Paper-Based Semi-quantitative Antimicrobial Susceptibility Testing.

Antimicrobial resistance is increasingly recognized as a major threat to global health. To combat this emerging threat, accessible antimicrobial susceptibility testing should be prioritized as a key component of stewardship efforts. In this work, we developed a user-friendly paper-based test that provides visual readout of bacterial antibiotic susceptibility in a semiquantitative format. We leveraged on-chip paper microfluidics to enable multiplexed testing of multiple antibiotic dilutions with a single sample addition step, replicating the functionality of traditional broth-dilution-based susceptibility testing in a simplified format. Our paper-based test offers several advantages including low sample volume requirement and lack of need for humidity control during incubation, an innovation that addresses a key limitation of conventional paper-microfluidic devices. Using several clinically relevant bacterial organisms and antimicrobial agents, we demonstrate that our colorimetric readout approach provides a strong predictor of susceptibility category.

Introduction

Antimicrobial resistance (AMR) is increasingly recognized as a substantial threat to global health.[1] Overuse and misuse of antimicrobials are contributing toward an increasing prevalence of antimicrobial-resistant and multidrug-resistant organisms. In the case of antibiotics, studies have shown that almost one in three prescriptions are inappropriate, either in choice of agent or duration.[2,3] Incorrect antimicrobial usage can facilitate the development of AMR genes through selective pressures.[4] In order to combat misuse, evidence-based antimicrobial agent selection based on antimicrobial susceptibility testing (AST) forms a key component of antimicrobial stewardship efforts.[5] Accessible susceptibility testing methods are crucial for facilitating stewardship, especially in point-of-care and limited resource settings. Clinically, the need for accessible testing methods remains unmet as many US hospitals outsource susceptibility testing to reference laboratories.[6] The preference for user-friendly testing methods is reflected in the fact that when testing is done in-house, hospital labs predominately favor the use of gradient diffusion,[6] which is relatively simple but can require subjective interpretation of results.[7] Reference laboratories on the other hand predominately favor use of broth microdilution,[6] which provides quantitative minimum inhibitory concentration (MIC) information but requires high operator input and availability of diagnostic instrumentation.

There have been ongoing research efforts to develop novel phenotypic AST methods.[8] Phenotypic AST, which involves direct measuring of organism growth in the presence of an antibiotic, is differentiated from genotypic AST, which involves detecting resistance genes to infer susceptibility profiles.[9] Developments in the area of microfluidic AST methods have attempted to improve upon conventional AST methods through automation and lower sample volume requirements.[10] Examples of novel microfluidic AST approaches include a self-loading chip device,[11] a pH-sensitive hydrogel sensor,[12] agarose channels that enable morphology tracking,[13] on-chip broth dilution,[14] and nanoliter arrays.[15] Despite these advances, microfluidic approaches often face limitations of manufacturing and readout complexity,[10] which have served as barriers to scalability and widespread adoption.

Paper microfluidics have the potential to address some of the scalability limitations of microfluidic AST. In paper microfluidics, 2D/3D channels are formed in a paper substrate through wax patterning to enable passive fluid transport.[16] Compared to the polymers commonly used to fabricate microfluidic devices, paper substrates offer greater manufacturing flexibility through compatibility with a number of patterning techniques, act as a natural medium for colorimetric tests, and can be easily disposed of via incineration.[16,17] Examples of paper-microfluidic AST approaches include a paper-polydimethylsiloxane (PDMS) hybrid disk diffusion culture device,[18] a paper-PDMS cell culture array,[19] and a paper-based β-lactamase test.[20] With paper-microfluidic devices, sample evaporation is often an issue that necessitates the use of external humidity control to resolve, which can limit the practicality of device deployment in point-of-care or limited resource settings.

In this work, we present a novel paper-microfluidic AST method that replicates the functionality of broth microdilution in a simplified format. Our method consists of a sealable paper-based test chip that provides a visual susceptibility readout. Printed wax channels allowed for antibiotics and growth-sensitive dye to be predried in spatially separate zones on the test chip, thereby enabling multiplexed testing of multiple antibiotic concentrations with a single sample addition step. Our chip was designed to eliminate the need for external humidity control during incubation, an improvement over a key limitation of conventional paper-microfluidic devices. We demonstrate the efficacy of our paper-based method through comparing on-chip and off-chip AST results for several clinically relevant bacterial organisms.

Results and Discussion

Paper-Based AST Design

As shown in Figure , we developed a sealable paper-based test chip that provides a visual readout of AST results. Hydrophobic wax channels were leveraged to create a network of test zones to enable susceptibility testing at multiple antibiotic concentrations with a single sample addition step. Serial-diluted antibiotics along with the colorimetric redox indicator resazurin were predried in the test zones, further simplifying testing workflow and replicating the functionality of broth-dilution-based AST. A single sample addition step reconstitutes dried reagents and allows bacteria to be incubated in the presence of a specific antibiotic concentration in each test zone. The redox indicator resazurin is reduced into resorufin by metabolically active bacteria,[21] thereby providing a visual indication of bacterial growth when the test zone antibiotic concentration is insufficient to inhibit growth (i.e., below MIC). When considered collectively, the number of “positive” test zones exhibiting color change is correlated with the bacterial strain’s MIC value and susceptibility category.

Figure 1

Overview of chip fabrication and AST process. (A) Wax patterns printed on chromatography paper is heated on a hot plate to form hydrophobic barriers through the thickness of the paper. (B) Each chip accommodates eight test zones that hold seven antibiotic concentrations along with a no-antibiotic control. Antibiotics and a resazurin-based dye, which provides visual indication of bacterial growth, are predried in the test zones prior to AST. Test zones increase in antibiotic concentration counterclockwise with the control zone at the three o’clock position (denoted by C). Prior to reagent dispensing, the back of the chip is sealed using sterile sealing film to prevent leakage. The GEN label denotes the chip containing the antibiotic gentamicin. (C) To initiate AST, the bacterial sample is dispensed in the center of the chip and water is dispensed in the peripheral region. The chip is then sealed using a sterile film to prevent contamination and trap evaporating water vapors, which helps maintain humidity during incubation. Increased fluid volume from the added water lowers sample evaporation during incubation. (D) Bacteria uninhibited by antibiotics reduce resazurin into resorufin, which induces a color change in the test zone. The number of zones with reduced resazurin is correlated with the magnitude of the bacterial strain’s MIC value. (E) Test chip with melted wax channels. (F) Sealed test chip ready for incubation.

To initiate on-chip AST, 80 μL of the sample is dispensed in the central region of the chip. The sample is then divided by the wax channels and flows into each of eight test zones via capillary action. We estimate that each test zone can accommodate 4–5 μL of sample volume. Because of these small sample volumes, we incorporated a peripheral water-holding region, which accommodates 120 μL of water, on the chip to minimize sample evaporation during incubation. Because the chip is sealed inside the nonpermeable film during incubation, evaporating water vapors raise the ambient humidity of the internal air pocket and the increased fluid volume from the water-hold region helps mitigate sample evaporation. As a result, this element of the chip design eliminates the need for external humidity control during incubation.

Colorimetric Detection of Susceptibility

Following incubation, AST results can be interpreted qualitatively by eye or quantitively through colorimetric image analysis. Quantitative colorimetric analysis of on-chip results is shown in Figure . The chip was imaged using a smartphone and analyzed in the hue, saturation, value (HSV) color-space, which we have found to be more precise for colorimetric analysis and less sensitive to lighting conditions compared to the red, blue, green (RGB) color-space.[23] For each test zone, the average hue value in a 160-pixel diameter circular region was calculated and a threshold (275 hue) was used to differentiate between positive (uninhibited growth) and negative (inhibited growth) zones. A lower number of positive test zones correlates to a lower MIC value and a higher number of positive test zones to a higher MIC. As shown in Figure B,D, a significant difference in the hue profile can be used to distinguish between ampicillin (AMP)-resistant (six positive zones) and AMP-susceptible (no-antibiotic control positive only) Escherichia coli strains.

Figure 2

Colorimetric detection of bacterial growth. (A) Post-incubation test chip of E. coli BAA-2452 (AMP resistant via off-chip AST) grown in the presence of varying concentrations of AMP. Test zones with uninhibited and viable bacteria exhibit colorimetric change from blue to pink as resazurin is reduced into resorufin. The chip was imaged using a smartphone and analyzed in the HSV color-space. (B) For each test zone, the average hue value in a 160-pixel diameter circular region was calculated and a threshold (275 hue) was used to differentiate between zones with reduced (indicates growth) and unreduced (indicates no growth) resazurin. Zones above the threshold (pink) are classified as positive and zones below (blue) as negative. (C) Post-incubation test chip of E. coli ATCC-25922 (AMP susceptible via off-chip AST) grown in the presence of AMP. (D) Test zone hue values of E. coli ATCC-25922.

On-Chip AST

To demonstrate expanded on-chip AST capabilities, we tested strains of E. coli ATCC-25922, E. coli BAA-2452, Klebsiella pneumoniae BAA-1903, and Acinetobacter baumannii BAA-1791 against four common antibiotics: gentamicin, ampicillin, ciprofloxacin, and meropenem (dilution concentrations for each antibiotic can be found in the Supporting Information). Strains were tested at a recommended starting concentration of 5 × 105 CFU/mL[22] and analyzed following overnight incubation. Each bacteria–antibiotic combination was tested in triplicate and the number of positive on-chip test zones, which showed consistency across triplicates for each combination, was correlated to off-chip AST determination (VITEK 2 automated AST). Comparisons of on-chip and off-chip AST results are shown in Figure . We observed that there is a statistically significant difference (P < 0.001) in the number of positive test zones for susceptible and resistant categories.

Figure 3

Comparison of on-chip and off-chip AST results. (A) For each antibiotic, on-chip AST results in the form of number of test zones with reduced resazurin (indicates growth) are plotted along with off-chip AST-determined susceptibility category. Square markers represent gentamicin (GEN), triangle markers represent ciprofloxacin (CIP), circle markers represent AMP, and diamond markers represent meropenem (MEM). R (pink) and S (blue) denote resistant and susceptible, respectively, as determined via off-chip AST. A statistically significant difference (P < 0.001) is observed in the number of positive test zones between susceptible and resistant organisms. Each bacteria–antibiotic combination was tested in triplicate, and the number of positive on-chip test zones showed consistency across triplicates for each combination. (B) Table summarizing AST results. On-chip results: number indicates number of test zones with reduced resazurin (including no-antibiotic control zone). Off-chip results: S and R denote vendor MIC interpretations, corresponding to susceptible and resistant, respectively.

Discussion

To combat the increasing prevalence of AMR, accessible AST should be prioritized as a key component of stewardship efforts. In this work, we developed a user-friendly paper-based test chip that provides visual readout of AST results in a semiquantitative format. We demonstrated on-chip that the number of positive test zones is a strong predictor of the susceptibility category. Incorporation of paper microfluidics enabled multiplexed testing of antibiotic dilutions with a single sample addition step, mimicking and simplifying the testing concept of traditional broth-dilution-based AST. We demonstrated the testing capability of our method using several clinically relevant bacterial organisms and antimicrobial agents.

Our paper-based AST method offers several advantages over conventional AST methods. First, predrying of reagents along with incorporating wax microfluidic channels on-chip greatly simplifies testing workflow and minimizes the number of operator steps needed to test multiple antibiotic concentrations. Second, the low sample volume requirement (averages to ∼10 μL per test zone) minimizes reagent consumption and is almost an order of magnitude lower than the per-well sample volume typically required in 96-well microdilution.[22] Third, spatially separate test zones facilitate semiquantitative interpretation of AST results, especially when aided by the resazurin-based colorimetric indicator. Imaging-based readout of results eliminates subjective interpretation and associated inaccuracies that can be present with visual-based readout of microdilution and gradient diffusion results.[24,25] Compared to PDMS-based microfluidic devices, our paper-based test chip is of lower fabrication complexity. In addition, because sealing of the chip from the external environment removes the need for humidity control, our chip is amenable to flexible incubating conditions with temperature control as the only requirement—meaning incubation can occur in a benchtop incubator, oven, or even on a hotplate. Eliminating the need for external humidity control during incubation marks an improvement over a key limitation of conventional paper-microfluidic devices.

We noticed that when drying antibiotics on paper, it was necessary to use higher concentrations of antibiotics than that typically used in liquid culture. This is a potential limitation as the MICs obtained on-chip cannot be directly compared with published MIC breakpoints without an intermediate calibration curve. We hypothesize that higher antibiotic concentrations are required because of (1) irreversible binding of dried antibiotics to paper and/or (2) diffusion of antibiotics out of the test zones into the feeder microfluidic channels which ultimately reduces the test zones’ final concentrations. This potential limitation can be mitigated as additional organisms with different susceptibility profiles are tested on-chip, and calibration curves are established linking on-chip and off-chip MIC values, which will enable fully quantitative AST in paper-based format. Additionally, modification of the paper substrate with blocking agents and/or surfactants may facilitate resolubilization of dried antibiotics.

In summary, we developed a user-friendly paper-based AST method that leverages paper microfluidics to simplifying testing and provide visual readout of AST results. Our method offers several advantages over conventional AST methods and has the potential to serve as an accessible alternative in point-of-care and limited-resource settings.

Experimental Section

Bacterial Strains

Quality control and multidrug-resistant reference strains with characterized AST profiles (via automated AST) were sourced from the American Type Culture Collection (ATCC). Strains used in this study consisted of E. coli ATCC 25922, E. coli BAA-2452, K. pneumoniae BAA-1903, and A. baumannii BAA-1791. Strains were received in the lyophilized format and revived according to vendor instructions.

Chip Design and Fabrication

2D microfluidic channels were designed using Adobe Illustrator and printed onto Whatman Grade 1 filter paper (GE Healthcare) using a ColorQube 8570 wax printer (Xerox). The channels were designed to separate the chip into two distinct regions: a central sample holding region that branches into eight test zones and a peripheral water holding region. Text labels were printed along the periphery of the chip to denote the antibiotic used and respective concentrations in each of the multiple zones. Following printing, each chip (5 cm length × 5 cm width) was cut out using scissors sterilized with 70% ethanol. In order to melt the wax channels through the thickness of the filter paper to form a hydrophobic barrier, each chip was heated on a hot plate (VWR) set to 150 °C for 1 min. After heating, the back side of each chip was sealed using the sterile microplate sealing film (VWR) to prevent leakage. Alternatively, the transparent tape can also be used.

To functionalize the chips for AST, antibiotics and the growth-sensitive dye resazurin were predried on the chip to enable testing with a single sample addition step. Antibiotics used consisted of AMP, MEM, gentamicin, and CIP (all sourced from MilliporeSigma). The growth-sensitive dye resazurin (PrestoBlue, Thermo Fisher) is a redox indicator that is reduced by metabolically active cells into resorufin, leading to significant colorimetric and fluorescent changes which provide a visual indication of bacterial growth.[21] With 10× PrestoBlue solution as the diluent, antibiotic stock solutions prepared at 10–50 mg/mL (depending on solubility in water) were initially diluted to 2.5–10 mg/mL and subsequently two-fold serially diluted (exact concentrations for each dilution of each antibiotic can be found in the Supporting Information). We noticed that when drying antibiotics on paper, it was necessary to use higher concentrations of antibiotics than that typically used in liquid culture. 4 μL of each antibiotic–dye mixture was dispensed in the test zones in increasing concentration counterclockwise. A control zone without antibiotic was dispensed at the three o’clock position. Chips were dried at 37 °C for 1 h and stored protected from light prior to use.

To prepare strains for on-chip AST, strains were first streaked on Mueller–Hinton II agar plates (Becton, Dickinson and Company) and incubated overnight at 37 °C in ambient air. Liquid cultures were then prepared by inoculating single colonies into cation-adjusted Mueller–Hinton Broth (CAMHB), followed by overnight incubation at 37 °C in ambient air. To achieve the recommended AST starting inoculum concentration of 5 × 105 CFU/mL,[22] the overnight culture was adjusted using CAMHB to the equivalent turbidity of a 0.5 McFarland standard, which represents a concentration of approximately 1 × 108 CFU/mL. Using a spectrophotometer (V-1200, VWR), the OD625 nm absorbance corresponding to a 0.5 McFarland standard was verified to be in the range of 0.08–0.13.[22] Cultures were subsequently diluted 1:100 in CAMHB to reach an approximate starting concentration of 5 × 105 CFU/mL. 80 μL of the diluted sample was then dispensed in the center of the test chip followed by 120 μL of distilled water in the peripheral water holding region. After the entire chip was saturated by the dispensed liquids, the chip was sealed using two pieces of the sterile microplate sealing film (VWR). Sealing of the test chip prevents contamination and helps to maintain humidity by trapping evaporating water vapors. The water-holding region increases the total volume of liquid stored on-chip, which helps to limit sample evaporation during incubation. Chips were incubated overnight (14–16 h) at 37 °C (ambient air) in a benchtop incubator protected from light.

Data Analysis

To quantitatively analyze color change in the test zones following incubation, test chips were imaged using a smartphone and analyzed in MATLAB (MathWorks). Images were analyzed in the HSV color-space which we determined to be less sensitive to lighting conditions compared to the RGB color-space. For each test zone, the average hue value in a 160-pixel diameter circular region was calculated and a threshold was used to differentiate between zones with reduced (growth) and unreduced (no growth) resazurin. The number of zones with reduced resazurin is correlated with the magnitude of the bacterial strain’s MIC value.

Statistical Analysis

The F-test was used to test for equality of population variances, and unpaired one-tailed t-test was used to test for equality of population means.