Moving microcapillary antibiotic susceptibility testing (mcAST) towards the clinic: unravelling kinetics of detection of uropathogenic E. coli, mass-manufacturing and usability for detection of urinary tract infections in human urine

Innovation in infection based point-of-care (PoC) diagnostics is vital to avoid unnecessary use of antibiotics and the development of antimicrobial resistance. Several groups including our research team have in recent years successfully miniaturised phenotypic antibiotic susceptibility tests (AST) of isolated bacterial strains, providing validation that miniaturised AST can match conventional microbiological methods. Some studies have also shown the feasibility of direct testing (without isolation or purification), specifically for urinary tract infections, paving the way for direct microfluidic AST systems at PoC. As rate of bacteria growth is intrinsically linked to the temperature of incubation, transferring miniaturised AST nearer the patient requires building new capabilities in terms of temperature control at PoC, furthermore widespread clinical use will require mass-manufacturing of microfluidic test strips and direct testing of urine samples. This study shows for the first-time application of microcapillary antibiotic susceptibility testing (mcAST) directly from clinical samples, using minimal equipment and simple liquid handling, and with kinetics of growth recorded using a smartphone camera. A complete PoC–mcAST system was presented and tested using 12 clinical samples sent to a clinical laboratory for microbiological analysis. The test showed 100% accuracy for determining bacteria in urine above the clinical threshold (5 out of 12 positive) and achieved 95% categorical agreement for 5 positive urines tested with 4 antibiotics (nitrofurantoin, ciprofloxacin, trimethoprim and cephalexin) within 6 h compared to the reference standard overnight AST method. A kinetic model is presented for metabolization of resazurin, demonstrating kinetics of degradation of resazurin in microcapillaries follow those observed for a microtiter plate, with time for AST dependent on the initial CFU ml−1 of uropathogenic bacteria in the urine sample. In addition, we show for the first time that use of air-drying for mass-manufacturing and deposition of AST reagents within the inner surface of mcAST strips matches results obtained with standard AST methods. These results take mcAST a step closer to clinical application, for example as PoC support for antibiotic prescription decisions within a day.


Microplate Broth Microdilution
Microplate BMD was performed as described (Needs et al., 2021c), 3-4 colonies per agar plate were inoculated in 1 mL of Mueller-Hinton Broth (MHB) and incubated shaking at 37 °C for several hours.
Cultures were diluted to a 0.5 McFarland standard using a spectrophotometer, diluted 1:100 in MHB.
Finally 50 µL of culture was added to microplate wells containing 50 µL of antibiotic solution and resazurin at a concentration of 0.5 mg/mL. Plates were incubated overnight at 37 °C.

Disc Diffusion
Clearly isolated single colonies from a freshly streaked (18-to 24-hour incubation) bacterial plate were suspended in a saline solution (0.8% NaCl), adjusted to a 0.5 McFarland, diluted 10-fold in a Mueller-Hinton broth (Sigma Aldrich) and further diluted 10-fold in a Mueller-Hinton broth.

Kinetic model for resazurin metabolization
We have modelled the rate of metabolization of resazurin as associated to the cell density as follows.
In the presence of unlimited substrate concentration and small cell density or CFU/ml, the rate of increase in bacteria density C X follows a standard exponential growth described by: where  is the growth rate of the bacteria (hr -1 ). On the other hand, metabolization of resazurin has been shown to be a biocatalytic process, with the rate of metabolization of resazurin being proportional to the biomass concentration or cell density, C X (in units of CFU). Assuming the rate of metabolization of resazurin dC R /dt is independent to the concentration of resazurin, C R , this can be represented a first order reaction: where k C is a kinetic constant representing the number of moles of resazurin converted per unit of time per CFU (mol CFU -1 hr -1 ).
Integration of Eq. (1) allows estimating the cell density, C x over time, t during exponential growth, considering the initial cell density in the inoculum, C X,0 : (3) Combining Eqs (2) and (3), yields: Conversion, X of resazurin can be defined as: (4) Where C R,0 is the initial concentration of resazurin and C R the resazurin concentration at a given time t. Note X varies between 0 (t = 0) and 1 (full conversion). The initial concentration of resazurin in the experiments is known (described in materials and methods). C R is not measured along the incubation time however it can be shown that concentration of resazurin is linearly proportional to the Abs/cm measured throughout the experiments, especially in the microcapillaries due to the very short light patch distance, in line with Lambert-Beer law.
Integrating Eq. (3) with boundary conditions C R,0 for t = 0 and C R for t = t and taking into account definition of X as described in Eq. (4), yields: Which is able to predict the degree of metabolization or conversion of resazurin during incubation time. With the growth rate  being estimated from experimental data, the kinetic model in Eq. (5) yields a single unknown kinetic constant k c (mol CFU -1 hr -1 ), which can be found by best-fitting the model to the experimental data using Excel's solver using as criteria minimum square differences.
Note that for simplicity often the term (1-X) vs time has been selected in order to match the dynamics of the experimental data, with absorbance linked to the amount of unconverted resazurin in the growing media.  (Table 2). Ciprofloxacin showed 95% agreement at the high bacterial load corresponding to 3x10 7 CFU/mL in urine, above the density recommended for AST measurements. Even at the clinical threshold of 10 5 CFU/mL there was 90% categorical agreement for ciprofloxacin. This issue is mostly due to the MIC of some of the UPEC isolates being within ± 1 doubling dilution of antibiotic. This would be a problem even when using that gold standard technique as MIC varies within labs.

SUPPORTING FIGURES
The portable hotbed stays with ± 2 degrees of the temperature set on the microcontroller.