New technologies for diagnosing bloodstream infection and measuring antimicrobial resistance

July 25, 2017
Editor’s note: MLO is pleased to present this outstanding research study by Drs. Adeyiga and Di Carlo of UCLA. In order to provide the article in full, we printed the first half in the July 2017 issue [MLO.2017;49(7):54-56]; here is “part two.”

Given that there are still opportunities for improvement with regard to technologies currently in stages of clinical translation, sepsis diagnostics is an area of active research within academia. There are key opportunities for development, for instance, in terms of increasing the yield of detection, defining other parameters by which we can accurately and precisely identify bloodstream infection, and furthering our ability to describe and understand how antimicrobial agents impede the growth and activity of pathogenic microbes and may impact other elements of the inflammatory response. In fact, several early-stage technologies for diagnostic clinical microbiology are emerging from academic research institutions.

These emerging technologies focus on several arenas, including high-throughput blood processing, high-sensitivity pathogen detection, and antibiotic susceptibility testing. Researchers are working to increase processing rates for large volume samples in order to increase the yield of a positive test indicating the presence of bloodstream infection without the use of blood culture. In addition, there is much research and development into technologies that increase sensitivity of pathogen detection. Also, several reported works focus on creating new methodologies for the rapid determination of antimicrobial susceptibility. This area of work has gained increased attention, especially given the national call to action in the battle to combat the development of antimicrobial resistance.47

High-throughput sample processing

In this section, the highlighted technologies make use of droplet microfluidics, inertial microfluidics, and dielectrophoresis in order to perform high-throughput sample processing. Given that current methods involve obtaining 10 milliliters of blood to test for bloodstream infection with blood culture, the capability of a platform to rapidly process a patient sample with high volume becomes imperative.

IC3D. One approach to improve the speed with which bloodstream cultures are identified, authored by Kang and colleagues,48 uses a technique called integrated comprehensive droplet digital detection (IC3D). This technology is designed to identify the presence of single cells in diluted whole blood within four hours, compared to the current time for blood culture positivity, which can range from eight to 48 hours.

There are three components to this technology: a microfluidic droplet generator, a particle counter system, and a bacterial detection method that utilizes DNAzyme sensor technology. The use of a microfluidic droplet generator allows for rapid processing of a large volume of blood into small volume liquid droplets. Generating small liquid volume droplets is useful for the purpose of increasing the likelihood that a single bacteria cell will be isolated within a small volume, leading to a high effective concentration of bacteria within each occupied droplet such that reaction with the sensor molecule leads to a high concentration of signal isolated within the drop. These conditions will have the effect of increasing the sensitivity of detection and increasing the signal-to-noise ratio for detection of the presence of a bacteria. Diluted blood is mixed with a solution that contains DNAzyme sensor solution and bacterial lysis buffer, which is then processed by a microfluidic droplet generator. DNAzyme sensors that detect the presence of bacteria can be detected using a 3D particle counter able to reliably count single fluorescent particles. Importantly, the capability of the sensor molecule to sensitively detect the presence of specific types of lysed bacteria will increase the accuracy of the technology.

The IC3D technology’s use of a DNAzyme sensor molecule to differentiate E. coli strains from two mammalian cell lines leads to a selective and specific signal. The authors perform further comparison experiments with other clinically relevant gram negative organisms, such as C. freundii and  . Here, while they are able to show a statistically significant difference in fluorescence intensity between the measured organisms, the magnitude of the difference in fluorescence intensity is small between organism types, showing the potential for cross-reactivity, among different organism types.

Inertial microfluidic separation. Another approach makes use of inertial microfluidic separation to purify bacteria from background cells. In this work,49 Hou and colleagues combine inertial microfluidics with a nucleic acid-based method of detection in order to detect pathogens spiked into diluted whole blood. An attractive feature of this platform is the use of a label-free method to separate smaller-sized bacteria from larger-sized blood cells, which are abundantly present in whole blood and certainly can obscure the presence of small-sized bacteria.

This method is label-free in that it does not require an affinity-based technique, such as antibody binding, to perform the separation. Instead, the separation of these different-sized particles is achieved by exploiting some of the unique physics exclusive to microfluidic fluid flow. For this work, the authors employed the physics of inertial microfluidics to allow for particle separation, where particle movement across streams that occurs within fluid flowing through a microfluidic channel is distinctive for each sized particle. The knowledge of these unique physics informs the authors’ creation of geometric designs for fluid channels that can segregate smaller and larger cells. In this case, the authors explore and determine the optimal microfluidic channel geometry that efficiently performs this separation, and also determine the optimal whole blood sample preparation necessary for use with their chip.

The microfluidic chip designed in this study has a spiral configuration, with two ports at the inlet and two ports at the outlet. Two types of fluid are introduced into the each inlet port: a diluted whole blood sample that has been spiked with bacterial pathogens, and a sheath fluid. From one outlet port, fluid is removed that contains separated red and white blood cells; from the other outlet port, a fluid containing bacteria, along with some platelets, is removed. After introducing blood containing bacterial pathogens at the inlet, the authors demonstrate that their chip is able to recover several types of bacteria at the outlet port, including E. coli, P. aeruginosa, E. faecalis, and S. aureus. The authors take the bacteria-containing fluid collected from the outlet port, perform an additional off-chip concentration and bacteria lysis steps, and then detect the bacteria of interest with another method.

Nucleic acid sequence molecules. In another example, described in a paper by Geiss and colleagues,50 the highlighted system detects bacteria using nucleic acid sequence reporter molecules that will bind complimentary nucleic acid sequence molecules present in the tested suspension. With this technique, pre-constructed probes that contain capture and reporter regions for specific nucleic acid segments of interest are allowed to mix in suspension and hybridize in the first step. Then, after affinity purification is performed to remove unhybridized probes, the suspension of hybridized complexes is allowed to adhere to a surface onto which a capture agent has been coated. These hybridized complexes can then be visualized and counted, given that the reporter region of the probe has a fluorescent segment, and an applied electric field orients and elongates each hybridized complex. In contrast to PCR, the cyclic steps needed for nucleic acid amplification are not employed. Here, the bacterial detection method chosen by the authors is used to detect bacterial rRNA instead of mRNA, with the rationale that rRNA is present in higher amounts and will allow a lower level limit of detection for their approach.

Finally, the authors propose use of a Transcriptional Susceptibility Score, as a quickly obtainable score that correlates with the antimicrobial susceptibility of an organism. The score is determined by first performing a short incubation of specific bacteria with an antimicrobial of interest. Then, the authors lyse bacteria and detect the mRNA present in solution using their chosen method of detection. The authors are able to show that their devised Transcriptional Susceptibility Score correlates with antimicrobial susceptibility measured using conventional culture techniques. This element of analysis adds an additional benefit to the proposed technology. The author’s chosen method of detecting bacteria addresses concerns such as the low levels of microorganisms often present in bloodstream infection. However, given that RNA detection is performed after centrifugation and lysis of the bacteria-rich solution obtained from processing using the designed chip, further work is necessary if this process is to be completely contained on-chip.

Dielectrophoresis. In another example, Cai and colleagues design a microfluidic chip where dielectrophoresis is the label-free method used to extract pathogens of interest.51 In contrast to the approach of Hou et al, where some process elements are performed off chip, in this system, diluted whole blood is loaded onto the chip, where pathogen extraction occurs. A wash step is performed, after which pathogens are extracted, and a PCR-based method of detection is used to identify extracted pathogens on chip.

These highlighted technologies are excellent examples of the work that can be performed using a multidisciplinary approach. The platforms ultimately resulting from these technologies that are translated into the clinical arena can surely have a large impact on the rapid diagnosis of bloodstream infection because of their ability to process large volumes on chip, and their demonstrated use to detect clinically relevant microorganisms. In addition, in each approach, the authors incorporated clinically relevant organisms and scenarios, demonstrating the applicability of the designed platforms in addressing the needs of clinical practitioners. These noted successes in this research focus area should hopefully inspire even more innovation that aims to solve this complex problem of rapidly acquiring pathogen-specific information from large volumes of blood in diagnosing bloodstream infection.

Improved pathogen detection sensitivity

In approaching this problem, researchers have looked at detecting compounds, such as proteins, nucleic acid, or other chemicals, exclusively produced by particular infectious agents, which may be present in an amount that is either easier to detect or more abundant in the clinical samples collected from patients with a suspected infection. Another strategy has been to develop more sensitive reporter molecules. To increase sensitivity, one approach, by Alatraktchi and associates,52 focuses on detecting a virulence factor of P. aeruginosa, pyocyanin, which is exclusively secreted by this organism. That this chemical is exclusively secreted by P. aeruginosa makes the detection of this chemical extremely useful for selective diagnosis of P. aeruginosa infection. Whereas the detection of the molecule is commonly performed with the use of spectrophotometry, obtaining reliable spectrophotometry sample measurements requires pretreatment of the sample to be tested, in order to reduce the presence of background noise. Pretreatment is necessary, given that pyocyanin is a redox active compound. The authors describe the detection of this molecule using amperometric electrochemical detection, where this chemical is detected by sensing a change in ions with the use of an electrical current.

This proposed technique has been designed to give a better signal-to-noise ratio that would result in the ability to detect the presence of P. aeruginosa in low amounts. The work described here was performed where pyocyanin was measured in artificial sputum. However, further work is required in order to transfer this technique for detecting this pyocyanin in patient samples such as human blood and respiratory samples, unless pre-testing sample preparation steps occur. Respiratory samples such as sputum can be highly viscous and complex, in terms of the presences of various substances such as glycoproteins, immunoglobulins, and inflammatory cells. Whole blood is even more complex, given the sheer abundance of cells and breadth of proteins present, as well as other compounds such as lipids and carbohydrates, which can also be redox active compounds. This example features an important direction in refining our ability to diagnose infection: the technique of measuring biomarkers that correlate with the presence of infection.

In two more techniques, authors design detection systems involving different reporter materials. The development of materials that can enhance our ability to detect markers associated with infection would advance the field. In one case, a unique detector system where cubic retroreflectors are modified with an antibody specific to the pathogen of interest was used.53 In this case, E. coli and MS2 virus was the target, and a sandwich type ELISA assay was performed with the functionalized cubic retroreflectors within qPCR tube reaction vessels that also contained antibody-modified tube caps. Immobilized cubes present within reaction vessels were imaged and counted. The authors describe additional modifications that can be made in order to make this a multiplex assay.

In another work, researchers employ microarray technology using functionalized nanotubes to allow the detection of multiple pathogens on one platform. The authors here54 describe the use of antibody-conjugated peptide nanotubes, where the specific pathogens of interest bind to modified nanotubes that have have an antibody to the particular pathogen.

The complexes consisting of modified nanotube bound to pathogens quickly settle onto the surface of the chip, where transducers have been patterned. As these complexes settle, a change in impedance occurs, generating a signal that can be used for pathogen detection. In this case, given that the functionalized nanotubes are not patterned onto the chip surface, functionalized nanotube-pathogen complexes that develop and settle onto the chip surface can be gently washed away. In this manner, there is the potential for chip reuse, which can be a cost- and resource-saving measure.

Nucleic acids have also been assayed using loop-mediated isothermal amplification, or LAMP, to assist with increasing pathogen detection sensitivity. In contrast to conventional PCR, which requires temperature cycling for its process, LAMP offers rapid amplification of nucleic acid sequences present within the sample of interest at a single temperature, thus eliminating the need for a thermal cycler, which is necessary for conventional PCR methods. In this reported study, the authors developed an integrated microfluidic electrochemical DNA (IMED) chip, for the purpose of detecting Salmonella serovars from the whole blood of infected mice.55 After introducing whole blood with LAMP reagents onto the IMED chip, LAMP is performed, and the products of amplification are detected using designed E-DNA probes that have been attached to a downstream region of the chip. The chip was designed so that minimal processing steps are needed, with the intention that it may be used in the point-of-care setting in a variety of commercial applications. In general, by investigating the utility of more sensitive reporter molecules, or looking at compounds that also correlate with the presence of infection, our ability to more accurately and specifically diagnose infection will be enhanced.

Detection of antimicrobial susceptibility

Much work is being conducted to advance our ability to quickly determine antimicrobial susceptibility. Given that current methods for determining antimicrobial susceptibility testing can take 24 hours or more to determine results, or can provide conflicting results across modalities, new methodologies for determining the ability of various antimicrobials to impede the growth of or to kill specific organisms are being developed. Current methodologies usually rely on an increase in the optical density of a suspension containing the microorganism of interest. This increase in optical density corresponds to increase in organism cell population, which then leads to the conclusion that the antimicrobial of interest has no effect on organism growth, and that the organism of interest is resistant to that antimicrobial. Currently, routinely performed measurements yield the minimum inhibitory concentration, whereas the minimum bactericidal concentration for a particular compound can be obtained with additional testing. Minimum inhibitory concentration measures are currently used in interpreting the antimicrobial susceptibility of various organisms.

To develop new methodologies for determining antimicrobial susceptibility of various organisms, an understanding of the mechanisms of antimicrobial resistance is needed. For example, what are the mechanisms involved that allow a particular organism to continue to replicate in the presence of a drug that should impair this ability?56-58 With this knowledge base, specific detection approaches can be applied to identify molecular markers underlying these mechanisms.

Instead of imaging turbidity to identify susceptibility, some have taken the approach of viewing the single-cell division events as a way of determining antimicrobial susceptibility testing more quickly. Several strategies have been employed to achieve this. In the microfluidic agarose channel (MAC) system, the growth behavior, in the form of the proportional growth rate of single bacteria cells immobilized with agarose, is microscopically observed using time-lapse microscopy as these cells are exposed to different concentrations of antimicrobials.59 Additionally, microfluidic devices have also been used to obtain single-cell growth data that can be used to build a pharmacodynamics model for bacteria exposed to different gradients of antimicrobials.60

The described approaches represent what can become a new paradigm in terms of understanding the impact of antimicrobial therapy at the single-cell level. Current methodologies, such as techniques where suspension turbidity is measured as a marker of antimicrobial resistance, obtain measures that very likely are more representative of population dynamics instead of single cell behavior. The analysis of single-cell behavior could potentially lead to more detailed studies of the cell-to-cell interactions that occur between inflammatory cells and pathogenic organisms in the presence of various antimicrobials, which may lead to interesting findings and also inform drug development in this area. The detection of genes from within single cells, and the proteins that are produced from them, give clinicians important information regarding whether a drug will or will not be effective against a particular pathogen. The ability to study single-cell behavior in cells exposed to various antimicrobials offers a new frontier in this area of clinical microbiology.

The new horizon

The described innovations currently ongoing in the area of diagnosing bloodstream infection and determining antimicrobial susceptibility reflect the high level of interest in developing new tools to address these issues in diagnostic microbiology. With regard to diagnosing bloodstream infection, given the complexity of this problem, a multidisciplinary approach will be necessary in order to make significant strides in this area. In terms of designing more tools for the purpose of performing antimicrobial susceptibility testing, there is much interest and development in this area, which will ultimately lead to progress and results. Exciting developments are on the horizon, with technology emerging that will allow the diagnosis of bloodstream infection within 24 hours, without the need for blood culture, and new knowledge with regard to understanding and measuring the way that microbial drug
resistance patterns occur.

The authors would like to acknowledge Dr. Jeffrey Klausner, who provided comments and feedback for this article.

Oladunni Adeyiga, MD, PhD, is an Assistant Clinical Professor in the Division of Infectious Diseases, Department of Medicine at UCLA. She plans to pursue a career in academic medicine developing diagnostics for use in clinical infectious diseases.

Dino Di Carlo, PhD, is a Professor in the Departments of Bioengineering and Mechanical Engineering in the Henry Samueli School of Engineering and Applied Science at UCLA.