Patent Application
20230230249
This invention relates devices, systems, and methods for rapid digital antimicrobial susceptibility testing by imaging and tracking single cells in a clinical sample.
Antimicrobial resistance is a rapidly growing threat to global public health, affecting millions of people annually. One cause for this global concern is the misuse and overuse of antimicrobials. Antimicrobial susceptibility testing (AST) methods typically used in clinical labs rely on overnight cell culture for pathogen infection detection and can involve additional isolation and sub-culture steps. Various emerging AST technologies have been proposed, which fall into two categories: genotypic and phenotypic approaches. Genotypic approaches can be used to detect antibiotic resistance genes. While sensitive, this approach requires prior knowledge of the pathogens, which can lead to false negatives when a new resistant strain emerges, and false positives because resistance genes do not necessarily produce resistant strains. Phenotypic approaches measure a phenotypic feature, such as size or number of bacterial cells. However, most AST technologies still require culture, isolation, and enrichment of bacterial cells.
For phenotypic AST, a common practice is to compare the bacterial cell growth in samples with and without antibiotics. In some cases, morphology changes (e.g., cell size), DNA/RNA copy changes, or cell number changes are used to quantify cell growth. However, for rapid AST with real samples, some features are not reliable. For example, bacteria continue growing in DNA and size at the initial phase with the presence of some antibiotics, and particulate contaminates can interfere with the DNA/cell number counting methods.
This disclosure relates systems and methods for rapid antimicrobial susceptibility testing (AST) using large volume scattering imaging (LVSi) for culture-free, rapid imaging and tracking single cells in clinical samples, including clinical samples with low bacterial counts. Single cell division events are tracked, allowing rapid identification of viable bacterial cells in the samples and AST without cell culturing. Single cell division measures the growth of live cells only, and is generally not sensitive to other impurities (e.g. crystals, cell debris, or dead bacterial cells).
In a first general aspect, detecting single bacterial cells in a sample includes collecting, from a sample provided to an imaging apparatus, a multiplicity of images of the sample over a length of time; assessing a trajectory of each bacterial cell in the sample; and assessing, based on the trajectory of each bacterial cell in the sample, a number of bacterial cell divisions that occur in the sample during the length of time.
Implementations of the first general aspect may include one or more of the following features.
The first general aspect may further include providing the sample to the imaging apparatus, collecting the sample from a subject, or both. The sample may be a bodily fluid (e.g., urine) from a subject. In some cases, the first general aspect further includes combining the sample with a culture medium. In certain cases, the first general aspect may further include diluting the sample, filtering the sample, or both.
In some implementations, the first general aspect further includes defining an infection threshold as a number of cell divisions, and comparing the number of bacterial cell divisions that occur in the sample during the length of time with the infection threshold. Some implementations further include identifying the sample as infection positive if the number of bacterial cell divisions that occur in the sample during the length of time exceeds the infection threshold or identifying the sample as infection negative if the infection threshold exceeds the number of bacterial cell divisions that occur in the sample during the length of time. In one example, the infection threshold is between 2 to 10 cell divisions. The length of time is typically in a range between 20 minutes and 120 minutes, or between 30 minutes and 60 minutes.
In some implementations, the sample is a first sample, the length of time is a first length of time, and the number of bacterial cell divisions that occur in the sample during the first length of time is a first number of bacterial cell divisions, and the first general aspect further includes collecting, from a second sample provided to the imaging apparatus, a multiplicity of images of the second sample over a second length of time; assessing a trajectory of each bacterial cell in the second sample; and assessing, based on the trajectory of each bacterial cell in the second sample, a second number of bacterial cell divisions that occur in the second sample during the second length of time. The first sample and the second sample may be obtained from a common source. In some cases, the first sample includes an antibiotic and the second sample is free of added antibiotic. Some implementations further include assessing a ratio of the first number of bacterial cell divisions to the second number of bacterial cell divisions. Some implementations further include defining a susceptibility threshold and comparing the ratio to the susceptibility threshold. Certain implementations further include identifying the first sample as resistant to the antibiotic if the susceptibility ratio exceeds the threshold or identifying the first sample as susceptible to the antibiotic if the susceptibility threshold exceeds the ratio. In some cases, the susceptibility threshold is in a range of 0.4 to 0.6, corresponding to inhibition of 40% to 60% of the bacterial cells, respectively.
In some implementations, a volume of the sample is in a range of 1 μL to 50 μL. A number of particles in the sample is typically less than about 2×105 particles/mL.
In some implementations, assessing the trajectory of each bacterial cell in the sample includes monitoring a position of each bacterial cell in a sequence of images. A magnification of the imaging apparatus is typically in a range of 0.5-10×.
Collecting the multiplicity of images may include irradiating the sample with light (e.g., infrared light). The sample may be a liquid sample. The sample is typically uncultured.
The first general aspect may further include, based on the number of bacterial cell divisions, administering an antibiotic to a subject (e.g., a mammalian subject or a human).
In a second general aspect, a system includes a light source, optics configured to focus light from the light source on a liquid sample in a container, an imaging device, and a controller operably coupled to the light source and the imaging device and configured to initiate collection of a series of images of the liquid sample over a length of time. Based on the images, the controller is further configured to assess a trajectory of each bacterial cell in the sample and to assess, based on the trajectory of each bacterial cell in the sample, a number of bacterial cell divisions that occur in the sample during the length of time.
In the AST systems and methods described, single cell sensitivity is achieved without immobilization or further processing (e.g., enrichment or culturing) of cells. Large sample volumes can be used in cuvettes or vials without additional reagents (e.g., DNA primers, enzymes, binding agents, etc.) or microfluidics. Results can be achieved within one hour with 97% accuracy, allowing precise antibiotic prescription and timely treatment of patients during clinic visits.
The described AST systems and methods overcome difficulties with traditional methods such as optical microscopy that can image bacterial cells but typically require immobilization of the cells on a surface. This feature of traditional optical microscopy, together with the small field of view of high-resolution optical microscopy, makes it difficult to locate bacterial cells in low concentration samples without enrichment. LVSi overcomes this difficulty by illuminating and imaging a large sample volume such that the presence of a few bacterial cells in a clinical sample can be tracked continuously.
The details of one or more embodiments of the subject matter of this disclosure are set forth in the accompanying drawings and the description. Other features, aspects, and advantages of the subject matter will become apparent from the description, the drawings, and the claims.
This disclosure describes rapid antimicrobial susceptibility testing (AST) systems and methods that implement large volume scattering imaging (LVSi) for real-time imaging of single cells with sensitivity and precision. These systems and methods work directly on urine samples in glass vials or cuvettes to image, track, and count the individual division events of single bacterial cells in clinical samples. These rapid AST systems and methods can image and count low bacterial concentration samples (e.g., 104 CFU/mL urine samples). To precisely track and count single division events of bacterial cells in the presence of various particles (e.g., crystals and dead skin cells) in a sample, a forward scattering optical imaging configuration and an imaging processing algorithm are implemented.
AST systems and methods described herein provide single cell precision in real-time without DNA primers, reagents, incubation, immobilization, or microfluidics. The elimination of microfluidics and associated pumps and valves simplifies the setup, removes clogging of microfluidic channels by air bubbles and impurities in real urine samples, and allows simultaneous tracking of multiple cells in parallel in free solution.
System 100 typically includes a controller (e.g., a computing device such as a laptop or desk top computer). The controller may be coupled to a network and one or more remote computing devices. The controller can be configured to control light source 102 and camera 106 and analyze images 118.
The controller may be used to implement automated division tracking to extract the trajectory of each individual cell and filter out single-cell division events over time for rapid antibiotic susceptibility determination. To track the real division events, each spot in the series of images or video is connected in time to form single-cell trajectories. For each trajectory, all temporal and spatial information is extracted, including trajectory start time, trajectory duration, trajectory end time, spot location (x, y) in each frame, spot mean intensity in each frame, and so on. With all of this information, the division events were filtered out depicted in the flow chart in
To estimate the single-cell division tracking accuracy, calibration is performed with the heat-deactivated E. coli cells, in which there should be no real growth associated division event. The division events tracked are called division over-counting for final division calibration. With different numbers of E. coli cells, the division over-counting result is shown in
The automated division tracking algorithm includes one or more of the following operations. In one operation, common background noise and image drift are corrected with temporal local minimal subtraction to improve the image contrast for detection of cell spots. In another operation, a Laplace of Gaussian (LOG) filter is used to detect individual cells. In another operation, directional linking of cell spots in adjacent frames is performed using a Karman algorithm to obtain single cell tracking trajectories of cells 120, such as those depicted in
In an exemplary process, a sample is collected from a subject (e.g., a human) and provided to an imaging apparatus (e.g., system 100 of
An infection threshold is defined as a number of cell divisions, and the number of bacterial cell divisions that occur in the sample during the length of time is compared with the infection threshold. The sample is identified as infection positive if the number of bacterial cell divisions that occur in the sample during the length of time exceeds the infection threshold. The sample is identified as infection negative if the infection threshold (e.g., 2 to 10 cell divisions) exceeds the number of bacterial cell divisions that occur in the sample during the length of time (e.g., between 20 minutes and 120 minutes or between 30 minutes and 60 minutes). Based on the number of bacterial cell divisions, an appropriate antibiotic may be administered to a subject
Results from a first sample that includes an antibiotic may be compared with a second sample from the same source that is free of antibiotic. A ratio of the first number of bacterial cell divisions to the second number of bacterial cell divisions can be assessed. A susceptibility threshold may be defined (e.g., 0.4 to 0.6, corresponding to inhibition of 40% to 60% of the bacterial cells) and compared to the ratio. The first sample may be identified as resistant to the antibiotic if the susceptibility ratio exceeds the threshold. The first sample may be identified as susceptible to the antibiotic if the susceptibility threshold exceeds the ratio.
In the examples below, a LVSi technique is used for detection of bacteria and determination of antimicrobial susceptibility directly in a real sample. By tracking the single-cell division events, growth of the viable cells is quantified with high sensitivity in a short time. For pure E. coli samples without sub-culture, direct AST with the cells from stationary phase was achieved in 60 minutes. Results revealed the variability in the growth rate of cells from different populations, demonstrating the existence of the persistent cells with the presence of antibiotics. The method allows single cell detection capability, enabling the study of heterogeneity of cell response to antibiotics and the antibiotic resistance evolution. For real samples, the technique was applied to 60 clinical urine samples and predicted 97% of the bacterial existence for the infection positive samples with 60 min. AST was also performed on these patient samples with ciprofloxacin, and achieved 100% categorical agreements within 60 min (sample-to-results).
The performance compares well with the existing culturing-based commercial technologies. The technique can test raw clinical samples without enrichment or culturing, and track the division events of individual viable bacterial cells in real time, which simplifies the testing procedures, improves the precision, and shortens the turnaround time from sample receipt to result determination. As the division tracking quantifies the bacterial cell growth, which is a universal phenotypic feature for AST, this technique is applicable to a wide range of bacteria.
E. coli ATCC 25922 were purchased from American Type Culture Collection (ATCC) and stored at −80° C. in 5% glycerol. Antibiotics, including ciprofloxacin and ampicillin were purchased from Sigma-Aldrich. The antibiotic powders were stored in the dark at 2 to 8° C. Frozen E. coli strains were thawed, and 50 μL of the cells were cultured in 5 mL of Luria-Bertani (LB) medium (per liter: 10 g peptone 140, 5 g yeast extract, and 5 g sodium chloride) at 37° C. and 150 rpm for 15 h. Then the overnight cultured E. coli was inoculated into LB broth directly. After dilution to appropriate cell concentrations, antibiotics were added to the E. coli suspensions.
De-identified clinical urine samples were obtained from the clinical microbiology laboratory of Mayo Clinic Hospital Arizona. Clinical samples were transported in an ice box and kept at 4° C. after receiving. The refrigerated urine samples were pre-warmed for 30 min at 37° C. before use. Then, the urine samples were passed through a 5 μm filter to remove the large substances and diluted with LB broth to a concentration of ˜2×105 CFU/mL. The diluted clinical urine samples (100 μL) were added to 96-well microtiter plates (Falcon, BD Biosciences) preloaded with LB broth (100 μL) with and without ciprofloxacin (2 μg/mL, final concentrations). After full mixing, 70 μL samples were transferred to cuvettes (Uvette, Eppendorf, Germany), and subjected to LVSi. A total of 60 urine samples with blind pathogens were tested using both optical division tracking and parallel validating plating. The results were compared with clinical microbiology culture results.
The dual channel LVSi system depicted in
Individual cells recorded by LVSi are resolved as bright spots. Before division tracking, each spot was detected with a Laplacian of Gaussian (LoG) filter with defined radius and threshold. Then, the spots from adjacent time frames were connected with a Karman filter for directional linking. So each bright spot became a single-cell trajectory. To track the division events, the newly appeared cell trajectories (those from the image edges excluded) were first filtered out as the child trajectory candidates, and for each child trajectory candidate, the nearby old trajectories (appeared before the child trajectory candidate) were checked. If there was a close-by old trajectory split at the start time of the child trajectory candidate, a potential division event was tracked. To filter out the merge/crossing induced splitting, the spots merging events were also checked and were filtered out from the potential division events. Finally, an intensity filter was used to evaluate the remaining division events, ensuring the parent cell intensity is about the summation of the two divided child ones, and the two child ones are similar in size.
To validate that the tracked division events were due to real cell growth, rather than an artifact of the particle merging or crossing, a division over-counting calibration test was designed. To rule out the real division events, the E. coli cells were heated at ˜65° C. for 15 min. Then, each 5-min video of the heated E. coli cells with different concentrations (2.0×104˜2.0×105 CFU/mL, corresponding to 100-1000 spots in image view) were analyzed for single-cell division tracking. Each concentration test was repeated three times. Then, the division over-counting calibration curve was extracted from the tracking results. The division over-counting is less an issue when the cell number is below 500, corresponding to the bacterial concentration of about 1.0×105 CFU/mL. When the cell number is above 500, some miss-counting of division events occurred. To rule out the cell density induced division over-counting, the final division events in all clinical samples were calibrated by subtracting the cell density associated over counted division events.
The statistical error of the division tracking was estimated by the mentioned division over-counting calibration. The final division events were calibrated by subtracting the fitted mean value of the division over-counting, in which the averaged standard error of the mean is calculated to be ˜1 in every 5-min video. To establish a 95% confidence interval, the error was multiplied by 2. For a 60 min detection, the cumulative standard error of the mean is ˜24. Since standard error of the mean in each 5-min is random, the final statistic error of the tracking was estimated by N1/2, where N is the cumulative standard error of the mean. The final threshold for infection identification (TI) was set to be 5 with a 95% confidence for the calibration. The susceptible threshold (TS) was set to be 0.5, corresponding to 50% of the growth inhibition in the antibiotic-treated samples.
Pure E. coli samples were used to establish a single-cell division tracking method. AST was performed directly from the stationary phase bacteria without culturing. The culture-independent capability was tested to mimic clinical urine samples in which the bacteria are likely in stationary phase due to environmental change and lack of nutrients. The pure E. coli samples, in which most of the cells were in stationary phase, were directly diluted into two equal volumes of culture medium (LB broth) without and with antibiotics at standard breakpoint concentration. The breakpoint concentration is the concentration of an antibiotic that defines whether a species of bacteria is susceptible or resistant to the antibiotic. If the minimum inhibitory concentration (MIC) (the lowest concentration of an antibiotic required to inhibit growth of an organism) is less than or equal to the susceptibility breakpoint, the bacteria is considered susceptible to the antibiotic. If the MIC is greater than this value, the bacteria is considered intermediate or resistant to the antibiotic. The diluted bacteria concentration is ˜1×105 CFU/mL. Then, a diluted sample of 70 μL was transferred to an imaging cuvette for direct LVSi for 60 min at 37° C. Image sequences containing hundreds of bacterial cells were obtained and trunked into 5-min videos. Individual cell division events were tracked in every 5-min video and cumulative division events were counted for the entire 60 min.
To validate the precision and broader application of the method, more tests with 2 μg/mL ciprofloxacin were performed, along with tests of an additional antibiotic, 16 μg/mL ampicillin. These tests were performed with different batch of E. coli samples. To compare the results from different experiments, the cumulative division events were normalized by the number of initial cells (No), showing the increase in cell growth. For clarity, 5 representative cumulative division tracking results of E. coli samples with and without antibiotics are plotted in
With the single-cell division tracking capability, some cells were still observed to divide in the presence of antibiotics, showing the cell-cell heterogeneity within a sample. Since these persistent cells in the antibiotics may eventually develop to a resistant strain, the digital counting method described herein provides a capability for early warning on potential drug resistance or tracking the progress of resistance development.
The E. coli samples from different batches were used to evaluate the influence of the sample variability on the robustness of the digital AST method. The effects of the bacterial initial status on the growth curve were observed. The samples were directly tested without sub-culture, with most of the cells are in stationary phase. For comparison, the susceptibility tests were also performed with E. coli from log phase (with 2 hr sub-culture). In log phase, sufficient division events occurred sooner and the total AST time was reduced to about 30 min. These results demonstrated that the cell status in the sample affect the total detection time. Effects of the environmental control were also examined. Comparison of the results from the optical system, and incubator with and without shaking suggests that continued measurement does not have a significant impact on the growth of the bacteria. To cross validate the division tracking accuracy, traditional plating detections were performed simultaneously on the same samples to verify the cell growth at each time point, and the results were consistent with the division tracking ones.
To demonstrate the capability of the digital AST method, the optical division tracking method was used with clinical urine samples for urinary tract infection (UTI) diagnosis and susceptibility determination. Digital AST was implemented directly with 60 de-identified clinical urine samples with blinded pathogens, including negative samples. Before detection, a simple filter (5 μm) was performed to remove the large substances. Then, the clinical urine samples were diluted with LB medium to a concentration around ˜1×105 CFU/mL and imaged for 60 min in a setup such as that depicted in
To explore the infection detection accuracy over time, comparison of a reference method (BD Phoenix) and digital AST at the time points of 30 min, 45 min and 60 min are shown in
Digital AST was performed for all of the 30 infection positive samples. The susceptibility profiles were determined by comparing the calibrated cumulative division in control (DC) and antibiotic (ciprofloxacin) treated tests (DABX). For a susceptible sample, such as that depicted in
For these samples, the susceptibility threshold (TS) was set to be 0.5, which indicates 50% of the cell growth inhibition. By comparing the DABX/DC ratio to the susceptibility threshold, 8 samples were determined to be resistant to ciprofloxacin and the other 22 samples were determined to be susceptible to ciprofloxacin. As shown in
Thus, LVSi with single-cell division tracking technology that rapidly detects the existence of bacteria and determines the antibiotic susceptibility was demonstrated. Use of a large image volume allowed use of a real sample directly without further enrichment. The single-cell division tracking and counting provided high sensitivity for rapid cell growth quantification. In clinical urine samples, there are particle impurities, some of which tend to precipitate during the test. Thus, in a spot counting method, the cell growth induced spot increases will be compromised by the sedimentation of impurities. The division tracking show advantages for measuring viable cell grow, reducing the total detection time compared to the spot counting results in clinical sample.
For accurate division tracking, one variation here is the final sample concentration. The single-cell division is based on the accurate particle detection and trajectory linking. When the particle density is too high, linkages may be assessed incorrectly. Based on the current linking algorithm, the linking accuracy is affected when the particle number is above 1000, which corresponds to the particle density of about 2.0×105/mL. Therefore, for optimal division tracking, the clinical sample is diluted to a concentration of about 1×105 CFU/mL before testing. Since there are urine particles in the samples, not all spots in the video are viable bacterial cells. To extract the dynamic range of the digital AST method, a minimal viable bacteria number needed for 60-min AST is assumed to be 50. When the viable bacteria number is below 50, a longer time is needed for digital AST. Then, the range of the viable bacterial cells in the video is 50-1000, and the dynamic range of the method is derived to be 1×104-2.0×105 CFU/mL.
Sample dilution is another factor for consideration. For optimal division tracking, clinical samples were diluted to make sure the total particle number in the video was below 1000. However, in some samples, the raw bacteria concentration is in the lower clinical range (between 1.0×104 to 1.0×105 CFU/mL). With an extra dilution step, these samples are likely to be over diluted, leading to a very small number of viable bacteria in the imaging system. A false negative result may be due to over dilution, with no viable bacterial cells after dilution based on the validation plating results. To reduce the influence of dilution on false infection negative results, the sample preparation process can be optimized to remove extra urine particles, the image volume can be increased, and the division tracking algorithm can be modified.
Additional details are provided in Exhibit 1.
Although this disclosure contains many specific embodiment details, these should not be construed as limitations on the scope of the subject matter or on the scope of what may be claimed, but rather as descriptions of features that may be specific to particular embodiments. Certain features that are described in this disclosure in the context of separate embodiments can also be implemented, in combination, in a single embodiment. Conversely, various features that are described in the context of a single embodiment can also be implemented in multiple embodiments, separately, or in any suitable sub-combination. Moreover, although previously described features may be described as acting in certain combinations and even initially claimed as such, one or more features from a claimed combination can, in some cases, be excised from the combination, and the claimed combination may be directed to a sub-combination or variation of a sub-combination.
Particular embodiments of the subject matter have been described. Other embodiments, alterations, and permutations of the described embodiments are within the scope of the following claims as will be apparent to those skilled in the art. While operations are depicted in the drawings or claims in a particular order, this should not be understood as requiring that such operations be performed in the particular order shown or in sequential order, or that all illustrated operations be performed (some operations may be considered optional), to achieve desirable results.
Accordingly, the previously described example embodiments do not define or constrain this disclosure. Other changes, substitutions, and alterations are also possible without departing from the spirit and scope of this disclosure.
This application claims the benefit of U.S. Patent Application No. 63/043,713 entitled “DIGITAL ANTIMICROBIAL SUSCEPTIBILITY TESTING” and filed on Jun. 24, 2020.
This invention was made with government support under R01 AI138993 awarded by the National Institutes of Health. The government has certain rights in the invention.
| Filing Document | Filing Date | Country | Kind |
|---|---|---|---|
| PCT/US2021/038750 | 6/23/2021 | WO |
| Number | Date | Country | |
|---|---|---|---|
| 63043713 | Jun 2020 | US |
