Development of a Broth Microdilution Method To Characterize Chlorhexidine MICs among Bacteria Collected from 2005 to 2019 at Three U.S. Sites

ABSTRACT Chlorhexidine bathing to prevent transmission of multidrug-resistant organisms has been adopted by many U.S. hospitals, but increasing chlorhexidine use has raised concerns about possible emergence of resistance. We sought to establish a broth microdilution method for determining chlorhexidine MICs and then used the method to evaluate chlorhexidine MICs for bacteria that can cause health care-associated infections. We adapted a broth microdilution method for determining chlorhexidine MICs, poured panels, established quality control ranges, and tested Staphylococcus aureus, Escherichia coli, Klebsiella pneumoniae, and Enterobacter cloacae complex isolates collected at three U.S. sites. Chlorhexidine MICs were determined for 535 isolates including 129 S. aureus, 156 E. coli, 142 K. pneumoniae, and 108 E. cloacae complex isolates. The respective MIC distributions for each species ranged from 1 to 8 mg/L (MIC50 = 2 mg/L and MIC90 = 4 mg/L), 1 to 64 mg/L (MIC50 = 2 mg/L and MIC90 = 4 mg/L), 4 to 64 mg/L (MIC50 = 16 mg/L and MIC90 = 32 mg/L), and 1 to >64 mg/L (MIC50 = 16 mg/L and MIC90 = 64 mg/L). We successfully adapted a broth microdilution procedure that several laboratories were able to use to determine the chlorhexidine MICs of bacterial isolates. This method could be used to investigate whether chlorhexidine MICs are increasing. IMPORTANCE Chlorhexidine bathing to prevent transmission of multidrug-resistant organisms and reduce health care-associated infections has been adopted by many hospitals. There is concern about the possible unintended consequences of using this agent widely. One possible unintended consequence is decreased susceptibility to chlorhexidine, but there are not readily available methods to perform this evaluation. We developed a method for chlorhexidine MIC testing that can be used to evaluate for possible unintended consequences.

C hlorhexidine is a topical antiseptic used for central venous catheter site preparation and maintenance to prevent infectious complications (1)(2)(3)(4). In addition, routine chlorhexidine bathing to prevent bloodstream infection and transmission of multidrug-resistant organisms has been supported by several cluster-randomized studies and is a strategy adopted by many U.S. hospitals (5)(6)(7). However, concerns have been raised about possible unintended consequences of widespread chlorhexidine use. These include an increase in chlorhexidine MICs or the prevalence of chlorhexidine resistance genes, antibiotic crossresistance, decolonization failure, or an alteration in the skin microbiome (8). However, studies of unintended consequences of chlorhexidine bathing are difficult to interpret (8)(9)(10). The definition of chlorhexidine resistance and the methods used to evaluate chlorhexidine susceptibility are not standardized, making comparisons difficult (8). Reported methods have included measuring MICs by broth microdilution or agar dilution, measuring minimum bactericidal concentrations, time-kill studies, surface disinfection tests, or biocide residue tests (10,11). There is also limited evidence of a correlation between reduced chlorhexidine susceptibility in vitro and decolonization failure in vivo.
In this study, we sought to validate and ensure the reproducibility of a broth microdilution method for determining chlorhexidine MICs. We then used this standardized method to test selected bacterial isolates to examine chlorhexidine MICs.

DISCUSSION
In this study, we adapted a broth microdilution procedure to measure chlorhexidine MICs that allowed as many as 190 panels capable of testing 1,520 isolates to be poured for use. These panels proved capable of being frozen, shipped, and used by multiple laboratories while achieving acceptable quality control (QC) results (12).
By using these panels, we were able to have three different laboratories perform chlorhexidine testing on an existing sample of S. aureus, E. coli, K. pneumoniae, and E. cloacae complex isolates collected from three different sites. The primary purpose of the chlorhexidine testing was to demonstrate the feasibility of different laboratories using this standardized method to test various species. We describe chlorhexidine MIC distributions for each organism (Fig. 1). When comparing our MIC distributions and MIC 50 and MIC 90 values to those reported by others, our E. coli MICs were lower, our K. pneumoniae and E. cloacae complex MICs were higher, and our S. aureus MICs were the same; overall, the MIC distributions were similar (13). We did not calculate epidemiological cutoff values because the isolates were mostly resistant to other antibiotics and thus may not have represented a true wildtype population. Others have demonstrated that sequence types associated with carbapenem-resistant isolates may have higher chlorhexidine MICs than other sequence types (14).
A limitation of the procedure that may impact its generalizability is that we used the Viaflo 96 instrument, which not all laboratories may have access to. Also, our study has several limitations regarding the selection of isolates for testing. We tested a convenience sample of mostly antibiotic-resistant isolates; susceptible isolates may have displayed a different MIC distribution. We did not test all relevant species; for example, Pseudomonas aeruginosa and Acinetobacter baumannii would have also been of interest. Also, most of the isolates were from sterile sites; isolates from nonsterile sites may have displayed a different MIC distribution.
In summary, we have adapted a broth microdilution method to determine the chlorhexidine MICs for health care-associated pathogens. We successfully demonstrated the feasibility of sharing these panels with multiple sites to generate chlorhexidine MIC data. This method could be used in surveillance studies to investigate whether chlorhexidine MICs are increasing. Surveillance for undesired effects is important as the practice of chlorhexidine bathing continues to increase.

MATERIALS AND METHODS
Description of isolates. A convenience sample of Staphylococcus aureus, Escherichia coli, Klebsiella pneumoniae, and Enterobacter cloacae complex isolates was selected for testing by the Orange County Health Care Agency (OCHCA), the Georgia Emerging Infections Program (EIP), and the Tennessee EIP. Each isolate was tested once. S. aureus isolates were all collected during 2005 to 2018 through EIP; all were from normally sterile sites and were methicillin-resistant (15). The Gram-negative bacilli were collected during 2011 to 2019 through the Shared Healthcare Intervention to Eliminate Life-Threatening Dissemination of multidrug-resistant organisms in Orange County or EIP surveillance (16,17). The EIP Gram-negative isolates were from normally sterile sites or urine, and the OCHCA isolates were from a variety of sterile and nonsterile sites. Participating laboratories included OCHCA (tested 164 Gram-negative isolates from OCHCA and 20 S. aureus isolates from Tennessee), Georgia EIP (tested 114 Gram-negative isolates and 60 S. aureus isolates from Georgia), and the CDC (tested 128 Gram-negative isolates and 49 S. aureus isolates from Tennessee).
Broth microdilution testing of bacteria for chlorhexidine MIC. A broth microdilution method, used by all three laboratories, was adapted from a procedure previously published (18). Adaptations were made to allow for multiple 96-well panels to be prepared and immediately frozen for later use. On the day of the chlorhexidine MIC panel pour, all involved plastics and reagents were warmed for 2 h in an ambient air incubator at 35°C to prevent inadvertent precipitation of insoluble chlorhexidine salts. These precipitates were noticed to occur during a pilot phase of this project for chlorhexidine concentrations at $128 mg/L when added to chilled (2 to 8°C) cation-adjusted Mueller-Hinton broth (CAMHB) or when the chlorhexidine-CAMHB mixture was frozen at chlorhexidine concentrations of $128 mg/L. To prevent foaming, a mechanical vortex  . The Viaflo aspirated the working dilutions with 96 sterile disposable 1,250-mL-capacity filter tips from a sterile 96-well square V-bottom block, with a 2-mL capacity per well of the block (Corning Inc., Salt Lake City, UT). This block was manually loaded with the working dilutions. All MIC panels were prepared at the CDC, and the final poured panels were kept frozen at 270°C until ready to use. The 16 quality control isolates ( Table 2) were tested once immediately on freshly poured panels and once 1 week later using frozen panels; when comparing fresh panels to frozen panels, MIC essential agreement was 93.8%. Frozen panels shipped to the Orange County Health Care Agency (OCHCA) were sealed in 7.5-mil Mylar foil bags to prevent a change in pH of the medium from carbon dioxide emitted by the dry ice used for shipping. The frozen panels for the Georgia Emerging Infections Program (EIP) were packed with frozen gel packs and delivered to that site within 1 h of packing. All isolate testing was performed in April to June 2019.
MIC panels were thawed at room temperature (2 to 4 h) before inoculating. An inoculum was prepared for each isolate by using a sterile swab to transfer bacteria from an 18-to 24-h blood agar plate culture into a tube of 5 mL of sterile saline. The solution was vortexed and adjusted until turbidity matched a 0.5 McFarland standard (using a turbidity meter). A 1:20 dilution was prepared by adding 2 mL of the adjusted inoculum into a sterile polypropylene tube containing 38 mL of sterile saline. The capped 1:20 dilution tube was gently inverted 8 to 10 times. Next, 5 to 10 mL of the 1:20 dilution was poured into a sterile 25-mL plastic reservoir. Eleven channels of a 12-channel pipettor were loaded with sterile aerosol barrier filter tips and used to aspirate 10 mL of the inoculum per pipette tip from the reservoir and then dispense 10 mL to each well per assigned row (columns 2 to 12). Column 1 was used as a negative control (Fig. 2). The 10 mL of inoculum was gently pipetted up and down 4 to 5 times to mix with the 100 mL of chlorhexidine-cation-adjusted Mueller-Hinton broth in each well of the designated MIC panel row, and then 10 mL of the inoculated mixture was withdrawn from each well of the row and discarded. The target final organism concentration per well should be approximately 2 Â 10 5 to 8 Â 10 5 CFU/mL when using this method (19). A 1-mL sterile loop of the inoculum from the plastic reservoir was used to inoculate a Trypticase soy agar plate with 5% sheep blood as a purity check. These steps were repeated for 8 isolates per panel. Completely inoculated panels were placed in an 8-inch by 8-inch resealable plastic bag with excess air removed to prevent moisture condensation (up to 4 panels per bag). The panels and purity plates were incubated at 35°C 6 2°C in an ambient air atmosphere for 16 to 20 h. After incubation, each panel was read by holding it against a black background, using a surface illuminated with reflected light. The lowest chlorhexidine concentration showing no visible growth in the well was read as the MIC endpoint. No trailing was observed when reading the panels.
Quality control. A set of 12 isolates with a range of chlorhexidine MICs was provided by Rush University Medical Center; there were another four isolates used for routine quality control (QC) ( Table 2). All 16 isolates were tested 10 times each at CDC to assess reproducibility and establish QC ranges using the RangeFinder MIC (https://clsi.org/meetings/microbiology/rangefinder/). The set of 12 was then tested once by OCHCA and the Georgia EIP prior to testing study isolates to establish that their results fell within the QC range  Table S1 in the supplemental material). The four isolates used for routine QC were tested on each day of testing to ensure conformity of the results with the tentatively established ranges (Table S1). All QC results fell within the expected ranges ( Table 2 and Table S1). Ethical approval. The protocol was reviewed by the CDC institutional review board (IRB) and all participating sites and was deemed nonresearch or received IRB approval with a waiver of informed consent.

SUPPLEMENTAL MATERIAL
Supplemental material is available online only. SUPPLEMENTAL FILE 1, XLSX file, 0.01 MB.