Introduction

Since its first identification as the causative agent of Legionnaires’ disease in 1976 [1], L. pneumophila became increasingly important, being considered a significant world health problem [2,3,4]. Legionnaires’ disease is caused by inhalation of aerosolized water contaminated with Legionella, which results in severe pneumonia in susceptible individuals [5]. Legionella bacteria proliferate in man-made environments like hot-water systems, spas, jacuzzis, cooling towers, air conditioning units, and decorative fountains [6,7,8,9].

The plate culture method, performed in accordance with the standard method, ISO 11731, isolates and quantifies Legionella species on buffered charcoal yeast extract (BCYE) agar medium supplemented with different combinations of selective agents [10, 11]. Pre-treatment steps (acid and heat) are also used to reduce other heterotrophic bacteria in the water sample that can interfere with the test by competing for nutrients or producing inhibitory secondary metabolites [12]. Because of the method complexity, the plate culture method has limitations for routine monitoring. The need for multiple treatments and plating conditions increases the time and resource requirements for analysis of each sample. Furthermore, the turnaround time for the conventional culture technique can be as low as 3 days for a qualitative positive result but it needs 7 days or more for the correct quantification or in the case of negative samples. In samples with high background flora, there are often false-negative or inconclusive (Too Numerous To Count-(TNTC)) results due to the overgrowth of the non-Legionella bacteria. Polymerase Chain Reaction (PCR) methods, based on the amplification of a specific gene from Legionella, are very fast and specific. However, there is the need to accurately compare them with plate culture methodologies to produce adequate warning and action levels for PCR, although a few reports have been published in the last few years giving guidance for these levels [13, 14]. In addition, the costs and need for 51 highly trained personnel are significant drawbacks of both techniques. Legiolert is a new culture method for the quantification of L. pneumophila and is based on quantification by MPN. The test differs significantly from traditional plate culture methods by providing results at 7 days, with rapid sample preparation and analysis, and a confirmed detection result. Like traditional culture methods, viable L. pneumophila from positive samples can be isolated if further investigations are required.

In this study, we report the outcome of a comparison between spread-plate culture using ISO 11731, an in-house viability qPCR (v-qPCR) method, and the Legiolert method for the detection and enumeration of L. pneumophila from naturally contaminated potable and non-potable water samples.

Materials and Methods

Sample Composition

A total of 177 water samples (165 potable and 12 non-potable waters) were analyzed between September 2017 and January 2018 for the presence and quantification of L. pneumophila by ISO 11731, v-qPCR, and Legiolert. Non-potable water samples originated from cooling towers, whereas potable water samples were mainly from showers and hot tap waters. Samples were selected from geographically diverse regions within Portugal.

Sample Collection

All samples were collected in sterile containers containing sodium thiosulfate. For hot-water samples, the sample comprises 2 L. The first liter of sample was collected in the first flush. After the first flush, the water temperature was monitored using a calibrated thermometer and when the maximum temperature was attained and stable, the second liter was collected. Cooling tower water samples were collected immersing the collection container in the cooling tower water bath.

Samples were transported to the laboratory within 8 h of collection. The transport was performed at (5 ± 3) °C, in accordance to ISO 19458:2006.

ISO 11731:2017

The ISO 11731:2017 procedure is based on the growth of presumptive Legionella colonies using selective glycine vancomycin polymyxin cycloheximide agar (GVPC, Oxoid). Briefly, 2 L of hot-water samples or 1 L of cooling tower water samples were filtered through a 0.2-μm pore size membrane (Millipore, USA) and the filter was placed in 5 mL of the original sample and resuspended by vortexing. After sample concentration, heat and acid treatments were performed in parallel with a direct inoculation of the concentrated sample (100 μL) onto BCYE and GVPC agar plates, in duplicate, according to the points 8–10 of the ISO 11731 matrix of decision and the plates were incubated at 36 °C for 10 days. Samples were observed every 2 days until the end of incubation period. Three colonies of each presumptive Legionella colonies were sub-cultured to buffered charcoal yeast extract (BCYE) agar plates with and without cysteine (Oxoid, UK). Isolates that presented growth on BCYE agar and absence of growth on BCYE agar without cysteine were considered Legionella. Latex 76 agglutination (Oxoid, UK) and qPCR were performed to confirm serotype and species, respectively. Results obtained by ISO 11731 were expressed in colony forming units (CFU)/L and calculated as the number of Legionella colony forming units in the original sample. This number is estimated by selecting the plate or set of plates showing the maximum number of confirmed colonies per water volume.

v-qPCR

Two milliliters of the concentrate obtained as described previously was centrifuged at 10000 g for 10 min, the supernatant was removed, and the pellet was suspended in 100 μL of sterile water. Samples were then treated with propidium monoazide (PMA) (Biotium Inc. USA) using a final concentration of 0.04 mM. Tubes were mixed gently by inverting several times and incubated on ice for 30 min before being transferred to the LED-based PhaST Blue System (IUL, Spain) and photo-activated at 100% light intensity for 15 min. DNA from viable cells was extracted at 95 °C for 10 min.

The qPCR assay for L. pneumophila was performed accordingly to ISO/TS 12869:2012, with primers Lp01: 5′-CCGATGCCACATCATTAGC-3´and LP02: 5′-CTATGAGTGGCGCTCAATTGG-3′ and probe LpP: 5′NED-CGGAAGCAATGGCTAAAGGCA-MGB 3′. The PCR mixture was composed by 125 µL of Maxima Probe/ROX qPCR Master Mix (Thermo Scientific, USA), 800 nmol of each primer, and 200 nmol of the probe. Each reaction was completed with DNA-free water to a final volume of 25 µL. qPCR conditions were 3 min at 95 °C, followed by 43 cycles of 20 s at 95 °C and 60 s at 60 °C. v-qPCR data were expressed in genome units (GU)/L. All the qPCR reactions were performed in a Applied Biosystems 7300 (Applied Biosystems, USA). In parallel to each batch of samples, positive (using L. pneumophila) and negative (using DNA/RNA-free water) controls were performed.

The qPCR assay for Legionella spp. was performed using the primers Lspp02: 5′-GCTTAACCTGGGACGGTCAGA-3′ and Lspp03: 5′-GCGCCACTAATTATTTTCATATAACCA-3′ and probe Lspp_Sonda: 5′FAM-CCTAATACTGACACTGAGGC‐BHQ1 3′. The qPCR mix and conditions were the same as described for the L. pneumophila protocol.

Legiolert™/Quanti-Tray®

Legiolert™ (IDEXX Laboratories, Westbrook, ME) was performed according to the manufacturer’s instructions.

For potable water samples, the hardness of each sample was first determined using Aquadur hardness test strips (Macherey–Nagel, Germany). Depending on the results obtained with this test, a volume of either 0.33 mL (for low hardness water) or 1 mL (for high hardness water) of hardness supplement was added to the 100 mL of the sample. A Legiolert blister pack was then added to the sample and the contents were shaken until completely dissolved. For non-potable samples, a 2 mL water sample was mixed with 2 mL of Legiolert pretreatment solution, mixed, and incubated for 1 min at room temperature. Following incubation, 2 mL of this solution was added to reagent pre-dissolved in 100 mL of sterile water. The solution was poured into a Quanti-tray/Legiolert tray that was immediately sealed and incubated at (39.0 ± 0.5) °C and (37.0 ± 0.5) °C, each with humidity, for potable and non-potable waters, respectively. After incubation for 7 days, the number of positive wells (corresponding to wells that are turbid and/or with a brown color) was counted and the most probable number was determined. Results with Legiolert were expressed in MPN/L. To confirm the specificity of the test, 10 percent of the total number of positive wells were sub-cultured on GVPC agar and cultivated as described previously for the speciation on ISO 11731.

Data Analysis

All data analysis was conducted with either Microsoft Excel 2016 (Microsoft Corporation, WA, US) or IBM SPSS Statistics 25 (IBM, NY, US). All data were transformed into a logarithmic scale by adding one to each value with subsequent log10 conversion. To evaluate the differences between the incidence of positive and negative samples by the methods tested, data were analyzed by the McNemar’s exact binomial test [15]. The differences in sensitivity of each method were also determined with the non-parametric Wilcoxon signed-rank test. To perform this test, data were excluded if the results for each paired comparison were zero. The test was chosen because the distribution of differences was found to be non-normal within the pairs. The normality of the data was conducted using the Shapiro–Wilk test. For the distinct method comparison, a ρ < 0.0001 was obtained, indicative of a non-normal distribution. To perform the Wilcoxon signed-rank test, the pair results of 126 were analyzed using a two-tailed test with a significant level of 0.05.

Results

L. pneumophila was found in 22% (40/177), 17% (30/177), and 6% (11/177) of the samples analyzed by Legiolert, v-qPCR, and ISO 11731, respectively. The mean concentration of L. pneumophila as detected by Legiolert was 2.70 log MPN/L (varying from 0 to 4.36 log MPN/L) similar to the mean concentration determined by v-qPCR (2.75 log GU/L, ranging between 0 and 5.38 log GU/L). The lowest result was obtained when samples were tested by ISO 11731, at a mean concentration of 2.20 log CFU/L (range from 0 to 2.81 log CFU/L). The median values for Legiolert, v-qPCR, and ISO 11731 were, respectively, 2.82 log MPN/L, 2.68 log GU/L, and 2.18 log CFU/L. The limit of detection for the Legiolert method was 1.00 log MPN/L for potable waters and 3.00 log MPN/L for non-potable waters.

The majority of samples were negative for all the methods tested (Fig. 1 and Table 1), with the largest number of samples ranging from 0 to 1.00 log CFU/L as measured by ISO 11731, corresponding to 165 samples (93%). Of these, 28 were positive using Legiolert (17%) and 23 positive by v-qPCR (14%), with concentrations up to 4.36 log MPN/L and up to 4.03 log GU/L for Legiolert and v-qPCR, respectively.

Fig. 1
figure 1

Bivariate plots showing pairwise comparison of methods. a ISO 11731 method with Legiolert; b Legiolert and v-qPCR; c ISO 11731 and v-qPCR

Full size image
Table 1 McNemar’s test analysis of the results for detection of L. pneumophila by 1L samples for Legiolert versus ISO 11731 and versus v-qPCR.
Full size table

There was good agreement between the different techniques for concentrations above 1.00 log CFU/L as measured by ISO 11731, with total agreement for ISO 11731 and v-qPCR for all concentration ranges (Fig. 1) and an accordance of 84% for the pair ISO 11731 and Legiolert. The presence/absence results from each method compared to Legiolert were analyzed using the McNemar’s test (Table 1), in the case of Legiolert compared to the ISO 11731, the number of positive samples for L. pneumophila is significantly higher than those obtained by the ISO 11731 (Table 1). For the other case, Legiolert vs v-qPCR, there was no significant difference in the detection of L. pneumophila by both methods. It is relevant that seven samples that were negative by ISO 11731 had Legiolert values above the action limit of 3.00 log CFU/L set by the European Centre for Disease Control and Preventions [16]. Similarly, v-qPCR vs ISO 11731, 3 v-qPCR positive/ISO 11731 negative samples were higher than the action level provided by the European Centre for Disease Control and Preventions [4] for culture.

Comparison of quantitative results of ISO 11731 versus Legiolert showed a statistically significant difference, with Legiolert being more sensitive according to the Wilcoxon signed-rank test (n = 12, ρ = 0.034) (Fig. 1). Similar results were determined when comparing Legiolert with v-qPCR, with the latter being more sensitive (n = 39, ρ = 0.000027) and which was equal to the comparison for ISO 11731:2017 and v-qPCR (n = 12, ρ = 0.010).

The Legiolert manufacturer indicates this method to be specific for the detection of L. pneumophila. To evaluate this claim, culture was extracted from 10% of the positive wells observed from each Legiolert test and each was confirmed for the presence of L. pneumophila by subculturing to BCYE and BCYE lacking cysteine, as per ISO 11731. The majority of the tested wells were found to be positive for L. pneumophila. Other non-Legionella microorganisms were also present in the wells, so that isolation of colonies was required. All wells were also tested for the presence of both Legionella spp. and L. pneumophila by qPCR, to cover the possibility of false results due to overgrowth of other microorganisms in the plates. Fifteen potable water samples, positive for Legiolert and tested for specific detection of L. pneumophila by qPCR were in fact only determined as being Legionella spp. rather than L. pneumophila. These results are indicative of wells providing false-positive results rather than a failure to recover L. pneumophila from the wells, although all the false-positive results were from Legionella spp. and not from any other bacterial isolate.

Discussion

The data generated in this study indicate that Legiolert detected L. pneumophila in a greater number of samples and at higher concentrations than ISO 11731. The presence/absence data are also congruent with the quantification results displaying a significant difference for the pair ISO 11731/Legiolert but no difference for the pair v-qPCR/Legiolert (as determined by the McNemar’s test). These ISO 17731/Legiolert findings agree with the previously published studies [17,18,19]).

Data for v-qPCR were also higher than those generated with ISO 11731. Moreover, the results have shown that a good agreement of positive samples was obtained with increasing concentrations of L. pneumophila as determined by ISO 11731. The observed difference in detection between v-qPCR and ISO 11731 may result from different issues already pointed out to the ISO 11731, including (i) the use of GVPC agar, which is more selective for L. pneumophila than for other species of Legionella with a possible inhibitory effect to some non-pneumophila species of Legionella [20], (ii) the fact that most of the samples tested were subjected to previous disinfection treatments (thermal and/or chlorine) which may render Legionella spp. and L. pneumophila viable but non-culturable in the GVPC 200 media, being detected by the system from Legiolert and by v-qPCR, and (iii) also to the overgrowth of other microorganisms that may mask or inhibit the presence of Legionella species.

Legiolert is described as providing confirmed results for L. pneumophila following seven days of incubation, and the results from this study corroborate that for the most part. The overall rate of false-positive wells was low. Barrette, 2019 [19] also detected a small percentage of false-positive results when studying water originating from cooling towers. V-qPCR results are available within one working day which presents an advantage for public-health and decision makers. However, the results obtained were not comparable to the culture method until now [21].

v-qPCR and Legiolert’s greater selectivity compared with ISO 11731 allows for the specific detection of L. pneumophila in every sample without any additional steps required by the ISO 11731 to enhance its specificity [11]. This increased specificity is even more relevant for water samples from cooling towers that present high levels of interfering microorganisms. With ISO 11731, heat and/or acid treatment, in addition to dilution of the sample, may decrease the number of interfering flora, but may still be insufficient to reduce it to acceptable levels to allow for accurate Legionella quantification, and may significantly increase the lower limit of detection [10, 22].

The present study demonstrated that both Legiolert and v-qPCR present similar results, with some advantages over the conventional culture method used to determine the levels of alert and action set by ECDC in Europe. These advantages include greater sensitivity with reduced interference from other species, an unequivocal positive signal that is easy to read and quantify, requiring less technical expertise and shorter total evaluation time than the ISO method so the final result can be achieved using fewer resources.