Standard methods are needed to reliably and efficiently assess bacterial contamination of processed medical devices. This article demonstrates a standard operating procedure (SOP) for fluorescence microscopy–based detection of residual bacteria on medical devices (BAC-VIS). BAC-VIS uses a 4',6-diamidino-2-phenylindole (DAPI) stain with fluorescent microscopy to quickly and cost-effectively detect bacterial contamination of processed medical device parts. The BAC-VIS protocol was optimized and achieved greater than 80% staining efficiency and a signal-to-noise ratio of more than 20 using four representative organisms. The SOP was first validated for use on a buildup biofilm model, accessory channels of contaminated clinically used devices, and inoculated endoscope end caps and O-rings. The buildup biofilm model was used to evaluate BAC-VIS after repeated treatment of adherent bacteria with three common high-level disinfectants: glutaraldehyde, ortho-phthalaldehyde, and peracetic acid. Next, BAC-VIS was used to assess clinically used endoscope parts that cultured positive for Gram-negative bacteria. DAPI-stained cells were found on all culture-positive devices, especially in grooves and imperfections on the surface. Finally, BAC-VIS was used to detect bacteria on inoculated endoscope device components. The results showed potential for BAC-VIS to be a valuable tool for industry and academic/medical researchers for investigations of contaminated medical devices. Results obtained using BAC-VIS can increase understanding of the role of design in cleanability, wear, and prevention of contamination and may lead to improvements in materials and design that could make processed endoscope use safer for patients. Of note, this protocol is not for detecting bacteria on scopes or scope parts that will be put back into clinical use.
Reusable medical devices are used by health professionals to diagnose and treat multiple patients.1 Among reusable devices that are processed, endoscopes—including duodenoscopes, colonoscopes, and gastroscopes—have been associated with frequent outbreaks of infections due to antibiotic resistant Gram-negative bacteria.2–5 Estimates indicate that 75 million endoscopies are performed annually in the United States,6 and alternative procedures are limited.
Due to widespread clinical need and frequent reports of infection, the importance of designing endoscopes for cleanability has come to the forefront. Studies have highlighted the challenges with cleaning current endoscopes adequately due to their complex design.7,,8 As a result, contamination may become an issue at some point in the life cycle of an endoscope. A 2016 U.S. Senate investigation identified 25 independent antibiotic-resistant infectious outbreaks linked to duodenoscopes worldwide from 2012 to 2015, infecting a total of 213 patients.9 In a study evaluating the processing of 45 endoscopes at three hospitals, microbial contamination was detected in 71% of the devices.7,,10 Another study reported that 60% of colonoscopes and gastroscopes still had microbial contamination despite full processing.11
The Food and Drug Administration (FDA) became aware of reports of infectious outbreaks potentially associated with duodenoscopes in 2013. Postmarket surveillance studies were ordered through Section 522 of the Federal Food, Drug, and Cosmetic Act. The results of the postmarket surveillance study showed contamination, though less extreme than found in some literature reports. As of July 2019, up to 4.4% of endoscope samples tested positive for low- to moderate-concern organisms (<100 colony-forming units [CFU]) and up to 6.1% of samples tested positive for high-concern organisms.12 The studies used a protocol for brushing/swabbing and culturing scopes,13 which is the gold standard practice for hospitals when an infection may be associated with an endoscopic procedure. The postmarket findings highlighted why an urgent need exists to better understand how device design may play a role in processed endoscopes' susceptibility to contamination.
Although swabbing and culturing or scraping and culturing are useful in detecting living microorganisms on a device, no standard method exists currently for identifying where cells are adhering to device components. Determining the locations of bacteria on scopes associated with infections can provide critical information about the device design to manufacturers. That information can be used in investigations of outbreaks and also to improve the device design. Detecting small amounts of cellular contamination in areas typically inaccessible to swabbing or scraping methods (e.g., underneath distal end caps or in in accessible elevator channels) is challenging, and current methods for detecting cellular contamination have considerable limitations. Borescopes often are used both during and following use of endoscopes to identify surface roughening or the presence of residual soil on the macro scale. However, the microscopic presence of bacterial cells or biofilm is not always visible or accessible with this technique. Microanalytical methods such as X-ray photoelectron spectroscopy, Fourier transform infrared spectroscopy, atomic force microscopy, and mass spectrometry can be used to study organic films on surfaces and may provide limited information to identify contamination on endoscope parts.8,14–17 However, definitively identifying cells with these analytical methods can be challenging.
In the current work, we respond to the need for a standard investigational procedure for medical device contamination by developing a microscopic approach to label and identify bacterial cells and biofilm directly on the surface of medical device components. Our standard operating procedure (SOP) uses widely available 4',6-diamidino-2-phenylindole (DAPI) and widefield or confocal fluorescence microscopy. We chose DAPI because it has a long shelf life, can stain a wide range of potential microbial contamination,18–20 has a good quantum yield, is inexpensive, and is widely used. In this work, the proposed SOP for fluorescence microscopy–based detection of residual bacteria on medical devices (BAC-VIS) was optimized for best detection efficiency and signal-to-noise ratio (S/N) using several common pathogenic bacterial strains. We validated the SOP using a buildup biofilm model, clinically used endoscope parts with and without positive culture results for pathogenic Gram-negative bacteria, and inoculated medical device components.
Staphylococcus epidermidis (ATCC 35984) and Escherichia coli (ATCC 53498) were purchased from ATCC (Manassas, VA). Green fluorescent protein (GFP)-producing Staphylococcus aureus AH254721 was provided by Alexander Horswill, PhD (University of Iowa, Iowa City), GFP-producing Pseudomonas aeruginosa PAO1 was provided by Abraham Joy, PhD (University of Akron, Akron, OH).22
Preparation and Use of DAPI for Staining Bacteria
DAPI preparation. A 20-mg/mL DAPI stock was prepared by diluting the solid reagent (Sigma-Aldrich, St. Louis, MO) in deionized sterile water and kept refrigerated. Working DAPI solutions (0.3 μM and 30 μM) were prepared by diluting this stock with deionized sterile water in a 15-mL Falcon tube (Corning, Corning, NY). The tube was wrapped in aluminum foil to protect it from light and kept at room temperature. Fresh DAPI solutions were made from the stock for each day's experiments. The DAPI reagent and stock can lose effectiveness over time; therefore, validating the shelf life for whatever storage conditions are used is important.
Bacterial preparation. Bacterial stocks were streaked onto nutrient agar plates and incubated at 37°C for 24 hours. Using a sterile inoculation loop, a colony of bacteria was transferred from the plate to a 15-mL Falcon tube containing 3 mL tryptic soy broth. This bacterial suspension was incubated while shaking at 37°C and 225 rpm for 16 to 18 hours. The bacterial suspension was then vortexed for 30 seconds and sonicated for five minutes. The suspension was kept upright for five minutes to allow the larger clusters of bacteria to settle. To obtain a single or double cell solution, 1 mL bacterial culture was withdrawn from the upper half of the suspension and subsequently pushed through a 5-μm syringe filter (BD, Franklin Lakes, NJ). The bacterial cell count was confirmed by hemocytometry and diluted to desired concentrations (e.g., ~108/mL).
Bacterial adhesion and staining. A glass microscope slide (Fisher Scientific, Hampton, NH) was sonicated in ethanol for five minutes and dried. Bacterial suspension (50 μL) was deposited on the slide and incubated at room temperature for 30 minutes before rinsing one time with phosphate-buffered saline (PBS). A total of 200 μL of the DAPI solution (either 0.3 μM or 30 μM) was deposited onto the slide so that the stain covered the entire site of initial bacterial deposition. The slide was incubated while protected from light at room temperature for 20 minutes. The slide then was immediately rinsed three times for five seconds with PBS. An additional 50 μL PBS was added, and the sample was covered with a coverslip.
Imaging. Samples were imaged within 30 minutes of rinsing. Widefield microscopy (SP8 Inverted; Leica Microsystems, Wetzlar, Germany) was performed using an arc lamp for excitation and a charge-coupled device camera for detection (blue filter set and 10× air objective). Confocal microscopy (SP8 Inverted) was also performed with the same 10× air objective to image samples. Confocal and transmitted light images were obtained using 405 nm diode laser excitation, with an emission wavelength of 436 nm selected using Leica's multiband spectrophotometer.
S/N was calculated, and a plot of the profile measuring grayscale values as a function of distance was obtained. The average background was calculated and removed from the maximum peak intensity to obtain a corrected peak (signal) value. The standard deviation of the background was calculated and averaged to determine the noise. S/N was determined by averaging the corrected peak divided by the calculated noise.
Microfluidic Buildup Biofilm Model
Bacterial suspension (P. aeruginosa, 108 CFU/mL), sterile water, air, and diluted high-level disinfectant (HLD) were cycled through the system at 0.72 mL/hour over eight days to simulate worst-case scenario endoscope processing and reuse. At the end of the eight days, the full concentration of HLD was applied to the channels, rinsed with sterile water, and stained using the BAC-VIS method. The channels were rinsed with sterile water once more and imaged using confocal microscopy.
To test whether the BAC-VIS was able to stain buildup biofilm bacteria, the channels were filled with 30 μM DAPI and protected from light for 20 minutes. The channels were then rinsed with 3 mL PBS each and filled with sterile water. Images of the same locations were taken on the confocal microscope both prior to and after the procedures. Image alignment was corrected using the StackReg ImageJ plugin.24
Real-World Validation Testing
Clinically used endoscope parts. We tested the real-world application of BAC-VIS to endoscope parts. Biopsy suction channels from a variety of different types of endoscopes, which cultured positive for Gram-negative bacteria, were evaluated using BAC-VIS. The culturing was done at Healthmark Industries. A total of 0.5 mL of the sample extracted from the channels was pipetted onto MacConkey agar. MacConkey agar is a selective and differential culture medium commonly used for the isolation of enteric Gram-negative bacteria. An L-shaped spreader was used to spread the sample extract evenly on the petri plate, which then was placed in an incubator at 37°C for 72 hours. After incubation, the petri plate was inspected for Gram-negative bacterial growth. To prepare samples for BAC-VIS, a 2-inch section of the channel was cut and sliced in half along the longitudinal (cylindrical) axis with a razor blade. Using sterile tweezers, the sample was incubated in a 15-mL Falcon tube containing 30 μM DAPI for 20 minutes while protected from light. The sample was removed from the tube and immediately rinsed with PBS three times for 30 seconds. Because most bacteria are expected on the interior (concave) surface of the tubing, the exposed inner surface of the tubing was placed down on a microscope cover glass and imaged with confocal and widefield fluorescence microscopy as described above.
Inoculated endoscope device components. Thermopolymer cap material (Viton) was obtained from Rubbersheet Warehouse (Huntington Beach, CA), and O-rings were obtained from Marco Rubber and Plastics (Seabrook, NH). Silicone O-rings and thermopolymer cap materials were inoculated overnight with an initial number of 108/mL E. coli cells. Control materials (no inoculation) and inoculated materials subjected to the optimized BAC-VIS method (30 μM DAPI for 20 minutes with PBS three times for 30 s) were imaged as described above.
Results and Discussion
We chose DAPI because it binds nucleic acids25,,26; thus, DAPI can stain human cells, dead bacteria, live bacteria, nucleic acids composing biofilms, and ribonucleic acid.18 For the purpose of detecting contamination on endoscope parts, the binding of DAPI to any of the above supports biological contamination. To optimize the concentration of DAPI stain for detecting bacteria in endoscope studies, we created glass slide surfaces with controlled amounts of bacteria stained with 0.3 μM or 30 μM DAPI (Figure 2A). A series of processing steps (see METHODS section) were used to produce well-defined images that could be used for quantitative analysis of staining. The goal was to determine if DAPI staining could achieve greater than 80% staining efficiency for all organisms and an S/N of more than 20.
To quantify the staining efficiency of DAPI for different types of bacteria, the total number of cells visibly stained by DAPI was compared with the total number of cells observed (after staining) by transmitted light (Figure 2B). The 0.3-μM DAPI concentration stained P. aeruginosa and S. aureus with about 90% efficiency. The 0.3-μM staining efficiency for S. epidermis (64%) and E. coli (19%) was much lower. At the higher 30-μM DAPI concentration, staining was more uniform across all four organisms, reaching greater than 90% for all but E. coli (83%). This is similar to the findings of Schallenberg et al.,27 who reported that a minimum of 18 μM DAPI was required to obtain reliable results for quantifying bacterial counts in environmental samples. Mycobacteria (results not shown) also were tested, but staining was not observed under the same conditions for up to 30 μM DAPI.
Next, the S/N was investigated. For all bacterial strains tested, S/N was observed to be substantially higher when the bacteria were stained with 30 μM DAPI compared with 0.3 μM DAPI (Figure 2C). At the 0.3-μM DAPI concentration, the two Gram-positive organisms S. aureus and S. epidermis had similar S/Ns, whereas the two Gram-negative organisms had slightly lower S/N. At the higher 30-μM DAPI concentration, E. coli, S. aureus, and P. aeruginosa had approximately a doubling of S/N, whereas S. epidermis had about a 2.5-fold increase in S/N. Although the 0.3-μM DAPI concentration was unable to meet the goal of greater than 80% staining efficiency, it did meet the goal of an S/N of more than 20 for all but P. aeruginosa. The 100-fold higher 30-μM DAPI concentration met the goal of greater than 80% staining efficiency and an S/N of more than 20 for all organisms. Because endoscope-related infections are commonly linked to Enterobacteriaceae, the staining efficiency for E. coli is important for real-world application. These results show that both Gram-positive and Gram-negative biological contamination can be detected using the BAC-VIS staining protocol. However, mycobacteria cannot be detected with the proposed SOP.
Microfluidic Buildup Biofilm Modeling
BAC-VIS was also tested on a buildup biofilm model intended to mimic real-world conditions. A microfluidic approach was used to create buildup biofilms (Figure 1 and Table 1). The biofilms created by this process have been exposed to conditions that replicate repeated cycles of contamination and disinfection found in processing. Buildup biofilms of P. aeruginosa were created using three HLDs in common use: glutaraldehyde (GA), ortho-phthalaldehyde (OPA), and peracetic acid (PAA).2 Because these disinfectants have different chemical actions on microbes, we wanted to ensure that the DAPI staining process would not result in removal of bacterial cells or biofilm, resulting in false-negative results for any of these chemistries. It was also important to confirm that staining with DAPI was robust for cells treated with disinfectants.28
The control channel without HLD treatment showed uniform sparse biofilm contamination (Figure 3A). The GA-treated channels had a high level of continuous interconnected biofilm buildup compared with the control channel (Figure 3B). This may be due to cross linking of biofilm during treatment with GA,29 making the biofilm more robust and difficult to remove with rinsing. The OPA-treated channels had substantially reduced amounts of buildup biofilm compared with the control and GA channels (Figure 3C). Where biofilm was observed, it was similar to GA in larger, more continuous buildup patches. This is consistent with the fact that OPA does cross link proteins like GA but at a lower density, resulting in less buildup of what is clinically observed as a “residue.”30 Images of PAA-treated buildup biofilm had similar GFP fluorescence from individual cells as the control channel and did not appear to have the cross-linked patches of biofilm like those from GA and OPA (Figure 3D). We observed continued presence of GFP fluorescence from biofilm in a flow cell several hours after all HLD treatments, while microscopic observation after the HLD process and thorough rinsing showed many bacterial cells swimming around the biofilm (video available at www.linkedin.com/posts/k-scott-phillips-a9524554_biofilms-activity-6717457488309968897-IvOu). These observations underscore the importance of developing improved cleaning and disinfection approaches for endoscopes. We also noted areas of lower-intensity background staining after the BAC-VIS method. These may be due to the fact that DAPI is capable of staining DNA in the biofilm extracellular matrix at a lower signal intensity than cellular DNA.31,,32
The process of quantifying buildup biofilm before and after staining was more challenging than testing staining efficiency of individual cells because of the higher density of buildup biofilms, which included some overlapping cells. To overcome this challenge, total fluorescence signal from the GFP-expressing P. aeruginosa buildup biofilm was quantified before and after the BAC-VIS SOP. Use of BAC-VIS on GA-treated buildup biofilms showed the largest loss of fluorescent signal from cells, with 90% remaining after rinsing, but had the highest consistency across measurements (±6.5%). Use of BAC-VIS on OPA-treated buildup biofilms showed a slight increase in total fluorescence from cells (103%) but had much larger variability across measurements (±29%). Similarly, use of BAC-VIS on PAA-treated buildup biofilms also showed an increase in total fluorescence from cells (121%), with large variability in measurements (±55%). The most likely explanation for the relatively small increase in cells after rinsing for OPA-treated chan nels and the much larger increase in cells after rinsing for PAA-treated channels is carryover of bacteria from the tubing and connections during rinsing. Although all connections and tubing are exposed to the same HLD treatment as the channels themselves, it is possible that buildup biofilm on these materials continued to slough off even after the process of buildup biofilm formation and HLD treatment and rinsing is complete. The fact that GA was observed to cross link biofilm and had very little other small background cells in channel images supports the lesser probability of cells sloughing off from surfaces after GA treatment. Because OPA also cross links cells to some degree, this could explain why it was intermediate between GA and PAA.
Use of BAC-VIS on Real-World Samples
The BAC-VIS SOP was applied to study real-world end-of-life accessory channels (Figure 4) from several different types of endoscopes. Channels were confirmed to be contaminated with Gram-negative bacteria (Table 2). Samples were cut to a length that would fit in a sample tube, stained and rinsed using the protocol, and then cut in half along the longitudinal axis and imaged. We tested standard widefield fluorescence (which is available in most labs) to assess whether BAC-VIS could provide reliable detection of bacteria (Figure 5). Samples that tested positive for Gram-negative contamination showed considerable numbers of DAPI-stained cells approximately 1 to 3 μm in size. On gastroscope 1, the distribution of cells appeared to be oriented with the flow direction of the channel, suggesting that grooves in the channel (seen in both new and used channels) may be favorable sites for buildup biofilm formation. The levels of contamination seen on these channels is especially concerning given that they were processed before removal from use.
One point to note is that for clear tubing only, sometimes artifacts were visible when focusing that were not from the surface imaged and were not found to be bacteria upon higher-resolution examination. These may have been other types of chemical soil contaminants or polymer components in the clear tubing. A careful comparison of control materials and clinically used materials showed that debris on the surfaces of clear channels could be seen in both transmitted light images and in fluorescence images, whereas the nonbacterial signals could be discriminated by the lack of details in the transmitted light image, as well as a change in the objects' size as the microscope focus was adjusted (Figure 6). The cells were either in focus around a size of 2 μm or were out of focus, whereas the artifact fluorescence would change size with different z-focal plains. Cells often were seen in grooves or other defects of the channels, similar to scanning electron microscope observations by Pajkos et al.33 We recommend that for clear materials, transmitted light images and fluorescence images should be overlaid to ensure that any fluorescent cells observed are not artifacts.
We also validated BAC-VIS using other device components (Figure 4) that can become contaminated. Unused distal end cap material and O-rings were inoculated with E. coli overnight, and BAC-VIS was used to examine them (Figure 7). Imaging of DAPI-stained control (uninoculated) silicone O-ring material showed no detectable cells (Figure 7A). After inoculation with E. coli, the presence of biofilm cells was detected (Figure 7B). Similarly, for endoscope end cap material, imaging of DAPI-stained control components showed no visible cells (Figure 7C), whereas imaging of DAPI-stained inoculated end cap materials very clearly showed the presence of cells (Figure 7D). When using BAC-VIS for testing device components, we recommend the use of a new, clean sample as a control (where possible) to ensure that any fluorescent cells observed are not artifacts.
In summary, we were able to use BAC-VIS with widefield and confocal fluorescence microscopy to observe bacterial cells on inoculated endoscope end cap material and O-rings, as well as clinically used accessory channels that had tested positive by culture. As expected, cells were not observed on clean, new materials. Thus, BAC-VIS can be used for a number of materials that might be found in processed devices.
Endoscope outbreaks often have been connected with damaged parts of instruments, such as the internal channels or loose biopsy port caps of endoscopes,34 and most published studies have focused on overall contamination rather than locating the sources and design features related to contamination. Therefore, developing approaches for imaging bacteria on processed device parts is important.
In the current work, we developed and characterized the BAC-VIS protocol. Of note, this protocol is for research purposes and not for detecting bacteria on endoscopes or scope parts that will be put back into clinical use. The protocol is as follows:
Stain test and clean control (when possible) samples for 20 minutes with a 30-uM DAPI solution in PBS.
Rinse samples thoroughly (10 seconds) at least three times with PBS.
To image, use a DAPI filter set for widefield microscopy or an excitation/emission setting of 405 nm/436 nm with confocal microscopy. A 10x objective should be sufficient, but higher power objectives can be used. Look for bacterial rods or cocci on the order of 1 to 3 μm and/or accompanying biofilm (less intense fluorescence).
For clear materials, overlaying transmission images with fluorescence images is recommended to verify that cell-like images are not due to artifacts. The outline of cells should correspond with the DAPI-stained fluorescent area, and both should be in focus simultaneously.
The BAC-VIS procedure reduces the burden for studying contamination on endoscope parts because it provides a rapid and sensitive approach to locating bacterial contamination using the widely available DAPI reagent and widefield fluorescence microscopy, which is a basic component of any biological or medical laboratory. The times, concentrations, and rinsing protocols described in this work can be used to study potentially contaminated devices to determine where and how contamination is occurring. Using BAC-VIS, bacteria were found on contaminated endoscope accessory channels, especially in grooves or defects of the channels, and we were able to detect bacteria on inoculated medical device parts.
The results showed why a good standard protocol is a valuable tool that can be used by industry and academic/medical researchers for additional, larger-scale investigations of endoscope parts and the role of design in cleanability, wear, and prevention of contamination. The findings obtained through this method may lead to improvements in materials and design that could make endoscope use safer for patients.
The findings and conclusions in this article have not been formally disseminated by the Food and Drug Administration and should not be construed to represent any agency determination or policy. The mention of commercial products, their sources, or their use in connection with material reported herein is not to be construed as either an actual or implied endorsement of such products by the Department of Health & Human Services.
K.S.P., A.K.M., S.P.H., and Y.W. conceived the research and R.B. and K.K. helped conceive the testing process for clinically used endoscope parts. M.W., Y.W., K.S.P., and H.W. conducted BAC-VIS–related laboratory experiments and/or performed data analysis. K.K. cultured clinically used endoscopy accessory channels. All authors worked on the manuscript.
The authors thank the Center for Devices and Radiological Health (CDRH) Critical Path and Food and Drug Administration (FDA) Office of Women's Health for support and Subha Maruvada, PhD, of the FDA's Office of Science and Engineering Laboratories for helpful discussions on the research. Clinically used endoscopy parts were obtained through a Research Collaboration Agreement between the FDA and Healthmark Industries (Fraser, MI).
This project was supported in part by an appointment to the Oak Ridge Institute for Science and Education (ORISE) Research Participation Program at FDA CDRH, which is administered by ORISE through an interagency agreement between the Department of Energy and FDA CDRH.
Michael Wong, BS, is a graduate student at the Center for Devices and Radiological Health of the Food and Drug Administration in Silver Spring, MD.
Yi Wang, PhD, is a research fellow at the Center for Devices and Radiological Health of the Food and Drug Administration in Silver Spring, MD.
Hao Wang, PhD, is a postdoctoral fellow at the Center for Devices and Radiological Health of the Food and Drug Administration in Silver Spring, MD.
April K. Marrone, PhD, MBA, is a regulatory chemist at the Center for Devices and Radiological Health of the Food and Drug Administration in Silver Spring, MD.
Shanil P. Haugen, PhD, is assistant director of the gastroenterology and endoscopy devices team at the Center for Devices and Radiological Health of the Food and Drug Administration in Silver Spring, MD.
Kaumudi Kulkarni, MS, is senior manager of research and development at Healthmark Industries in Fraser, MI.
Ralph Basile, MBA, is vice president of marketing and regulatory affairs at Healthmark Industries in Fraser, MI.
K. Scott Phillips, PhD, is leader of the Biofilms Research Group at the Center for Devices and Radiological Health of the Food and Drug Administration in Silver Spring, MD. Email: email@example.comCorresponding author