Effectiveness of ATP bioluminescence assay for presumptive identification of microorganisms in hospital water sources

The presence of organisms in drinking water is assessed in laboratory analysis by qualitative and quantitative criteria [1]. Initial investigations have included consistent but arduous methods, such as the multiple tube method, involving serial analysis of water samples. Subsequent studies have included alternative, less laborious methods of quantitative and qualitative analysis that use chromogenic and fluorogenic substrates and maintain high specificity and sensitivity along with speed [2, 3]. Among these methods, the membrane filter method stands out. It is considered a feasible laboratory procedure that allows for bacterial colony counts [1].

Another methodology, similar to the ATP bioluminescence system, has been proposed as a new method of measuring cellular material in water by measuring levels of adenosine triphosphate (ATP) [4].

One of the major advantages of ATP detection is that it is fast and easy to carry out [5‐7]. However, there are also some disadvantages, such as low sensitivity and inability to differentiate extracellular and intracellular ATP [7‐10]. Dead cells, blood, excretions, human secretions, food and other organic materials are considered extracellular ATP, whereas intracellular ATP is detected by means of viable microorganisms [7, 10, 11].

Studies comparing the traditional methodology (colony-forming units, − CFU) with ATP detection have shown divergent results, indicating the need for more research to establish better understanding of the relationship between microorganism counts and ATP. Lee et al. [12] and Bushon et al. [13] found good correlations between ATP and CFU, but other studies [14, 15] found no significant association.

Considering the high incidence of waterborne diseases in the general population, it is necessary to establish faster and more specific protocols for microbiological investigation of water, in order to adopt control measures. The objective of the present study was to verify whether ATP bioluminescence can be used as an alternative methodology for presumptive testing for presence of microorganisms in water, by confirming that there is a correlation between RLU and the presence of these microorganisms.

Table 1 shows the distribution of the results obtained, according to the origin of the samples. Eighty-three (94.3%) samples presented bacterial growth: 69 (83.1%) from faucets and 14 (16.9%) from purifiers. Of these, 13 (15.6%) presented heterotrophic organism levels above those recommended by the Brazilian Ministry of Health (≤500 CFU/mL). There was also fungal contamination in 49 (55.6%) water samples.

In relation to the data for RLU, 36 samples presented exclusive bacterial growth. This can be explained by the fact that no fungal growth was observed in the Sabouraud agar medium plates, which favor fungal growth. If neither yeast nor mold grew in the plates with a medium that is selective for fungi, it can be concluded that there was only bacterial growth in the plate count agar medium, with a mean RLU value of 14.4. Only two samples presented exclusive fungal growth, with a mean RLU of 23. Water samples that presented simultaneous growth of two types of microorganisms (N = 47) had a mean RLU value of 16.5. For the samples that showed no microorganism growth (N = 3), the mean RLU value was 8.6. The minimum and maximum RLU values found in the samples were 4 (3rd and 7th floor) and 141 (2nd floor). An ROC curve was used to establish a cutoff point of RLU values to detect contaminated water. However, the sensitivity and specificity of the samples were not significant, as most had results lower than 90% (Fig. 1).

Pearson’s correlation coefficient between RLU measurement and microbial concentration (absorbance) in distilled water revealed results with significant differences, p < 0.0001 (Figs. 2 and 3).

Minimum and maximum limits of detection are 10⁻¹⁰ and 10. For S. aureus, the corresponding RLU value for the 10⁻¹⁰ dilution was 18; for the concentration of 10, it was 349,333. For C. parapsilosis, the corresponding RLU values were 37 and 780,198 for the minimum and maximum limits of detection, respectively.

Lack of significant correlations between RLU and the presence of bacteria and fungi (Fig. 1) could be due to several factors, such as variations in microbial cell size and cell development stages, as has been described by other authors [7]. A sample with few yeast cells can result in high levels of RLU (ATP), and one with many yeast cells can present low levels of RLU. The effect of environmental stress on RLU emission has been investigated in other studies [19, 20]. On study examined the effects of hydric stress and nutritional conditions on Pseudomonas paucimobilis present in soil. The findings demonstrated that ATP content increased in dry soil or with the introduction of glucose and ammonium salts [19].

Furthermore, in the present study, RLU was detected even in samples in which fungi and bacteria were not isolated. This may be explained by the presence of other microorganisms, such as protozoa or anaerobic bacteria. In addition, the chlorine present in water (Table 1) could inhibit some microorganisms for culture, but ATP was present. Similarly, some bacteria that would grow at lower or higher temperatures did not grow in the culture media, but ATP was found.

This lack of correlation could also have occurred because the system does not make a distinction between intracellular and extracellular ATP. Both intracellular and extracellular ATP were measured, which does not correspond to the microorganisms present in the samples. Biological systems have other nonbacterial sources of ATP and, consequently, microbial ATP represents only a portion of the ATP responsible for bioluminescence. Interference from ATP of nonmicrobial origin may limit the sensitivity of the method and lead to false positives [21‐23]. However, whether the samples included intracellular or extracellular ATP, the results of the present study indicate that, in addition to including viable microorganisms, the samples also contained secretions and excretions, making the water unfit for consumption and implying health problems.

In relation to the average RLU in the samples with exclusive fungal growth, there was higher-than-average RLU in samples with exclusive bacterial growth, or growth of both bacteria and fungi. A possible explanation is that in the samples with exclusive fungal growth, free chlorine residual content was within the established limits (Table 1), and some bacteria that were present could not be grown, but their ATP was detected, indicating high levels of RLU.

Perhaps for these reasons, discrepancies between the results of the ATP bioluminescence and the presence of microorganism in the samples indicate that this correlation is still questionable and needs more research in order to gain consensus.

When compared to samples of distilled water (control test) contaminated with ATCC from S. aureus and C. parapsilosis, and considering the correspondence of these samples’ absorbance with RLU, the results were significant; when the inoculums increased, so did RLU levels (Figs. 2 and 3). In this situation, distilled water does not have interferents or microorganisms other than those that have been introduced, which could explain the significant results obtained.

The ATP bioluminescence system used in this study showed low correlations with microbiological cultures. Therefore, this technology cannot be used as an alternative methodology for presumptive testing for the presence of microorganisms in water.