Effect of peracetic acid–ultraviolet combination treatment on microbial and endotoxin levels in a pharmaceutical water system

Water is one of the industry's fundamental necessities and has a variety of uses in the pharmaceutical industry, such as product components, excipients, or solvents in the production process, or as a cleaning agent for hygienic reasons (Strade et al. 2020). In the pharmaceutical industry, two highly regulated types of water are purified water (PW) and water for injection (WFI) (Collentro 2016; Selim et al. 2020). According to the United States Pharmacopeia (USP) definition (Pharmacopeia 2014), ‘Purified Water must meet the requirements for ionic and organic chemical purity and must be protected from microbial contamination.’ WFI ‘is used as an excipient in producing parenteral and other preparations where product endotoxin content must be controlled and in other pharmaceutical applications, such as cleaning certain equipment and parenteral product contact components’ (Pharmacopeia 2014). Since water is a suitable medium for microbial growth, the presence of microbes and endotoxins in water systems is unavoidable (Snyder et al. 2003). So, the most challenging quality attribute in PW and WFI is the microbial content of water (Ashour et al. 2011; Shintani 2016). PW should meet the TOC, conductivity, and bioburden test standards, and WFI must also meet the endotoxin tests and bacteriological purity requirements (Williams 2004). Endotoxins are lipopolysaccharides (LPSs) produced by most Gram-negative bacteria and certain cyanobacteria. They are the most prominent pyrogens in the pharmaceutical industry (Sandle 2012). Endotoxins are stoutly embedded in the cell wall membrane. However, they are continuously released into the surrounding environment during normal cell growth and division, as well as cell disintegration (Mehmood 2019; Azargun et al. 2020). Two major components of LPS are hydrophilic (polysaccharide group) and hydrophobic (a lipid group called lipid A), which induce most of the inflammatory signaling and promote inflammatory mediator release even at very low dosage (Zhang et al. 2019; Eyvazi et al. 2020) and trigger a serious and wide range of health risks such as vasoconstriction, vasodilation, cardiovascular disorder, multiple organ damage or failure, and death (Trent et al. 2006). Therefore, to prevent the adverse effects of endotoxin on human health, its detection is needed by the U.S. Food and Drug Administration (FDA 1987; Eyvazi et al. 2020) and the USP (Pharmacopeia 2014). Due to the acid/alkali and heat resistance nature of endotoxins, they are not destroyed by alteration in temperature, pH, and even sterilization or distillation, and extreme acid/alkali and high temperature are needed to destroy them (Mehmood 2019). In pharmaceutical industries, various technologies are used to treat water, including sand filtration, softening (to lower hardness), activated carbon (to remove chlorates, chlorine, unpleasant odors and tastes, and organics), ultraviolet (UV) irradiation (to minimize microorganism contamination), deionization, reverse osmosis, and final filtration and distillation (Keyashian 2014; Farrag & Abdelbasier 2019). Periodic disinfection of the whole water system is crucial to keep it clean and avoid undesirable contamination or biofilms. The effects of some traditional water disinfection treatments on microbes and endotoxins have been studied (Anderson et al. 2003; Oh et al. 2014; Ren et al. 2019). Peracetic acid (PAA) (CH₃CO₃H) is a clear colorless liquid organic acid with strong oxidizing action and a broad antimicrobial spectrum. PAA acts through the production of reactive free radicals, mainly hydroxyl radical and reactive peroxidase which oxidize bacteria and biofilms (Farjami et al. 2022). It is a high-level disinfectant with rapid biocidal activity against viruses, fungi, spores, and bacteria including mycobacteria, and it is increasingly used as a chemosterilant of medical equipment (Denyer et al. 2008). Wide-spectrum microbial activity, absence of disinfection byproducts, low residuals, low dependency on pH, and short contact time are remarkable advantages of PAA for disinfection of water systems (Hugo & Russell 1998). UV irradiation, when combined with oxidizing agents like PAA (PAA–UV combination), produces a highly reactive hydroxyl radical, which drives the synergistic disinfecting action (Zhang et al. 2020). A well-designed microbiological monitoring program, comprising complete sampling and enumeration of total viable bacteria for each system component, should be specified to ensure treated water quality. The program should be carried out by qualified personnel and carefully documented (Clontz 2008; Tarhriz et al. 2014; Shintani 2016; Selim et al. 2020). Microbial limits for pharmaceutical-grade waters are addressed by the USP (Pharmacopeia 2014) and World Health Organization guidelines (2005). However, establishing in-house action or alert limits is extensively needed for effective microbial control (Cholayudth 2006). The alert limits must be established to distinguish single abnormal high counts from acceptable normal counts and should be low enough to warn of notable alteration from normal conditions (Cholayudth 2006; Roesti 2019). A common way to determine alert level using historical data is to use an X/Y graph to show microbial counts with an X-axis representative of timescale and a Y-axis representative of microbial counts and the level is calculated using statistical methods (Hussong & Madsen 2004; Roesti 2019). This work presents a monitoring program for microbial disinfection and sanitation of the PW and WFI system in a pharmaceutical plant, Tabriz, Iran. The objective was to investigate the effect of PAA–UV combination treatment following brushing and steaming throughout the water system on microbial count and endotoxin levels in PW and WFI.

The materials used in the pharmaceutical water purification system have the following characteristics: multi-media water filters containing three layers of media for multi-media filtration, including anthracite, sand, and garnet, and remove particles down to 10 μm. Activated carbon with the size of 12 × 40 mesh and capacity of 0.1 pound of organics per 1 pound of carbon at a flow rate of 1 gallon per minute per cubic foot (gpm/cu.ft.) and a bed depth of 3 ft. Sand filters with a size of 0.4 mm and a uniformity coefficient of 1.5. Polymeric cation and anion exchange resins with an average diameter of 0.5 mm and UV irradiation of 150 mJ/cm².

Microbial enumeration was done using the pour plate technique, with R2A (Reasoner's 2A agar) and TSA (trypticase soy agar) as culture media. 10 ml of sample were diluted with sterile PBS (phosphate-buffered saline) to bring the volume to 100 ml. Then, 1 ml of diluted sample was pipetted directly onto Petri dishes, 15 ml of melted TSA at 45 °C were added, and incubated at 30–35 °C for a minimum of 3 days. The same procedure was applied with the R2A medium, and samples were incubated at 20–25 °C for a minimum of 5 days. Before testing the water samples, 1 ml of sterile PBS was added to the plate as a negative control, and the same protocols were followed. Microbial counts were done with a colony counter (Biotest, Frankfurt, Germany) and compared with control limits determined for water system monitoring. Membrane filtration was used for microbial enumeration of samples from the WFI storage tank and point of use (Figure 2). 10 ml of aliquots from diluted samples were vacuum-filtered onto a 0.2 μm pore membrane. Membrane filters were plated onto TSA and R2A agar and incubated at 37 °C for 72 ± 4 h and 20–25 °C for 3–5 days, respectively.

Individual control charts (I-chart) were plotted using microbial counts versus sample collection days, with upper control limits (UCLs). The UCL was calculated as the mean microbial count plus three times the standard deviation.

Endotoxin evaluations were done in distilled water daily and reported weekly over 6 months, equal to 24 weeks. (The first 12 weeks were treated with UV radiation alone, whereas the 15–26 months were treated with a PAA–UV combination. At the 13th and 14th weeks, no test was done since the pharmaceutical facility was undergoing cleaning and sanitation procedures.) Each week, 15 endotoxin tests were conducted (n_week = 15). The LAL gel-clot method was used to detect endotoxins in water samples. The sampling procedure was performed aseptically to prevent further microbial contamination, and samples were collected in sterile, pyrogen-free Falcons and stored at 4 °C until assayed. Analysis was done using the LAL gel-clot method with a commercial kit (Pyrogent plus 200, BioWhittaker Company, USA) containing 10 ng of standard endotoxin Escherichia coli 055:B5 as control and LA lysate in lyophilized form with the sensitivity of LAL 0.125 EU/ml. Both reagents were reconstituted by LAL reagent water (containing endotoxin < 0.005 IU/ml). For the work, 0.1 ml of the reconstituted lysate solution was added to 0.1 ml of the test water sample in sterile, pyrogen-free tubes, and the contents of the tubes were thoroughly agitated using a vortex mixer. The tubes were covered by aluminum foil and incubated in a water bath for 60 min at 37 °C. The tubes were then removed from the water bath and inverted. Clot formation in the tube is representative of LAL >0.125 IU/ml and endotoxins presence in the sample and a positive result, and vice versa (Mehmood 2019). In Figure 5, the reported endotoxin value for each week is the sum of the positive endotoxin test measured over a week.

The findings from three independent experiments are reported as mean ± standard deviation (SD). Whenever possible and relevant, the data were analyzed for statistical significance using the unpaired Student's t-test. P-value < 0.05 was used as the statistical significance threshold. GraphPad Prism Software ver. 6.0 was utilized to analyze the data (GraphPad Software Inc.).

Table 2 shows the in-house target, alert, and action limits for each purification step specified in the pharmaceutical qualification documents of the water system. According to obtained results in Table 1, no sample in this study exceeded the alert and action limit set for the water system in both treatment periods (UV treatment alone and PAA–UV combination treatment). Also, the microbial counts of no water sample exceeded the European Pharmacopeia's action limit (1,000 CFU/ml for PW and 10 CFU/100 ml for WFI) (Sandle 2017).

Figure 5 displays the weekly number of positive endotoxin tests conducted over a period of 24 weeks (tests for endotoxins were not conducted in weeks 13 and 14). The frequency of endotoxin-positive samples after UV treatment was markedly higher than those after PAA–UV combination treatment. Water samples were mainly LAL-positive (11 positive weekly reports out of 12 total reports) when treated with UV alone but rarely LAL-positive after the PAA–UV combination treatment (2 positive weekly reports out of 12 total reports).

This research investigated how combining PAA (0.1% V/V) and UV light (>150 mJ/cm²) affects microbial and endotoxin levels in a pharmaceutical water system. Our finding demonstrated that the microbial counts in R2A media were higher than TSA's in most test samples. The type of selected culture media significantly affects the level of measured bacteria (Sandle 2017). R2A is the suggested medium for enumerating heterotrophic plates in water samples. It promotes the survival and development of under-stressed or sub-lethally injured microorganisms by giving a lower incubation temperature (20–25 °C) and low amounts of nutrients such as yeast extracts, dextrose, peptone, and a longer incubation duration than traditional media such as TSA (Ashour et al. 2011; Collentro 2016). In this work, the high microbial count in the R2A medium suggests that the water's bacterial flora has adapted to a carbon- and nutrient-poor environment. Similar to what we found, Sandle et al. (2015) indicated microbial counts of water samples in R2A media were higher than in TSA media. In addition to the type of medium, a decrease in temperature and an extension of the incubation period contributed to an increase in the number of colonies. It seems that R2A medium numbers are more indicative of endogenous water flora originating from raw water and secondary contamination in an industrial environment (Pisitkun et al. 2004). TSA count is also important here because it can be considered a sign of process-derived contaminations (Maltais et al. 2017; Alemu 2020).

As seen in Figure 3, the microbial counts of samples from the well and break tank (pretreatment stages) were less than 1 CFU/ml. Their relatively low microbial count indicated that the entering raw water was of reasonably high quality. It is related to the location of the researched plant and the depth of the underground water reservoir. Samples collected from sampling point 2 (activated carbon) showed a higher proportion of microbial count than samples collected from pretreatment stages (well and break tank) and sand filtration (Table 1). Activated carbon beads are a potent tool for removing color, odor, impurities, and chlorine-related compounds in water. They are also highly susceptible to the build of microbial growth due to the adsorption of organic material. Therefore, bioburden growth is more likely to happen due to the adsorption of contaminants onto the beads (Çeçen & Aktas 2011). Like our findings, Ashour et al. (2011) reported that carbon beds are ideal environments for bacterial growth and show high bacterial contamination in a water system.

UV irradiation is an effective disinfectant due to its strong germicidal ability. Samples instantly collected after UV irradiation (sampling point 3) showed average microbial counts of less than 1 CFU/ml. However, the average count of 16 CFU/ml was observed in sampling point 4 (cation resin 1). Therefore, it seems regrowths of suppressed bacteria have happened. Consistent with our result, Zhang et al. (2015) reported that UV light induces a viable but non-culturable E. coli. They indicated that the antimicrobial action of UV light is reversible, and regrowth of microorganisms can occur. Therefore, in this study, microbial count increase in cation resin 1 after UV irradiation can be interpreted in part with the reversible antimicrobial mechanism of UV light (Choi & Choi 2010; Zhang et al. 2015), which led to the regrowth of inhibited bacteria.

High microbial counts are observed in samples collected from all deionization steps, including sampling points 4–7. Cation/anion resins are microbial friendly and highly susceptible to building microbial growth and biofilm formation since microbes can use them as a surface for attachment and colonization (Keyashian 2014; Shintani 2016). Therefore, extensive care, especially disinfection procedures, is needed to prevent more colonization and growth of viable bioburden in water purification systems (Reddy et al. 2014).

PAA was utilized for its synergistic antibacterial effect with UV light. The lowered microbial counts of water samples treated with a combination of PAA and UV, as compared to those treated with UV alone (Table 1), demonstrated that PAA had an extra effect. For instance, in the case of the sampling point of 4 (cation resin 1), the mean count of 16 CFU/ ml decreased to 8 CFU/ ml after PAA–UV combination treatment (P-value = 0.004). Also, in the case of the sampling point of 8 (mixed bed), this treatment reduced the mean count from 16 to 7 CFU/ml (P-value = 0.004). To sum up, statistical analysis (unpaired t-test) of our data (P-values < 0.05 in Table 1) revealed that disinfection of the water system by a combination of PAA–UV significantly reduced the mean microbial count in all treatment steps, including sand filtration, activated carbon, both cation and anion resins, mixed bed, 20 μm filter, and deionization tank. Our data demonstrated that a combination of PAA–UV has more disinfectant power than UV alone in the water system. In line with our data, Sun and his colleagues (2018) found PAA–UV inactivated E. coli more than PAA or UV alone owing to the production of hydroxyl radicals by UV-activated PAA. Weng et al. (2018) also indicated that PAA–UV combined treatment, especially at a high PAA dose, effectively reduced MS2 bacteriophage and murine norovirus in wastewater more than PAA or UV alone.

Combination of PAA as an oxidizing agent with UV irradiation results in the vigorous formation of hydroxyl radical with organic, inorganic, and biological matters and works as disinfectant. Indeed, the O–O bond in the molecule of PAA is interrupted in the presence of UV and leads to the generation of highly reactive hydroxyl radicals throughout the UV disinfection. The production of hydroxyl radicals is considered the main factor that causes the synergistic effect of PAA–UV combination treatment. Consistent with our findings, Zhang et al. (2020) found that the combination of PAA–UV efficiently inactivates E. coli. They showed radical generation (mostly •OH and CH₃C (O) OO•) and diffusion of PAA into bacterial cells had a significant role in the disinfection procedure (Zhang et al. 2020). Similarly, Sun et al. (2018) investigated disinfection mechanisms under PAA–UV intervention. Their results showed that in addition to the generation of hydroxyl radicals (•OH), their formation location is also a crucial factor. They found that the alkyl moiety of PAA provides hydrophobic properties, enabling it to adsorb on and/or diffuse into E. coli, and causes the production of hydroxyl radicals inside E. coli cells. Therefore, hydroxyl radicals generated under PAA–UV combination treatment have a strong disinfectant capacity.

Figure 4 shows several days (<3 days) among 90 days for each treatment (UV alone or PAA–UV combination), the microbial counts of samples taken from activated carbon, cation resin 2, and deionized water tank were above the calculated UCLs. One of the major reasons for spontaneous high counts can probably be related to a detachment of biofilms in the water system (Hugo & Russell 1998; Pharmacopeia 2014). The slope of I-charts was non-zero in both treatments, meaning microbial counts increased in time and exhibited a positive trend in microbial counts of each treatment period (Roesti 2019). However, sample locations treated with a combination of PAA–UV had a smaller slope than those treated with UV alone. For instance, the slope of the I-chart of cation resin 2 when treated with UV alone is 0.0815, but it decreased to 0.0041 when treated with the combination of PAA–UV (Figure 4(b)). It can be inferred that the disinfection power of the PAA–UV combination is higher than UV to reduce the mean microbiological count.

According to the USP requirement (Pharmacopeia 2014), WFI is needed to contain ≤0.25 unit/ml of US Endotoxin. WFI and PW specifications are the same except for the endotoxin's purity, which is obligatory for the WFI system. Figure 5 shows the results of the LAL test by the gel-clot method. Positive endotoxin tests were observed in WFI samples after both treatments. However, their microbial counts are zero, as shown in Table 1. The main reason is the Limulus assay can identify both viable and nonviable bacteria. UV irradiation kills most of the existing bacterial population but does not necessarily remove or detoxify bacterial endotoxins, which are released during the disintegration and death of bacterial cells in addition to the normal cell growth and division step (Jorgensen et al. 1979). Therefore, the endotoxin level of treated WFI samples likely reflects that the dead Gram-negative bacteria still possess endotoxin activity measurable by the Limulus assay. In addition, it can be related to the detachment of biofilms. Sandle (2017) showed that the most commonly occurring species were Gram-negative bacteria in a pharmaceutical water system. Also, our previous study showed that the isolated bacteria from a pharmaceutical water system were E. coli, Gram-negative bacteria, which generated and released endotoxins (Farjami et al. 2022). Therefore, positive endotoxin results in WFI samples can be related partly to the detachment of biofilms containing Gram-negative bacteria and endotoxin release in the water system.

As shown in Figure 5, WFI samples were mostly LAL-positive for most consecutive weeks over 12 weeks when treated with UV (150 mJ/cm²) (11 positive weekly reports out of 12 total reports). In line with our results, Ren et al. (2019) indicated a low-intensity UV lamp (150 mJ/cm² and less) cannot remove the LPS activity. In addition, samples collected over 12 weeks after disinfection with PAA–UV were mostly LAL-negative (10 negative weekly reports out of 12 total reports). Therefore, adding PAA to the disinfection process resulted in a remarkable decrease in both endotoxin levels and microbial counts. Gorke & Kittel (2002) used PAA > 0.1% solution for bimonthly disinfection of the dialysis fluid system. Consistent with our results, they found that samples collected after PAA treatment were mostly LAL-negative (27 negative samples out of 28 samples).

The other disinfectants' effect was investigated on the endotoxin level of water systems. Huang et al. (2011) showed that chlorination had no significant impact on decreasing the endotoxin activity assessed by the LAL assay. Ren et al. (2019) reported that ozonation could not inhibit LPS release of Gram-negative bacteria and reduce endotoxin activity. Thus, according to our findings, the combination of PAA–UV disinfection policy reduced both microbial and endotoxin levels in water systems.

The water system's microbial and endotoxin levels were assessed for 6 months in the pharmaceutical factory in Tabriz, Iran, following two treatment policies (UV alone and PAA–UV combination). We found that microbial counts of samples significantly decreased (P < 0.05) after PAA–UV combination treatment compared with UV treatment alone. The microbial counts of no water sample exceeded the action limits specified by the European Pharmacopeia (100 CFU/ml for PW and 10 CFU/100 ml for WFI). Also, WFI samples were mostly endotoxin-negative after PAA–UV combination treatment, whereas they were mostly endotoxin-positive after UV treatment alone. Overall, PAA–UV combination can be used as a potent disinfection policy to improve the water system's microbial quality. Therefore, rotation between more than one disinfectant policy and periodic disinfection of the water system with the assessed approach is recommended to inhibit pharmaceutical water and product contamination as well as bacterial infection.

This article does not contain any studies with human participants or animals performed by any of the authors.

The financial support was received from Tabriz University of Medical Sciences.

F.L. conceived the ideas and supervised all works. S.E. recorded the data. A.F., H.H., and M.S.-S. analyzed and interpreted the obtained results. A.F. wrote the manuscript and the manuscript has been reviewed and edited by all authors.

We acknowledge the Shahid Ghazi Pharmaceutical Company for the close cooperation in taking samples and testing performance.

All relevant data are included in the paper or its Supplementary Information.

The authors declare there is no conflict.