterium spp.

Table 3

Comparison of culture method and the in‐house PCR

Total = 200	Legionella spp. (sample ID)‡	Legionella pneumophila (sample ID)
Culture* + and PCR† +	126	117
Culture − and PCR −	66	77
Culture + and PCR −	8 (ID 141, 142, 143, 144, 174, 175, 264, 290)	1 (ID 290)
Culture − and PCR +	0 (no samples)	5 (ID 168,169,170,172,173)

Culture describes confirmation on specific agar. + and − describe the final confirmation result.

PCR describes confirmation using PCR. + and − describe the final confirmation result.

For samples with divergent PCR and culture method result, sample IDs are given (for more information, see Table 4).

Table 4

Comparison of samples with divergent in‐house PCR and culture results

Sample ID	Culture method (ISO 11731:2017)	In‐house PCR (laboratory 2)	MALDI‐TOF and agglutination test (Swiss reference laboratory)
141	Legionella spp.	No Legionella spp.	Acinetobacter
142	Legionella spp.	No Legionella spp.	Acinetobacter
143	Legionella spp.	No Legionella spp.	Acinetobacter
144	Legionella spp.	No Legionella spp.	Acinetobacter
168	Legionella spp.	Legionella pneumophila	Legionella pneumophila serogroup 2–14
169	Legionella spp.	Legionella pneumophila	Legionella pneumophila serogroup 2–14
170	Legionella spp.	Legionella pneumophila	Legionella pneumophila serogroup 2–14
172	Legionella spp.	Legionella pneumophila	Legionella pneumophila serogroup 2–14
173	Legionella spp.	Legionella pneumophila	Mycobacterium spp.
174	Legionella spp.	No Legionella spp.	Bacillus spp.
175	Legionella spp.	No Legionella spp.	Bacillus spp.
264	Legionella pneumophila	No Legionella spp.	Bacillus spp.
290	Legionella spp.	No Legionella spp.	Bacillus spp.

Comparison of commercial PCR and in‐house PCR (interlaboratory comparison)

In total, 140 samples (90 samples of laboratory 1 and 50 samples of laboratory 2) were analysed with PCR by both laboratories. Two samples have been excluded due to inconclusive labelling. The results of the interlaboratory comparison can be seen in Table 5.

Table 5

Interlaboratory comparison of commercial PCR kit and in‐house PCR

Total = 138	Legionella spp. (sample ID)†	Legionella pneumophila (sample ID)
Commercial PCR* + and in‐house PCR +	71	57
Commercial PCR − and in‐house PCR −	63	77
Commercial PCR + and in‐house PCR −	3 (ID 014,058,067)	3 (ID 003,014,083)
Commercial PCR − and in‐house PCR +	1 (ID 057)	1 (ID 090)

PCR describes confirmation using PCR. + and − describe the final confirmation result.

For samples with divergent PCR result between the laboratories, sample IDs are given (for more information, see Table 6)

Table 6

Samples with divergent PCR results in the interlaboratory comparison

Sample ID	Commercial PCR kit (laboratory 1)	In‐house PCR (laboratory 2)	Culture method (ISO 11731:2017)
003	Legionella pneumophila serogroup 1* (Legionella spp.)†	Legionella spp.	Legionella spp.‡
014	Legionella pneumophila serogroup 2‐14	No Legionella spp.	Legionella pneumophila serogroup 2‐14
057	No Legionella spp.	Legionella spp.	No Legionella spp.
058	Legionella spp.	No Legionella spp.	Legionella spp.
067	Legionella spp.	No Legionella spp.	Legionella spp.
083	Legionella pneumophila serogroup 1	Legionella spp.	Legionella pneumophila serogroup 1
090	Legionella spp.	Legionella pneumophila	Legionella spp.

Original result.

Result after recultivation (just for information purpose, was not included in the statistical evaluation).

MALDI‐TOF result was Legionella anisa (available due to different results during the comparison of culture method and the commercial PCR kit).

In the statistical evaluation, the results of the commercial PCR kit have been stated as reference method. For Legionella spp., the interlaboratory comparison showed an accuracy of 97%, a specificity of 98% and a sensitivity of 96%. For L. pneumophila, the method comparison showed an accuracy of 97%, a specificity of 99% and a sensitivity of 95%. With a statistical agreement of 0·94 for Legionella spp. and L. pneumophila, a significant correlation was found.

Seven of the 138 samples showed divergent results between the commercial PCR kit and the in‐house PCR (see Table 6). As described above, one sample of laboratory 1 showed a discrepancy between the PCR result of laboratory 1 before and after recultivation for the analysis in the Swiss national laboratory. For the statistical evaluation, the diverging first result has been used in the method comparisons.

Discussion

ISO 11731:2017 (Appendix G) indicates the possibility to use the PCR method to identify Legionella colonies. In previous studies, the PCR method has also been successfully applied to confirm presumptive colonies (Saint 1998; Giglio et al. 2005; Tabatabaei et al. 2016). In contrast to these studies, we used a multiplex PCR that enables us to check presumptive colonies for Legionella spp. and identify L. pneumophila simultaneously with the same measurement experiment. Additionally, the commercial PCR kit delivers information on L. pneumophila serogroup 1.

The PCR‐confirmation of presumptive colonies directly from the cultivation plate has various advantages compared to the standard procedure involving subculturing and subsequent agglutination test:

The specificity is significantly higher. The subculture yields a substantially higher proportion of false‐positive results than the PCR‐method. Of the 18 colonies verified by the Swiss reference laboratory, 10 of the culture method results were false‐positive regarding Legionella spp. Furthermore, the agglutination method has failed to identify L. pneumophila positive samples in seven cases, where PCR showed positive results.

The confirmation by PCR needs less analysis time and final results can be obtained 2–5 days faster. Subculturing on blood‐agar or BCYE‐Cys‐Agar takes additional 2–5 days for incubation. In contrast, PCR identification of presumptive colonies directly from the primary plate delivers results within 2–4 h, including both DNA extraction and PCR run time. Different kit manufacturers for multiplex PCR recently offer even shorter DNA extraction procedures and PCR run times, especially optimized for identification purposes, only. However, for quantification of Legionella according to ISO 11731:2017, the primary plate must be incubated for at least 7–10 days to give all possible Legionella spp. enough time to grow.

The two‐step procedure of subculturing and subsequently agglutination test can be reduced to one step using PCR confirmation. The PCR detection has the potential for automatization and to reduce the time for procedure. In the future, multiplex PCR can be further expanded to other Legionella strains and serogroups. For example, Biotecon Diagnostics GmbH already offers a commercial kit for the determination of Legionella spp., L. pneumophila and L. pneumophila serogroup 1. Additionally, the infection potential of an isolate can be determined by multiplex PCR for pathogenesis markers (Huang et al. 2006).

Our comparison study shows that taking presumptive colonies directly from the culture plate and confirming them by multiplex PCR has a faster response time and is more specific than the standard confirmation method by subcultivation according to ISO 11731:2017. Thereby, it is also possible to directly confirm the presence and absence of L. pneumophila serogroup 1 already after 7–10 days while still reporting the result in CFU as most legislations still demand.

The method comparison of PCR‐identification vs the subculture identification described in ISO 11731:2017 was performed according to ISO 16140:2003. Molecular detection of Legionella by PCR met all criteria for the qualitative validation of an alternative method. According to our results, the confirmation step of ISO 11731:2017 can be safely substituted by PCR.

In the future, we expect further applications of PCR methods emerging in routine analysis of Legionella. It needs to be examined whether a PCR threshold value in genomic units per litre (GU l−1) can be defined for which the current legal limit for Legionella spp. (1000 CFU per litre) can be reliably met. Such thresholds were already proposed by several studies (Lee et al. 2011; Collins et al. 2017). An alternative could be the establishment of real‐time PCR as a cost‐effective and fast screening method for routine inspections, e.g., via a first PCR screening and subsequent cultivation only for positive samples or in case a specific threshold value of GU l−1 is exceeded. Similar procedures are already accepted or mentioned as possible in some countries such as Canada (Public Works and Government Services Canada 2016) and Italy (Ministero della Salute 2015).

Several PCR‐methods have been developed that favour DNA detection and quantification of viable (meaning intact) Legionella cells (Nocker et al. 2006; Chen and Chang 2010). These qPCRs are directly applied without precultivation. They are affected by different factors such as density of the cells and methods applied (Taylor et al. 2014; Braun et al. 2019). Another approach to differentiate between viable and non‐viable Legionella cells is to detect precursor RNA produced only by viable cells on contact with fresh nutrients during cultivation (Boss et al. 2018). Although real‐time PCR seems promising, additional studies are needed to develop a stable viability qPCR‐method for different environmental samples. In the future, the combined application of conventional culture methods and new culture‐independent molecular biological methods will be purposeful and helpful to detect and assess microbial contamination in drinking water installations.

Conflict of Interest

The authors declare no conflict of interest.

Authors contribution

Dana Eble and Vanessa Gehrig contributed data or analysis tools, performed the analysis and wrote the paper. Patricia Schubert‐Ullrich conceived and designed the analysis and wrote the paper. René Köppel conceived and designed the analysis, contributed data or analysis tools, performed the analysis and wrote the paper. Hans Peter Füchslin conceived and designed the analysis, collected the data, contributed data or analysis tools, performed the analysis and wrote the paper.