Applicability of solid-phase cytometry and epifluorescence microscopy for rapid assessment of the microbiological quality of dialysis water

Abstract

Currently, the gold standard to assess the microbiological quality of dialysis water is the determination of heterotrophic plate counts (HPC). The long waiting time of the HPC method and the fact that most bacteria are not culturable on agar plates provokes the search for rapid alternative methods for monitoring the microbiological quality of dialysis water. We tested the applicability of total viable counts (TVC) and total direct counts (TDC), determined via solid-phase cytometry and epifluorescence microscopy (EFM), in comparison to the standard HPC determination method in 113 samples from 13 dialysis water treatment units (59 drinking water and 54 dialysis water samples). Additionally, for a set of dialysis water samples (n = 22) endotoxin concentrations were also determined. TVC showed high correlation with HPC and results were of comparable magnitude for most investigated dialysis water samples [median: 3 cells/colony forming units (CFU) 100 mL(-1)]. However, in one dialysis water sample, HPC values (5800 CFU 100 mL(-1)) were >35-fold lower than TVC values (2.05 × 10(5) cells 100 mL(-1)) indicating severe limits of the HPC method to assess the microbiological quality of dialysis water. For drinking water, TVC (median: 4.8 × 10(4) cells 100 mL(-1)) was on average one order of magnitude higher than HPC (median: 2.5 × 10(3) cells 100 mL(-1)). TDCs (median dialysis water: 1.1 × 10(4) cells 100 mL(-1) and median drinking water: 4.9 × 10(6) cells 100 mL(-1)) were always several orders of magnitude higher than HPC or TVC. We propose that the TVC/solid-phase cytometry approach is a reliable and rapid alternative to the culture-dependent approach for assessment of the microbiological quality of dialysis water, especially when fast results are needed. TDC determined via EFM lacks sensitivity and reliability for assessing microbial concentrations in low-cell dialysis water samples since the limits of detection and quantification are high.